| blob | 1310ef77f7ee68c5d238176efce4dfa6da9b7891 |
1 ;;; cc-engine.el --- core syntax guessing engine for CC mode -*- coding: utf-8 -*-
3 ;; Copyright (C) 1985, 1987, 1992-2016 Free Software Foundation, Inc.
5 ;; Authors: 2001- Alan Mackenzie
6 ;; 1998- Martin Stjernholm
7 ;; 1992-1999 Barry A. Warsaw
8 ;; 1987 Dave Detlefs
9 ;; 1987 Stewart Clamen
10 ;; 1985 Richard M. Stallman
11 ;; Maintainer: bug-cc-mode@gnu.org
12 ;; Created: 22-Apr-1997 (split from cc-mode.el)
13 ;; Keywords: c languages
14 ;; Package: cc-mode
16 ;; This file is part of GNU Emacs.
18 ;; GNU Emacs is free software: you can redistribute it and/or modify
19 ;; it under the terms of the GNU General Public License as published by
20 ;; the Free Software Foundation, either version 3 of the License, or
21 ;; (at your option) any later version.
23 ;; GNU Emacs is distributed in the hope that it will be useful,
24 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
25 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
26 ;; GNU General Public License for more details.
28 ;; You should have received a copy of the GNU General Public License
29 ;; along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>.
31 ;;; Commentary:
33 ;; The functions which have docstring documentation can be considered
34 ;; part of an API which other packages can use in CC Mode buffers.
35 ;; Otoh, undocumented functions and functions with the documentation
36 ;; in comments are considered purely internal and can change semantics
37 ;; or even disappear in the future.
38 ;;
39 ;; (This policy applies to CC Mode as a whole, not just this file. It
40 ;; probably also applies to many other Emacs packages, but here it's
41 ;; clearly spelled out.)
43 ;; Hidden buffer changes
44 ;;
45 ;; Various functions in CC Mode use text properties for caching and
46 ;; syntactic markup purposes, and those of them that might modify such
47 ;; properties but still don't modify the buffer in a visible way are
48 ;; said to do "hidden buffer changes". They should be used within
49 ;; `c-save-buffer-state' or a similar function that saves and restores
50 ;; buffer modifiedness, disables buffer change hooks, etc.
51 ;;
52 ;; Interactive functions are assumed to not do hidden buffer changes,
53 ;; except in the specific parts of them that do real changes.
54 ;;
55 ;; Lineup functions are assumed to do hidden buffer changes. They
56 ;; must not do real changes, though.
57 ;;
58 ;; All other functions that do hidden buffer changes have that noted
59 ;; in their doc string or comment.
60 ;;
61 ;; The intention with this system is to avoid wrapping every leaf
62 ;; function that do hidden buffer changes inside
63 ;; `c-save-buffer-state'. It should be used as near the top of the
64 ;; interactive functions as possible.
65 ;;
66 ;; Functions called during font locking are allowed to do hidden
67 ;; buffer changes since the font-lock package run them in a context
68 ;; similar to `c-save-buffer-state' (in fact, that function is heavily
69 ;; inspired by `save-buffer-state' in the font-lock package).
71 ;; Use of text properties
72 ;;
73 ;; CC Mode uses several text properties internally to mark up various
74 ;; positions, e.g. to improve speed and to eliminate glitches in
75 ;; interactive refontification.
76 ;;
77 ;; Note: This doc is for internal use only. Other packages should not
78 ;; assume that these text properties are used as described here.
79 ;;
80 ;; 'category
81 ;; Used for "indirection". With its help, some other property can
82 ;; be cheaply and easily switched on or off everywhere it occurs.
83 ;;
84 ;; 'syntax-table
85 ;; Used to modify the syntax of some characters. It is used to
86 ;; mark the "<" and ">" of angle bracket parens with paren syntax, to
87 ;; "hide" obtrusive characters in preprocessor lines, and to mark C++
88 ;; raw strings to enable their fontification.
89 ;;
90 ;; This property is used on single characters and is therefore
91 ;; always treated as front and rear nonsticky (or start and end open
92 ;; in XEmacs vocabulary). It's therefore installed on
93 ;; `text-property-default-nonsticky' if that variable exists (Emacs
94 ;; >= 21).
95 ;;
96 ;; 'c-is-sws and 'c-in-sws
97 ;; Used by `c-forward-syntactic-ws' and `c-backward-syntactic-ws' to
98 ;; speed them up. See the comment blurb before `c-put-is-sws'
99 ;; below for further details.
100 ;;
101 ;; 'c-type
102 ;; This property is used on single characters to mark positions with
103 ;; special syntactic relevance of various sorts. Its primary use is
104 ;; to avoid glitches when multiline constructs are refontified
105 ;; interactively (on font lock decoration level 3). It's cleared in
106 ;; a region before it's fontified and is then put on relevant chars
107 ;; in that region as they are encountered during the fontification.
108 ;; The value specifies the kind of position:
109 ;;
110 ;; 'c-decl-arg-start
111 ;; Put on the last char of the token preceding each declaration
112 ;; inside a declaration style arglist (typically in a function
113 ;; prototype).
114 ;;
115 ;; 'c-decl-end
116 ;; Put on the last char of the token preceding a declaration.
117 ;; This is used in cases where declaration boundaries can't be
118 ;; recognized simply by looking for a token like ";" or "}".
119 ;; `c-type-decl-end-used' must be set if this is used (see also
120 ;; `c-find-decl-spots').
121 ;;
122 ;; 'c-<>-arg-sep
123 ;; Put on the commas that separate arguments in angle bracket
124 ;; arglists like C++ template arglists.
125 ;;
126 ;; 'c-decl-id-start and 'c-decl-type-start
127 ;; Put on the last char of the token preceding each declarator
128 ;; in the declarator list of a declaration. They are also used
129 ;; between the identifiers cases like enum declarations.
130 ;; 'c-decl-type-start is used when the declarators are types,
131 ;; 'c-decl-id-start otherwise.
132 ;;
133 ;; 'c-awk-NL-prop
134 ;; Used in AWK mode to mark the various kinds of newlines. See
135 ;; cc-awk.el.
137 ;;; Code:
144 load-path)))
153 \f
154 ;; Make declarations for all the `c-lang-defvar' variables in cc-langs.
166 \f
167 ;;; Internal state variables.
169 ;; Internal state of hungry delete key feature
173 ;; The electric flag (toggled by `c-toggle-electric-state').
174 ;; If t, electric actions (like automatic reindentation, and (if
175 ;; c-auto-newline is also set) auto newlining) will happen when an electric
176 ;; key like `{' is pressed (or an electric keyword like `else').
180 ;; Internal state of auto newline feature.
184 ;; Included in the mode line to indicate the active submodes.
185 ;; (defvar c-submode-indicators nil)
186 ;; (make-variable-buffer-local 'c-submode-indicators)
189 ;; Calculate the new state of PREVSTATE, t or nil, based on arg. If
190 ;; arg is nil or zero, toggle the state. If arg is negative, turn
191 ;; the state off, and if arg is positive, turn the state on
197 \f
198 ;; Basic handling of preprocessor directives.
200 ;; This is a dynamically bound cache used together with
201 ;; `c-query-macro-start' and `c-query-and-set-macro-start'. It only
202 ;; works as long as point doesn't cross a macro boundary.
211 c-macro-start))
219 c-macro-start))
221 ;; One element macro cache to cope with continual movement within very large
222 ;; CPP macros.
225 ;; Nil or cons of the bounds of the most recent CPP form probed by
226 ;; `c-beginning-of-macro', `c-end-of-macro' or `c-syntactic-end-of-macro'.
227 ;; The cdr will be nil if we know only the start of the CPP form.
230 ;; The starting position from where we determined `c-macro-cache'.
233 ;; Either nil, or the syntactic end of the macro currently represented by
234 ;; `c-macro-cache'.
237 ;; Either nil, or the last character of the macro currently represented by
238 ;; `c-macro-cache' which isn't in a comment. */
241 ;; Called from a before-change function. If the change region is before or
242 ;; in the macro characterized by `c-macro-cache' etc., nullify it
243 ;; appropriately. BEG and END are the standard before-change-functions
244 ;; parameters. END isn't used.
249 c-macro-cache-start-pos nil
250 c-macro-cache-syntactic nil
251 c-macro-cache-no-comment nil))
256 c-macro-cache-syntactic nil
257 c-macro-cache-no-comment nil))))
260 ;; Check that the ostensible CPP construct at point is a real one. In
261 ;; particular, if point is on the first line of a narrowed buffer, make sure
262 ;; that the "#" isn't, say, the second character of a "##" operator. Return
263 ;; t when the macro is real, nil otherwise.
274 t)
278 "Go to the beginning of a preprocessor directive.
279 Leave point at the beginning of the directive and return t if in one,
280 otherwise return nil and leave point unchanged.
282 Note that this function might do hidden buffer changes. See the
283 comment at the start of cc-engine.el for more info."
295 t))
297 c-macro-cache-start-pos nil
298 c-macro-cache-syntactic nil
299 c-macro-cache-no-comment nil)
312 c-macro-cache-start-pos here)
313 t)
315 nil))))))
318 "Go to the end of a preprocessor directive.
319 More accurately, move the point to the end of the closest following
320 line that doesn't end with a line continuation backslash - no check is
321 done that the point is inside a cpp directive to begin with.
323 Note that this function might do hidden buffer changes. See the
324 comment at the start of cc-engine.el for more info."
333 c-macro-cache-start-pos nil
334 c-macro-cache-syntactic nil
335 c-macro-cache-no-comment nil))
341 t)))
346 ;; Go to the end of a CPP directive, or a "safe" pos just before.
347 ;;
348 ;; This is normally the end of the next non-escaped line. A "safe"
349 ;; position is one not within a string or comment. (The EOL on a line
350 ;; comment is NOT "safe").
351 ;;
352 ;; This function must only be called from the beginning of a CPP construct.
353 ;;
354 ;; Note that this function might do hidden buffer changes. See the comment
355 ;; at the start of cc-engine.el for more info.
358 s)
371 ;; Go to the end of a CPP directive, or a pos just before which isn't in a
372 ;; comment. For this purpose, open strings are ignored.
373 ;;
374 ;; This function must only be called from the beginning of a CPP construct.
375 ;;
376 ;; Note that this function might do hidden buffer changes. See the comment
377 ;; at the start of cc-engine.el for more info.
380 s)
394 ;; Assuming point is at the "#" that introduces a preprocessor
395 ;; directive, it's moved forward to the end of the identifier which is
396 ;; "#define"d (or whatever c-opt-cpp-macro-define specifies). Non-nil
397 ;; is returned in this case, in all other cases nil is returned and
398 ;; point isn't moved.
399 ;;
400 ;; This function might do hidden buffer changes.
406 ;; Assuming point is at the "#" that introduces a preprocessor
407 ;; directive, it's moved forward to the start of the definition body
408 ;; if it's a "#define" (or whatever c-opt-cpp-macro-define
409 ;; specifies). Non-nil is returned in this case, in all other cases
410 ;; nil is returned and point isn't moved.
411 ;;
412 ;; This function might do hidden buffer changes.
418 \f
419 ;;; Basic utility functions.
422 ;; If ELT is a member of the (possibly dotted) list DLIST, remove all
423 ;; occurrences of it (except for any in the last cdr of DLIST).
424 ;;
425 ;; Call this as (setq DLIST (c-delq-from-dotted-list ELT DLIST)), as
426 ;; sometimes the original structure is changed, sometimes it's not.
427 ;;
428 ;; This function is needed in Emacs < 24.5, and possibly XEmacs, because
429 ;; `delq' throws an error in these versions when given a dotted list.
438 dlist))
441 ;; Return the given region as a string where all syntactic
442 ;; whitespace is removed or, where necessary, replaced with a single
443 ;; space. If PAREN-LEVEL is given then all parens in the region are
444 ;; collapsed to "()", "[]" etc.
445 ;;
446 ;; This function might do hidden buffer changes.
490 ;; Shift the indentation of the current line with the specified
491 ;; amount (positive inwards). The buffer is modified only if
492 ;; SHIFT-AMT isn't equal to zero.
495 tmp-char-inserted)
497 nil
498 ;; If we're on an empty line inside a macro, we take the point
499 ;; to be at the current indentation and shift it to the
500 ;; appropriate column. This way we don't treat the extra
501 ;; whitespace out to the line continuation as indentation.
517 ;; If initial point was within line's indentation and we're not on
518 ;; a line with a line continuation in a macro, position after the
519 ;; indentation. Else stay at same point in text.
527 ;; Return non-nil if the string KEYWORD is a known keyword. More
528 ;; precisely, the value is the symbol for the keyword in
529 ;; `c-keywords-obarray'.
533 ;; Return non-nil if the symbol KEYWORD-SYM, as returned by
534 ;; `c-keyword-sym', is a member of LANG-CONSTANT, which is the name
535 ;; of a language constant that ends with "-kwds". If KEYWORD-SYM is
536 ;; nil then the result is nil.
539 ;; String syntax chars, suitable for skip-syntax-(forward|backward).
544 ;; Regexp matching string limit syntax.
549 ;; Regexp matching WS followed by string limit syntax.
553 ;; Holds formatted error strings for the few cases where parse errors
554 ;; are reported.
561 c-parsing-error)
563 ;; Faces given to comments and string literals. This is used in some
564 ;; situations to speed up recognition; it isn't mandatory that font
565 ;; locking is in use. This variable is extended with the face in
566 ;; `c-doc-face-name' when fontification is activated in cc-fonts.el.
570 ;; New in Emacs 22.
574 ;; Put a c-type property with the given value at POS.
578 ;; Remove all occurrences of the c-type property that has the given
579 ;; value in the region between FROM and TO. VALUE is assumed to not
580 ;; be nil.
581 ;;
582 ;; Note: This assumes that c-type is put on single chars only; it's
583 ;; very inefficient if matching properties cover large regions.
592 \f
593 ;; Some debug tools to visualize various special positions. This
594 ;; debug code isn't as portable as the rest of CC Mode.
630 \f
631 ;; `c-beginning-of-statement-1' and accompanying stuff.
633 ;; KLUDGE ALERT: c-maybe-labelp is used to pass information between
634 ;; c-crosses-statement-barrier-p and c-beginning-of-statement-1. A
635 ;; better way should be implemented, but this will at least shut up
636 ;; the byte compiler.
639 ;; New awk-compatible version of c-beginning-of-statement-1, ACM 2002/6/22
641 ;; Macros used internally in c-beginning-of-statement-1 for the
642 ;; automaton actions.
645 stack)))
650 t
651 ,do-if-done
656 ;; Throw nil if stack is empty, else throw non-nil.
674 "No matching `%s' found for `%s' on line %d"
681 noerror comma-delim)
682 "Move to the start of the current statement or declaration, or to
683 the previous one if already at the beginning of one. Only
684 statements/declarations on the same level are considered, i.e. don't
685 move into or out of sexps (not even normal expression parentheses).
687 If point is already at the earliest statement within braces or parens,
688 this function doesn't move back into any whitespace preceding it; it
689 returns `same' in this case.
693 is within the continuation. If starting at such a token, move to the
694 corresponding statement start. If at the beginning of a statement,
695 move to the closest containing statement if there is any. This might
696 also stop at a continuation clause.
698 Labels are treated as part of the following statements if
699 IGNORE-LABELS is non-nil. (FIXME: Doesn't work if we stop at a known
700 statement start keyword.) Otherwise, each label is treated as a
701 separate statement.
703 Macros are ignored \(i.e. skipped over) unless point is within one, in
704 which case the content of the macro is treated as normal code. Aside
705 from any normal statement starts found in it, stop at the first token
708 macros before leaving them.
710 Return:
712 `same' if stopped at the beginning of the current statement;
713 `up' if stepped to a containing statement;
714 `previous' if stepped to a preceding statement;
715 `beginning' if stepped from a statement continuation clause to
716 its start clause; or
717 `macro' if stepped to a macro start.
718 Note that `same' and not `label' is returned if stopped at the same
719 label without crossing the colon character.
721 LIM may be given to limit the search. If the search hits the limit,
722 point will be left at the closest following token, or at the start
723 position if that is less (`same' is returned in this case).
725 NOERROR turns off error logging to `c-parsing-error'.
727 Normally only `;' and virtual semicolons are considered to delimit
728 statements, but if COMMA-DELIM is non-nil then `,' is treated
729 as a delimiter too.
731 Note that this function might do hidden buffer changes. See the
732 comment at the start of cc-engine.el for more info."
734 ;; The bulk of this function is a pushdown automaton that looks at statement
735 ;; boundaries and the tokens (such as "while") in c-opt-block-stmt-key. Its
736 ;; purpose is to keep track of nested statements, ensuring that such
737 ;; statements are skipped over in their entirety (somewhat akin to what C-M-p
738 ;; does with nested braces/brackets/parentheses).
739 ;;
740 ;; Note: The position of a boundary is the following token.
741 ;;
742 ;; Beginning with the current token (the one following point), move back one
743 ;; sexp at a time (where a sexp is, more or less, either a token or the
744 ;; entire contents of a brace/bracket/paren pair). Each time a statement
745 ;; boundary is crossed or a "while"-like token is found, update the state of
746 ;; the PDA. Stop at the beginning of a statement when the stack (holding
747 ;; nested statement info) is empty and the position has been moved.
748 ;;
749 ;; The following variables constitute the PDA:
750 ;;
751 ;; sym: This is either the "while"-like token (e.g. 'for) we've just
752 ;; scanned back over, 'boundary if we've just gone back over a
753 ;; statement boundary, or nil otherwise.
754 ;; state: takes one of the values (nil else else-boundary while
755 ;; while-boundary catch catch-boundary).
756 ;; nil means "no "while"-like token yet scanned".
757 ;; 'else, for example, means "just gone back over an else".
758 ;; 'else-boundary means "just gone back over a statement boundary
759 ;; immediately after having gone back over an else".
760 ;; saved-pos: A vector of either saved positions (tok ptok pptok, etc.) or
761 ;; of error reporting information.
762 ;; stack: The stack onto which the PDA pushes its state. Each entry
763 ;; consists of a saved value of state and saved-pos. An entry is
764 ;; pushed when we move back over a "continuation" token (e.g. else)
765 ;; and popped when we encounter the corresponding opening token
766 ;; (e.g. if).
767 ;;
768 ;;
769 ;; The following diagram briefly outlines the PDA.
770 ;;
771 ;; Common state:
772 ;; "else": Push state, goto state `else'.
773 ;; "while": Push state, goto state `while'.
774 ;; "catch" or "finally": Push state, goto state `catch'.
775 ;; boundary: Pop state.
776 ;; other: Do nothing special.
777 ;;
778 ;; State `else':
779 ;; boundary: Goto state `else-boundary'.
780 ;; other: Error, pop state, retry token.
781 ;;
782 ;; State `else-boundary':
783 ;; "if": Pop state.
784 ;; boundary: Error, pop state.
785 ;; other: See common state.
786 ;;
787 ;; State `while':
788 ;; boundary: Save position, goto state `while-boundary'.
789 ;; other: Pop state, retry token.
790 ;;
791 ;; State `while-boundary':
792 ;; "do": Pop state.
793 ;; boundary: Restore position if it's not at start, pop state. [*see below]
794 ;; other: See common state.
795 ;;
796 ;; State `catch':
797 ;; boundary: Goto state `catch-boundary'.
798 ;; other: Error, pop state, retry token.
799 ;;
800 ;; State `catch-boundary':
801 ;; "try": Pop state.
802 ;; "catch": Goto state `catch'.
803 ;; boundary: Error, pop state.
804 ;; other: See common state.
805 ;;
806 ;; [*] In the `while-boundary' state, we had pushed a 'while state, and were
807 ;; searching for a "do" which would have opened a do-while. If we didn't
808 ;; find it, we discard the analysis done since the "while", go back to this
809 ;; token in the buffer and restart the scanning there, this time WITHOUT
810 ;; pushing the 'while state onto the stack.
811 ;;
812 ;; In addition to the above there is some special handling of labels
813 ;; and macros.
817 macro-start
820 c-stmt-delim-chars-with-comma
821 c-stmt-delim-chars))
822 c-in-literal-cache c-maybe-labelp after-case:-pos saved
823 ;; Current position.
824 pos
825 ;; Position of last stmt boundary character (e.g. ;).
826 boundary-pos
827 ;; The position of the last sexp or bound that follows the
828 ;; first found colon, i.e. the start of the nonlabel part of
829 ;; the statement. It's `start' if a colon is found just after
830 ;; the start.
831 after-labels-pos
832 ;; Like `after-labels-pos', but the first such position inside
833 ;; a label, i.e. the start of the last label before the start
834 ;; of the nonlabel part of the statement.
835 last-label-pos
836 ;; The last position where a label is possible provided the
837 ;; statement started there. It's nil as long as no invalid
838 ;; label content has been found (according to
839 ;; `c-nonlabel-token-key'). It's `start' if no valid label
840 ;; content was found in the label. Note that we might still
841 ;; regard it a label if it starts with `c-label-kwds'.
842 label-good-pos
843 ;; Putative positions of the components of a bitfield declaration,
844 ;; e.g. "int foo : NUM_FOO_BITS ;"
845 bitfield-type-pos bitfield-id-pos bitfield-size-pos
846 ;; Symbol just scanned back over (e.g. 'while or 'boundary).
847 ;; See above.
848 sym
849 ;; Current state in the automaton. See above.
850 state
851 ;; Current saved positions. See above.
852 saved-pos
853 ;; Stack of conses (state . saved-pos).
854 stack
855 ;; Regexp which matches "for", "if", etc.
858 ;; Return value.
860 ;; Positions of the last three sexps or bounds we've stopped at.
861 tok ptok pptok)
871 ;; Try to skip back over unary operator characters, to register
872 ;; that we've moved.
876 ;; Protect post-++/-- operators just before a virtual semicolon.
880 ;; Skip back over any semicolon here. If it was a bare semicolon, we're
881 ;; done. Later on we ignore the boundaries for statements that don't
882 ;; contain any sexp. The only thing that is affected is that the error
883 ;; checking is a little less strict, and we really don't bother.
893 pos saved)
895 ;; Begin at start and not pos to detect macros if we stand
896 ;; directly after the #.
899 ;; Record this as the first token if not starting inside it.
902 ;; The following while loop goes back one sexp (balanced parens,
903 ;; etc. with contents, or symbol or suchlike) each iteration. This
904 ;; movement is accomplished with a call to c-backward-sexp approx 170
905 ;; lines below.
906 ;;
907 ;; The loop is exited only by throwing nil to the (catch 'loop ...):
908 ;; 1. On reaching the start of a macro;
909 ;; 2. On having passed a stmt boundary with the PDA stack empty;
910 ;; 3. On reaching the start of an Objective C method def;
911 ;; 4. From macro `c-bos-pop-state'; when the stack is empty;
912 ;; 5. From macro `c-bos-pop-state-and-retry' when the stack is empty.
916 ;; Are we in a macro, just after the opening #?
931 ;; Stop at the first token in the content of the macro.
935 ret 'macro
936 ignore-labels t))
939 ;; Do a round through the automaton if we've just passed a
940 ;; statement boundary or passed a "while"-like token.
948 ;; The PDA state handling.
949 ;;
950 ;; Refer to the description of the PDA in the opening
951 ;; comments. In the following OR form, the first leaf
952 ;; attempts to handles one of the specific actions detailed
953 ;; (e.g., finding token "if" whilst in state `else-boundary').
954 ;; We drop through to the second leaf (which handles common
955 ;; state) if no specific handler is found in the first cond.
956 ;; If a parsing error is detected (e.g. an "else" with no
957 ;; preceding "if"), we throw to the enclosing catch.
958 ;;
959 ;; Note that the (eq state 'else) means
960 ;; "we've just passed an else", NOT "we're looking for an
961 ;; else".
978 ;; Since this can cause backtracking we do a
979 ;; little more careful analysis to avoid it:
980 ;; If there's a label in front of the while
981 ;; it can't be part of a do-while.
1010 ;; This is state common. We get here when the previous
1011 ;; cond statement found no particular state handler.
1013 ;; If we have a boundary at the start
1014 ;; position we push a frame to go to the
1015 ;; previous statement.
1024 ;; Is this a real while, or a do-while?
1025 ;; The next `when' triggers unless we are SURE that
1026 ;; the `while' is not the tail end of a `do-while'.
1029 ;; The following kludge is to prevent
1030 ;; infinite recursion when called from
1031 ;; c-awk-after-if-for-while-condition-p,
1032 ;; or the like.
1036 ;; Since this can cause backtracking we do a
1037 ;; little more careful analysis to avoid it: If
1038 ;; the while isn't followed by a (possibly
1039 ;; virtual) semicolon it can't be a do-while.
1048 ;; We're either past a statement boundary or at the
1049 ;; start of a statement, so throw away any label data
1050 ;; for the previous one.
1052 last-label-pos nil
1053 c-maybe-labelp nil))))
1055 ;; Step to the previous sexp, but not if we crossed a
1056 ;; boundary, since that doesn't consume an sexp.
1060 ;; HERE IS THE SINGLE PLACE INSIDE THE PDA LOOP WHERE WE MOVE
1061 ;; BACKWARDS THROUGH THE SOURCE.
1065 ;; The end position of the area to search for statement
1066 ;; barriers in this round.
1069 ;; Go back over exactly one logical sexp, taking proper
1070 ;; account of macros and escaped EOLs.
1074 ;; Give up if we hit an unbalanced block. Since the
1075 ;; stack won't be empty the code below will report a
1076 ;; suitable error.
1079 ;; Have we moved into a macro?
1082 ;; Have we crossed a statement boundary? If not,
1083 ;; keep going back until we find one or a "real" sexp.
1090 ;; Have we just gone back over an escaped NL? This
1091 ;; doesn't count as a sexp.
1094 ;; Have we crossed a statement boundary?
1097 ;; Are we at a macro beginning?
1099 c-opt-cpp-prefix
1105 ;; Just gone back over a brace block?
1114 ;; Just gone back over some paren block?
1118 before-sws-pos)))
1121 ;; Just gone back over an ordinary symbol of some sort?
1127 ptok tok
1128 tok boundary-pos
1130 ;; Like a C "continue". Analyze the next sexp.
1133 ;; ObjC method def?
1140 ;; Might we have a bitfield declaration, "<type> <id> : <size>"?
1143 ;; The : <size> and <id> fields?
1153 ;; The <type> field?
1161 bitfield-id-pos nil)))))
1163 ;; Handle labels.
1166 ;; `c-crosses-statement-barrier-p' has found a colon, so we
1167 ;; might be in a label now. Have we got a real label
1168 ;; (including a case label) or something like C++'s "public:"?
1169 ;; A case label might use an expression rather than a token.
1172 ;; Catch C++'s inheritance construct "class foo : bar".
1186 ; been found.
1188 ;; We're in a potential label and it's the first
1189 ;; time we've found something that isn't allowed in
1190 ;; one.
1193 ;; We've moved back by a sexp, so update the token positions.
1195 pptok ptok
1196 ptok tok
1202 ;; If the stack isn't empty there might be errors to report.
1211 ;; Need to investigate closer whether we've crossed
1212 ;; between a substatement and its containing statement.
1216 pptok)
1218 ptok)
1226 after-labels-pos
1233 last-label-pos
1234 pos))
1236 ;; We're in a label. Maybe we should step to the statement
1237 ;; after it.
1242 ;; Might have jumped over several labels. Go to the last one.
1245 ;; Have we got "case <expression>:"?
1255 ;; Skip over the unary operators that can start the statement.
1258 ;; protect AWK post-inc/decrement operators, etc.
1263 ret)))
1266 "Return non-nil if there is a non-comment non-macro punctuation character
1267 between FROM and TO. FROM must not be in a string or comment. The returned
1268 value is the position of the first such character."
1279 "Return non-nil if buffer positions FROM to TO cross one or more
1280 statement or declaration boundaries. The returned value is actually
1281 the position of the earliest boundary char. FROM must not be within
1282 a string or comment.
1284 The variable `c-maybe-labelp' is set to the position of the first `:' that
1285 might start a label (i.e. not part of `::' and not preceded by `?'). If a
1286 single `?' is found, then `c-maybe-labelp' is cleared.
1288 For AWK, a statement which is terminated by an EOL (not a ; or a }) is
1290 the line. If this virtual semicolon is _at_ from, the function recognizes it.
1292 Note that this function might do hidden buffer changes. See the
1293 comment at the start of cc-engine.el for more info."
1295 ;; If the current language has CPP macros, insert # into skip-chars.
1298 c-opt-cpp-symbol ; usually "#"
1300 c-stmt-delim-chars))
1303 lit-range lit-start vsemi-pos)
1310 skip-chars
1314 ;; Virtual semicolon?
1324 ;; In a string/comment?
1331 ;; Ignore scope operators.
1335 ;; A question mark. Can't be a label, so stop
1336 ;; looking for more : and ?.
1339 ;; At a CPP construct or a "#" or "##" operator?
1350 ;; In trailing space after an as yet undetected virtual semicolon?
1353 ; unterminated string/regexp.
1358 nil))))))
1361 "Return non-nil if the point is at the first token in a statement
1362 or somewhere in the syntactic whitespace before it.
1365 Any kind of declaration-like construct that occur outside function
1368 Note that this function might do hidden buffer changes. See the
1369 comment at the start of cc-engine.el for more info."
1373 c-maybe-labelp)
1384 "Return non-nil if the point is at the first token in an expression or
1385 statement, or somewhere in the syntactic whitespace before it.
1388 grammar sense: It's any sequence of expression tokens except commas,
1389 unless they are enclosed inside parentheses of some kind. Also, an
1390 expression never continues past an enclosing parenthesis, but it might
1391 contain parenthesis pairs of any sort except braces.
1393 Since expressions never cross statement boundaries, this function also
1394 recognizes statement beginnings, just like `c-at-statement-start-p'.
1396 Note that this function might do hidden buffer changes. See the
1397 comment at the start of cc-engine.el for more info."
1402 c-maybe-labelp)
1410 \f
1411 ;; A set of functions that covers various idiosyncrasies in
1412 ;; implementations of `forward-comment'.
1414 ;; Note: Some emacsen considers incorrectly that any line comment
1415 ;; ending with a backslash continues to the next line. I can't think
1416 ;; of any way to work around that in a reliable way without changing
1417 ;; the buffer, though. Suggestions welcome. ;) (No, temporarily
1418 ;; changing the syntax for backslash doesn't work since we must treat
1419 ;; escapes in string literals correctly.)
1422 "Move forward past whitespace and the closest following comment, if any.
1423 Return t if a comment was found, nil otherwise. In either case, the
1424 point is moved past the following whitespace. Line continuations,
1425 i.e. a backslashes followed by line breaks, are treated as whitespace.
1426 The line breaks that end line comments are considered to be the
1427 comment enders, so the point will be put on the beginning of the next
1428 line if it moved past a line comment.
1430 This function does not do any hidden buffer changes."
1438 ;; Some emacsen (e.g. XEmacs 21) return t when moving
1439 ;; forwards at eob.
1440 nil
1442 ;; Emacs includes the ending newline in a b-style (c++)
1443 ;; comment, but XEmacs doesn't. We depend on the Emacs
1444 ;; behavior (which also is symmetric).
1448 t))))
1451 "Move forward past all following whitespace and comments.
1452 Line continuations, i.e. a backslashes followed by line breaks, are
1453 treated as whitespace.
1455 Note that this function might do hidden buffer changes. See the
1456 comment at the start of cc-engine.el for more info."
1459 ;; If forward-comment in at least XEmacs 21 is given a large
1460 ;; positive value, it'll loop all the way through if it hits
1461 ;; eob.
1463 ;; Some emacsen (e.g. XEmacs 21) return t when moving
1464 ;; forwards at eob.
1469 t))))
1472 "Move backward past whitespace and the closest preceding comment, if any.
1473 Return t if a comment was found, nil otherwise. In either case, the
1474 point is moved past the preceding whitespace. Line continuations,
1475 i.e. a backslashes followed by line breaks, are treated as whitespace.
1476 The line breaks that end line comments are considered to be the
1477 comment enders, so the point cannot be at the end of the same line to
1478 move over a line comment.
1480 This function does not do any hidden buffer changes."
1483 ;; When we got newline terminated comments, forward-comment in all
1484 ;; supported emacsen so far will stop at eol of each line not
1485 ;; ending with a comment when moving backwards. This corrects for
1486 ;; that, and at the same time handles line continuations.
1494 ;; Some emacsen (e.g. Emacs 19.34) return t when moving
1495 ;; backwards at bob.
1496 nil
1498 ;; Leave point after the closest following newline if we've
1499 ;; backed up over any above, since forward-comment won't move
1500 ;; backward over a line comment if point is at the end of the
1501 ;; same line.
1506 ;; If forward-comment above succeeded and we're at eol
1507 ;; then the newline we moved over above didn't end a
1508 ;; line comment, so we give it another go.
1510 t))
1512 ;; Emacs <= 20 and XEmacs move back over the closer of a
1513 ;; block comment that lacks an opener.
1516 t)))))
1519 "Move backward past all preceding whitespace and comments.
1520 Line continuations, i.e. a backslashes followed by line breaks, are
1521 treated as whitespace. The line breaks that end line comments are
1522 considered to be the comment enders, so the point cannot be at the end
1523 of the same line to move over a line comment. Unlike
1524 c-backward-syntactic-ws, this function doesn't move back over
1525 preprocessor directives.
1527 Note that this function might do hidden buffer changes. See the
1528 comment at the start of cc-engine.el for more info."
1532 ;; `forward-comment' in some emacsen (e.g. XEmacs 21.4)
1533 ;; return t when moving backwards at bob.
1539 ;; Cope specifically with ^M^J here -
1540 ;; forward-comment sometimes gets stuck after ^Ms,
1541 ;; sometimes after ^M^J.
1545 t)
1549 t))))
1550 moved-comment)
1552 ;; Emacs <= 20 and XEmacs move back over the
1553 ;; closer of a block comment that lacks an opener.
1555 t)
1557 ;; XEmacs treats line continuations as whitespace but
1558 ;; only in the backward direction, which seems a bit
1559 ;; odd. Anyway, this is necessary for Emacs.
1564 t))))))
1566 \f
1567 ;; Tools for skipping over syntactic whitespace.
1569 ;; The following functions use text properties to cache searches over
1570 ;; large regions of syntactic whitespace. It works as follows:
1571 ;;
1572 ;; o If a syntactic whitespace region contains anything but simple
1573 ;; whitespace (i.e. space, tab and line breaks), the text property
1574 ;; `c-in-sws' is put over it. At places where we have stopped
1575 ;; within that region there's also a `c-is-sws' text property.
1576 ;; That since there typically are nested whitespace inside that
1577 ;; must be handled separately, e.g. whitespace inside a comment or
1578 ;; cpp directive. Thus, from one point with `c-is-sws' it's safe
1579 ;; to jump to another point with that property within the same
1580 ;; `c-in-sws' region. It can be likened to a ladder where
1581 ;; `c-in-sws' marks the bars and `c-is-sws' the rungs.
1582 ;;
1583 ;; o The `c-is-sws' property is put on the simple whitespace chars at
1584 ;; a "rung position" and also maybe on the first following char.
1585 ;; As many characters as can be conveniently found in this range
1586 ;; are marked, but no assumption can be made that the whole range
1587 ;; is marked (it could be clobbered by later changes, for
1588 ;; instance).
1589 ;;
1590 ;; Note that some part of the beginning of a sequence of simple
1591 ;; whitespace might be part of the end of a preceding line comment
1592 ;; or cpp directive and must not be considered part of the "rung".
1593 ;; Such whitespace is some amount of horizontal whitespace followed
1594 ;; by a newline. In the case of cpp directives it could also be
1595 ;; two newlines with horizontal whitespace between them.
1596 ;;
1597 ;; The reason to include the first following char is to cope with
1598 ;; "rung positions" that don't have any ordinary whitespace. If
1599 ;; `c-is-sws' is put on a token character it does not have
1600 ;; `c-in-sws' set simultaneously. That's the only case when that
1601 ;; can occur, and the reason for not extending the `c-in-sws'
1602 ;; region to cover it is that the `c-in-sws' region could then be
1603 ;; accidentally merged with a following one if the token is only
1604 ;; one character long.
1605 ;;
1606 ;; o On buffer changes the `c-in-sws' and `c-is-sws' properties are
1607 ;; removed in the changed region. If the change was inside
1608 ;; syntactic whitespace that means that the "ladder" is broken, but
1609 ;; a later call to `c-forward-sws' or `c-backward-sws' will use the
1610 ;; parts on either side and use an ordinary search only to "repair"
1611 ;; the gap.
1612 ;;
1613 ;; Special care needs to be taken if a region is removed: If there
1614 ;; are `c-in-sws' on both sides of it which do not connect inside
1615 ;; the region then they can't be joined. If e.g. a marked macro is
1616 ;; broken, syntactic whitespace inside the new text might be
1617 ;; marked. If those marks would become connected with the old
1618 ;; `c-in-sws' range around the macro then we could get a ladder
1619 ;; with one end outside the macro and the other at some whitespace
1620 ;; within it.
1621 ;;
1622 ;; The main motivation for this system is to increase the speed in
1623 ;; skipping over the large whitespace regions that can occur at the
1624 ;; top level in e.g. header files that contain a lot of comments and
1625 ;; cpp directives. For small comments inside code it's probably
1626 ;; slower than using `forward-comment' straightforwardly, but speed is
1627 ;; not a significant factor there anyway.
1629 ; (defface c-debug-is-sws-face
1630 ; '((t (:background "GreenYellow")))
1631 ; "Debug face to mark the `c-is-sws' property.")
1632 ; (defface c-debug-in-sws-face
1633 ; '((t (:underline t)))
1634 ; "Debug face to mark the `c-in-sws' property.")
1636 ; (defun c-debug-put-sws-faces ()
1637 ; ;; Put the sws debug faces on all the `c-is-sws' and `c-in-sws'
1638 ; ;; properties in the buffer.
1639 ; (interactive)
1640 ; (save-excursion
1641 ; (c-save-buffer-state (in-face)
1642 ; (goto-char (point-min))
1643 ; (setq in-face (if (get-text-property (point) 'c-is-sws)
1644 ; (point)))
1645 ; (while (progn
1646 ; (goto-char (next-single-property-change
1647 ; (point) 'c-is-sws nil (point-max)))
1648 ; (if in-face
1649 ; (progn
1650 ; (c-debug-add-face in-face (point) 'c-debug-is-sws-face)
1651 ; (setq in-face nil))
1652 ; (setq in-face (point)))
1653 ; (not (eobp))))
1654 ; (goto-char (point-min))
1655 ; (setq in-face (if (get-text-property (point) 'c-in-sws)
1656 ; (point)))
1657 ; (while (progn
1658 ; (goto-char (next-single-property-change
1659 ; (point) 'c-in-sws nil (point-max)))
1660 ; (if in-face
1661 ; (progn
1662 ; (c-debug-add-face in-face (point) 'c-debug-in-sws-face)
1663 ; (setq in-face nil))
1664 ; (setq in-face (point)))
1665 ; (not (eobp)))))))
1668 ;;`(message ,@args)
1669 )
1672 ;; This macro does a hidden buffer change.
1679 ;; This macro does a hidden buffer change.
1686 ;; This macro does a hidden buffer change.
1693 ;; This macro does a hidden buffer change.
1700 ;; This macro does a hidden buffer change.
1708 ;; Called from `after-change-functions'. Note that if
1709 ;; `c-forward-sws' or `c-backward-sws' are used outside
1710 ;; `c-save-buffer-state' or similar then this will remove the cache
1711 ;; properties right after they're added.
1712 ;;
1713 ;; This function does hidden buffer changes.
1716 ;; Adjust the end to remove the properties in any following simple
1717 ;; ws up to and including the next line break, if there is any
1718 ;; after the changed region. This is necessary e.g. when a rung
1719 ;; marked empty line is converted to a line comment by inserting
1720 ;; "//" before the line break. In that case the line break would
1721 ;; keep the rung mark which could make a later `c-backward-sws'
1722 ;; move into the line comment instead of over it.
1732 ;; Ensure that an `c-in-sws' range gets broken. Note that it isn't
1733 ;; safe to keep a range that was continuous before the change. E.g:
1734 ;;
1735 ;; #define foo
1736 ;; \
1737 ;; bar
1738 ;;
1739 ;; There can be a "ladder" between "#" and "b". Now, if the newline
1740 ;; after "foo" is removed then "bar" will become part of the cpp
1741 ;; directive instead of a syntactically relevant token. In that
1742 ;; case there's no longer syntactic ws from "#" to "b".
1749 ;; Used by `c-forward-syntactic-ws' to implement the unbounded search.
1750 ;;
1751 ;; This function might do hidden buffer changes.
1754 ;; unmarked part of the simple ws region.
1756 rung-is-marked next-rung-is-marked simple-ws-end
1757 ;; `safe-start' is set when it's safe to cache the start position.
1758 ;; It's not set if we've initially skipped over comments and line
1759 ;; continuations since we might have gone out through the end of a
1760 ;; macro then. This provision makes `c-forward-sws' not populate the
1761 ;; cache in the majority of cases, but otoh is `c-backward-sws' by far
1762 ;; more common.
1763 safe-start)
1765 ;; Skip simple ws and do a quick check on the following character to see
1766 ;; if it's anything that can't start syntactic ws, so we can bail out
1767 ;; early in the majority of cases when there just are a few ws chars.
1776 ;; Find the last rung position to avoid setting properties in all
1777 ;; the cases when the marked rung is complete.
1778 ;; (`next-single-property-change' is certain to move at least one
1779 ;; step forward.)
1782 ;; Got no marked rung here. Since the simple ws might have started
1783 ;; inside a line comment or cpp directive we must set `rung-pos' as
1784 ;; high as possible.
1790 ;; In the following while form, we move over a "ladder" and
1791 ;; following simple WS each time round the loop, appending the WS
1792 ;; onto the ladder, joining adjacent ladders, and terminating when
1793 ;; there is no more WS or we reach EOB.
1798 ;; The following search is the main reason that `c-in-sws'
1799 ;; and `c-is-sws' aren't combined to one property.
1803 ;; If the `c-in-sws' region extended past the last
1804 ;; `c-is-sws' char we have to go back a bit.
1811 "c-forward-sws cached move %s -> %s (max %s)"
1818 ;; We'll loop here if there is simple ws after the last rung.
1819 ;; That means that there's been some change in it and it's
1820 ;; possible that we've stepped into another ladder, so extend
1821 ;; the previous one to join with it if there is one, and try to
1822 ;; use the cache again.
1824 "c-forward-sws extending rung with [%s..%s] (max %s)"
1827 ;; Remove any `c-in-sws' property from the last char of
1828 ;; the rung before we mark it with `c-is-sws', so that we
1829 ;; won't connect with the remains of a broken "ladder".
1835 last-put-in-sws-pos rung-pos)))
1837 ;; Now move over any comments (x)or a CPP construct.
1843 ;; Skipped over comments. Don't cache at eob in case the buffer
1844 ;; is narrowed.
1855 ;; Skip a preprocessor directive.
1862 ;; Don't cache at eob in case the buffer is narrowed.
1867 ;; Skip over a noise macro.
1872 ;; We've searched over a piece of non-white syntactic ws. See if this
1873 ;; can be cached.
1879 ;; Cache if we haven't skipped comments only, and if we started
1880 ;; either from a marked rung or from a completely uncached
1881 ;; position.
1886 ;; See if there's a marked rung in the encountered simple ws. If
1887 ;; so then we can cache, unless `safe-start' is nil. Even then
1888 ;; we need to do this to check if the cache can be used for the
1889 ;; next step.
1893 safe-start))
1897 "c-forward-sws caching [%s..%s] - [%s..%s] (max %s)"
1901 ;; Remove the properties for any nested ws that might be cached.
1902 ;; Only necessary for `c-is-sws' since `c-in-sws' will be set
1903 ;; anyway.
1911 last-put-in-sws-pos rung-pos))
1913 ;; Remove any `c-in-sws' property from the last char of
1914 ;; the rung before we mark it with `c-is-sws', so that we
1915 ;; won't connect with the remains of a broken "ladder".
1918 rung-end-pos))
1921 "c-forward-sws not caching [%s..%s] - [%s..%s] (max %s)"
1925 ;; Set `rung-pos' for the next rung. It's the same thing here as
1926 ;; initially, except that the rung position is set as early as
1927 ;; possible since we can't be in the ending ws of a line comment or
1928 ;; cpp directive now.
1935 ;; Make sure that the newly marked `c-in-sws' region doesn't connect to
1936 ;; another one after the point (which might occur when editing inside a
1937 ;; comment or macro).
1941 "c-forward-sws clearing at %s for cache separation"
1942 last-put-in-sws-pos)
1946 ;; If at eob we have to clear the last character before the end
1947 ;; instead since the buffer might be narrowed and there might
1948 ;; be a `c-in-sws' after (point-max). In this case it's
1949 ;; necessary to clear both properties.
1951 "c-forward-sws clearing thoroughly at %s for cache separation"
1954 last-put-in-sws-pos))))
1955 ))))
1958 ;; Used by `c-backward-syntactic-ws' to implement the unbounded search.
1959 ;;
1960 ;; This function might do hidden buffer changes.
1963 ;; part of the simple ws region.
1965 rung-is-marked simple-ws-beg cmt-skip-pos)
1967 ;; Skip simple horizontal ws and do a quick check on the preceding
1968 ;; character to see if it's anything that can't end syntactic ws, so we can
1969 ;; bail out early in the majority of cases when there just are a few ws
1970 ;; chars. Newlines are complicated in the backward direction, so we can't
1971 ;; skip over them.
1981 ;; Try to find a rung position in the simple ws preceding point, so that
1982 ;; we can get a cache hit even if the last bit of the simple ws has
1983 ;; changed recently.
1989 ;; `rung-pos' will be the earliest marked position, which means that
1990 ;; there might be later unmarked parts in the simple ws region.
1991 ;; It's not worth the effort to fix that; the last part of the
1992 ;; simple ws is also typically edited often, so it could be wasted.
1999 ;; Each time round the next while form, we move back over a ladder
2000 ;; and append any simple WS preceding it, if possible joining with
2001 ;; the previous ladder.
2007 ;; The following search is the main reason that `c-in-sws'
2008 ;; and `c-is-sws' aren't combined to one property.
2012 ;; If the `c-in-sws' region extended past the first
2013 ;; `c-is-sws' char we have to go forward a bit.
2018 "c-backward-sws cached move %s <- %s (min %s)"
2030 t
2032 nil))
2034 ;; We'll loop here if there is simple ws before the first rung.
2035 ;; That means that there's been some change in it and it's
2036 ;; possible that we've stepped into another ladder, so extend
2037 ;; the previous one to join with it if there is one, and try to
2038 ;; use the cache again.
2040 "c-backward-sws extending rung with [%s..%s] (min %s)"
2043 ;; Remove any `c-in-sws' property from the last char of
2044 ;; the rung before we mark it with `c-is-sws', so that we
2045 ;; won't connect with the remains of a broken "ladder".
2048 rung-pos)
2052 last-put-in-sws-pos rung-pos))
2061 ;; Inside a cpp directive. See if it should be skipped over.
2064 ;; Move back over all line continuations in the region skipped
2065 ;; over by `c-backward-comments'. If we go past it then we
2066 ;; started inside the cpp directive.
2075 ;; Don't move past the cpp directive if we began inside
2076 ;; it. Note that the position at the end of the last line
2077 ;; of the macro is also considered to be within it.
2079 nil)
2081 ;; It's worthwhile to spend a little bit of effort on finding
2082 ;; the end of the macro, to get a good `simple-ws-beg'
2083 ;; position for the cache. Note that `c-backward-comments'
2084 ;; could have stepped over some comments before going into
2085 ;; the macro, and then `simple-ws-beg' must be kept on the
2086 ;; same side of those comments.
2093 ;; Might happen if comments after the macro were skipped
2094 ;; over.
2098 t)))
2103 simple-ws-beg)
2104 ;; Skipped over comments. Must put point at the end of
2105 ;; the simple ws at point since we might be after a line
2106 ;; comment or cpp directive that's been partially
2107 ;; narrowed out, and we can't risk marking the simple ws
2108 ;; at the end of it.
2110 t)
2117 ;; Skipped over a noise macro
2119 t)))
2121 ;; We've searched over a piece of non-white syntactic ws. See if this
2122 ;; can be cached.
2127 ;; Cache if we started either from a marked rung or from a
2128 ;; completely uncached position.
2129 rung-is-marked
2132 ;; Cache if there's a marked rung in the encountered simple ws.
2140 "c-backward-sws caching [%s..%s] - [%s..%s] (min %s)"
2145 ;; Remove the properties for any nested ws that might be cached.
2146 ;; Only necessary for `c-is-sws' since `c-in-sws' will be set
2147 ;; anyway.
2152 ;; Remove any `c-in-sws' property from the last char of
2153 ;; the rung before we mark it with `c-is-sws', so that we
2154 ;; won't connect with the remains of a broken "ladder".
2157 rung-end-pos)
2160 last-put-in-sws-pos simple-ws-beg)
2161 rung-pos)
2166 "c-backward-sws not caching [%s..%s] - [%s..%s] (min %s)"
2172 ))
2174 ;; Make sure that the newly marked `c-in-sws' region doesn't connect to
2175 ;; another one before the point (which might occur when editing inside a
2176 ;; comment or macro).
2180 "c-backward-sws clearing at %s for cache separation"
2183 last-put-in-sws-pos))
2185 ;; If at bob and the buffer is narrowed, we have to clear the
2186 ;; character we're standing on instead since there might be a
2187 ;; `c-in-sws' before (point-min). In this case it's necessary
2188 ;; to clear both properties.
2190 "c-backward-sws clearing thoroughly at %s for cache separation"
2191 last-put-in-sws-pos)
2194 ))))
2196 \f
2197 ;; Other whitespace tools
2199 ;; Is the region (beg end) WS, and is there WS (or BOB/EOB) next to the
2200 ;; region? This is a "heuristic" function. .....
2201 ;;
2202 ;; The motivation for the second bit is to check whether removing this
2203 ;; region would coalesce two symbols.
2204 ;;
2205 ;; FIXME!!! This function doesn't check virtual semicolons in any way. Be
2206 ;; careful about using this function for, e.g. AWK. (2007/3/7)
2215 \f
2216 ;; A system for finding noteworthy parens before the point.
2219 ;; A maximum comfortable scanning distance, e.g. between
2220 ;; `c-state-cache-good-pos' and "HERE" (where we call c-parse-state). When
2221 ;; this distance is exceeded, we take "emergency measures", e.g. by clearing
2222 ;; the cache and starting again from point-min or a beginning of defun. This
2223 ;; value can be tuned for efficiency or set to a lower value for testing.
2227 ;; The state cache used by `c-parse-state' to cut down the amount of
2228 ;; searching. It's the result from some earlier `c-parse-state' call. See
2229 ;; `c-parse-state''s doc string for details of its structure.
2230 ;;
2231 ;; The use of the cached info is more effective if the next
2232 ;; `c-parse-state' call is on a line close by the one the cached state
2233 ;; was made at; the cache can actually slow down a little if the
2234 ;; cached state was made very far back in the buffer. The cache is
2235 ;; most effective if `c-parse-state' is used on each line while moving
2236 ;; forward.
2240 ;; This is a position where `c-state-cache' is known to be correct, or
2241 ;; nil (see below). It's a position inside one of the recorded unclosed
2242 ;; parens or the top level, but not further nested inside any literal or
2243 ;; subparen that is closed before the last recorded position.
2244 ;;
2245 ;; The exact position is chosen to try to be close to yet earlier than
2246 ;; the position where `c-state-cache' will be called next. Right now
2247 ;; the heuristic is to set it to the position after the last found
2248 ;; closing paren (of any type) before the line on which
2249 ;; `c-parse-state' was called. That is chosen primarily to work well
2250 ;; with refontification of the current line.
2251 ;;
2252 ;; 2009-07-28: When `c-state-point-min' and the last position where
2253 ;; `c-parse-state' or for which `c-invalidate-state-cache' was called, are
2254 ;; both in the same literal, there is no such "good position", and
2255 ;; c-state-cache-good-pos is then nil. This is the ONLY circumstance in which
2256 ;; it can be nil. In this case, `c-state-point-min-literal' will be non-nil.
2257 ;;
2258 ;; 2009-06-12: In a brace desert, c-state-cache-good-pos may also be in
2259 ;; the middle of the desert, as long as it is not within a brace pair
2260 ;; recorded in `c-state-cache' or a paren/bracket pair.
2262 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2263 ;; We maintain a simple cache of positions which aren't in a literal, so as to
2264 ;; speed up testing for non-literality.
2266 ;; The approximate interval between entries in `c-state-nonlit-pos-cache'.
2270 ;; A list of buffer positions which are known not to be in a literal or a cpp
2271 ;; construct. This is ordered with higher positions at the front of the list.
2272 ;; Only those which are less than `c-state-nonlit-pos-cache-limit' are valid.
2276 ;; An upper limit on valid entries in `c-state-nonlit-pos-cache'. This is
2277 ;; reduced by buffer changes, and increased by invocations of
2278 ;; `c-state-literal-at'.
2282 ;; A list of elements which are either buffer positions (when such positions
2283 ;; are not in literals) or lists of the form (POS TYPE START), where POS is
2284 ;; a buffer position inside a literal, TYPE is the type of the literal
2285 ;; ('string, 'c, or 'c++) and START is the start of the literal.
2289 ;; An upper limit on valid entries in `c-state-semi-nonlit-pos-cache'. This
2290 ;; is reduced by buffer changes, and increased by invocations of
2291 ;; `c-parse-ps-state-below'.
2294 ;; Truncate the upper bound of the cache `c-state-semi-nonlit-pos-cache' to
2295 ;; POS, if it is higher than that position.
2300 ;; Do a parse-partial-sexp from a position in the buffer before HERE which
2301 ;; isn't in a literal, and return information about HERE, either:
2302 ;; (STATE TYPE BEG) if HERE is in a literal; or
2303 ;; (STATE) otherwise,
2304 ;; where STATE is the parsing state at HERE, TYPE is the type of the literal
2305 ;; enclosing HERE, (one of 'string, 'c, 'c++) and BEG is the starting
2306 ;; position of that literal (including the delimiter).
2307 ;;
2308 ;; Unless NOT-IN-DELIMITER is non-nil, when TO is inside a two-character
2309 ;; comment opener, this is recognized as being in a comment literal.
2310 ;;
2311 ;; Only elements 3 (in a string), 4 (in a comment), 5 (following a quote), 7
2312 ;; (comment type), and 8 (start of comment/string), and possibly 10 (in
2313 ;; newer Emacsen only, the syntax of a position after a potential first char
2314 ;; of a two char construct) of STATE are valid.
2323 ty)
2344 ;; This function will supersede c-state-pp-to-literal.
2345 ;;
2346 ;; Do a parse-partial-sexp from a position in the buffer before HERE which
2347 ;; isn't in a literal, and return information about HERE, either:
2348 ;; (STATE TYPE (BEG . END)) if HERE is in a literal; or
2349 ;; (STATE) otherwise,
2350 ;; where STATE is the parsing state at HERE, TYPE is the type of the literal
2351 ;; enclosing HERE, (one of 'string, 'c, 'c++) and (BEG . END) is the
2352 ;; boundaries of that literal (including the delimiters).
2353 ;;
2354 ;; Unless NOT-IN-DELIMITER is non-nil, when TO is inside a two-character
2355 ;; comment opener, this is recognized as being in a comment literal.
2356 ;;
2357 ;; Only elements 3 (in a string), 4 (in a comment), 5 (following a quote), 7
2358 ;; (comment type), and 8 (start of comment/string), and possibly 10 (in
2359 ;; newer Emacsen only, the syntax of a position after a potential first char
2360 ;; of a two char construct) of STATE are valid.
2369 ty start)
2378 nil ; TARGETDEPTH
2379 nil ; STOPBEFORE
2380 s ; OLDSTATE
2398 ;; Do a parse-partial-sexp from FROM to TO, returning either
2399 ;; (STATE TYPE (BEG . END)) if TO is in a literal; or
2400 ;; (STATE) otherwise,
2401 ;; where STATE is the parsing state at TO, TYPE is the type of the literal
2402 ;; (one of 'c, 'c++, 'string) and (BEG . END) is the boundaries of the literal,
2403 ;; including the delimiters.
2404 ;;
2405 ;; Unless NOT-IN-DELIMITER is non-nil, when TO is inside a two-character
2406 ;; comment opener, this is recognized as being in a comment literal.
2407 ;;
2408 ;; Only elements 3 (in a string), 4 (in a comment), 5 (following a quote),
2409 ;; 7 (comment type) and 8 (start of comment/string) (and possibly 9) of
2410 ;; STATE are valid.
2414 ty co-st)
2422 nil ; TARGETDEPTH
2423 nil ; STOPBEFORE
2424 s ; OLDSTATE
2441 ;; Create a list suitable to use as the old-state parameter to
2442 ;; `parse-partial-sexp', out of ELT. ELT is either just a number, a buffer
2443 ;; position, or it is a list (POS TYPE STARTING-POS). Here POS is the
2444 ;; buffer position the other elements are pertinent for, TYPE is either 'c
2445 ;; or 'c++ (for a comment) or a character (for a string delimiter) or t
2446 ;; (meaning a string fence opened the string), STARTING-POS is the starting
2447 ;; position of the comment or string.
2462 elt)))
2464 in-string in-comment after-quote
2465 min-depth com-style com-str-start
2466 intermediate nil))
2470 ;; Convert STATE, a `parse-partial-sexp' state valid at POINT, to an element
2471 ;; for the `c-state-semi-nonlit-pos-cache' cache. This is either POINT
2472 ;; (when point is not in a literal) or a list (POINT TYPE STARTING-POS),
2473 ;; where TYPE is the type of the literal, either 'string, 'c, or 'c++, and
2474 ;; STARTING-POS is the starting position of the comment or string.
2486 ;; Get the buffer position from ELT, an element from the cache
2487 ;; `c-state-semi-nonlit-pos-cache'.
2489 elt
2493 ;; Given a buffer position HERE, Return a cons (CACHE-POS . STATE), where
2494 ;; CACHE-POS is a position not very far before HERE for which the
2495 ;; parse-partial-sexp STATE is valid. Note that the only valid elements of
2496 ;; STATE are those concerning comments and strings; STATE is the state of a
2497 ;; null `parse-partial-sexp' scan when CACHE-POS is not in a comment or
2498 ;; string.
2503 elt state pos npos high-elt)
2504 ;; Trim the cache to take account of buffer changes.
2506 c-state-semi-nonlit-pos-cache-limit))
2524 ;; Add an element to `c-state-semi-nonlit-pos-cache' each iteration.
2538 ;; Return a buffer position before HERE which is "safe", i.e. outside any
2539 ;; string, comment, or macro.
2540 ;;
2541 ;; NOTE: This function manipulates `c-state-nonlit-pos-cache'. This cache
2542 ;; MAY NOT contain any positions within macros, since macros are frequently
2543 ;; turned into comments by use of the `c-cpp-delimiter' category properties.
2544 ;; We cannot rely on this mechanism whilst determining a cache pos since
2545 ;; this function is also called from outwith `c-parse-state'.
2550 pos npos high-pos lit macro-beg macro-end)
2551 ;; Trim the cache to take account of buffer changes.
2563 ;; Add an element to `c-state-nonlit-pos-cache' each iteration.
2569 ;; Test for being in a literal. If so, go to after it.
2576 ;; Test for being in a macro. If so, go to after it.
2591 ;; Add one extra element above HERE so as to to avoid the previous
2592 ;; expensive calculation when the next call is close to the current
2593 ;; one. This is especially useful when inside a large macro.
2600 pos))))
2603 ;; If position HERE is inside a literal, return (START . END), the
2604 ;; boundaries of the literal (which may be outside the accessible bit of the
2605 ;; buffer). Otherwise, return nil.
2606 ;;
2607 ;; This function is almost the same as `c-literal-limits'. Previously, it
2608 ;; differed in that it was a lower level function, and that it rigorously
2609 ;; followed the syntax from BOB. `c-literal-limits' is now (2011-12)
2610 ;; virtually identical to this function.
2618 ;; Return the start of the literal containing POS, or POS itself.
2620 pos))
2623 ;; Return a position outside of a string/comment/macro at or before POS.
2624 ;; STATE is the parse-partial-sexp state at POS.
2628 pos)))
2633 res))))
2635 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2636 ;; Stuff to do with point-min, and coping with any literal there.
2639 ;; This is (point-min) when `c-state-cache' was last calculated. A change of
2640 ;; narrowing is likely to affect the parens that are visible before the point.
2646 ;; These two variables define the literal, if any, containing point-min.
2647 ;; Their values are, respectively, 'string, c, or c++, and the start of the
2648 ;; literal. If there's no literal there, they're both nil.
2652 ;; This is the earliest buffer-pos from which scanning can be done. It is
2653 ;; either the end of the literal containing point-min, or point-min itself.
2654 ;; It becomes nil if the buffer is changed earlier than this point.
2656 ;; Return the lowest valid scanning pos. This will be the end of the
2657 ;; literal enclosing point-min, or point-min itself.
2669 ;; Determine the properties of any literal containing POINT-MIN, setting the
2670 ;; variables `c-state-point-min-lit-type', `c-state-point-min-lit-start',
2671 ;; and `c-state-min-scan-pos' accordingly. The return value is meaningless.
2673 lit)
2688 c-state-point-min-lit-start nil
2689 c-state-min-scan-pos p-min)))))
2692 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2693 ;; A variable which signals a brace dessert - helpful for reducing the number
2694 ;; of fruitless backward scans.
2697 ;; Used only in `c-append-lower-brace-pair-to-state-cache'. It is set when
2698 ;; that defun has searched backwards for a brace pair and not found one. Its
2699 ;; value is either nil or a cons (PA . FROM), where PA is the position of the
2700 ;; enclosing opening paren/brace/bracket which bounds the backwards search (or
2701 ;; nil when at top level) and FROM is where the backward search started. It
2702 ;; is reset to nil in `c-invalidate-state-cache'.
2705 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2706 ;; Lowish level functions/macros which work directly on `c-state-cache', or a
2707 ;; list of like structure.
2709 ;; Return the address of the top left brace/bracket/paren recorded in CACHE
2710 ;; (default `c-state-cache') (or nil).
2717 ;; Return the address of the latest brace/bracket/paren (whether left or
2718 ;; right) recorded in CACHE (default `c-state-cache') or nil.
2725 ;; Return the position just after the latest brace/bracket/paren (whether
2726 ;; left or right) recorded in CACHE (default `c-state-cache') or nil.
2734 ;; From the state-cache, determine the buffer position from which we might
2735 ;; scan forward to HERE to update this cache. This position will be just
2736 ;; after a paren/brace/bracket recorded in the cache, if possible, otherwise
2737 ;; return the earliest position in the accessible region which isn't within
2738 ;; a literal. If the visible portion of the buffer is entirely within a
2739 ;; literal, return NIL.
2741 ;(while (>= (or (c-state-cache-top-lparen c) 1) here)
2757 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2758 ;; Variables which keep track of preprocessor constructs.
2767 ;; These are the limits of the macro containing point at the previous call of
2768 ;; `c-parse-state', or nil.
2770 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2771 ;; Defuns which analyze the buffer, yet don't change `c-state-cache'.
2773 ;; Return a start position for building `c-state-cache' from
2774 ;; scratch. This will be at the top level, 2 defuns back.
2776 ;; Go back 2 bods, but ignore any bogus positions returned by
2777 ;; beginning-of-defun (i.e. open paren in column zero).
2787 ;; Return the position of the opening paren/brace/bracket before HERE- which
2788 ;; matches the outermost close p/b/b between HERE+ and TOP. Except when
2789 ;; there's a macro, HERE- and HERE+ are the same. Like this:
2790 ;;
2791 ;; ............................................
2792 ;; | |
2793 ;; ( [ ( .........#macro.. ) ( ) ] )
2794 ;; ^ ^ ^ ^
2795 ;; | | | |
2796 ;; return HERE- HERE+ TOP
2797 ;;
2798 ;; If there aren't enough opening paren/brace/brackets, return the position
2799 ;; of the outermost one found, or HERE- if there are none. If there are no
2800 ;; closing p/b/bs between HERE+ and TOP, return HERE-. HERE-/+ and TOP
2801 ;; must not be inside literals. Only the accessible portion of the buffer
2802 ;; will be scanned.
2804 ;; PART 1: scan from `here+' up to `top', accumulating ")"s which enclose
2805 ;; `here'. Go round the next loop each time we pass over such a ")". These
2806 ;; probably match "("s before `here-'.
2818 ;; PART 2: Scan back before `here-' searching for the "("s
2819 ;; matching/mismatching the ")"s found above. We only need to direct the
2820 ;; caller to scan when we've encountered unmatched right parens.
2829 pos))
2832 ;; Determine the scanning strategy for adjusting `c-parse-state', attempting
2833 ;; to minimize the amount of scanning. HERE is the pertinent position in
2834 ;; the buffer, GOOD-POS is a position where `c-state-cache' (possibly with
2835 ;; its head trimmed) is known to be good, or nil if there is no such
2836 ;; position.
2837 ;;
2838 ;; The return value is a list, one of the following:
2839 ;;
2840 ;; o - ('forward START-POINT) - scan forward from START-POINT,
2841 ;; which is not less than the highest position in `c-state-cache' below HERE,
2842 ;; which is after GOOD-POS.
2843 ;; o - ('backward nil) - scan backwards (from HERE).
2844 ;; o - ('back-and-forward START-POINT) - like 'forward, but when HERE is earlier
2845 ;; than GOOD-POS.
2846 ;; o - ('BOD START-POINT) - scan forwards from START-POINT, which is at the
2847 ;; top level.
2848 ;; o - ('IN-LIT nil) - point is inside the literal containing point-min.
2850 BOD-pos ; position of 2nd BOD before HERE.
2851 strategy ; 'forward, 'backward, 'BOD, or 'IN-LIT.
2852 start-point
2858 start-point nil
2859 cache-pos nil
2870 start-point cache-pos
2873 ;; Might we be better off starting from the top level, two defuns back,
2874 ;; instead? This heuristic no longer works well in C++, where
2875 ;; declarations inside namespace brace blocks are frequently placed at
2876 ;; column zero. (2015-11-10): Remove the condition on C++ Mode.
2878 open-paren-in-column-0-is-defun-start)
2879 ;; (not (c-major-mode-is 'c++-mode))
2884 start-point BOD-pos)))
2889 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2890 ;; Routines which change `c-state-cache' and associated values.
2892 ;; The region (more precisely, point-min) has changed since we
2893 ;; calculated `c-state-cache'. Amend `c-state-cache' accordingly.
2895 ;; If point-min has MOVED BACKWARDS then we drop the state completely.
2896 ;; It would be possible to do a better job here and recalculate the top
2897 ;; only.
2901 c-state-cache-good-pos c-state-min-scan-pos
2902 c-state-brace-pair-desert nil))
2904 ;; point-min has MOVED FORWARD.
2906 ;; Is the new point-min inside a (different) literal?
2911 ;; Cut off a bit of the tail from `c-state-cache'.
2913 pa)
2922 ; inside a recorded
2923 ; brace pair.
2925 c-state-cache-good-pos c-state-min-scan-pos)
2928 )))
2933 ;; If there is a brace pair preceding FROM in the buffer, at the same level
2934 ;; of nesting (not necessarily immediately preceding), push a cons onto
2935 ;; `c-state-cache' to represent it. FROM must not be inside a literal. If
2936 ;; UPPER-LIM is non-nil, we append the highest brace pair whose "}" is below
2937 ;; UPPER-LIM.
2938 ;;
2939 ;; Return non-nil when this has been done.
2940 ;;
2941 ;; The situation it copes with is this transformation:
2942 ;;
2943 ;; OLD: { (.) {...........}
2944 ;; ^ ^
2945 ;; FROM HERE
2946 ;;
2947 ;; NEW: { {....} (.) {.........
2948 ;; ^ ^ ^
2949 ;; LOWER BRACE PAIR HERE or HERE
2950 ;;
2951 ;; This routine should be fast. Since it can get called a LOT, we maintain
2952 ;; `c-state-brace-pair-desert', a small cache of "failures", such that we
2953 ;; reduce the time wasted in repeated fruitless searches in brace deserts.
2963 ;; Don't start scanning in the middle of a CPP construct unless
2964 ;; it contains HERE - these constructs, in Emacs, are "commented
2965 ;; out" with category properties.
2968 macro-start-or-from
2969 from))
2973 ;; If we're essentially repeating a fruitless search, just give up.
2979 ;; DESERT-LIM. Avoid repeated searching through the cached desert.
2985 ;; CACHE-LIM. This limit will be necessary when an opening
2986 ;; paren at `cache-pos' has just had its matching close paren
2987 ;; inserted into the buffer. `cache-pos' continues to be a
2988 ;; search bound, even though the algorithm below would skip
2989 ;; over the new paren pair.
2998 ;; The top limit is EOB to ensure that `bra' is inside the
2999 ;; accessible part of the buffer at the next scan operation.
3002 ;; In the next pair of nested loops, the inner one moves back past a
3003 ;; pair of (mis-)matching parens or brackets; the outer one moves
3004 ;; back over a sequence of unmatched close brace/paren/bracket each
3005 ;; time round.
3021 ; brace, ignore it.
3024 ;; We've found the desired brace-pair.
3036 ;; We haven't found a brace pair. Record this in the cache.
3039 bra
3044 ;; If BRA+1 is nil, do nothing. Otherwise, BRA+1 is the buffer position
3045 ;; following a {, and that brace has a (mis-)matching } (or ]), and we
3046 ;; "push" "a" brace pair onto `c-state-cache'.
3047 ;;
3048 ;; Here "push" means overwrite the top element if it's itself a brace-pair,
3049 ;; otherwise push it normally.
3050 ;;
3051 ;; The brace pair we push is normally the one surrounding BRA+1, but if the
3052 ;; latter is inside a macro, not being a macro containing
3053 ;; MACRO-START-OR-HERE, we scan backwards through the buffer for a non-macro
3054 ;; base pair. This latter case is assumed to be rare.
3055 ;;
3056 ;; Note: POINT is not preserved in this routine.
3066 c-state-cache)))
3067 ;; N.B. This defsubst codes one method for the simple, normal case,
3068 ;; and a more sophisticated, slower way for the general case. Don't
3069 ;; eliminate this defsubst - it's a speed optimization.
3073 ;; Scan the buffer from FROM to HERE, adding elements into `c-state-cache'
3074 ;; for braces etc. Return a candidate for `c-state-cache-good-pos'.
3075 ;;
3076 ;; FROM must be after the latest brace/paren/bracket in `c-state-cache', if
3077 ;; any. Typically, it is immediately after it. It must not be inside a
3078 ;; literal.
3084 here)))
3087 ; Is actually the pos. to scan for a (/{/[ from,
3088 ; which sometimes is after a silly )/}/].
3091 ; good-pos.
3099 ;; Each time round the following loop, we enter a successively deeper
3100 ;; level of brace/paren nesting. (Except sometimes we "continue at
3101 ;; the existing level".) `pa+1' is a pos inside an opening
3102 ;; brace/paren/bracket, usually just after it.
3105 ;; Each time round the next loop moves forward over an opening then
3106 ;; a closing brace/bracket/paren. This loop is white hot, so it
3107 ;; plays ugly tricks to go fast. DON'T PUT ANYTHING INTO THIS
3108 ;; LOOP WHICH ISN'T ABSOLUTELY NECESSARY!!! It terminates when a
3109 ;; call of `scan-lists' signals an error, which happens when there
3110 ;; are no more b/b/p's to scan.
3121 ;; We've just entered a deeper nesting level.
3123 ;; Insert the brace pair (if present) and the single open
3124 ;; paren/brace/bracket into `c-state-cache' It cannot be
3125 ;; inside a macro, except one around point, because of what
3126 ;; `c-neutralize-syntax-in-CPP' has done.
3128 ;; Insert the opening brace/bracket/paren position.
3130 ;; Clear admin stuff for the next more nested part of the scan.
3134 ;; All open p/b/b's at this nesting level, if any, have probably
3135 ;; been closed by matching/mismatching ones. We're probably
3136 ;; finished - we just need to check for having found an
3137 ;; unmatched )/}/], which we ignore. Such a )/}/] can't be in a
3138 ;; macro, due the action of `c-neutralize-syntax-in-CPP'.
3141 ;; Record the final, innermost, brace-pair if there is one.
3144 ;; Determine a good pos
3152 here-bol))
3161 ;; Remove stale entries from the `c-cache-state', i.e. those which will
3162 ;; not be in it when it is amended for position HERE. This may involve
3163 ;; replacing a CONS element for a brace pair containing HERE with its car.
3164 ;; Additionally, the "outermost" open-brace entry before HERE will be
3165 ;; converted to a cons if the matching close-brace is below HERE.
3166 ;;
3167 ;; START-POINT is a "maximal" "safe position" - there must be no open
3168 ;; parens/braces/brackets between START-POINT and HERE.
3169 ;;
3170 ;; As a second thing, calculate the result of parse-partial-sexp at
3171 ;; PPS-POINT, w.r.t. START-POINT. The motivation here is that
3172 ;; `c-state-cache-good-pos' may become PPS-POINT, but the caller may need to
3173 ;; adjust it to get outside a string/comment. (Sorry about this! The code
3174 ;; needs to be FAST).
3175 ;;
3176 ;; Return a list (GOOD-POS SCAN-BACK-POS CONS-SEPARATED PPS-STATE), where
3177 ;; o - GOOD-POS is a position where the new value `c-state-cache' is known
3178 ;; to be good (we aim for this to be as high as possible);
3179 ;; o - SCAN-BACK-POS, if not nil, indicates there may be a brace pair
3180 ;; preceding POS which needs to be recorded in `c-state-cache'. It is a
3181 ;; position to scan backwards from. It is the position of the "{" of the
3182 ;; last element to be removed from `c-state-cache', when that elt is a
3183 ;; cons, otherwise nil.
3184 ;; o - CONS-SEPARATED is t when a cons element in `c-state-cache' has been
3185 ;; replaced by its car because HERE lies inside the brace pair represented
3186 ;; by the cons.
3187 ;; o - PPS-STATE is the parse-partial-sexp state at PPS-POINT.
3197 ; start-point or nil
3204 ; HERE), or nil.
3210 pps-state ; Will be 9 or 10 elements long.
3211 pos
3212 upper-lim ; ,beyond which `c-state-cache' entries are removed
3213 scan-back-pos
3214 cons-separated
3215 pair-beg pps-point-state target-depth)
3217 ;; Remove entries beyond HERE. Also remove any entries inside
3218 ;; a macro, unless HERE is in the same macro.
3223 here
3229 ;; If `upper-lim' is inside the last recorded brace pair, remove its
3230 ;; RBrace and indicate we'll need to search backwards for a previous
3231 ;; brace pair.
3237 cons-separated t))
3239 ;; The next loop jumps forward out of a nested level of parens each
3240 ;; time round; the corresponding elements in `c-state-cache' are
3241 ;; removed. `pos' is just after the brace-pair or the open paren at
3242 ;; (car c-state-cache). There can be no open parens/braces/brackets
3243 ;; between `start-point'/`start-point-actual-macro-start' and HERE,
3244 ;; due to the interface spec to this function.
3247 in-macro-start)))
3249 ; marked by a `category' text property.
3250 start-point))
3261 ;; Do nothing when we've merely reached pps-point.
3262 )
3264 ;; Scan!
3268 target-depth
3269 nil pps-state))
3279 ;; We've just moved out of the paren pair containing the brace-pair
3280 ;; at (car c-state-cache). `pair-beg' is where the open paren is,
3281 ;; and is potentially where the open brace of a cons in
3282 ;; c-state-cache will be.
3288 ; containing Lparen
3292 ; together with enclosed brace pair.
3293 ;; (t nil) ; Ignore an unmated Rparen.
3294 )))
3299 nil nil ; TARGETDEPTH, STOPBEFORE
3300 pps-state)))
3302 ;; If the last paren pair we moved out of was actually a brace pair,
3303 ;; insert it into `c-state-cache'.
3308 c-state-cache)))
3313 ;; Strip stale elements of `c-state-cache' by moving backwards through the
3314 ;; buffer, and inform the caller of the scenario detected.
3315 ;;
3316 ;; HERE is the position we're setting `c-state-cache' for.
3317 ;; CACHE-POS (a locally bound variable) is just after the latest recorded
3318 ;; position in `c-state-cache' before HERE, or a position at or near
3319 ;; point-min which isn't in a literal.
3320 ;;
3321 ;; This function must only be called only when (> `c-state-cache-good-pos'
3322 ;; HERE). Usually the gap between CACHE-POS and HERE is large. It is thus
3323 ;; optimized to eliminate (or minimize) scanning between these two
3324 ;; positions.
3325 ;;
3326 ;; Return a three element list (GOOD-POS SCAN-BACK-POS FWD-FLAG), where:
3327 ;; o - GOOD-POS is a "good position", where `c-state-cache' is valid, or
3328 ;; could become so after missing elements are inserted into
3329 ;; `c-state-cache'. This is JUST AFTER an opening or closing
3330 ;; brace/paren/bracket which is already in `c-state-cache' or just before
3331 ;; one otherwise. exceptionally (when there's no such b/p/b handy) the BOL
3332 ;; before `here''s line, or the start of the literal containing it.
3333 ;; o - SCAN-BACK-POS, if non-nil, indicates there may be a brace pair
3334 ;; preceding POS which isn't recorded in `c-state-cache'. It is a position
3335 ;; to scan backwards from.
3336 ;; o - FWD-FLAG, if non-nil, indicates there may be parens/braces between
3337 ;; POS and HERE which aren't recorded in `c-state-cache'.
3338 ;;
3339 ;; The comments in this defun use "paren" to mean parenthesis or square
3340 ;; bracket (as contrasted with a brace), and "(" and ")" likewise.
3341 ;;
3342 ;; . {..} (..) (..) ( .. { } ) (...) ( .... . ..)
3343 ;; | | | | | |
3344 ;; CP E here D C good
3347 pa ren ; positions of "(" and ")"
3348 dropped-cons ; whether the last element dropped from `c-state-cache'
3349 ; was a cons (representing a brace-pair)
3350 good-pos ; see above.
3351 lit ; (START . END) of a literal containing some point.
3352 here-lit-start here-lit-end ; bounds of literal containing `here'
3353 ; or `here' itself.
3358 ;; Remove completely irrelevant entries from `c-state-cache'.
3364 ;; At this stage, (>= pos here);
3365 ;; (< (c-state-cache-top-lparen) here) (or is nil).
3370 ;; CASE 1: The top of the cache is a brace pair which now encloses
3371 ;; `here'. As good-pos, return the address. of the "{". Since we've no
3372 ;; knowledge of what's inside these braces, we have no alternative but
3373 ;; to direct the caller to scan the buffer from the opening brace.
3377 ; entry into a { entry, so the caller needs to
3378 ; search for a brace pair before the {.
3380 ;; `here' might be inside a literal. Check for this.
3385 ;; Has `here' just "newly entered" a macro?
3398 ;; `here' might be nested inside any depth of parens (or brackets but
3399 ;; not braces). Scan backwards to find the outermost such opening
3400 ;; paren, if there is one. This will be the scan position to return.
3407 ;; CASE 2: Address of outermost ( or [ which now encloses `here', but
3408 ;; didn't enclose the (previous) `c-state-cache-good-pos'. If there is
3409 ;; a brace pair preceding this, it will already be in `c-state-cache',
3410 ;; unless there was a brace pair after it, i.e. there'll only be one to
3411 ;; scan for if we've just deleted one.
3414 ;; `here' isn't enclosed in a (previously unrecorded) bracket/paren.
3415 ;; Further forward scanning isn't needed, but we still need to find a
3416 ;; GOOD-POS. Step out of all enclosing "("s on HERE's line.
3428 ;; CASE 3: After a }/)/] before `here''s BOL.
3433 ;; CASE 3.5: Just after an existing entry in `c-state-cache' on `here''s
3434 ;; line or the previous line.
3438 ;; CASE 4; Best of a bad job: BOL before `here-bol', or beginning of
3439 ;; literal containing it.
3443 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3444 ;; Externally visible routines.
3448 c-state-cache-good-pos 1
3449 c-state-nonlit-pos-cache nil
3450 c-state-nonlit-pos-cache-limit 1
3451 c-state-semi-nonlit-pos-cache nil
3452 c-state-semi-nonlit-pos-cache-limit 1
3453 c-state-brace-pair-desert nil
3454 c-state-point-min 1
3455 c-state-point-min-lit-type nil
3456 c-state-point-min-lit-start nil
3457 c-state-min-scan-pos 1
3458 c-state-old-cpp-beg nil
3459 c-state-old-cpp-end nil)
3462 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3463 ;; Debugging routines to dump `c-state-cache' in a "replayable" form.
3464 ;; (defmacro c-sc-de (elt) ; "c-state-cache-dump-element"
3465 ;; `(format ,(concat "(setq " (symbol-name elt) " %s) ") ,elt))
3466 ;; (defmacro c-sc-qde (elt) ; "c-state-cache-quote-dump-element"
3467 ;; `(format ,(concat "(setq " (symbol-name elt) " '%s) ") ,elt))
3468 ;; (defun c-state-dump ()
3469 ;; ;; For debugging.
3470 ;; ;(message
3471 ;; (concat
3472 ;; (c-sc-qde c-state-cache)
3473 ;; (c-sc-de c-state-cache-good-pos)
3474 ;; (c-sc-qde c-state-nonlit-pos-cache)
3475 ;; (c-sc-de c-state-nonlit-pos-cache-limit)
3476 ;; (c-sc-qde c-state-brace-pair-desert)
3477 ;; (c-sc-de c-state-point-min)
3478 ;; (c-sc-de c-state-point-min-lit-type)
3479 ;; (c-sc-de c-state-point-min-lit-start)
3480 ;; (c-sc-de c-state-min-scan-pos)
3481 ;; (c-sc-de c-state-old-cpp-beg)
3482 ;; (c-sc-de c-state-old-cpp-end)))
3483 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3486 ;; Invalidate all info on `c-state-cache' that applies to the buffer at HERE
3487 ;; or higher and set `c-state-cache-good-pos' accordingly. The cache is
3488 ;; left in a consistent state.
3489 ;;
3490 ;; This is much like `c-whack-state-after', but it never changes a paren
3491 ;; pair element into an open paren element. Doing that would mean that the
3492 ;; new open paren wouldn't have the required preceding paren pair element.
3493 ;;
3494 ;; This function is called from c-before-change.
3496 ;; The caches of non-literals:
3497 ;; Note that we use "<=" for the possibility of the second char of a two-char
3498 ;; comment opener being typed; this would invalidate any cache position at
3499 ;; HERE.
3504 ;; `c-state-cache':
3505 ;; Case 1: if `here' is in a literal containing point-min, everything
3506 ;; becomes (or is already) nil.
3510 c-state-cache-good-pos nil
3511 c-state-min-scan-pos nil)
3513 ;; Truncate `c-state-cache' and set `c-state-cache-good-pos' to a value
3514 ;; below `here'. To maintain its consistency, we may need to insert a new
3515 ;; brace pair.
3517 too-high-pa ; recorded {/(/[ next above or just below here, or nil.
3518 dropped-cons ; was the last removed element a brace pair?
3519 pa)
3520 ;; The easy bit - knock over-the-top bits off `c-state-cache'.
3527 ;; Do we need to add in an earlier brace pair, having lopped one off?
3534 ;; The brace-pair desert marker:
3542 ;; Find and record all noteworthy parens between some good point earlier in
3543 ;; the file and point. That good point is at least the beginning of the
3544 ;; top-level construct we are in, or the beginning of the preceding
3545 ;; top-level construct if we aren't in one.
3546 ;;
3547 ;; The returned value is a list of the noteworthy parens with the last one
3548 ;; first. If an element in the list is an integer, it's the position of an
3549 ;; open paren (of any type) which has not been closed before the point. If
3550 ;; an element is a cons, it gives the position of a closed BRACE paren
3551 ;; pair[*]; the car is the start brace position and the cdr is the position
3552 ;; following the closing brace. Only the last closed brace paren pair
3553 ;; before each open paren and before the point is recorded, and thus the
3554 ;; state never contains two cons elements in succession. When a close brace
3555 ;; has no matching open brace (e.g., the matching brace is outside the
3556 ;; visible region), it is not represented in the returned value.
3557 ;;
3558 ;; [*] N.B. The close "brace" might be a mismatching close bracket or paren.
3559 ;; This defun explicitly treats mismatching parens/braces/brackets as
3560 ;; matching. It is the open brace which makes it a "brace" pair.
3561 ;;
3562 ;; If POINT is within a macro, open parens and brace pairs within
3563 ;; THIS macro MIGHT be recorded. This depends on whether their
3564 ;; syntactic properties have been suppressed by
3565 ;; `c-neutralize-syntax-in-CPP'. This might need fixing (2008-12-11).
3566 ;;
3567 ;; Currently no characters which are given paren syntax with the
3568 ;; syntax-table property are recorded, i.e. angle bracket arglist
3569 ;; parens are never present here. Note that this might change.
3570 ;;
3571 ;; BUG: This function doesn't cope entirely well with unbalanced
3572 ;; parens in macros. (2008-12-11: this has probably been resolved
3573 ;; by the function `c-neutralize-syntax-in-CPP'.) E.g. in the
3574 ;; following case the brace before the macro isn't balanced with the
3575 ;; one after it:
3576 ;;
3577 ;; {
3578 ;; #define X {
3579 ;; }
3580 ;;
3581 ;; Note to maintainers: this function DOES get called with point
3582 ;; within comments and strings, so don't assume it doesn't!
3583 ;;
3584 ;; This function might do hidden buffer changes.
3587 strategy ; 'forward, 'backward etc..
3588 ;; Candidate positions to start scanning from:
3589 cache-pos ; highest position below HERE already existing in
3590 ; cache (or 1).
3591 good-pos
3592 start-point ; (when scanning forward) a place below HERE where there
3593 ; are no open parens/braces between it and HERE.
3594 bopl-state
3595 res
3596 cons-separated
3598 ;; If POINT-MIN has changed, adjust the cache
3602 ;; Strategy?
3609 c-state-cache-good-pos start-point))
3611 ;; SCAN!
3619 ; start-point)
3629 good-pos)))
3641 good-pos)))
3645 c-state-cache-good-pos nil))))
3647 c-state-cache)
3650 ;; This is a wrapper over `c-invalidate-state-cache-1'.
3651 ;;
3652 ;; It suppresses the syntactic effect of the < and > (template) brackets and
3653 ;; of all parens in preprocessor constructs, except for any such construct
3654 ;; containing point. We can then call `c-invalidate-state-cache-1' without
3655 ;; worrying further about macros and template delimiters.
3657 ;; Emacs
3662 c-state-old-cpp-beg
3663 c-state-old-cpp-end
3667 ;; XEmacs
3671 ;; If PLACE is non-nil, return a marker marking it, otherwise nil.
3672 ;; We (re)use MARKER.
3678 ;; This is a wrapper over `c-parse-state-1'. See that function for a
3679 ;; description of the functionality and return value.
3680 ;;
3681 ;; It suppresses the syntactic effect of the < and > (template) brackets and
3682 ;; of all parens in preprocessor constructs, except for any such construct
3683 ;; containing point. We can then call `c-parse-state-1' without worrying
3684 ;; further about macros and template delimiters.
3691 here-cpp-beg)
3693 ;; FIXME!!! Put in a `condition-case' here to protect the integrity of the
3694 ;; subsystem.
3697 ;; Emacs
3701 here-cpp-beg here-cpp-end
3705 ;; XEmacs
3709 c-state-old-cpp-end
3712 ;; Debug tool to catch cache inconsistencies. This is called from
3713 ;; 000tests.el.
3736 c-state-cache-good-pos
3737 c-state-nonlit-pos-cache
3738 c-state-nonlit-pos-cache-limit
3739 c-state-semi-nonlit-pos-cache
3740 c-state-semi-nonlit-pos-cache-limit
3741 c-state-brace-pair-desert
3742 c-state-point-min
3743 c-state-point-min-lit-type
3744 c-state-point-min-lit-start
3745 c-state-min-scan-pos
3746 c-state-old-cpp-beg
3747 c-state-old-cpp-end
3748 c-parse-state-point))))
3770 conses-not-ok))
3787 ;; The cache can actually go further back due to the ad-hoc way
3788 ;; the first paren is found, so try to whack off a bit of its
3789 ;; start before complaining.
3790 ;; (save-excursion
3791 ;; (goto-char (or (c-least-enclosing-brace res2) (point)))
3792 ;; (c-beginning-of-defun-1)
3793 ;; (while (not (or (bobp) (eq (char-after) ?{)))
3794 ;; (c-beginning-of-defun-1))
3795 ;; (unless (equal (c-whack-state-before (point) res1) res2)
3796 ;; (message (concat "c-parse-state inconsistency at %s: "
3797 ;; "using cache: %s, from scratch: %s")
3798 ;; here res1 res2)))
3801 here res1 res2 min-point)
3807 here res1)
3812 res2 ; res1 correct a cascading series of errors ASAP
3813 ))
3819 'c-debug-parse-state
3827 \f
3829 ;; Whack off any state information from PAREN-STATE which lies
3830 ;; before BUFPOS. Not destructive on PAREN-STATE.
3833 car)
3844 ;; Whack off any state information from PAREN-STATE which lies at or
3845 ;; after BUFPOS. Not destructive on PAREN-STATE.
3850 ;; just check the car, because in a balanced brace
3851 ;; expression, it must be impossible for the corresponding
3852 ;; close brace to be before point, but the open brace to
3853 ;; be after.
3855 nil ; whack it off
3857 ;; its possible that the open brace is before
3858 ;; bufpos, but the close brace is after. In that
3859 ;; case, convert this to a non-cons element. The
3860 ;; rest of the state is before bufpos, so we're
3861 ;; done.
3863 ;; we know that both the open and close braces are
3864 ;; before bufpos, so we also know that everything else
3865 ;; on state is before bufpos.
3868 nil ; whack it off
3869 ;; it's before bufpos, so everything else should too.
3872 nil)))
3875 ;; Return the bufpos of the innermost enclosing open paren before
3876 ;; bufpos, or nil if none was found.
3886 enclosingp))
3889 ;; Return the bufpos of the outermost enclosing open paren, or nil
3890 ;; if none was found.
3897 pos))
3900 ;; Return the closest "safe" position recorded on PAREN-STATE that
3901 ;; is higher up than BUFPOS. Return nil if PAREN-STATE doesn't
3902 ;; contain any. Return nil if BUFPOS is nil, which is useful to
3903 ;; find the closest limit before a given limit that might be nil.
3904 ;;
3905 ;; A "safe" position is a position at or after a recorded open
3906 ;; paren, or after a recorded close paren. The returned position is
3907 ;; thus either the first position after a close brace, or the first
3908 ;; position after an enclosing paren, or at the enclosing paren in
3909 ;; case BUFPOS is immediately after it.
3919 ;; See below.
3922 ;; elem is the position at and not after the opening paren, so
3923 ;; we can go forward one more step unless it's equal to
3924 ;; bufpos. This is useful in some cases avoid an extra paren
3925 ;; level between the safe position and bufpos.
3930 ;; This is used for `font-lock-beginning-of-syntax-function'. It
3931 ;; goes to the closest previous point that is known to be outside
3932 ;; any string literal or comment. `c-state-cache' is used if it has
3933 ;; a position in the vicinity.
3935 elem
3938 ;; Note: Similar code in `c-safe-position'. The
3939 ;; difference is that we accept a safe position at
3940 ;; the point and don't bother to go forward past open
3941 ;; parens.
3956 ;; The position is far back. Try `c-beginning-of-defun-1'
3957 ;; (although we can't be entirely sure it will go to a position
3958 ;; outside a comment or string in current emacsen). FIXME:
3959 ;; Consult `syntax-ppss' here.
3964 \f
3965 ;; Tools for scanning identifiers and other tokens.
3968 "Return non-nil if the point is on or directly after an identifier.
3969 Keywords are recognized and not considered identifiers. If an
3970 identifier is detected, the returned value is its starting position.
3971 If an identifier ends at the point and another begins at it \(can only
3972 happen in Pike) then the point for the preceding one is returned.
3974 Note that this function might do hidden buffer changes. See the
3975 comment at the start of cc-engine.el for more info."
3977 ;; FIXME: Shouldn't this function handle "operator" in C++?
3984 ;; Check for a normal (non-keyword) identifier.
3990 ;; Handle the `<operator> syntax in Pike.
3994 t
4001 ;; Handle the "operator +" syntax in C++.
4018 )))
4021 ;; If the point is at the end of a symbol then skip backward to the
4022 ;; beginning of it. Don't move otherwise. Return non-nil if point
4023 ;; moved.
4024 ;;
4025 ;; This function might do hidden buffer changes.
4028 ;; Handle the `<operator> syntax in Pike.
4034 t
4036 nil)))))
4039 ;; Move to the beginning of the current token. Do not move if not
4040 ;; in the middle of one. BACK-LIMIT may be used to bound the
4041 ;; backward search; if given it's assumed to be at the boundary
4042 ;; between two tokens. Return non-nil if the point is moved, nil
4043 ;; otherwise.
4044 ;;
4045 ;; This function might do hidden buffer changes.
4052 ;; `c-nonsymbol-token-regexp' should always match
4053 ;; since we've skipped backward over punctuation
4054 ;; or paren syntax, but consume one char in case
4055 ;; it doesn't so that we don't leave point before
4056 ;; some earlier incorrect token.
4063 ;; Move to the end of the current token. Do not move if not in the
4064 ;; middle of one. BACK-LIMIT may be used to bound the backward
4065 ;; search; if given it's assumed to be at the boundary between two
4066 ;; tokens. Return non-nil if the point is moved, nil otherwise.
4067 ;;
4068 ;; This function might do hidden buffer changes.
4076 ;; `c-nonsymbol-token-regexp' should always match since
4077 ;; we've skipped backward over punctuation or paren
4078 ;; syntax, but move forward in case it doesn't so that
4079 ;; we don't leave point earlier than we started with.
4095 "Move forward by tokens.
4096 A token is defined as all symbols and identifiers which aren't
4098 treated properly). Point is always either left at the beginning of a
4099 token or not moved at all. COUNT specifies the number of tokens to
4100 move; a negative COUNT moves in the opposite direction. A COUNT of 0
4101 moves to the next token beginning only if not already at one. If
4102 BALANCED is true, move over balanced parens, otherwise move into them.
4103 Also, if BALANCED is true, never move out of an enclosing paren.
4105 LIMIT sets the limit for the movement and defaults to the point limit.
4106 The case when LIMIT is set in the middle of a token, comment or macro
4107 is handled correctly, i.e. the point won't be left there.
4109 Return the number of tokens left to move \(positive or negative). If
4110 BALANCED is true, a move over a balanced paren counts as one. Note
4111 that if COUNT is 0 and no appropriate token beginning is found, 1 will
4112 be returned. Thus, a return value of 0 guarantees that point is at
4113 the requested position and a return value less \(without signs) than
4114 COUNT guarantees that point is at the beginning of some token.
4116 Note that this function might do hidden buffer changes. See the
4117 comment at the start of cc-engine.el for more info."
4124 c-jump-syntax-balanced
4125 c-jump-syntax-unbalanced))
4130 ;; If count is zero we should jump if in the middle of a token.
4137 ;; Skip whitespace. Count this as a move if we did in
4138 ;; fact move.
4142 ;; Moved out of bounds. Make sure the returned count isn't zero.
4147 ;; Use `condition-case' to avoid having the limit tests
4148 ;; inside the loop.
4156 t)
4159 t)
4160 ;; `c-nonsymbol-token-regexp' above should always
4161 ;; match if there are correct tokens. Try to
4162 ;; widen to see if the limit was set in the
4163 ;; middle of one, else fall back to treating
4164 ;; the offending thing as a one character token.
4169 nil)
4172 t))))
4182 count)))
4185 "Move backward by tokens.
4186 See `c-forward-token-2' for details."
4194 c-jump-syntax-balanced
4195 c-jump-syntax-unbalanced))
4199 ;; The count is zero so try to skip to the beginning of the
4200 ;; current token.
4204 ;; The limit is inside the same token, so return 1.
4207 ;; We're not in the middle of a token. If there's
4208 ;; whitespace after the point then we must move backward,
4209 ;; so set count to 1 in that case.
4211 ;; If we're looking at a '#' that might start a cpp
4212 ;; directive then we have to do a more elaborate check.
4225 ;; Use `condition-case' to avoid having to check for buffer
4226 ;; limits in `backward-char', `scan-sexps' and `goto-char' below.
4235 ;; This can be very inefficient if there's a long
4236 ;; sequence of operator tokens without any separation.
4237 ;; That doesn't happen in practice, anyway.
4247 count)))
4250 "Like `c-forward-token-2' but doesn't treat multicharacter operator
4252 characters are jumped over character by character. This function is
4253 for compatibility only; it's only a wrapper over `c-forward-token-2'."
4258 "Like `c-backward-token-2' but doesn't treat multicharacter operator
4260 characters are jumped over character by character. This function is
4261 for compatibility only; it's only a wrapper over `c-backward-token-2'."
4265 \f
4266 ;; Tools for doing searches restricted to syntactically relevant text.
4269 paren-level not-inside-token
4270 lookbehind-submatch)
4271 "Like `re-search-forward', but only report matches that are found
4272 in syntactically significant text. I.e. matches in comments, macros
4273 or string literals are ignored. The start point is assumed to be
4274 outside any comment, macro or string literal, or else the content of
4275 that region is taken as syntactically significant text.
4277 If PAREN-LEVEL is non-nil, an additional restriction is added to
4278 ignore matches in nested paren sexps. The search will also not go
4279 outside the current list sexp, which has the effect that if the point
4280 should be moved to BOUND when no match is found \(i.e. NOERROR is
4281 neither nil nor t), then it will be at the closing paren if the end of
4282 the current list sexp is encountered first.
4284 If NOT-INSIDE-TOKEN is non-nil, matches in the middle of tokens are
4285 ignored. Things like multicharacter operators and special symbols
4287 constants.
4289 If LOOKBEHIND-SUBMATCH is non-nil, it's taken as a number of a
4290 subexpression in REGEXP. The end of that submatch is used as the
4291 position to check for syntactic significance. If LOOKBEHIND-SUBMATCH
4292 isn't used or if that subexpression didn't match then the start
4294 subexpression is never tested before the starting position, so it
4298 subexpression can match the end of nonwhite syntactic whitespace,
4299 i.e. the end of comments or cpp directives. This since the function
4300 skips over such things before resuming the search. It's on the other
4302 matches syntactic whitespace.
4304 Bug: Unbalanced parens inside cpp directives are currently not handled
4305 correctly \(i.e. they don't get ignored as they should) when
4306 PAREN-LEVEL is set.
4308 Note that this function might do hidden buffer changes. See the
4309 comment at the start of cc-engine.el for more info."
4314 ;;(message "c-syntactic-re-search-forward %s %s %S" (point) bound regexp)
4317 tmp
4318 ;; Start position for the last search.
4319 search-pos
4320 ;; The `parse-partial-sexp' state between the start position
4321 ;; and the point.
4322 state
4323 ;; The current position after the last state update. The next
4324 ;; `parse-partial-sexp' continues from here.
4326 ;; The position at which to check the state and the state
4327 ;; there. This is separate from `state-pos' since we might
4328 ;; need to back up before doing the next search round.
4329 check-pos check-state
4330 ;; Last position known to end a token.
4332 ;; Set when a valid match is found.
4333 found)
4341 t
4342 ;; Without the following, when PAREN-LEVEL it non-nil, and
4343 ;; NOERROR is not nil or t, and the very first search above
4344 ;; has just failed, point would end up at BOUND rather than
4345 ;; just before the next close paren.
4347 paren-level
4352 nil))
4363 state-pos check-pos paren-level nil state))
4365 check-state state))
4367 ;; NOTE: If we got a look behind subexpression and get
4368 ;; an insignificant match in something that isn't
4369 ;; syntactic whitespace (i.e. strings or in nested
4370 ;; parentheses), then we can never skip more than a
4371 ;; single character from the match start position
4372 ;; (i.e. `state-pos' here) before continuing the
4373 ;; search. That since the look behind subexpression
4374 ;; might match the end of the insignificant region in
4375 ;; the next search.
4379 ;; Match inside a line comment. Skip to eol. Use
4380 ;; `re-search-forward' instead of `skip-chars-forward' to get
4381 ;; the right bound behavior.
4385 ;; Match inside a block comment. Skip to the '*/'.
4392 ;; Match in the middle of the opener of a block or line
4393 ;; comment.
4398 ;; The last `parse-partial-sexp' above might have
4399 ;; stopped short of the real check position if the end
4400 ;; of the current sexp was encountered in paren-level
4401 ;; mode. The checks above are always false in that
4402 ;; case, and since they can do better skipping in
4403 ;; lookbehind-submatch mode, we do them before
4404 ;; checking the paren level.
4408 ;; Check the paren level first since we're short of the
4409 ;; syntactic checking position if the end of the
4410 ;; current sexp was encountered by `parse-partial-sexp'.
4413 ;; Inside a nested paren sexp.
4415 ;; See the NOTE above.
4417 ;; Skip out of the paren quickly.
4421 ;; Have exited the current paren sexp.
4424 ;; The last `parse-partial-sexp' call above
4425 ;; has left us just after the closing paren
4426 ;; in this case, so we can modify the bound
4427 ;; to leave the point at the right position
4428 ;; upon return.
4430 nil)
4434 ;; Match inside a string.
4437 ;; See the NOTE above.
4439 ;; Skip to the end of the string before continuing.
4448 continue)))
4453 ;; Match inside a macro. Skip to the end of it.
4461 (< check-pos
4467 ;; Inside a token.
4469 ;; See the NOTE above.
4474 ;; A real match.
4476 nil)))
4478 ;; Should loop to search again, but take care to avoid
4479 ;; looping on the same spot.
4483 nil
4486 t))))
4492 ;;(message "c-syntactic-re-search-forward done %s" (or (match-end 0) (point)))
4499 ;; Search failed. Set point as appropriate.
4503 nil)))
4508 ;; Return the start of the literal point is in, or nil.
4509 ;; We read and write the variables `safe-pos', `safe-pos-list', `state'
4510 ;; bound in the caller.
4512 ;; Use `parse-partial-sexp' from a safe position down to the point to check
4513 ;; if it's outside comments and strings.
4516 ;; Pick a safe position as close to the point as possible.
4517 ;;
4518 ;; FIXME: Consult `syntax-ppss' here if our cache doesn't give a good
4519 ;; position.
4531 ;; Cache positions along the way to use if we have to back up more. We
4532 ;; cache every closing paren on the same level. If the paren cache is
4533 ;; relevant in this region then we're typically already on the same
4534 ;; level as the target position. Note that we might cache positions
4535 ;; after opening parens in case safe-pos is in a nested list. That's
4536 ;; both uncommon and harmless.
4544 ;; If the state contains the start of the containing sexp we cache that
4545 ;; position too, so that parse-partial-sexp in the next run has a bigger
4546 ;; chance of starting at the same level as the target position and thus
4547 ;; will get more good safe positions into the list.
4553 ;; Inside string or comment. Continue search at the
4554 ;; beginning of it.
4558 "Like `skip-chars-backward' but only look at syntactically relevant chars,
4559 i.e. don't stop at positions inside syntactic whitespace or string
4560 literals. Preprocessor directives are also ignored, with the exception
4561 of the one that the point starts within, if any. If LIMIT is given,
4562 it's assumed to be at a syntactically relevant position.
4564 If PAREN-LEVEL is non-nil, the function won't stop in nested paren
4565 sexps, and the search will also not go outside the current paren sexp.
4566 However, if LIMIT or the buffer limit is reached inside a nested paren
4567 then the point will be left at the limit.
4569 Non-nil is returned if the point moved, nil otherwise.
4571 Note that this function might do hidden buffer changes. See the
4572 comment at the start of cc-engine.el for more info."
4576 state-2
4577 ;; A list of syntactically relevant positions in descending
4578 ;; order. It's used to avoid scanning repeatedly over
4579 ;; potentially large regions with `parse-partial-sexp' to verify
4580 ;; each position. Used in `c-ssb-lit-begin'
4581 safe-pos-list
4582 ;; The result from `c-beginning-of-macro' at the start position or the
4583 ;; start position itself if it isn't within a macro. Evaluated on
4584 ;; demand.
4585 start-macro-beg
4586 ;; The earliest position after the current one with the same paren
4587 ;; level. Used only when `paren-level' is set.
4588 lit-beg
4593 ;; The next loop "tries" to find the end point each time round,
4594 ;; loops when it hasn't succeeded.
4600 ;; Don't stop inside a literal.
4603 t)))
4615 ;; Not at the right level.
4618 ;; We stop above if we go out of a paren.
4619 ;; Now check whether it precedes or is
4620 ;; nested in the starting sexp.
4624 pps-end-pos paren-level-pos
4625 nil nil state-2))
4628 ;; We've stopped short of the starting position
4629 ;; so the hit was inside a nested list. Go up
4630 ;; until we are at the right level.
4639 nil)
4640 t))
4643 nil))
4645 ;; The hit was outside the list at the start
4646 ;; position. Go to the start of the list and exit.
4648 nil))
4651 ;; Inside a macro.
4659 t
4661 ;; It's inside the same macro we started in so it's
4662 ;; a relevant match.
4664 nil))))))
4668 ;; Skip syntactic ws afterwards so that we don't stop at the
4669 ;; end of a comment if `skip-chars' is something like "^/".
4673 ;; We might want to extend this with more useful return values in
4674 ;; the future.
4677 ;; The following is an alternative implementation of
4678 ;; `c-syntactic-skip-backward' that uses backward movement to keep
4679 ;; track of the syntactic context. It turned out to be generally
4680 ;; slower than the one above which uses forward checks from earlier
4681 ;; safe positions.
4682 ;;
4683 ;;(defconst c-ssb-stop-re
4684 ;; ;; The regexp matching chars `c-syntactic-skip-backward' needs to
4685 ;; ;; stop at to avoid going into comments and literals.
4686 ;; (concat
4687 ;; ;; Match comment end syntax and string literal syntax. Also match
4688 ;; ;; '/' for block comment endings (not covered by comment end
4689 ;; ;; syntax).
4690 ;; "\\s>\\|/\\|\\s\""
4691 ;; (if (memq 'gen-string-delim c-emacs-features)
4692 ;; "\\|\\s|"
4693 ;; "")
4694 ;; (if (memq 'gen-comment-delim c-emacs-features)
4695 ;; "\\|\\s!"
4696 ;; "")))
4697 ;;
4698 ;;(defconst c-ssb-stop-paren-re
4699 ;; ;; Like `c-ssb-stop-re' but also stops at paren chars.
4700 ;; (concat c-ssb-stop-re "\\|\\s(\\|\\s)"))
4701 ;;
4702 ;;(defconst c-ssb-sexp-end-re
4703 ;; ;; Regexp matching the ending syntax of a complex sexp.
4704 ;; (concat c-string-limit-regexp "\\|\\s)"))
4705 ;;
4706 ;;(defun c-syntactic-skip-backward (skip-chars &optional limit paren-level)
4707 ;; "Like `skip-chars-backward' but only look at syntactically relevant chars,
4708 ;;i.e. don't stop at positions inside syntactic whitespace or string
4709 ;;literals. Preprocessor directives are also ignored. However, if the
4710 ;;point is within a comment, string literal or preprocessor directory to
4711 ;;begin with, its contents is treated as syntactically relevant chars.
4712 ;;If LIMIT is given, it limits the backward search and the point will be
4713 ;;left there if no earlier position is found.
4714 ;;
4715 ;;If PAREN-LEVEL is non-nil, the function won't stop in nested paren
4716 ;;sexps, and the search will also not go outside the current paren sexp.
4717 ;;However, if LIMIT or the buffer limit is reached inside a nested paren
4718 ;;then the point will be left at the limit.
4719 ;;
4720 ;;Non-nil is returned if the point moved, nil otherwise.
4721 ;;
4722 ;;Note that this function might do hidden buffer changes. See the
4723 ;;comment at the start of cc-engine.el for more info."
4724 ;;
4725 ;; (save-restriction
4726 ;; (when limit
4727 ;; (narrow-to-region limit (point-max)))
4728 ;;
4729 ;; (let ((start (point)))
4730 ;; (catch 'done
4731 ;; (while (let ((last-pos (point))
4732 ;; (stop-pos (progn
4733 ;; (skip-chars-backward skip-chars)
4734 ;; (point))))
4735 ;;
4736 ;; ;; Skip back over the same region as
4737 ;; ;; `skip-chars-backward' above, but keep to
4738 ;; ;; syntactically relevant positions.
4739 ;; (goto-char last-pos)
4740 ;; (while (and
4741 ;; ;; `re-search-backward' with a single char regexp
4742 ;; ;; should be fast.
4743 ;; (re-search-backward
4744 ;; (if paren-level c-ssb-stop-paren-re c-ssb-stop-re)
4745 ;; stop-pos 'move)
4746 ;;
4747 ;; (progn
4748 ;; (cond
4749 ;; ((looking-at "\\s(")
4750 ;; ;; `paren-level' is set and we've found the
4751 ;; ;; start of the containing paren.
4752 ;; (forward-char)
4753 ;; (throw 'done t))
4754 ;;
4755 ;; ((looking-at c-ssb-sexp-end-re)
4756 ;; ;; We're at the end of a string literal or paren
4757 ;; ;; sexp (if `paren-level' is set).
4758 ;; (forward-char)
4759 ;; (condition-case nil
4760 ;; (c-backward-sexp)
4761 ;; (error
4762 ;; (goto-char limit)
4763 ;; (throw 'done t))))
4764 ;;
4765 ;; (t
4766 ;; (forward-char)
4767 ;; ;; At the end of some syntactic ws or possibly
4768 ;; ;; after a plain '/' operator.
4769 ;; (let ((pos (point)))
4770 ;; (c-backward-syntactic-ws)
4771 ;; (if (= pos (point))
4772 ;; ;; Was a plain '/' operator. Go past it.
4773 ;; (backward-char)))))
4774 ;;
4775 ;; (> (point) stop-pos))))
4776 ;;
4777 ;; ;; Now the point is either at `stop-pos' or at some
4778 ;; ;; position further back if `stop-pos' was at a
4779 ;; ;; syntactically irrelevant place.
4780 ;;
4781 ;; ;; Skip additional syntactic ws so that we don't stop
4782 ;; ;; at the end of a comment if `skip-chars' is
4783 ;; ;; something like "^/".
4784 ;; (c-backward-syntactic-ws)
4785 ;;
4786 ;; (< (point) stop-pos))))
4787 ;;
4788 ;; ;; We might want to extend this with more useful return values
4789 ;; ;; in the future.
4790 ;; (/= (point) start))))
4792 \f
4793 ;; Tools for handling comments and string literals.
4796 "Return the type of literal point is in, if any.
4797 The return value is `c' if in a C-style comment, `c++' if in a C++
4798 style comment, `string' if in a string literal, `pound' if DETECT-CPP
4799 is non-nil and in a preprocessor line, or nil if somewhere else.
4800 Optional LIM is used as the backward limit of the search. If omitted,
4801 or nil, `c-beginning-of-defun' is used.
4803 The last point calculated is cached if the cache is enabled, i.e. if
4804 `c-in-literal-cache' is bound to a two element vector.
4806 Note that this function might do hidden buffer changes. See the
4807 comment at the start of cc-engine.el for more info."
4817 "Return a cons of the beginning and end positions of the comment or
4818 string surrounding point (including both delimiters), or nil if point
4820 to start parsing from. If NEAR is non-nil, then the limits of any
4822 spaces and tabs between point and the literal. The search for such a
4823 literal is done first in forward direction. If NOT-IN-DELIMITER is
4824 non-nil, the case when point is inside a starting delimiter won't be
4825 recognized. This only has effect for comments which have starting
4826 delimiters with more than one character.
4828 Note that this function might do hidden buffer changes. See the
4829 comment at the start of cc-engine.el for more info."
4840 nil nil
4841 s
4851 ;; Search forward for a literal.
4862 ;; Search backward.
4873 ;; Block comment. Due to the nature of line
4874 ;; comments, they will always be covered by the
4875 ;; normal case above.
4878 ;; If LIM is bogus, beg will be bogus.
4882 ))))
4885 "Return the start of the string or comment surrounding point, or nil if
4886 point isn't in one. SAFE-POS, if non-nil, is a position before point which is
4894 ;; In case external callers use this; it did have a docstring.
4898 "If the argument is a cons of two buffer positions (such as returned by
4899 `c-literal-limits'), and that range contains a C++ style line comment,
4900 then an extended range is returned that contains all adjacent line
4901 comments (i.e. all comments that starts in the same column with no
4902 empty lines or non-whitespace characters between them). Otherwise the
4903 argument is returned.
4905 Note that this function might do hidden buffer changes. See the
4906 comment at the start of cc-engine.el for more info."
4917 ;; Got to take care in the backward direction to handle
4918 ;; comments which are preceded by code.
4932 range)
4936 "Convenience function that given the result of `c-literal-limits',
4937 returns nil or the type of literal that the range surrounds, one
4938 of the symbols `c', `c++' or `string'. It's much faster than using
4939 `c-in-literal' and is intended to be used when you need both the
4940 type of a literal and its limits.
4942 Note that this function might do hidden buffer changes. See the
4943 comment at the start of cc-engine.el for more info."
4954 range))
4957 ;; Get a "safe place" approximately TRY-SIZE characters before START.
4958 ;; This defsubst doesn't preserve point.
4964 ;; Return a buffer position HOW-FAR-BACK non-literal characters from START
4965 ;; (default point). This is done by going back further in the buffer then
4966 ;; searching forward for literals. The position found won't be in a
4967 ;; literal. We start searching for the sought position TRY-SIZE (default
4968 ;; twice HOW-FAR-BACK) bytes back from START. This function must be fast.
4969 ;; :-)
4977 stack elt size
4980 ;; Move forward one literal each time round this loop.
4981 ;; Move forward to the start of a comment or string.
4983 pos
4984 start
4985 nil ; target-depth
4986 nil ; stop-before
4987 s ; state
4990 ;; Gather details of the non-literal-bit - starting pos and size.
4994 pos))
4998 ;; Move forward to the end of the comment/string.
5002 start
5003 nil ; target-depth
5004 nil ; stop-before
5005 s ; state
5009 ;; Now try and find enough non-literal characters recorded on the stack.
5010 ;; Go back one recorded literal each time round this loop.
5012 stack)
5017 ;; Have we found enough yet?
5027 ;; Return a buffer position about HOW-FAR non-literal characters forward
5028 ;; from START-POS (default point), which must not be inside a literal.
5035 ;; Scan over counted characters.
5037 pos
5039 nil ; target-depth
5040 nil ; stop-before
5041 s ; state
5045 ;; Scan over literal characters.
5048 pos
5050 nil ; target-depth
5051 nil ; stop-before
5052 s ; state
5057 \f
5058 ;; `c-find-decl-spots' and accompanying stuff.
5060 ;; Variables used in `c-find-decl-spots' to cache the search done for
5061 ;; the first declaration in the last call. When that function starts,
5062 ;; it needs to back up over syntactic whitespace to look at the last
5063 ;; token before the region being searched. That can sometimes cause
5064 ;; moves back and forth over a quite large region of comments and
5065 ;; macros, which would be repeated for each changed character when
5066 ;; we're called during fontification, since font-lock refontifies the
5067 ;; current line for each change. Thus it's worthwhile to cache the
5068 ;; first match.
5069 ;;
5070 ;; `c-find-decl-syntactic-pos' is a syntactically relevant position in
5071 ;; the syntactic whitespace less or equal to some start position.
5072 ;; There's no cached value if it's nil.
5073 ;;
5074 ;; `c-find-decl-match-pos' is the match position if
5075 ;; `c-find-decl-prefix-search' matched before the syntactic whitespace
5076 ;; at `c-find-decl-syntactic-pos', or nil if there's no such match.
5087 ; (defface c-debug-decl-spot-face
5088 ; '((t (:background "Turquoise")))
5089 ; "Debug face to mark the spots where `c-find-decl-spots' stopped.")
5090 ; (defface c-debug-decl-sws-face
5091 ; '((t (:background "Khaki")))
5092 ; "Debug face to mark the syntactic whitespace between the declaration
5093 ; spots and the preceding token end.")
5109 ;; Macro used inside `c-find-decl-spots'. It ought to be a defun,
5110 ;; but it contains lots of free variables that refer to things
5111 ;; inside `c-find-decl-spots'. The point is left at `cfd-match-pos'
5112 ;; if there is a match, otherwise at `cfd-limit'.
5113 ;;
5114 ;; The macro moves point forward to the next putative start of a declaration
5115 ;; or cfd-limit. This decl start is the next token after a "declaration
5116 ;; prefix". The declaration prefix is the earlier of `cfd-prop-match' and
5117 ;; `cfd-re-match'. `cfd-match-pos' is set to the decl prefix.
5118 ;;
5119 ;; This macro might do hidden buffer changes.
5122 ;; Find the next property match position if we haven't got one already.
5133 ;; Find the next `c-decl-prefix-or-start-re' match if we haven't
5134 ;; got one already.
5140 ;; Each time round, the next `while' moves forward over a pseudo match
5141 ;; of `c-decl-prefix-or-start-re' which is either inside a literal, or
5142 ;; is a ":" not preceded by "public", etc.. `cfd-re-match' and
5143 ;; `cfd-re-match-end' get set.
5150 ;; No match. Finish up and exit the loop.
5152 nil)
5155 ;; Matched the end of a token preceding a decl spot.
5159 ;; Matched a token that start a decl spot.
5162 c-literal-faces)
5163 ;; Pseudo match inside a comment or string literal. Skip out
5164 ;; of comments and string literals.
5176 ;; Found, e.g., "::" in C++
5177 t)
5183 ;; Found a ":" which isn't part of "public:", etc.
5184 t)
5187 ;; If our match was at the decl start, we have to back up over the
5188 ;; preceding syntactic ws to set `cfd-match-pos' and to catch
5189 ;; any decl spots in the syntactic ws.
5194 ;; Choose whichever match is closer to the start.
5197 cfd-re-match nil)
5199 cfd-prop-match nil))
5204 ;; Skip forward past comments only so we don't skip macros.
5206 ;; Set the position to continue at. We can avoid going over
5207 ;; the comments skipped above a second time, but it's possible
5208 ;; that the comment skipping has taken us past `cfd-prop-match'
5209 ;; since the property might be used inside comments.
5215 ;; Call CFD-FUN for each possible spot for a declaration, cast or
5216 ;; label from the point to CFD-LIMIT.
5217 ;;
5218 ;; CFD-FUN is called with point at the start of the spot. It's passed two
5219 ;; arguments: The first is the end position of the token preceding the spot,
5220 ;; or 0 for the implicit match at bob. The second is a flag that is t when
5221 ;; the match is inside a macro. Point should be moved forward by at least
5222 ;; one token.
5223 ;;
5224 ;; If CFD-FUN adds `c-decl-end' properties somewhere below the current spot,
5225 ;; it should return non-nil to ensure that the next search will find them.
5226 ;;
5227 ;; Such a spot is:
5228 ;; o The first token after bob.
5229 ;; o The first token after the end of submatch 1 in
5230 ;; `c-decl-prefix-or-start-re' when that submatch matches. This
5231 ;; submatch is typically a (L or R) brace or paren, a ;, or a ,.
5232 ;; o The start of each `c-decl-prefix-or-start-re' match when
5233 ;; submatch 1 doesn't match. This is, for example, the keyword
5234 ;; "class" in Pike.
5235 ;; o The start of a previously recognized declaration; "recognized"
5236 ;; means that the last char of the previous token has a `c-type'
5237 ;; text property with the value `c-decl-end'; this only holds
5238 ;; when `c-type-decl-end-used' is set.
5239 ;;
5240 ;; Only a spot that match CFD-DECL-RE and whose face is in the
5241 ;; CFD-FACE-CHECKLIST list causes CFD-FUN to be called. The face
5242 ;; check is disabled if CFD-FACE-CHECKLIST is nil.
5243 ;;
5244 ;; If the match is inside a macro then the buffer is narrowed to the
5245 ;; end of it, so that CFD-FUN can investigate the following tokens
5246 ;; without matching something that begins inside a macro and ends
5247 ;; outside it. It's to avoid this work that the CFD-DECL-RE and
5248 ;; CFD-FACE-CHECKLIST checks exist.
5249 ;;
5250 ;; The spots are visited approximately in order from top to bottom.
5251 ;; It's however the positions where `c-decl-prefix-or-start-re'
5252 ;; matches and where `c-decl-end' properties are found that are in
5253 ;; order. Since the spots often are at the following token, they
5254 ;; might be visited out of order insofar as more spots are reported
5255 ;; later on within the syntactic whitespace between the match
5256 ;; positions and their spots.
5257 ;;
5258 ;; It's assumed that comments and strings are fontified in the
5259 ;; searched range.
5260 ;;
5261 ;; This is mainly used in fontification, and so has an elaborate
5262 ;; cache to handle repeated calls from the same start position; see
5263 ;; the variables above.
5264 ;;
5265 ;; All variables in this function begin with `cfd-' to avoid name
5266 ;; collision with the (dynamically bound) variables used in CFD-FUN.
5267 ;;
5268 ;; This function might do hidden buffer changes.
5272 ;; The end of the token preceding the decl spot last found
5273 ;; with `c-decl-prefix-or-start-re'. `cfd-limit' if there's
5274 ;; no match.
5275 cfd-re-match
5276 ;; The end position of the last `c-decl-prefix-or-start-re'
5277 ;; match. If this is greater than `cfd-continue-pos', the
5278 ;; next regexp search is started here instead.
5280 ;; The end of the last `c-decl-end' found by
5281 ;; `c-find-decl-prefix-search'. `cfd-limit' if there's no
5282 ;; match. If searching for the property isn't needed then we
5283 ;; disable it by setting it to `cfd-limit' directly.
5285 ;; The end of the token preceding the decl spot last found by
5286 ;; `c-find-decl-prefix-search'. 0 for the implicit match at
5287 ;; bob. `cfd-limit' if there's no match. In other words,
5288 ;; this is the minimum of `cfd-re-match' and `cfd-prop-match'.
5290 ;; The position to continue searching at.
5291 cfd-continue-pos
5292 ;; The position of the last "real" token we've stopped at.
5293 ;; This can be greater than `cfd-continue-pos' when we get
5294 ;; hits inside macros or at `c-decl-end' positions inside
5295 ;; comments.
5297 ;; The end position of the last entered macro.
5300 ;; Initialize by finding a syntactically relevant start position
5301 ;; before the point, and do the first `c-decl-prefix-or-start-re'
5302 ;; search unless we're at bob.
5305 ;; Must back up a bit since we look for the end of the previous
5306 ;; statement or declaration, which is earlier than the first
5307 ;; returned match.
5309 ;; This `cond' moves back over any literals or macros. It has special
5310 ;; handling for when the region being searched is entirely within a
5311 ;; macro. It sets `cfd-continue-pos' (unless we've reached
5312 ;; `cfd-limit').
5314 ;; First we need to move to a syntactically relevant position.
5315 ;; Begin by backing out of comment or string literals.
5316 ;;
5317 ;; This arm of the cond actually triggers if we're in a literal,
5318 ;; and cfd-limit is at most at BONL.
5320 ;; This arm of the `and' moves backwards out of a literal when
5321 ;; the face at point is a literal face. In this case, its value
5322 ;; is always non-nil.
5324 ;; Try to use the faces to back up to the start of the
5325 ;; literal. FIXME: What if the point is on a declaration
5326 ;; inside a comment?
5332 ;; XEmacs doesn't fontify the quotes surrounding string
5333 ;; literals.
5342 ;; Don't trust the literal to contain only literal faces
5343 ;; (the font lock package might not have fontified the
5344 ;; start of it at all, for instance) so check that we have
5345 ;; arrived at something that looks like a start or else
5346 ;; resort to `c-literal-limits'.
5350 )
5354 ;; The start is in a literal. If the limit is in the same
5355 ;; one we don't have to find a syntactic position etc. We
5356 ;; only check that if the limit is at or before bonl to save
5357 ;; time; it covers the by far most common case when font-lock
5358 ;; refontifies the current line only.
5369 ;; Completely inside a literal. Set up variables to trig the
5370 ;; (< cfd-continue-pos cfd-start-pos) case below and it'll
5371 ;; find a suitable start position.
5374 ;; Check if the region might be completely inside a macro, to
5375 ;; optimize that like the completely-inside-literal above.
5382 ;; (Maybe) completely inside a macro. Only need to trig the
5383 ;; (< cfd-continue-pos cfd-start-pos) case below to make it
5384 ;; set things up.
5386 start-in-macro t))
5388 ;; The default arm of the `cond' moves back over any macro we're in
5389 ;; and over any syntactic WS. It sets `c-find-decl-syntactic-pos'.
5391 ;; Back out of any macro so we don't miss any declaration
5392 ;; that could follow after it.
5396 ;; Now we're at a proper syntactically relevant position so we
5397 ;; can use the cache. But first clear it if it applied
5398 ;; further down.
5403 ;; Don't have to do this if the cache is relevant here,
5404 ;; typically if the same line is refontified again. If
5405 ;; we're just some syntactic whitespace further down we can
5406 ;; still use the cache to limit the skipping.
5409 ;; If we hit `c-find-decl-syntactic-pos' and
5410 ;; `c-find-decl-match-pos' is set then we install the cached
5411 ;; values. If we hit `c-find-decl-syntactic-pos' and
5412 ;; `c-find-decl-match-pos' is nil then we know there's no decl
5413 ;; prefix in the whitespace before `c-find-decl-syntactic-pos'
5414 ;; and so we can continue the search from this point. If we
5415 ;; didn't hit `c-find-decl-syntactic-pos' then we're now in
5416 ;; the right spot to begin searching anyway.
5418 c-find-decl-match-pos)
5420 cfd-continue-pos syntactic-pos)
5425 ;; Always consider bob a match to get the first
5426 ;; declaration in the file. Do this separately instead of
5427 ;; letting `c-decl-prefix-or-start-re' match bob, so that
5428 ;; regexp always can consume at least one character to
5429 ;; ensure that we won't get stuck in an infinite loop.
5434 ;; Do an initial search now. In the bob case above it's
5435 ;; only done to search for a `c-decl-end' spot.
5441 ;; Advance `cfd-continue-pos' if it's before the start position.
5442 ;; The closest continue position that might have effect at or
5443 ;; after the start depends on what we started in. This also
5444 ;; finds a suitable start position in the special cases when the
5445 ;; region is completely within a literal or macro.
5450 ;; If we're in a macro then it's the closest preceding token
5451 ;; in the macro. Check this before `start-in-literal',
5452 ;; since if we're inside a literal in a macro, the preceding
5453 ;; token is earlier than any `c-decl-end' spot inside the
5454 ;; literal (comment).
5456 ;; The only syntactic ws in macros are comments.
5462 ;; If we're in a comment it can only be the closest
5463 ;; preceding `c-decl-end' position within that comment, if
5464 ;; any. Go back to the beginning of such a property so that
5465 ;; `c-find-decl-prefix-search' will find the end of it.
5466 ;; (Can't stop at the end and install it directly on
5467 ;; `cfd-prop-match' since that variable might be cleared
5468 ;; after `cfd-fun' below.)
5469 ;;
5470 ;; Note that if the literal is a string then the property
5471 ;; search will simply skip to the beginning of it right
5472 ;; away.
5484 ;; Didn't find any property inside the comment, so we can
5485 ;; skip it entirely. (This won't skip past a string, but
5486 ;; that'll be handled quickly by the next
5487 ;; `c-find-decl-prefix-search' anyway.)
5493 ;; If we started in normal code, the only match that might
5494 ;; apply before the start is what we already got in
5495 ;; `cfd-match-pos' so we can continue at the start position.
5496 ;; (Note that we don't get here if the first match is below
5497 ;; it.)
5500 ;; Delete found matches if they are before our new continue
5501 ;; position, so that `c-find-decl-prefix-search' won't back up
5502 ;; to them later on.
5510 ;; This is the normal case and we got a proper syntactic
5511 ;; position. If there's a match then it's always outside
5512 ;; macros and comments, so advance to the next token and set
5513 ;; `cfd-token-pos'. The loop below will later go back using
5514 ;; `cfd-continue-pos' to fix declarations inside the
5515 ;; syntactic ws.
5523 ;; Have one of the special cases when the region is completely
5524 ;; within a literal or macro. `cfd-continue-pos' is set to a
5525 ;; good start position for the search, so do it.
5528 ;; Now loop, one decl spot per iteration. We already have the first
5529 ;; match in `cfd-match-pos'.
5531 ;; Go forward over "false matches", one per iteration.
5536 ;; Kludge to filter out matches on the "<" that
5537 ;; aren't open parens, for the sake of languages
5538 ;; that got `c-recognize-<>-arglists' set.
5543 ;; If `cfd-continue-pos' is less or equal to
5544 ;; `cfd-token-pos', we've got a hit inside a macro
5545 ;; that's in the syntactic whitespace before the last
5546 ;; "real" declaration we've checked. If they're equal
5547 ;; we've arrived at the declaration a second time, so
5548 ;; there's nothing to do.
5552 ;; If `cfd-continue-pos' is less than `cfd-token-pos'
5553 ;; we're still searching for declarations embedded in
5554 ;; the syntactic whitespace. In that case we need
5555 ;; only to skip comments and not macros, since they
5556 ;; can't be nested, and that's already been done in
5557 ;; `c-find-decl-prefix-search'.
5562 ;; Continue if the following token fails the
5563 ;; CFD-DECL-RE and CFD-FACE-CHECKLIST checks.
5570 t)))
5581 ;; Narrow to the end of the macro if we got a hit inside
5582 ;; one, to avoid recognizing things that start inside the
5583 ;; macro and end outside it.
5585 ;; Not in the same macro as in the previous round.
5596 t
5599 t)
5600 ;; The matched token was the last thing in the macro,
5601 ;; so the whole match is bogus.
5610 ;; Restore limits if we did macro narrowing above.
5617 ; cfd-match-pos, etc.
5619 \f
5620 ;; A cache for found types.
5622 ;; Buffer local variable that contains an obarray with the types we've
5623 ;; found. If a declaration is recognized somewhere we record the
5624 ;; fully qualified identifier in it to recognize it as a type
5625 ;; elsewhere in the file too. This is not accurate since we do not
5626 ;; bother with the scoping rules of the languages, but in practice the
5627 ;; same name is seldom used as both a type and something else in a
5628 ;; file, and we only use this as a last resort in ambiguous cases (see
5629 ;; `c-forward-decl-or-cast-1').
5630 ;;
5631 ;; Not every type need be in this cache. However, things which have
5632 ;; ceased to be types must be removed from it.
5633 ;;
5634 ;; Template types in C++ are added here too but with the template
5635 ;; arglist replaced with "<>" in references or "<" for the one in the
5636 ;; primary type. E.g. the type "Foo<A,B>::Bar<C>" is stored as
5637 ;; "Foo<>::Bar<". This avoids storing very long strings (since C++
5638 ;; template specs can be fairly sized programs in themselves) and
5639 ;; improves the hit ratio (it's a type regardless of the template
5640 ;; args; it's just not the same type, but we're only interested in
5641 ;; recognizing types, not telling distinct types apart). Note that
5642 ;; template types in references are added here too; from the example
5643 ;; above there will also be an entry "Foo<".
5648 ;; Clears `c-found-types'.
5652 ;; Add the given region as a type in `c-found-types'. If the region
5653 ;; doesn't match an existing type but there is a type which is equal
5654 ;; to the given one except that the last character is missing, then
5655 ;; the shorter type is removed. That's done to avoid adding all
5656 ;; prefixes of a type as it's being entered and font locked. This
5657 ;; doesn't cover cases like when characters are removed from a type
5658 ;; or added in the middle. We'd need the position of point when the
5659 ;; font locking is invoked to solve this well.
5660 ;;
5661 ;; This function might do hidden buffer changes.
5668 ;; Remove the "NAME" from c-found-types, if present.
5672 ;; Return non-nil if the given region contains a type in
5673 ;; `c-found-types'.
5674 ;;
5675 ;; This function might do hidden buffer changes.
5677 c-found-types))
5680 ;; Return all the types in `c-found-types' as a sorted list of
5681 ;; strings.
5685 type-list)))
5686 c-found-types)
5689 ;; Shut up the byte compiler.
5693 ;; An after change function which, in conjunction with the info in
5694 ;; c-maybe-stale-found-type (set in c-before-change), removes a type
5695 ;; from `c-found-types', should this type have become stale. For
5696 ;; example, this happens to "foo" when "foo \n bar();" becomes
5697 ;; "foo(); \n bar();". Such stale types, if not removed, foul up
5698 ;; the fontification.
5699 ;;
5700 ;; Have we, perhaps, added non-ws characters to the front/back of a found
5701 ;; type?
5721 ;; Changing the amount of (already existing) whitespace - don't do anything.
5726 ;; The syntactic relationship which defined a "found type" has been
5727 ;; destroyed.
5730 ;; ((eq (car c-maybe-stale-found-type) 'c-decl-type-start) FIXME!!!
5731 )))
5733 \f
5734 ;; Setting and removing syntax properties on < and > in languages (C++
5735 ;; and Java) where they can be template/generic delimiters as well as
5736 ;; their normal meaning of "less/greater than".
5738 ;; Normally, < and > have syntax 'punctuation'. When they are found to
5739 ;; be delimiters, they are marked as such with the category properties
5740 ;; c-<-as-paren-syntax, c->-as-paren-syntax respectively.
5742 ;; STRATEGY:
5743 ;;
5744 ;; It is impossible to determine with certainty whether a <..> pair in
5745 ;; C++ is two comparison operators or is template delimiters, unless
5746 ;; one duplicates a lot of a C++ compiler. For example, the following
5747 ;; code fragment:
5748 ;;
5749 ;; foo (a < b, c > d) ;
5750 ;;
5751 ;; could be a function call with two integer parameters (each a
5752 ;; relational expression), or it could be a constructor for class foo
5753 ;; taking one parameter d of templated type "a < b, c >". They are
5754 ;; somewhat easier to distinguish in Java.
5755 ;;
5756 ;; The strategy now (2010-01) adopted is to mark and unmark < and
5757 ;; > IN MATCHING PAIRS ONLY. [Previously, they were marked
5758 ;; individually when their context so indicated. This gave rise to
5759 ;; intractable problems when one of a matching pair was deleted, or
5760 ;; pulled into a literal.]
5761 ;;
5762 ;; At each buffer change, the syntax-table properties are removed in a
5763 ;; before-change function and reapplied, when needed, in an
5764 ;; after-change function. It is far more important that the
5765 ;; properties get removed when they they are spurious than that they
5766 ;; be present when wanted.
5767 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
5769 ;; POS (default point) is at a < character. If it is marked with
5770 ;; open paren syntax-table text property, remove the property,
5771 ;; together with the close paren property on the matching > (if
5772 ;; any).
5787 ;; POS (default point) is at a > character. If it is marked with
5788 ;; close paren syntax-table property, remove the property, together
5789 ;; with the open paren property on the matching < (if any).
5804 ;; POS (default point) is at a < or > character. If it has an
5805 ;; open/close paren syntax-table property, remove this property both
5806 ;; from the current character and its partner (which will also be
5807 ;; thusly marked).
5817 ;; POS (default point) is at a < character. If it is both marked
5818 ;; with open/close paren syntax-table property, and has a matching >
5819 ;; (also marked) which is after LIM, remove the property both from
5820 ;; the current > and its partner. Return t when this happens, nil
5821 ;; when it doesn't.
5835 t)))
5838 ;; POS (default point) is at a > character. If it is both marked
5839 ;; with open/close paren syntax-table property, and has a matching <
5840 ;; (also marked) which is before LIM, remove the property both from
5841 ;; the current < and its partner. Return t when this happens, nil
5842 ;; when it doesn't.
5856 t)))
5858 ;; Set by c-common-init in cc-mode.el.
5861 ;; Set by c-after-change in cc-mode.el.
5866 ;; Unmark certain pairs of "< .... >" which are currently marked as
5867 ;; template/generic delimiters. (This marking is via syntax-table text
5868 ;; properties), and expand the (c-new-BEG c-new-END) region to include all
5869 ;; unmarked < and > operators within the certain bounds (see below).
5870 ;;
5871 ;; These pairs are those which are in the current "statement" (i.e.,
5872 ;; the region between the {, }, or ; before BEG and the one after
5873 ;; END), and which enclose any part of the interval (BEG END).
5874 ;;
5875 ;; Note that in C++ (?and Java), template/generic parens cannot
5876 ;; enclose a brace or semicolon, so we use these as bounds on the
5877 ;; region we must work on.
5878 ;;
5879 ;; This function is called from before-change-functions (via
5880 ;; c-get-state-before-change-functions). Thus the buffer is widened,
5881 ;; and point is undefined, both at entry and exit.
5882 ;;
5883 ;; FIXME!!! This routine ignores the possibility of macros entirely.
5884 ;; 2010-01-29.
5889 new-beg new-end beg-limit end-limit)
5890 ;; Locate the earliest < after the barrier before the changed region,
5891 ;; which isn't already marked as a paren.
5895 ;; Remove the syntax-table/category properties from each pertinent <...>
5896 ;; pair. Firstly, the ones with the < before beg and > after beg....
5905 ;; ...Then the ones with < before end and > after end.
5916 ;; Extend the fontification region, if needed.
5925 ;; This is called from `after-change-functions' when
5926 ;; c-recognize-<>-arglists' is set. It ensures that no "<" or ">"
5927 ;; chars with paren syntax become part of another operator like "<<"
5928 ;; or ">=".
5929 ;;
5930 ;; This function might do hidden buffer changes.
5963 ;; This function is called as an after-change function. It restores the
5964 ;; category/syntax-table properties on template/generic <..> pairs between
5965 ;; c-new-BEG and c-new-END. It may do hidden buffer changes.
5985 \f
5986 ;; Functions to handle C++ raw strings.
5987 ;;
5988 ;; A valid C++ raw string looks like
5989 ;; R"<id>(<contents>)<id>"
5990 ;; , where <id> is an identifier from 0 to 16 characters long, not containing
5991 ;; spaces, control characters, double quote or left/right paren. <contents>
5992 ;; can include anything which isn't the terminating )<id>", including new
5993 ;; lines, "s, parentheses, etc.
5994 ;;
5995 ;; CC Mode handles C++ raw strings by the use of `syntax-table' text
5996 ;; properties as follows:
5997 ;;
5998 ;; (i) On a validly terminated raw string, no `syntax-table' text properties
5999 ;; are applied to the opening and closing delimiters, but any " in the
6000 ;; contents is given the property value "punctuation" (`(1)') to prevent it
6001 ;; interacting with the "s in the delimiters.
6002 ;;
6003 ;; The font locking routine `c-font-lock-c++-raw-strings' (in cc-fonts.el)
6004 ;; recognizes valid raw strings, and fontifies the delimiters (apart from
6005 ;; the parentheses) with the default face and the parentheses and the
6006 ;; <contents> with font-lock-string-face.
6007 ;;
6008 ;; (ii) A valid, but unterminated, raw string opening delimiter gets the
6009 ;; "punctuation" value (`(1)') of the `syntax-table' text property, and the
6010 ;; open parenthesis gets the "string fence" value (`(15)').
6011 ;;
6012 ;; `c-font-lock-c++-raw-strings' puts c-font-lock-warning-face on the entire
6013 ;; unmatched opening delimiter (from the R up to the open paren), and allows
6014 ;; the rest of the buffer to get font-lock-string-face, caused by the
6015 ;; unmatched "string fence" `syntax-table' text property value.
6016 ;;
6017 ;; (iii) Inside a macro, a valid raw string is handled as in (i). An
6018 ;; unmatched opening delimiter is handled slightly differently. In addition
6019 ;; to the "punctuation" and "string fence" properties on the delimiter,
6020 ;; another "string fence" `syntax-table' property is applied to the last
6021 ;; possible character of the macro before the terminating linefeed (if there
6022 ;; is such a character after the "("). This "last possible" character is
6023 ;; never a backslash escaping the end of line. If the character preceding
6024 ;; this "last possible" character is itself a backslash, this preceding
6025 ;; character gets a "punctuation" `syntax-table' value. If the "(" is
6026 ;; already at the end of the macro, it gets the "punctuation" value, and no
6027 ;; "string fence"s are used.
6028 ;;
6029 ;; The effect on the fontification of either of these tactics is that rest of
6030 ;; the macro (if any) after the "(" gets font-lock-string-face, but the rest
6031 ;; of the file is fontified normally.
6035 ;; Get POINT's relationship to any containing raw string.
6036 ;; If point isn't in a raw string, return nil.
6037 ;; Otherwise, return the following list:
6038 ;;
6039 ;; (POS B\" B\( E\) E\")
6040 ;;
6041 ;; , where POS is the symbol `open-delim' if point is in the opening
6042 ;; delimiter, the symbol `close-delim' if it's in the closing delimiter, and
6043 ;; nil if it's in the string body. B\", B\(, E\), E\" are the positions of
6044 ;; the opening and closing quotes and parentheses of a correctly terminated
6045 ;; raw string. (N.B.: E\) and E\" are NOT on the "outside" of these
6046 ;; characters.) If the raw string is not terminated, E\) and E\" are set to
6047 ;; nil.
6048 ;;
6049 ;; Note: this routine is dependant upon the correct syntax-table text
6050 ;; properties being set.
6052 open-quote-pos open-paren-pos close-paren-pos close-quote-pos id)
6084 open-quote-pos open-paren-pos close-paren-pos close-quote-pos))))
6087 ;; Point is immediately after a raw string opening delimiter. Remove any
6088 ;; `syntax-table' text properties associated with the delimiter (if it's
6089 ;; unmatched) or the raw string.
6090 ;;
6091 ;; ID, a string, is the delimiter's identifier. OPEN-QUOTE and OPEN-PAREN
6092 ;; are the buffer positions of the delimiter's components. BOUND is the
6093 ;; bound for searching for a matching closing delimiter; it is usually nil,
6094 ;; but if we're inside a macro, it's the end of the macro.
6095 ;;
6096 ;; Point is moved to after the (terminated) raw string, or left after the
6097 ;; unmatched opening delimiter, as the case may be. The return value is of
6098 ;; no significance.
6102 ;; A terminated raw string
6109 closing-paren)
6111 )
6116 ))))
6119 ;; An unterminated raw string either not in a macro, or in a macro with
6120 ;; the open parenthesis right up against the end of macro
6125 ;; An unterminated string in a macro, with at least one char after the
6126 ;; open paren
6136 ))))
6139 ;; Remove any `syntax-table' text properties associated with C++ raw strings
6140 ;; contained in the region (START FINISH). Point is undefined at entry and
6141 ;; exit, and the return value has no significance.
6146 c-anchored-cpp-prefix ; 2
6150 finish t))
6154 ;; We've found a raw string
6160 ;; We've found a CPP construct. Search for raw strings within it.
6175 ;; This function clears `syntax-table' text properties from C++ raw strings
6176 ;; in the region (c-new-BEG c-new-END). BEG and END are the standard
6177 ;; arguments supplied to any before-change function.
6178 ;;
6179 ;; Point is undefined on both entry and exit, and the return value has no
6180 ;; significance.
6181 ;;
6182 ;; This function is called as a before-change function solely due to its
6183 ;; membership of the C++ value of `c-get-state-before-change-functions'.
6187 beg
6191 ; Optimize this so that we don't call
6192 ; `c-raw-string-pos' twice when once
6193 ; will do. (2016-06-02).
6195 end
6197 )
6209 ;; Point is immediately after a raw string opening delimiter. Apply any
6210 ;; pertinent `syntax-table' text properties to the delimiter and also the
6211 ;; raw string, should there be a valid matching closing delimiter.
6212 ;;
6213 ;; ID, a string, is the delimiter's identifier. OPEN-QUOTE and OPEN-PAREN
6214 ;; are the buffer positions of the delimiter's components. BOUND is the
6215 ;; bound for searching for a matching closing delimiter; it is usually nil,
6216 ;; but if we're inside a macro, it's the end of the macro.
6217 ;;
6218 ;; Point is moved to after the (terminated) raw string, or left after the
6219 ;; unmatched opening delimiter, as the case may be. The return value is of
6220 ;; no significance.
6235 ;; In a CPP construct, we try to apply a generic-string `syntax-table'
6236 ;; text property to the last possible character in the string, so that
6237 ;; only characters within the macro get "stringed out".
6243 ;; This regular expression matches either an escape pair (which
6244 ;; isn't an escaped NL) (submatch 5) or a non-escaped character
6245 ;; (which isn't itself a backslash) (submatch 10). The long
6246 ;; preambles to these (respectively submatches 2-4 and 6-9)
6247 ;; ensure that we have the correct parity for sequences of
6248 ;; backslashes, etc..
6269 ;; This function applies `syntax-table' text properties to C++ raw strings
6270 ;; beginning in the region (c-new-BEG c-new-END). BEG, END, and OLD-LEN are
6271 ;; the standard arguments supplied to any after-change function.
6272 ;;
6273 ;; Point is undefined on both entry and exit, and the return value has no
6274 ;; significance.
6275 ;;
6276 ;; This function is called as an after-change function solely due to its
6277 ;; membership of the C++ value of `c-before-font-lock-functions'.
6279 ;; If the region (c-new-BEG c-new-END) has expanded, remove
6280 ;; `syntax-table' text-properties from the new piece(s).
6284 c-new-BEG
6287 c-old-BEG))))
6293 c-old-END)
6294 c-new-END)))
6300 c-anchored-cpp-prefix ; 2
6304 c-new-END t))
6308 ;; We've found a raw string.
6314 ;; We've found a CPP construct. Search for raw strings within it.
6328 \f
6329 ;; Handling of small scale constructs like types and names.
6331 ;; Dynamically bound variable that instructs `c-forward-type' to also
6332 ;; treat possible types (i.e. those that it normally returns 'maybe or
6333 ;; 'found for) as actual types (and always return 'found for them).
6334 ;; This means that it records them in `c-record-type-identifiers' if
6335 ;; that is set, and that it adds them to `c-found-types'.
6338 ;; Dynamically bound variable that instructs `c-forward-<>-arglist' to
6339 ;; mark up successfully parsed arglists with paren syntax properties on
6340 ;; the surrounding angle brackets and with `c-<>-arg-sep' in the
6341 ;; `c-type' property of each argument separating comma.
6342 ;;
6343 ;; Setting this variable also makes `c-forward-<>-arglist' recurse into
6344 ;; all arglists for side effects (i.e. recording types), otherwise it
6345 ;; exploits any existing paren syntax properties to quickly jump to the
6346 ;; end of already parsed arglists.
6347 ;;
6348 ;; Marking up the arglists is not the default since doing that correctly
6349 ;; depends on a proper value for `c-restricted-<>-arglists'.
6352 ;; Dynamically bound variable that instructs `c-forward-<>-arglist' to
6353 ;; not accept arglists that contain binary operators.
6354 ;;
6355 ;; This is primarily used to handle C++ template arglists. C++
6356 ;; disambiguates them by checking whether the preceding name is a
6357 ;; template or not. We can't do that, so we assume it is a template
6358 ;; if it can be parsed as one. That usually works well since
6359 ;; comparison expressions on the forms "a < b > c" or "a < b, c > d"
6360 ;; in almost all cases would be pointless.
6361 ;;
6362 ;; However, in function arglists, e.g. in "foo (a < b, c > d)", we
6363 ;; should let the comma separate the function arguments instead. And
6364 ;; in a context where the value of the expression is taken, e.g. in
6365 ;; "if (a < b || c > d)", it's probably not a template.
6368 ;; Dynamically bound variables that instructs
6369 ;; `c-forward-keyword-clause', `c-forward-<>-arglist',
6370 ;; `c-forward-name', `c-forward-type', `c-forward-decl-or-cast-1', and
6371 ;; `c-forward-label' to record the ranges of all the type and
6372 ;; reference identifiers they encounter. They will build lists on
6373 ;; these variables where each element is a cons of the buffer
6374 ;; positions surrounding each identifier. This recording is only
6375 ;; activated when `c-record-type-identifiers' is non-nil.
6376 ;;
6377 ;; All known types that can't be identifiers are recorded, and also
6378 ;; other possible types if `c-promote-possible-types' is set.
6379 ;; Recording is however disabled inside angle bracket arglists that
6380 ;; are encountered inside names and other angle bracket arglists.
6381 ;; Such occurrences are taken care of by `c-font-lock-<>-arglists'
6382 ;; instead.
6383 ;;
6384 ;; Only the names in C++ template style references (e.g. "tmpl" in
6385 ;; "tmpl<a,b>::foo") are recorded as references, other references
6386 ;; aren't handled here.
6387 ;;
6388 ;; `c-forward-label' records the label identifier(s) on
6389 ;; `c-record-ref-identifiers'.
6393 ;; This variable will receive a cons cell of the range of the last
6394 ;; single identifier symbol stepped over by `c-forward-name' if it's
6395 ;; successful. This is the range that should be put on one of the
6396 ;; record lists above by the caller. It's assigned nil if there's no
6397 ;; such symbol in the name.
6402 ;; Always true.
6412 ;; Always true.
6420 ;; Dynamically bound variable that instructs `c-forward-type' to
6421 ;; record the ranges of types that only are found. Behaves otherwise
6422 ;; like `c-record-type-identifiers'.
6426 ;; Used internally in `c-forward-keyword-clause' to move forward
6427 ;; over a type (if TYPE is 'type) or a name (otherwise) which
6428 ;; possibly is prefixed by keywords and their associated clauses.
6429 ;; Try with a type/name first to not trip up on those that begin
6430 ;; with a keyword. Return t if a known or found type is moved
6431 ;; over. The point is clobbered if nil is returned. If range
6432 ;; recording is enabled, the identifier is recorded on as a type
6433 ;; if TYPE is 'type or as a reference if TYPE is 'ref.
6434 ;;
6435 ;; This macro might do hidden buffer changes.
6441 nil
6452 t)))
6455 ;; Used internally in `c-forward-keyword-clause' to move forward
6456 ;; over a comma separated list of types or names using
6457 ;; `c-forward-keyword-prefixed-id'.
6458 ;;
6459 ;; This macro might do hidden buffer changes.
6470 ;; Point is at a c-noise-macro-with-parens-names macro identifier. Go
6471 ;; forward over this name, any parenthesis expression which follows it, and
6472 ;; any syntactic WS, ending up at the next token. If there is an unbalanced
6473 ;; paren expression, leave point at it. Always Return t.
6478 t)
6481 ;; Submatch MATCH in the current match data is assumed to surround a
6482 ;; token. If it's a keyword, move over it and any immediately
6483 ;; following clauses associated with it, stopping at the start of
6484 ;; the next token. t is returned in that case, otherwise the point
6485 ;; stays and nil is returned. The kind of clauses that are
6486 ;; recognized are those specified by `c-type-list-kwds',
6487 ;; `c-ref-list-kwds', `c-colon-type-list-kwds',
6488 ;; `c-paren-nontype-kwds', `c-paren-type-kwds', `c-<>-type-kwds',
6489 ;; and `c-<>-arglist-kwds'.
6490 ;;
6491 ;; This function records identifier ranges on
6492 ;; `c-record-type-identifiers' and `c-record-ref-identifiers' if
6493 ;; `c-record-type-identifiers' is non-nil.
6494 ;;
6495 ;; Note that for `c-colon-type-list-kwds', which doesn't necessary
6496 ;; apply directly after the keyword, the type list is moved over
6497 ;; only when there is no unaccounted token before it (i.e. a token
6498 ;; that isn't moved over due to some other keyword list). The
6499 ;; identifier ranges in the list are still recorded if that should
6500 ;; be done, though.
6501 ;;
6502 ;; This function might do hidden buffer changes.
6505 ;; The call to `c-forward-<>-arglist' below is made after
6506 ;; `c-<>-sexp-kwds' keywords, so we're certain they actually
6507 ;; are angle bracket arglists and `c-restricted-<>-arglists'
6508 ;; should therefore be nil.
6520 ;; There's a type directly after a keyword in `c-type-list-kwds'.
6525 ;; There's a name directly after a keyword in `c-ref-list-kwds'.
6530 ;; There's an open paren after a keyword in `c-paren-any-kwds'.
6537 ;; Use `c-forward-type' on every identifier we can find
6538 ;; inside the paren, to record the types.
6563 ;; If we are at the colon already, we move over the type
6564 ;; list after it.
6570 ;; Not at the colon, so stop here. But the identifier
6571 ;; ranges in the type list later on should still be
6572 ;; recorded.
6575 ;; If a keyword matched both one of the types above and
6576 ;; this one, we match `c-colon-type-list-re' after the
6577 ;; clause matched above.
6584 ;; There's a type after the `c-colon-type-list-re' match
6585 ;; after a keyword in `c-colon-type-list-kwds'.
6589 t)))
6591 ;; cc-mode requires cc-fonts.
6595 ;; The point is assumed to be at a "<". Try to treat it as the open
6596 ;; paren of an angle bracket arglist and move forward to the
6597 ;; corresponding ">". If successful, the point is left after the
6598 ;; ">" and t is returned, otherwise the point isn't moved and nil is
6599 ;; returned. If ALL-TYPES is t then all encountered arguments in
6600 ;; the arglist that might be types are treated as found types.
6601 ;;
6602 ;; The variable `c-parse-and-markup-<>-arglists' controls how this
6603 ;; function handles text properties on the angle brackets and argument
6604 ;; separating commas.
6605 ;;
6606 ;; `c-restricted-<>-arglists' controls how lenient the template
6607 ;; arglist recognition should be.
6608 ;;
6609 ;; This function records identifier ranges on
6610 ;; `c-record-type-identifiers' and `c-record-ref-identifiers' if
6611 ;; `c-record-type-identifiers' is non-nil.
6612 ;;
6613 ;; This function might do hidden buffer changes.
6616 ;; If `c-record-type-identifiers' is set then activate
6617 ;; recording of any found types that constitute an argument in
6618 ;; the arglist.
6626 ;; `nconc' doesn't mind that the tail of
6627 ;; `c-record-found-types' is t.
6629 t)
6632 nil)))
6635 ;; Recursive part of `c-forward-<>-arglist'.
6636 ;;
6637 ;; This function might do hidden buffer changes.
6639 ;; Cover this so that any recorded found type ranges are
6640 ;; automatically lost if it turns out to not be an angle
6641 ;; bracket arglist. It's propagated through the return value
6642 ;; on successful completion.
6644 ;; List that collects the positions after the argument
6645 ;; separating ',' in the arglist.
6646 arg-start-pos)
6647 ;; If the '<' has paren open syntax then we've marked it as an angle
6648 ;; bracket arglist before, so skip to the end.
6656 t
6657 ;; Got unmatched paren angle brackets. We don't clear the paren
6658 ;; syntax properties and retry, on the basis that it's very
6659 ;; unlikely that paren angle brackets become operators by code
6660 ;; manipulation. It's far more likely that it doesn't match due
6661 ;; to narrowing or some temporary change.
6663 nil))
6668 ;; go forward one non-alphanumeric character (group) per iteration of
6669 ;; this loop.
6683 ;; All encountered identifiers are types, so set the
6684 ;; promote flag and parse the type.
6702 ;; Note: These regexps exploit the match order in \| so
6703 ;; that "<>" is matched by "<" rather than "[^>:-]>".
6705 ;; Stop on ',', '|', '&', '+' and '-' to catch
6706 ;; common binary operators that could be between
6707 ;; two comparison expressions "a<b" and "c>d".
6708 ;; 2016-02-11: C++11 templates can now contain arithmetic
6709 ;; expressions, so template detection in C++ is now less
6710 ;; robust than it was.
6711 c-<>-notable-chars-re
6712 nil t t))
6716 ;; Either an operator starting with '>' or the end of
6717 ;; the angle bracket arglist.
6726 ;; The angle bracket arglist is finished.
6738 ;; Either an operator starting with '<' or a nested arglist.
6742 ;; The '<' begins a multi-char operator.
6745 ;; We're at a nested <.....>
6750 ;; There's always an identifier before an angle
6751 ;; bracket arglist, or a keyword in `c-<>-type-kwds'
6752 ;; or `c-<>-arglist-kwds'.
6769 subres)
6770 ;; It was an angle bracket arglist.
6773 ;; Record the identifier before the template as a type
6774 ;; or reference depending on whether the arglist is last
6775 ;; in a qualified identifier.
6785 ;; At a "less than" operator.
6787 ;; (forward-char) ; NO! We've already gone over the <.
6788 )))
6800 ;; Just another argument. Record the position. The
6801 ;; type check stuff that made us stop at it is at
6802 ;; the top of the loop.
6806 ;; Got a character that can't be in an angle bracket
6807 ;; arglist argument. Abort using `throw', since
6808 ;; it's useless to try to find a surrounding arglist
6809 ;; if we're nested.
6815 ;; The point is assumed to be directly after a ">". Try to treat it
6816 ;; as the close paren of an angle bracket arglist and move back to
6817 ;; the corresponding "<". If successful, the point is left at
6818 ;; the "<" and t is returned, otherwise the point isn't moved and
6819 ;; nil is returned. ALL-TYPES is passed on to
6820 ;; `c-forward-<>-arglist'.
6821 ;;
6822 ;; If the optional LIMIT is given, it bounds the backward search.
6823 ;; It's then assumed to be at a syntactically relevant position.
6824 ;;
6825 ;; This is a wrapper around `c-forward-<>-arglist'. See that
6826 ;; function for more details.
6835 t
6836 ;; See corresponding note in `c-forward-<>-arglist'.
6838 nil)
6845 t
6846 ;; Stopped at bob or a char that isn't allowed in an
6847 ;; arglist, so we've failed.
6849 nil)
6854 ;; If we moved then the "<" was part of some
6855 ;; multicharacter token.
6856 t
6862 ;; Matched the arglist. Break the while.
6864 nil)
6866 ;; We started from a non-paren ">" inside an
6867 ;; arglist.
6869 nil)
6871 ;; Matched a shorter arglist. Can be a nested
6872 ;; one so continue looking.
6874 t))
6875 t))))))
6880 ;; Move forward over a complete name if at the beginning of one,
6881 ;; stopping at the next following token. A keyword, as such,
6882 ;; doesn't count as a name. If the point is not at something that
6883 ;; is recognized as a name then it stays put.
6884 ;;
6885 ;; A name could be something as simple as "foo" in C or something as
6886 ;; complex as "X<Y<class A<int>::B, BIT_MAX >> b>, ::operator<> ::
6887 ;; Z<(a>b)> :: operator const X<&foo>::T Q::G<unsigned short
6888 ;; int>::*volatile const" in C++ (this function is actually little
6889 ;; more than a `looking-at' call in all modes except those that,
6890 ;; like C++, have `c-recognize-<>-arglists' set).
6891 ;;
6892 ;; Return
6893 ;; o - nil if no name is found;
6894 ;; o - 'template if it's an identifier ending with an angle bracket
6895 ;; arglist;
6896 ;; o - 'operator of it's an operator identifier;
6897 ;; o - t if it's some other kind of name.
6898 ;;
6899 ;; This function records identifier ranges on
6900 ;; `c-record-type-identifiers' and `c-record-ref-identifiers' if
6901 ;; `c-record-type-identifiers' is non-nil.
6902 ;;
6903 ;; This function might do hidden buffer changes.
6906 ;; Turn off `c-promote-possible-types' here since we might
6907 ;; call `c-forward-<>-arglist' and we don't want it to promote
6908 ;; every suspect thing in the arglist to a type. We're
6909 ;; typically called from `c-forward-type' in this case, and
6910 ;; the caller only wants the top level type that it finds to
6911 ;; be promoted.
6912 c-promote-possible-types)
6918 ;; Check for keyword. We go to the last symbol in
6919 ;; `c-identifier-key' first.
6932 ;; "template" is only valid inside an
6933 ;; identifier if preceded by "::".
6938 t))
6940 ;; Handle a C++ operator or template identifier.
6944 ;; Got "... ::template".
6948 res subres))))
6951 ;; Got a cast operator.
6955 ;; Now we should match a sequence of either
6956 ;; '*', '&' or a name followed by ":: *",
6957 ;; where each can be followed by a sequence
6958 ;; of `c-opt-type-modifier-key'.
6961 t)
6971 t))))
6979 ;; Got some other operator.
6987 nil)
6989 ;; `id-start' is equal to `id-end' if we've jumped over
6990 ;; an identifier that doesn't end with a symbol token.
6991 ;; That can occur e.g. for Java import directives on the
6992 ;; form "foo.bar.*".
6999 res t)))
7009 ;; Got a concatenated identifier. This handles the
7010 ;; cases with tricky syntactic whitespace that aren't
7011 ;; covered in `c-identifier-key'.
7014 t)
7018 ;; Maybe an angle bracket arglist.
7031 ;; Continue if there's an identifier concatenation
7032 ;; operator after the template argument.
7038 t)
7044 nil)))
7045 )))))
7048 res))
7051 ;; Move forward over a type spec if at the beginning of one,
7052 ;; stopping at the next following token. The keyword "typedef"
7053 ;; isn't part of a type spec here.
7054 ;;
7055 ;; BRACE-BLOCK-TOO, when non-nil, means move over the brace block in
7056 ;; constructs like "struct foo {...} bar ;" or "struct {...} bar;".
7057 ;; The current (2009-03-10) intention is to convert all uses of
7058 ;; `c-forward-type' to call with this parameter set, then to
7059 ;; eliminate it.
7060 ;;
7061 ;; Return
7062 ;; o - t if it's a known type that can't be a name or other
7063 ;; expression;
7064 ;; o - 'known if it's an otherwise known type (according to
7065 ;; `*-font-lock-extra-types');
7066 ;; o - 'prefix if it's a known prefix of a type;
7067 ;; o - 'found if it's a type that matches one in `c-found-types';
7068 ;; o - 'maybe if it's an identifier that might be a type;
7069 ;; o - 'decltype if it's a decltype(variable) declaration; - or
7070 ;; o - nil if it can't be a type (the point isn't moved then).
7071 ;;
7072 ;; The point is assumed to be at the beginning of a token.
7073 ;;
7074 ;; Note that this function doesn't skip past the brace definition
7075 ;; that might be considered part of the type, e.g.
7076 ;; "enum {a, b, c} foo".
7077 ;;
7078 ;; This function records identifier ranges on
7079 ;; `c-record-type-identifiers' and `c-record-ref-identifiers' if
7080 ;; `c-record-type-identifiers' is non-nil.
7081 ;;
7082 ;; This function might do hidden buffer changes.
7090 ;; Skip leading type modifiers. If any are found we know it's a
7091 ;; prefix of a type.
7110 ; "typedef".
7129 ;; In many languages the name can be used without the
7130 ;; prefix, so we add it to `c-found-types'.
7133 c-last-identifier-range)
7155 id-range c-last-identifier-range))))
7167 id-end)
7169 ;; Looking at a primitive or known type identifier. We've
7170 ;; checked for a name first so that we don't go here if the
7171 ;; known type match only is a prefix of another name.
7182 ;; There might be more keywords for the type.
7210 ;; A normal identifier.
7220 ;; It's an identifier that has been used as
7221 ;; a type somewhere else.
7222 'found
7223 ;; It's an identifier that might be a type.
7226 ;; A template is sometimes a type.
7232 ;; It's an identifier that has been used as
7233 ;; a type somewhere else.
7234 'found
7235 ;; It's an identifier that might be a type.
7237 t)))
7239 ;; Otherwise it's an operator identifier, which is not a type.
7244 ;; Skip trailing type modifiers. If any are found we know it's
7245 ;; a type.
7252 ;; Step over any type suffix operator. Do not let the existence
7253 ;; of these alter the classification of the found type, since
7254 ;; these operators typically are allowed in normal expressions
7255 ;; too.
7262 ;; Look for a trailing operator that concatenates the type
7263 ;; with a following one, and if so step past that one through
7264 ;; a recursive call. Note that we don't record concatenated
7265 ;; types in `c-found-types' - it's the component types that
7266 ;; are recorded when appropriate.
7269 c-promote-possible-types))
7270 ;; If we can't promote then set `c-record-found-types' so that
7271 ;; we can merge in the types from the second part afterwards if
7272 ;; it turns out to be a known type there.
7275 subres)
7284 ;; If either operand certainly is a type then both are, but we
7285 ;; don't let the existence of the operator itself promote two
7286 ;; uncertain types to a certain one.
7305 ;; Merge in the ranges of any types found by the second
7306 ;; `c-forward-type'.
7308 ;; `nconc' doesn't mind that the tail of
7309 ;; `c-record-found-types' is t.
7311 c-record-type-identifiers))))
7319 ;;(message "c-forward-type %s -> %s: %s" start (point) res)
7321 res))
7324 ;; Used for Java code only at the moment. Assumes point is on the @, moves
7325 ;; forward an annotation and returns t. Leaves point unmoved and returns
7326 ;; nil if there is no annotation at point.
7335 t)
7338 t))
7342 ;; Pull the next open brace from PS (which has the form of paren-state),
7343 ;; skipping over any brace pairs. Returns NIL when PS is exhausted.
7351 ;; Point is putatively just after a "compound identifier", i.e. something
7352 ;; looking (in C++) like this "FQN::of::base::Class". Move to the start of
7353 ;; this construct and return t. If the parsing fails, return nil, leaving
7354 ;; point unchanged.
7356 end
7357 )
7359 nil
7366 c-opt-identifier-concat-key
7373 t)))
7376 ;; Point is just after a closing brace/parenthesis. Try to parse this as a
7377 ;; C++ member initializer list, going back to just after the introducing ":"
7378 ;; and returning t. Otherwise return nil, leaving point unchanged.
7403 res))
7406 ;; Go back over a list of elements, each looking like:
7407 ;; <symbol> (<expression>) ,
7408 ;; or <symbol> {<expression>} ,
7409 ;; when we are putatively immediately after a comma. Stop when we don't see
7410 ;; a comma. If either of <symbol> or bracketed <expression> is missing,
7411 ;; throw nil to 'level. If the terminating } or ) is unmatched, throw nil
7412 ;; to 'done. This is not a general purpose macro!
7426 ;; Test whether we are in a C++ member initializer list, and if so, go back
7427 ;; to the introducing ":", returning the position of the opening paren of
7428 ;; the function's arglist. Otherwise return nil, leaving point unchanged.
7431 pos level-plausible at-top-level res)
7432 ;; Assume tentatively that we're at the top level. Try to go back to the
7433 ;; colon we seek.
7471 res))
7473 \f
7474 ;; Handling of large scale constructs like statements and declarations.
7476 ;; Macro used inside `c-forward-decl-or-cast-1'. It ought to be a
7477 ;; defsubst or perhaps even a defun, but it contains lots of free
7478 ;; variables that refer to things inside `c-forward-decl-or-cast-1'.
7480 ;; `c-forward-decl-or-cast-1' can consume an arbitrary length list
7481 ;; of types when parsing a declaration, which means that it
7482 ;; sometimes consumes the identifier in the declaration as a type.
7483 ;; This is used to "backtrack" and make the last type be treated as
7484 ;; an identifier instead.
7487 ;; These identifiers are bound only in the inner let.
7489 identifier-start type-start
7490 got-parens nil
7491 got-identifier t
7492 got-suffix t
7493 got-suffix-after-parens id-start
7497 t
7498 backup-at-type))
7500 id-start backup-id-start)
7502 id-start start-pos))
7504 ;; When these flags already are set we've found specifiers that
7505 ;; unconditionally signal these attributes - backtracking doesn't
7506 ;; change that. So keep them set in that case.
7513 ;; This identifier is bound only in the inner let.
7517 ;; Assuming point is at the start of a declarator, move forward over it,
7518 ;; leaving point at the next token after it (e.g. a ) or a ; or a ,).
7519 ;;
7520 ;; Return a list (ID-START ID-END BRACKETS-AFTER-ID GOT-INIT), where ID-START and
7521 ;; ID-END are the bounds of the declarator's identifier, and
7522 ;; BRACKETS-AFTER-ID is non-nil if a [...] pair is present after the id.
7523 ;; GOT-INIT is non-nil when the declarator is followed by "=" or "(".
7524 ;;
7525 ;; If ACCEPT-ANON is non-nil, move forward over any "anonymous declarator",
7526 ;; i.e. something like the (*) in int (*), such as might be found in a
7527 ;; declaration. In such a case ID-START and ID-END in the return value are
7528 ;; both set to nil. A "null" "anonymous declarator" gives a non-nil result.
7529 ;;
7530 ;; If no declarator is found, leave point unmoved and return nil. LIMIT is
7531 ;; an optional limit for forward searching.
7532 ;;
7533 ;; Note that the global variable `c-last-identifier-range' is written to, so
7534 ;; the caller should bind it if necessary.
7536 ;; Inside the following "condition form", we move forward over the
7537 ;; declarator's identifier up as far as any opening bracket (for array
7538 ;; size) or paren (for parameters of function-type) or brace (for
7539 ;; array/struct initialization) or "=" or terminating delimiter
7540 ;; (e.g. "," or ";" or "}").
7542 id-start id-end brackets-after-id paren-depth)
7547 ;; The following form moves forward over the declarator's
7548 ;; identifier (and what precedes it), returning t. If there
7549 ;; wasn't one, it returns nil.
7552 ;; Skip over type decl prefix operators, one for each iteration
7553 ;; of the while. These are, e.g. "*" in "int *foo" or "(" and
7554 ;; "*" in "int (*foo) (void)" (Note similar code in
7555 ;; `c-forward-decl-or-cast-1'.)
7566 ;; If the third submatch matches in C++ then
7567 ;; we're looking at an identifier that's a
7568 ;; prefix only if it specifies a member pointer.
7573 ;; We only check for a trailing "::" and
7574 ;; let the "*" that should follow be
7575 ;; matched in the next round.
7576 t
7577 ;; It turned out to be the real identifier,
7578 ;; so flag that and stop.
7580 nil))
7581 t))
7588 t)))
7590 ;; If we haven't passed the identifier already, do it now.
7601 t)
7604 ;; Skip out of the parens surrounding the identifier. If closing
7605 ;; parens are missing, this form returns nil.
7611 ;; Skip over any trailing bit, such as "__attribute__".
7621 ;; Search syntactically to the end of the declarator (";",
7622 ;; ",", a closing paren, eob etc) or to the beginning of an
7623 ;; initializer or function prototype ("=" or "\\s\(").
7624 ;; Note that square brackets are now not also treated as
7625 ;; initializers, since this broke when there were also
7626 ;; initializing brace lists.
7635 found))
7639 nil)))
7642 ;; Move forward over a declaration or a cast if at the start of one.
7643 ;; The point is assumed to be at the start of some token. Nil is
7644 ;; returned if no declaration or cast is recognized, and the point
7645 ;; is clobbered in that case.
7646 ;;
7647 ;; If a declaration is parsed:
7648 ;;
7649 ;; The point is left at the first token after the first complete
7650 ;; declarator, if there is one. The return value is a list of 4 elements,
7651 ;; where the first is the position of the first token in the declarator.
7652 ;; (See below for the other three.)
7653 ;; Some examples:
7654 ;;
7655 ;; void foo (int a, char *b) stuff ...
7656 ;; car ^ ^ point
7657 ;; float (*a)[], b;
7658 ;; car ^ ^ point
7659 ;; unsigned int a = c_style_initializer, b;
7660 ;; car ^ ^ point
7661 ;; unsigned int a (cplusplus_style_initializer), b;
7662 ;; car ^ ^ point (might change)
7663 ;; class Foo : public Bar {}
7664 ;; car ^ ^ point
7665 ;; class PikeClass (int a, string b) stuff ...
7666 ;; car ^ ^ point
7667 ;; enum bool;
7668 ;; car ^ ^ point
7669 ;; enum bool flag;
7670 ;; car ^ ^ point
7671 ;; void cplusplus_function (int x) throw (Bad);
7672 ;; car ^ ^ point
7673 ;; Foo::Foo (int b) : Base (b) {}
7674 ;; car ^ ^ point
7675 ;;
7676 ;; auto foo = 5;
7677 ;; car ^ ^ point
7678 ;; auto cplusplus_11 (int a, char *b) -> decltype (bar):
7679 ;; car ^ ^ point
7680 ;;
7681 ;;
7682 ;;
7683 ;; The second element of the return value is non-nil when a
7684 ;; `c-typedef-decl-kwds' specifier is found in the declaration.
7685 ;; Specifically it is a dotted pair (A . B) where B is t when a
7686 ;; `c-typedef-kwds' ("typedef") is present, and A is t when some
7687 ;; other `c-typedef-decl-kwds' (e.g. class, struct, enum)
7688 ;; specifier is present. I.e., (some of) the declared
7689 ;; identifier(s) are types.
7690 ;;
7691 ;; The third element of the return value is non-nil when the declaration
7692 ;; parsed might be an expression. The fourth element is the position of
7693 ;; the start of the type identifier.
7694 ;;
7695 ;; If a cast is parsed:
7696 ;;
7697 ;; The point is left at the first token after the closing paren of
7698 ;; the cast. The return value is `cast'. Note that the start
7699 ;; position must be at the first token inside the cast parenthesis
7700 ;; to recognize it.
7701 ;;
7702 ;; PRECEDING-TOKEN-END is the first position after the preceding
7703 ;; token, i.e. on the other side of the syntactic ws from the point.
7704 ;; Use a value less than or equal to (point-min) if the point is at
7705 ;; the first token in (the visible part of) the buffer.
7706 ;;
7707 ;; CONTEXT is a symbol that describes the context at the point:
7708 ;; 'decl In a comma-separated declaration context (typically
7709 ;; inside a function declaration arglist).
7710 ;; '<> In an angle bracket arglist.
7711 ;; 'arglist Some other type of arglist.
7712 ;; nil Some other context or unknown context. Includes
7713 ;; within the parens of an if, for, ... construct.
7714 ;;
7715 ;; LAST-CAST-END is the first token after the closing paren of a
7716 ;; preceding cast, or nil if none is known. If
7717 ;; `c-forward-decl-or-cast-1' is used in succession, it should be
7718 ;; the position after the closest preceding call where a cast was
7719 ;; matched. In that case it's used to discover chains of casts like
7720 ;; "(a) (b) c".
7721 ;;
7722 ;; This function records identifier ranges on
7723 ;; `c-record-type-identifiers' and `c-record-ref-identifiers' if
7724 ;; `c-record-type-identifiers' is non-nil.
7725 ;;
7726 ;; This function might do hidden buffer changes.
7729 ;; first type, i.e. after any leading specifiers. It might
7730 ;; also point at the beginning of the preceding syntactic
7731 ;; whitespace.
7733 ;; Set to the result of `c-forward-type'.
7734 at-type
7735 ;; The position of the first token in what we currently
7736 ;; believe is the type in the declaration or cast, after any
7737 ;; specifiers and their associated clauses.
7738 type-start
7739 ;; The position of the first token in what we currently
7740 ;; believe is the declarator for the first identifier. Set
7741 ;; when the type is found, and moved forward over any
7742 ;; `c-decl-hangon-kwds' and their associated clauses that
7743 ;; occurs after the type.
7744 id-start
7745 ;; These store `at-type', `type-start' and `id-start' of the
7746 ;; identifier before the one in those variables. The previous
7747 ;; identifier might turn out to be the real type in a
7748 ;; declaration if the last one has to be the declarator in it.
7749 ;; If `backup-at-type' is nil then the other variables have
7750 ;; undefined values.
7751 backup-at-type backup-type-start backup-id-start
7752 ;; This stores `kwd-sym' of the symbol before the current one.
7753 ;; This is needed to distinguish the C++11 version of "auto" from
7754 ;; the pre C++11 meaning.
7755 backup-kwd-sym
7756 ;; Set if we've found a specifier (apart from "typedef") that makes
7757 ;; the defined identifier(s) types.
7758 at-type-decl
7759 ;; Set if we've a "typedef" keyword.
7760 at-typedef
7761 ;; Set if we've found a specifier that can start a declaration
7762 ;; where there's no type.
7763 maybe-typeless
7764 ;; Save the value of kwd-sym between loops of the "Check for a
7765 ;; type" loop. Needed to distinguish a C++11 "auto" from a pre
7766 ;; C++11 one.
7767 prev-kwd-sym
7768 ;; If a specifier is found that also can be a type prefix,
7769 ;; these flags are set instead of those above. If we need to
7770 ;; back up an identifier, they are copied to the real flag
7771 ;; variables. Thus they only take effect if we fail to
7772 ;; interpret it as a type.
7773 backup-at-type-decl backup-maybe-typeless
7774 ;; Whether we've found a declaration or a cast. We might know
7775 ;; this before we've found the type in it. It's 'ids if we've
7776 ;; found two consecutive identifiers (usually a sure sign, but
7777 ;; we should allow that in labels too), and t if we've found a
7778 ;; specifier keyword (a 100% sure sign).
7779 at-decl-or-cast
7780 ;; Set when we need to back up to parse this as a declaration
7781 ;; but not as a cast.
7782 backup-if-not-cast
7783 ;; For casts, the return position.
7784 cast-end
7785 ;; Have we got a new-style C++11 "auto"?
7786 new-style-auto
7787 ;; Set when the symbol before `preceding-token-end' is known to
7788 ;; terminate the previous construct, or when we're at point-min.
7789 at-decl-start
7790 ;; Save `c-record-type-identifiers' and
7791 ;; `c-record-ref-identifiers' since ranges are recorded
7792 ;; speculatively and should be thrown away if it turns out
7793 ;; that it isn't a declaration or cast.
7796 ;; Set when we parse a declaration which might also be an expression,
7797 ;; such as "a *b". See CASE 16 and CASE 17.
7798 maybe-expression)
7807 c-pre-start-tokens)))))
7812 ;; Check for a type. Unknown symbols are treated as possible
7813 ;; types, but they could also be specifiers disguised through
7814 ;; macros like __INLINE__, so we recognize both types and known
7815 ;; specifiers after them too.
7820 ;; Look for a specifier keyword clause.
7838 ;; Found a known or possible type or a prefix of a known type.
7847 ;; Got two identifiers with nothing but whitespace
7848 ;; between them. That can only happen in declarations.
7852 ;; If the previous identifier is a found type we
7853 ;; record it as a real one; it might be some sort of
7854 ;; alias for a prefix like "unsigned".
7861 backup-type-start type-start
7862 backup-id-start id-start
7863 backup-kwd-sym kwd-sym
7864 at-type found-type
7865 type-start start
7867 ;; The previous ambiguous specifier/type turned out
7868 ;; to be a type since we've parsed another one after
7869 ;; it, so clear these backup flags.
7870 backup-at-type-decl nil
7871 backup-maybe-typeless nil))
7875 ;; Handle known specifier keywords and
7876 ;; `c-decl-hangon-kwds' which can occur after known
7877 ;; types.
7880 noise-start)
7881 ;; It's a hang-on keyword or noise clause that can occur
7882 ;; anywhere.
7885 ;; Move the identifier start position if
7886 ;; we've passed a type.
7888 ;; Otherwise treat this as a specifier and
7889 ;; move the fallback position.
7893 ;; It's an ordinary specifier so we know that
7894 ;; anything before this can't be the type.
7896 start-pos kwd-clause-end)
7899 ;; It's ambiguous whether this keyword is a
7900 ;; specifier or a type prefix, so set the backup
7901 ;; flags. (It's assumed that `c-forward-type'
7902 ;; moved further than `c-forward-keyword-clause'.)
7910 ;; This test only happens after we've scanned a type.
7911 ;; So, with valid syntax, kwd-sym can't be 'typedef.
7916 ;; Haven't matched a type so it's an unambiguous
7917 ;; specifier keyword and we know we're in a
7918 ;; declaration.
7924 ;; If the type isn't known we continue so that we'll jump
7925 ;; over all specifiers and type identifiers. The reason
7926 ;; to do this for a known type prefix is to make things
7927 ;; like "unsigned INT16" work.
7932 ;; If a known type was found, we still need to skip over any
7933 ;; hangon keyword clauses after it. Otherwise it has already
7934 ;; been done in the loop above.
7944 ;; A prefix type is itself a primitive type when it's not
7945 ;; followed by another type.
7949 ;; Got no type but set things up to continue anyway to handle
7950 ;; the various cases when a declaration doesn't start with a
7951 ;; type.
7956 ;; If it's C++ then check if the last "type" ends on the form
7957 ;; "foo::foo" or "foo::~foo", i.e. if it's the name of a
7958 ;; (con|de)structor.
7969 ;; Cheating in the handling of syntactic ws below.
7976 name))
7977 ;; It is a (con|de)structor name. In that case the
7978 ;; declaration is typeless so zap out any preceding
7979 ;; identifier(s) that we might have taken as types.
7982 backup-at-type nil
7983 id-start type-start))))))
7985 ;; Check for and step over a type decl expression after the thing
7986 ;; that is or might be a type. This can't be skipped since we
7987 ;; need the correct end position of the declarator for
7988 ;; `max-type-decl-end-*'.
7990 ;; True if there's a non-open-paren match of
7991 ;; `c-type-decl-prefix-key'.
7992 got-prefix
7993 ;; True if the declarator is surrounded by a parenthesis pair.
7994 got-parens
7995 ;; True if there is an identifier in the declarator.
7996 got-identifier
7997 ;; True if there's a non-close-paren match of
7998 ;; `c-type-decl-suffix-key'.
7999 got-suffix
8000 ;; True if there's a prefix match outside the outermost
8001 ;; paren pair that surrounds the declarator.
8002 got-prefix-before-parens
8003 ;; True if there's a suffix match outside the outermost
8004 ;; paren pair that surrounds the declarator. The value is
8005 ;; the position of the first suffix match.
8006 got-suffix-after-parens
8007 ;; True if we've parsed the type decl to a token that is
8008 ;; known to end declarations in this context.
8009 at-decl-end
8010 ;; The earlier values of `at-type' and `type-start' if we've
8011 ;; shifted the type backwards.
8012 identifier-type identifier-start
8013 ;; If `c-parse-and-markup-<>-arglists' is set we need to
8014 ;; turn it off during the name skipping below to avoid
8015 ;; getting `c-type' properties that might be bogus. That
8016 ;; can happen since we don't know if
8017 ;; `c-restricted-<>-arglists' will be correct inside the
8018 ;; arglist paren that gets entered.
8019 c-parse-and-markup-<>-arglists
8020 ;; Start of the identifier for which `got-identifier' was set.
8021 name-start)
8025 ;; Skip over type decl prefix operators. (Note similar code in
8026 ;; `c-forward-declarator'.)
8036 ;; If the third submatch matches in C++ then
8037 ;; we're looking at an identifier that's a
8038 ;; prefix only if it specifies a member pointer.
8043 ;; We only check for a trailing "::" and
8044 ;; let the "*" that should follow be
8045 ;; matched in the next round.
8047 ;; It turned out to be the real identifier,
8048 ;; so stop.
8049 nil))
8050 t))
8064 ;; Skip over an identifier.
8071 ;; Skip over type decl suffix operators and trailing noise macros.
8083 t)
8087 t)
8094 ;; No suffix matched. We might have matched the
8095 ;; identifier as a type and the open paren of a
8096 ;; function arglist as a type decl prefix. In that
8097 ;; case we should "backtrack": Reinterpret the last
8098 ;; type as the identifier, move out of the arglist and
8099 ;; continue searching for suffix operators.
8100 ;;
8101 ;; Do this even if there's no preceding type, to cope
8102 ;; with old style function declarations in K&R C,
8103 ;; (con|de)structors in C++ and `c-typeless-decl-kwds'
8104 ;; style declarations. That isn't applicable in an
8105 ;; arglist context, though.
8110 maybe-typeless
8111 backup-maybe-typeless
8118 t)))
8127 at-type))
8128 ;; Have found no identifier but `c-typeless-decl-kwds' has
8129 ;; matched so we know we're inside a declaration. The
8130 ;; preceding type must be the identifier instead.
8133 ;; Prepare the "-> type;" for fontification later on.
8143 at-decl-or-cast
8146 ;; CASE 1
8148 ;; Encountered something inside parens that isn't matched by
8149 ;; the `c-type-decl-*' regexps, so it's not a type decl
8150 ;; expression. Try to skip out to the same paren depth to
8151 ;; not confuse the cast check below.
8153 ;; If we've found a specifier keyword then it's a
8154 ;; declaration regardless.
8162 ;; Now we've collected info about various characteristics of
8163 ;; the construct we're looking at. Below follows a decision
8164 ;; tree based on that. It's ordered to check more certain
8165 ;; signs before less certain ones.
8170 ;; CASE 2
8173 ;; Got another identifier directly after the type, so it's a
8174 ;; declaration.
8179 ;; (not got-suffix-after-parens)
8181 maybe-typeless
8182 backup-maybe-typeless
8184 ;; Check whether we have "bar (gnu);" where we
8185 ;; are directly inside a class (etc.) called "bar".
8195 type-start
8201 ;; Got a declaration of the form "foo bar (gnu);" or "bar
8202 ;; (gnu);" where we've recognized "bar" as the type and "gnu"
8203 ;; as the declarator, and in the latter case, checked that
8204 ;; "bar (gnu)" appears directly inside the class "bar". In
8205 ;; this case it's however more likely that "bar" is the
8206 ;; declarator and "gnu" a function argument or initializer
8207 ;; (if `c-recognize-paren-inits' is set), since the parens
8208 ;; around "gnu" would be superfluous if it's a declarator.
8209 ;; Shift the type one step backward.
8212 ;; Found no identifier.
8217 ;; CASE 3
8219 ;; Got a plain list of identifiers. If a colon follows it's
8220 ;; a valid label, or maybe a bitfield. Otherwise the last
8221 ;; one probably is the declared identifier and we should
8222 ;; back up to the previous type, providing it isn't a cast.
8226 ;; If we've found a specifier keyword then it's a
8227 ;; declaration regardless.
8238 ;; CASE 4
8242 ;; Got a plain list of identifiers followed by some suffix.
8243 ;; If this isn't a cast then the last identifier probably is
8244 ;; the declared one and we should back up to the previous
8245 ;; type.
8249 ;; CASE 5
8251 ;; If the type is known we know that there can't be any
8252 ;; identifier somewhere else, and it's only in declarations in
8253 ;; e.g. function prototypes and in casts that the identifier may
8254 ;; be left out.
8258 ;; Only got a single identifier (parsed as a type so far).
8259 ;; CASE 6
8261 ;; Check that the identifier isn't at the start of an
8262 ;; expression.
8263 at-decl-end
8266 ;; Inside an arglist that contains declarations. If K&R
8267 ;; style declarations and parenthesis style initializers
8268 ;; aren't allowed then the single identifier must be a
8269 ;; type, else we require that it's known or found
8270 ;; (primitive types are handled above).
8275 ;; Inside a template arglist. Accept known and found
8276 ;; types; other identifiers could just as well be
8277 ;; constants in C++.
8280 ;; CASE 7
8281 ;; Can't be a valid declaration or cast, but if we've found a
8282 ;; specifier it can't be anything else either, so treat it as
8283 ;; an invalid/unfinished declaration or cast.
8291 maybe-typeless
8292 backup-maybe-typeless
8296 c-after-suffixed-type-maybe-decl-key))))))
8297 ;; Got an empty paren pair and a preceding type that probably
8298 ;; really is the identifier. Shift the type backwards to make
8299 ;; the last one the identifier. This is analogous to the
8300 ;; "backtracking" done inside the `c-type-decl-suffix-key' loop
8301 ;; above.
8302 ;;
8303 ;; Exception: In addition to the conditions in that
8304 ;; "backtracking" code, do not shift backward if we're not
8305 ;; looking at either `c-after-suffixed-type-decl-key' or "[;,]".
8306 ;; Since there's no preceding type, the shift would mean that
8307 ;; the declaration is typeless. But if the regexp doesn't match
8308 ;; then we will simply fall through in the tests below and not
8309 ;; recognize it at all, so it's better to try it as an abstract
8310 ;; declarator instead.
8313 ;; Still no identifier.
8314 ;; CASE 8
8316 ;; Require `got-prefix' together with either `got-parens' or
8317 ;; `got-suffix' to recognize it as an abstract declarator:
8318 ;; `got-parens' only is probably an empty function call.
8319 ;; `got-suffix' only can build an ordinary expression together
8320 ;; with the preceding identifier which we've taken as a type.
8321 ;; We could actually accept on `got-prefix' only, but that can
8322 ;; easily occur temporarily while writing an expression so we
8323 ;; avoid that case anyway. We could do a better job if we knew
8324 ;; the point when the fontification was invoked.
8327 ;; CASE 9
8331 got-suffix-after-parens
8333 ;; Got a type, no declarator but a paren suffix. I.e. it's a
8334 ;; normal function call after all (or perhaps a C++ style object
8335 ;; instantiation expression).
8338 ;; CASE 10
8340 ;; By now we've located the type in the declaration that we know
8341 ;; we're in.
8344 ;; CASE 11
8352 ;; Shift the type backward in the case that there's a
8353 ;; single identifier inside parens. That can only
8354 ;; occur in K&R style function declarations so it's
8355 ;; more likely that it really is a function call.
8356 ;; Therefore we only do this after
8357 ;; `c-after-suffixed-type-decl-key' has matched.
8359 got-suffix-after-parens))
8360 ;; A declaration according to `c-after-suffixed-type-decl-key'.
8363 ;; CASE 12
8366 ;; It's a declaration if a known type precedes it and it can't be a
8367 ;; function call.
8370 ;; If we get here we can't tell if this is a type decl or a normal
8371 ;; expression by looking at it alone. (That's under the assumption
8372 ;; that normal expressions always can look like type decl expressions,
8373 ;; which isn't really true but the cases where it doesn't hold are so
8374 ;; uncommon (e.g. some placements of "const" in C++) it's not worth
8375 ;; the effort to look for them.)
8377 ;;; 2008-04-16: commented out the next form, to allow the function to recognize
8378 ;;; "foo (int bar)" in CC (an implicit type (in class foo) without a semicolon)
8379 ;;; as a(n almost complete) declaration, enabling it to be fontified.
8380 ;; CASE 13
8381 ;; (unless (or at-decl-end (looking-at "=[^=]"))
8382 ;; If this is a declaration it should end here or its initializer(*)
8383 ;; should start here, so check for allowed separation tokens. Note
8384 ;; that this rule doesn't work e.g. with a K&R arglist after a
8385 ;; function header.
8386 ;;
8387 ;; *) Don't check for C++ style initializers using parens
8388 ;; since those already have been matched as suffixes.
8389 ;;
8390 ;; If `at-decl-or-cast' is then we've found some other sign that
8391 ;; it's a declaration or cast, so then it's probably an
8392 ;; invalid/unfinished one.
8393 ;; (throw 'at-decl-or-cast at-decl-or-cast))
8395 ;; Below are tests that only should be applied when we're certain to
8396 ;; not have parsed halfway through an expression.
8398 ;; CASE 14
8400 ;; The expression starts with a known type so treat it as a
8401 ;; declaration.
8404 ;; CASE 15
8406 ;; In C++ we check if the identifier is a known type, since
8407 ;; (con|de)structors use the class name as identifier.
8408 ;; We've always shifted over the identifier as a type and
8409 ;; then backed up again in this case.
8410 identifier-type
8413 ;; `at-type' probably won't be 'found for
8414 ;; destructors since the "~" is then part of the
8415 ;; type name being checked against the list of
8416 ;; known types, so do a check without that
8417 ;; operator.
8422 c-identifier-syntax-table
8426 ;; We have already parsed the type earlier,
8427 ;; so it'd be possible to cache the end
8428 ;; position instead of redoing it here, but
8429 ;; then we'd need to keep track of another
8430 ;; position everywhere.
8438 ;; CASE 16
8440 at-type
8444 at-decl-start))
8445 ;; Got something like "foo * bar;". Since we're not inside
8446 ;; an arglist it would be a meaningless expression because
8447 ;; the result isn't used. We therefore choose to recognize
8448 ;; it as a declaration. We only allow a suffix (which makes
8449 ;; the construct look like a function call) when
8450 ;; `at-decl-start' provides additional evidence that we do
8451 ;; have a declaration.
8455 ;; CASE 17
8460 ;; Got something like "a (*b) (c);" or "a (b) = c;". It could
8461 ;; be an odd expression or it could be a declaration. Treat
8462 ;; it as a declaration if "a" has been used as a type
8463 ;; somewhere else (if it's a known type we won't get here).
8467 ;; CASE 18
8473 ;; Got a type followed by an abstract declarator. If `got-prefix'
8474 ;; is set it's something like "a *" without anything after it. If
8475 ;; `got-parens' or `got-suffix' is set it's "a()", "a[]", "a()[]",
8476 ;; or similar, which we accept only if the context rules out
8477 ;; expressions.
8480 ;; If we had a complete symbol table here (which rules out
8481 ;; `c-found-types') we should return t due to the disambiguation rule
8482 ;; (in at least C++) that anything that can be parsed as a declaration
8483 ;; is a declaration. Now we're being more defensive and prefer to
8484 ;; highlight things like "foo (bar);" as a declaration only if we're
8485 ;; inside an arglist that contains declarations.
8486 ;; CASE 19
8489 ;; The point is now after the type decl expression.
8492 ;; Check for a cast.
8495 c-cast-parens
8497 ;; Should be the first type/identifier in a cast paren.
8501 ;; The closing paren should follow.
8506 ;; There should be a primary expression after it.
8515 ;; Check if the expression begins with a prefix keyword.
8518 ;; Expression begins with an ambiguous operator. Treat
8519 ;; it as a cast if it's a type decl or if we've
8520 ;; recognized the type somewhere else.
8523 ;; Unless it's a keyword, it's the beginning of a primary
8524 ;; expression.
8526 ;; If `c-primary-expr-regexp' matched a nonsymbol token, check
8527 ;; that it matched a whole one so that we don't e.g. confuse
8528 ;; the operator '-' with '->'. It's ok if it matches further,
8529 ;; though, since it e.g. can match the float '.5' while the
8530 ;; operator regexp only matches '.'.
8534 ;; There should either be a cast before it or something that isn't an
8535 ;; identifier or close paren.
8543 ;; It's a symbol. Accept it only if it's one of the
8544 ;; keywords that can precede an expression (without
8545 ;; surrounding parens).
8548 ;; Check that it isn't a close paren (block close is ok,
8549 ;; though).
8551 ;; Check that it isn't a nonsymbol identifier.
8554 ;; Handle the cast.
8564 ;; We're at a declaration. Highlight the type and the following
8565 ;; declarators.
8571 ;; Make sure to propagate the `c-decl-arg-start' property to
8572 ;; the next argument if it's set in this one, to cope with
8573 ;; interactive refontification.
8576 ;; Record the type's coordinates in `c-record-type-identifiers' for
8577 ;; later fontification.
8579 ;; There seems no reason to exclude a token from
8580 ;; fontification just because it's "a known type that can't
8581 ;; be a name or other expression". 2013-09-18.
8582 )
8591 maybe-expression
8592 type-start))
8595 ;; False alarm. Restore the recorded ranges.
8597 c-record-ref-identifiers save-rec-ref-ids)
8598 nil))))
8601 ;; Assuming that point is at the beginning of a token, check if it starts a
8602 ;; label and if so move over it and return non-nil (t in default situations,
8603 ;; specific symbols (see below) for interesting situations), otherwise don't
8604 ;; move and return nil. "Label" here means "most things with a colon".
8605 ;;
8606 ;; More precisely, a "label" is regarded as one of:
8607 ;; (i) a goto target like "foo:" - returns the symbol `goto-target';
8608 ;; (ii) A case label - either the entire construct "case FOO:", or just the
8609 ;; bare "case", should the colon be missing. We return t;
8610 ;; (iii) a keyword which needs a colon, like "default:" or "private:"; We
8611 ;; return t;
8612 ;; (iv) One of QT's "extended" C++ variants of
8613 ;; "private:"/"protected:"/"public:"/"more:" looking like "public slots:".
8614 ;; Returns the symbol `qt-2kwds-colon'.
8615 ;; (v) QT's construct "signals:". Returns the symbol `qt-1kwd-colon'.
8616 ;; (vi) One of the keywords matched by `c-opt-extra-label-key' (without any
8617 ;; colon). Currently (2006-03), this applies only to Objective C's
8618 ;; keywords "@private", "@protected", and "@public". Returns t.
8619 ;;
8620 ;; One of the things which will NOT be recognized as a label is a bit-field
8621 ;; element of a struct, something like "int foo:5".
8622 ;;
8623 ;; The end of the label is taken to be just after the colon, or the end of
8624 ;; the first submatch in `c-opt-extra-label-key'. The point is directly
8625 ;; after the end on return. The terminating char gets marked with
8626 ;; `c-decl-end' to improve recognition of the following declaration or
8627 ;; statement.
8628 ;;
8629 ;; If ASSUME-MARKUP is non-nil, it's assumed that the preceding
8630 ;; label, if any, has already been marked up like that.
8631 ;;
8632 ;; If PRECEDING-TOKEN-END is given, it should be the first position
8633 ;; after the preceding token, i.e. on the other side of the
8634 ;; syntactic ws from the point. Use a value less than or equal to
8635 ;; (point-min) if the point is at the first token in (the visible
8636 ;; part of) the buffer.
8637 ;;
8638 ;; The optional LIMIT limits the forward scan for the colon.
8639 ;;
8640 ;; This function records the ranges of the label symbols on
8641 ;; `c-record-ref-identifiers' if `c-record-type-identifiers' (!) is
8642 ;; non-nil.
8643 ;;
8644 ;; This function might do hidden buffer changes.
8647 label-end
8648 qt-symbol-idx
8649 macro-start ; if we're in one.
8650 label-type
8651 kwd)
8653 ;; "case" or "default" (Doesn't apply to AWK).
8656 ;; Record only the keyword itself for fontification, since in
8657 ;; case labels the following is a constant expression and not
8658 ;; a label.
8662 ;; Find the label end.
8666 ;; Stop on chars that aren't allowed in expressions,
8667 ;; and on operator chars that would be meaningless
8668 ;; there. FIXME: This doesn't cope with ?: operators.
8676 t)
8678 ;; It's an unfinished label. We consider the keyword enough
8679 ;; to recognize it as a label, so that it gets fontified.
8680 ;; Leave the point at the end of it, but don't put any
8681 ;; `c-decl-end' marker.
8683 t))))
8685 ;; @private, @protected, @public, in Objective C, or similar.
8688 ;; For a `c-opt-extra-label-key' match, we record the whole
8689 ;; thing for fontification. That's to get the leading '@' in
8690 ;; Objective-C protection labels fontified.
8697 ;; All other cases of labels.
8700 ;; A colon label must have something before the colon.
8703 ;; Check that we're not after a token that can't precede a label.
8705 ;; Trivially succeeds when there's no preceding token.
8706 ;; Succeeds when we're at a virtual semicolon.
8715 ;; Check if we're after a label, if we're after a closing
8716 ;; paren that belong to statement, and with
8717 ;; `c-label-prefix-re'. It's done in different order
8718 ;; depending on `assume-markup' since the checks have
8719 ;; different expensiveness.
8746 ;; Might be after another label, so check it recursively.
8750 ;; Essentially the same as the
8751 ;; `c-syntactic-re-search-forward' regexp below.
8757 ;; Note: the following should work instead of the
8758 ;; narrow-to-region above. Investigate why not,
8759 ;; sometime. ACM, 2006-03-31.
8760 ;; (c-syntactic-skip-backward "^-]:?;}=*/%&|,<>!@+"
8761 ;; macro-start t)
8763 ;; If the caller turned on recording for us,
8764 ;; it shouldn't apply when we check the
8765 ;; preceding label.
8766 c-record-type-identifiers)
8767 ;; A label can't start at a cpp directive. Check for
8768 ;; this, since c-forward-syntactic-ws would foul up on it.
8773 ;; Point is still at the beginning of the possible label construct.
8774 ;;
8775 ;; Check that the next nonsymbol token is ":", or that we're in one
8776 ;; of QT's "slots" declarations. Allow '(' for the sake of macro
8777 ;; arguments. FIXME: Should build this regexp from the language
8778 ;; constants.
8780 ;; public: protected: private:
8790 ;; QT double keyword like "protected slots:" or goto target.
8809 'qt-1kwd-colon
8821 ;; Check that `c-nonlabel-token-key' doesn't match anywhere.
8831 t)
8834 ;; Record the identifiers in the label for fontification, unless
8835 ;; it begins with `c-label-kwds' in which case the following
8836 ;; identifiers are part of a (constant) expression that
8837 ;; shouldn't be fontified.
8849 ;; Not a label.
8851 label-type))
8854 ;; Assuming the point is at the beginning of a token, try to move
8855 ;; forward to the end of the Objective-C directive that starts
8856 ;; there. Return t if a directive was fully recognized, otherwise
8857 ;; the point is moved as far as one could be successfully parsed and
8858 ;; nil is returned.
8859 ;;
8860 ;; This function records identifier ranges on
8861 ;; `c-record-type-identifiers' and `c-record-ref-identifiers' if
8862 ;; `c-record-type-identifiers' is non-nil.
8863 ;;
8864 ;; This function might do hidden buffer changes.
8867 start-char
8869 lim
8870 ;; Turn off recognition of angle bracket arglists while parsing
8871 ;; types here since the protocol reference list might then be
8872 ;; considered part of the preceding name or superclass-name.
8883 t)
8892 ;; Handle the name of the class itself.
8894 ;; (c-forward-token-2) ; 2006/1/13 This doesn't move if the token's
8895 ;; at EOB.
8902 ;; Look for ": superclass-name" or "( category-name )".
8913 ;; Look for a protocol reference list.
8919 t))))
8925 t)
8928 nil)))
8931 ;; Go to the first non-whitespace after the colon that starts a
8932 ;; multiple inheritance introduction. Optional LIM is the farthest
8933 ;; back we should search.
8934 ;;
8935 ;; This function might do hidden buffer changes.
8943 ;; Return nil if we aren't in a method definition, otherwise the
8944 ;; position of the initial [+-].
8945 ;;
8946 ;; This function might do hidden buffer changes.
8952 ))
8954 ;; Contributed by Kevin Ryde <user42@zip.com.au>.
8956 ;; Return non-nil if point is within a gcc \"asm\" block.
8957 ;;
8958 ;; This should be called with point inside an argument list.
8959 ;;
8960 ;; Only one level of enclosing parentheses is considered, so for
8961 ;; instance nil is returned when in a function call within an asm
8962 ;; operand.
8963 ;;
8964 ;; This function might do hidden buffer changes.
8976 a function.
8978 More precisely, being at the top-level means that point is either
8979 outside any enclosing block (such as a function definition), or
8980 directly inside a class, namespace or other block that contains
8981 another declaration level.
8983 If point is not at the top-level (e.g. it is inside a method
8984 definition), then nil is returned. Otherwise, if point is at a
8985 top-level not enclosed within a class definition, t is returned.
8986 Otherwise, a 2-vector is returned where the zeroth element is the
8987 buffer position of the start of the class declaration, and the first
8988 element is the buffer position of the enclosing class's opening
8989 brace.
8991 Note that this function might do hidden buffer changes. See the
8992 comment at the start of cc-engine.el for more info."
8993 ;; Note to maintainers: this function consumes a great mass of CPU cycles.
8994 ;; Its use should thus be minimized as far as possible.
9000 ;; Return non-nil if the point is in the region after the argument
9001 ;; list of a function and its opening brace (or semicolon in case it
9002 ;; got no body). If there are K&R style argument declarations in
9003 ;; that region, the point has to be inside the first one for this
9004 ;; function to recognize it.
9005 ;;
9006 ;; If successful, the point is moved to the first token after the
9007 ;; function header (see `c-forward-decl-or-cast-1' for details) and
9008 ;; the position of the opening paren of the function arglist is
9009 ;; returned.
9010 ;;
9011 ;; The point is clobbered if not successful.
9012 ;;
9013 ;; LIM is used as bound for backward buffer searches.
9014 ;;
9015 ;; This function might do hidden buffer changes.
9028 ;; There should not be a '=' or ',' between beg and the
9029 ;; start of the declaration since that means we were in the
9030 ;; "expression part" of the declaration.
9035 ;; Check that there's an arglist paren in the
9036 ;; declaration.
9039 ;; The declarator is a paren expression, so skip past it
9040 ;; so that we don't get stuck on that instead of the
9041 ;; function arglist.
9045 ;; Don't trip up on "operator ()".
9052 ;; Return the position of the first argument declaration if point is
9053 ;; inside a K&R style argument declaration list, nil otherwise.
9054 ;; `c-recognize-knr-p' is not checked. If LIM is non-nil, it's a
9055 ;; position that bounds the backward search for the argument list. This
9056 ;; function doesn't move point.
9057 ;;
9058 ;; Point must be within a possible K&R region, e.g. just before a top-level
9059 ;; "{". It must be outside of parens and brackets. The test can return
9060 ;; false positives otherwise.
9061 ;;
9062 ;; This function might do hidden buffer changes.
9065 ;; If we're in a macro, our search range is restricted to it. Narrow to
9066 ;; the searchable range.
9070 before-lparen after-rparen
9073 ; we give up.
9074 ids ; List of identifiers in the parenthesized list.
9075 id-start after-prec-token decl-or-cast decl-res
9076 c-last-identifier-range identifier-ok)
9079 ;; Search backwards for the defun's argument list. We give up if we
9080 ;; encounter a "}" (end of a previous defun) an "=" (which can't be in
9081 ;; a knr region) or BOB.
9082 ;;
9083 ;; The criterion for a paren structure being the arg list is:
9084 ;; o - there is non-WS stuff after it but before any "{"; AND
9085 ;; o - the token after it isn't a ";" AND
9086 ;; o - it is preceded by either an identifier (the function name) or
9087 ;; a macro expansion like "DEFUN (...)"; AND
9088 ;; o - its content is a non-empty comma-separated list of identifiers
9089 ;; (an empty arg list won't have a knr region).
9090 ;;
9091 ;; The following snippet illustrates these rules:
9092 ;; int foo (bar, baz, yuk)
9093 ;; int bar [] ;
9094 ;; int (*baz) (my_type) ;
9095 ;; int (*(* yuk) (void)) (void) ;
9096 ;; {
9097 ;;
9098 ;; Additionally, for a knr list to be recognized:
9099 ;; o - The identifier of each declarator up to and including the
9100 ;; one "near" point must be contained in the arg list.
9111 ; parens/brackets.
9115 ;; We're inside a paren. Could it be our argument list....?
9121 ;; FIXME!!! What about macros between the parens? 2007/01/20
9124 ;; It can't be the arg list if next token is ; or {
9129 ;; Is the thing preceding the list an identifier (the
9130 ;; function name), or a macro expansion?
9140 ;; Have we got a non-empty list of comma-separated
9141 ;; identifiers?
9154 ids)
9162 ;; Are all the identifiers in the k&r list up to the
9163 ;; current one also in the argument list?
9171 after-prec-token
9172 nil ; Or 'arglist ???
9173 nil)))
9178 nil))
9186 ids))
9192 identifier-ok))
9193 ;; ...Yes. We've identified the function's argument list.
9198 ;; ...No. The current parens aren't the function's arg list.
9205 ;; skip forward over conditional at point, including any predicate
9206 ;; statements in parentheses. No error checking is performed.
9207 ;;
9208 ;; This function might do hidden buffer changes.
9210 ;; else if()
9215 ;; do, else, try, finally
9220 ;; for, if, while, switch, catch, synchronized, foreach
9224 ;; If looking at the token after a conditional then return the
9225 ;; position of its start, otherwise return nil.
9226 ;;
9227 ;; This function might do hidden buffer changes.
9238 ;; If the point is after an operator identifier that isn't handled
9239 ;; like an ordinary symbol (i.e. like "operator =" in C++) then the
9240 ;; position of the start of that identifier is returned. nil is
9241 ;; returned otherwise. The point may be anywhere in the syntactic
9242 ;; whitespace after the last token of the operator identifier.
9243 ;;
9244 ;; This function might do hidden buffer changes.
9256 ;; Assuming point is at a brace that opens a statement block of some
9257 ;; kind, move to the proper anchor point for that block. It might
9258 ;; need to be adjusted further by c-add-stmt-syntax, but the
9259 ;; position at return is suitable as start position for that
9260 ;; function.
9261 ;;
9262 ;; This function might do hidden buffer changes.
9269 ;; Assuming point is at a brace that opens the block of a top level
9270 ;; declaration of some kind, move to the proper anchor point for
9271 ;; that block.
9272 ;;
9273 ;; This function might do hidden buffer changes.
9278 ;; Search forward for the end of the "header" of the current
9279 ;; declaration. That's the position where the definition body
9280 ;; starts, or the first variable initializer, or the ending
9281 ;; semicolon. I.e. search forward for the closest following
9282 ;; (syntactically relevant) '{', '=' or ';' token. Point is left
9283 ;; _after_ the first found token, or at point-max if none is found.
9284 ;;
9285 ;; This function might do hidden buffer changes.
9290 ;; In C++ we need to take special care to handle operator
9291 ;; tokens and those pesky template brackets.
9296 ;; Handle operator identifiers, i.e. ignore any
9297 ;; operator token preceded by "operator".
9304 t
9306 nil)))))
9315 ;; Go to the beginning of the current declaration, or the beginning
9316 ;; of the previous one if already at the start of it. Point won't
9317 ;; be moved out of any surrounding paren. Return a cons cell of the
9318 ;; form (MOVE . KNR-POS). MOVE is like the return value from
9319 ;; `c-beginning-of-statement-1'. If point skipped over some K&R
9320 ;; style argument declarations (and they are to be recognized) then
9321 ;; KNR-POS is set to the start of the first such argument
9322 ;; declaration, otherwise KNR-POS is nil. If LIM is non-nil, it's a
9323 ;; position that bounds the backward search.
9324 ;;
9325 ;; NB: Cases where the declaration continues after the block, as in
9326 ;; "struct foo { ... } bar;", are currently recognized as two
9327 ;; declarations, e.g. "struct foo { ... }" and "bar;" in this case.
9328 ;;
9329 ;; This function might do hidden buffer changes.
9335 ;; `c-beginning-of-statement-1' stops at a block start, but we
9336 ;; want to continue if the block doesn't begin a top level
9337 ;; construct, i.e. if it isn't preceded by ';', '}', ':', bob,
9338 ;; or an open paren.
9340 ;; Go back one "statement" each time round the loop until we're just
9341 ;; after a ;, }, or :, or at BOB or the start of a macro or start of
9342 ;; an ObjC method. This will move over a multiple declaration whose
9343 ;; components are comma separated.
9345 ;; Must check with c-opt-method-key in ObjC mode.
9356 ;; Check that we don't move from the first thing in a
9357 ;; macro to its header.
9363 move tentative-move))
9368 ;; Handle K&R argdecls. Back up after the "statement" jumped
9369 ;; over by `c-beginning-of-statement-1', unless it was the
9370 ;; function body, in which case we're sitting on the opening
9371 ;; brace now. Then test if we're in a K&R argdecl region and
9372 ;; that we started at the other side of the first argdecl in
9373 ;; it.
9383 'previous
9385 knr-argdecl-start))
9388 ;; `c-beginning-of-statement-1' counts each brace block as a separate
9389 ;; statement, so the result will be 'previous if we've moved over any.
9390 ;; So change our result back to 'same if necessary.
9391 ;;
9392 ;; If they were brace list initializers we might not have moved over a
9393 ;; declaration boundary though, so change it to 'same if we've moved
9394 ;; past a '=' before '{', but not ';'. (This ought to be integrated
9395 ;; into `c-beginning-of-statement-1', so we avoid this extra pass which
9396 ;; potentially can search over a large amount of text.). Take special
9397 ;; pains not to get mislead by C++'s "operator=", and the like.
9400 c++-template-syntax-table
9406 ; no more, or get to one which isn't an "operator ="
9409 c-overloadable-operators-regexp
9410 c-opt-op-identifier-prefix
9425 ;; Assuming point is at the start of a declaration (as detected by
9426 ;; e.g. `c-beginning-of-decl-1'), go to the end of it. Unlike
9427 ;; `c-beginning-of-decl-1', this function handles the case when a
9428 ;; block is followed by identifiers in e.g. struct declarations in C
9429 ;; or C++. If a proper end was found then t is returned, otherwise
9430 ;; point is moved as far as possible within the current sexp and nil
9431 ;; is returned. This function doesn't handle macros; use
9432 ;; `c-end-of-macro' instead in those cases.
9433 ;;
9434 ;; This function might do hidden buffer changes.
9437 c++-template-syntax-table
9445 ;; Stopped at the ';' in a K&R argdecl section which is
9446 ;; detected using the same criteria as in
9447 ;; `c-beginning-of-decl-1'. Move to the following block
9448 ;; start.
9452 ;; Encountered a block in the declaration. Jump over it.
9463 ;; Check for `c-opt-block-decls-with-vars-key'
9464 ;; before the first paren.
9467 c-opt-block-decls-with-vars-key
9469 lim t t t)
9472 ;; Check that the first following paren is
9473 ;; the block.
9475 lim t t t)
9477 ;; The declaration doesn't have any of the
9478 ;; `c-opt-block-decls-with-vars' keywords in the
9479 ;; beginning, so it ends here at the end of the block.
9487 nil)))
9490 ;; Assuming the point is at an open brace, check if it starts a
9491 ;; block that contains another declaration level, i.e. that isn't a
9492 ;; statement block or a brace list, and if so return non-nil.
9493 ;;
9494 ;; If the check is successful, the return value is the start of the
9495 ;; keyword that tells what kind of construct it is, i.e. typically
9496 ;; what `c-decl-block-key' matched. Also, if GOTO-START is set then
9497 ;; the point will be at the start of the construct, before any
9498 ;; leading specifiers, otherwise it's at the returned position.
9499 ;;
9500 ;; The point is clobbered if the check is unsuccessful.
9501 ;;
9502 ;; CONTAINING-SEXP is the position of the open of the surrounding
9503 ;; paren, or nil if none.
9504 ;;
9505 ;; The optional LIMIT limits the backward search for the start of
9506 ;; the construct. It's assumed to be at a syntactically relevant
9507 ;; position.
9508 ;;
9509 ;; If any template arglists are found in the searched region before
9510 ;; the open brace, they get marked with paren syntax.
9511 ;;
9512 ;; This function might do hidden buffer changes.
9519 ;; Could be at the end of a template arglist.
9527 ;; Skip back over noise clauses.
9529 c-opt-cpp-prefix
9539 t)
9541 nil))))
9543 ;; Note: Can't get bogus hits inside template arglists below since they
9544 ;; have gotten paren syntax above.
9546 ;; If `goto-start' is set we begin by searching for the
9547 ;; first possible position of a leading specifier list.
9548 ;; The `c-decl-block-key' search continues from there since
9549 ;; we know it can't match earlier.
9552 open-brace t t)
9554 t)
9555 t)
9561 ;; Exclude cases where we matched what would ordinarily
9562 ;; be a block declaration keyword, except where it's not
9563 ;; legal because it's part of a "compound keyword" like
9564 ;; "enum class". Of course, if c-after-brace-list-key
9565 ;; is nil, we can skip the test.
9575 ;; Found a keyword that can't be a type?
9578 ;; Can be a type too, in which case it's the return type of a
9579 ;; function (under the assumption that no declaration level
9580 ;; block construct starts with a type).
9583 ;; Jumped over a type, but it could be a declaration keyword
9584 ;; followed by the declared identifier that we've jumped over
9585 ;; instead (e.g. in "class Foo {"). If it indeed is a type
9586 ;; then we should be at the declarator now, so check for a
9587 ;; valid declarator start.
9588 ;;
9589 ;; Note: This doesn't cope with the case when a declared
9590 ;; identifier is followed by e.g. '(' in a language where '('
9591 ;; also might be part of a declarator expression. Currently
9592 ;; there's no such language.
9596 ;; In Pike a list of modifiers may be followed by a brace
9597 ;; to make them apply to many identifiers. Note that the
9598 ;; match data will be empty on return in this case.
9604 ;; Use this variant to avoid yet another special regexp.
9608 t)))
9610 ;; Got a match.
9613 ;; Back up over any preceding specifiers and their clauses
9614 ;; by going forward from `first-specifier-pos', which is the
9615 ;; earliest possible position where the specifier list can
9616 ;; start.
9622 ;; Accept any plain symbol token on the ground that
9623 ;; it's a specifier masked through a macro (just
9624 ;; like `c-forward-decl-or-cast-1' skip forward over
9625 ;; such tokens).
9626 ;;
9627 ;; Could be more restrictive wrt invalid keywords,
9628 ;; but that'd only occur in invalid code so there's
9629 ;; no use spending effort on it.
9635 ;; Can't parse a declaration preamble and is still
9636 ;; before `kwd-start'. That means `first-specifier-pos'
9637 ;; was in some earlier construct. Search again.
9641 ;; Got no preamble before the block declaration keyword.
9647 kwd-start)))
9650 ;; Check whether point is directly inside a brace block which is the brace
9651 ;; block of a class, struct, or union which is called NAME, a string.
9654 )
9658 ; *c-looking-at-decl-block
9659 ; containing-sexp goto-start &optional
9660 ; limit)
9663 nil)
9670 ;; Check if the closest containing paren sexp is a declaration
9671 ;; block, returning a 2 element vector in that case. Aref 0
9672 ;; contains the bufpos at boi of the class key line, and aref 1
9673 ;; contains the bufpos of the open brace. This function is an
9674 ;; obsolete wrapper for `c-looking-at-decl-block'.
9675 ;;
9676 ;; This function might do hidden buffer changes.
9684 nil))
9689 ;; Return the buffer position of the most enclosing decl-block brace (in the
9690 ;; sense of c-looking-at-decl-block) in the PAREN-STATE structure, or nil if
9691 ;; none was found.
9700 open-brace))
9703 ;; Return the position of the L-brace if point is inside a brace list
9704 ;; initialization of an array, etc. This is an approximate function,
9705 ;; designed for speed over accuracy. It will not find every bracelist, but
9706 ;; a non-nil result is reliable. We simply search for "= {" (naturally with
9707 ;; syntactic whitespace allowed). PAREN-STATE is the normal thing that it
9708 ;; is everywhere else.
9717 b-pos)))
9720 ;; We're at a "{" which might be the opening brace of a enum which is
9721 ;; strongly typed (by a ":" followed by a type). If this is the case, leave
9722 ;; point before the colon and return t. Otherwise leave point unchanged and return nil.
9723 ;; Match data will be clobbered.
9743 nil)
9745 t)
9747 t)
9751 t)))
9754 ;; We're at a "{". Move back to the enum-like keyword that starts this
9755 ;; declaration and return t, otherwise don't move and return nil.
9757 up-sexp-pos before-identifier)
9773 t)
9785 t)
9791 ;; return the buffer position of the beginning of the brace list
9792 ;; statement if we're inside a brace list, otherwise return nil.
9793 ;; CONTAINING-SEXP is the buffer pos of the innermost containing
9794 ;; paren. PAREN-STATE is the remainder of the state of enclosing
9795 ;; braces
9796 ;;
9797 ;; N.B.: This algorithm can potentially get confused by cpp macros
9798 ;; placed in inconvenient locations. It's a trade-off we make for
9799 ;; speed.
9800 ;;
9801 ;; This function might do hidden buffer changes.
9803 ;; This will pick up brace list declarations.
9807 ;; this will pick up array/aggregate init lists, even if they are nested.
9810 ;; Pike can have class definitions anywhere, so we must
9811 ;; check for the class key here.
9813 c-decl-block-key))
9814 bufpos braceassignp lim next-containing macro-start)
9816 containing-sexp)
9827 ;; We're in an in-expression block of some kind. Do not
9828 ;; check nesting. We deliberately set the limit to the
9829 ;; containing sexp, so that c-looking-at-inexpr-block
9830 ;; doesn't check for an identifier before it.
9832 ;; see if the open brace is preceded by = or [...] in
9833 ;; this statement, but watch out for operator=
9836 ;; Checks to do only on the first sexp before the brace.
9839 ;; In Java, an initialization brace list may follow
9840 ;; directly after "new Foo[]", so check for a "new"
9841 ;; earlier.
9847 ;; Carry on looking if this is an
9848 ;; identifier (may contain "." in Java)
9849 ;; or another "[]" sexp.
9852 ;; Checks to do on all sexps before the brace, up to the
9853 ;; beginning of the statement.
9861 ;; We've seen a =, but must check earlier tokens so
9862 ;; that it isn't something that should be ignored.
9868 ;; Check for operator =
9871 nil)
9872 ;; Check for `<opchar>= in Pike.
9875 ;; Special case for Pikes
9876 ;; `[]=, since '[' is not in
9877 ;; the punctuation class.
9880 nil)
9882 ;; make sure we're not in a C++ template
9883 ;; argument assignment
9895 ))))
9896 nil)
9903 ;; We've hit the beginning of the aggregate list.
9908 ;; Brace lists can't contain a semicolon, so we're done.
9913 ;; We've a macro whose expansion starts with the '{'.
9914 ;; Heuristically, if we have a ';' in it we've not got a
9915 ;; brace list, otherwise we have.
9922 containing-sexp nil)
9925 ;; Go up one level
9927 lim nil
9928 next-containing nil)))))
9930 bufpos))
9931 ))
9934 ;; If we're looking at the start of a pike-style list, i.e., `({Â })',
9935 ;; `([Â ])', `(<Â >)', etc., a cons of a cons of its starting and ending
9936 ;; positions and its entry in c-special-brace-lists is returned, nil
9937 ;; otherwise. The ending position is nil if the list is still open.
9938 ;; LIM is the limit for forward search. The point may either be at
9939 ;; the `(' or at the following paren character. Tries to check the
9940 ;; matching closer, but assumes it's correct if no balanced paren is
9941 ;; found (i.e. the case `({ ... } ... )' is detected as _not_ being
9942 ;; a special brace list).
9943 ;;
9944 ;; This function might do hidden buffer changes.
9949 inner-beg end type)
9964 nil)))))
9974 ;; Check balancing of the inner paren
9975 ;; below.
9978 t)
9979 ;; If the inner char isn't a paren then
9980 ;; we can't check balancing, so just
9981 ;; check the char before the outer
9982 ;; closing paren.
9997 ;; Return non-nil if between two statements or declarations, assuming
9998 ;; point is not inside a literal or comment.
9999 ;;
10000 ;; Obsolete - `c-at-statement-start-p' or `c-at-expression-start-p'
10001 ;; are recommended instead.
10002 ;;
10003 ;; This function might do hidden buffer changes.
10008 ;; Return non-nil if we're looking at the beginning of a block
10009 ;; inside an expression. The value returned is actually a cons of
10010 ;; either 'inlambda, 'inexpr-statement or 'inexpr-class and the
10011 ;; position of the beginning of the construct.
10012 ;;
10013 ;; LIM limits the backward search. CONTAINING-SEXP is the start
10014 ;; position of the closest containing list. If it's nil, the
10015 ;; containing paren isn't used to decide whether we're inside an
10016 ;; expression or not. If both LIM and CONTAINING-SEXP are used, LIM
10017 ;; needs to be farther back.
10018 ;;
10019 ;; If CHECK-AT-END is non-nil then extra checks at the end of the
10020 ;; brace block might be done. It should only be used when the
10021 ;; construct can be assumed to be complete, i.e. when the original
10022 ;; starting position was further down than that.
10023 ;;
10024 ;; This function might do hidden buffer changes.
10028 haskell-op-pos
10030 ;; Look at the character after point only as a last resort
10031 ;; when we can't disambiguate.
10034 ;; Search for a C++11 "->" which suggests a lambda declaration.
10065 ;; If the class definition is at the start of
10066 ;; a statement, we don't consider it an
10067 ;; in-expression class.
10075 ;; Also, in Pike we treat it as an
10076 ;; in-expression class if it's used in an
10077 ;; object clone expression.
10098 nil)
10108 ;; Accept several square bracket sexps for
10109 ;; Java array initializations.
10116 ;; C++11 lambda function declaration
10132 block-follows
10144 block-follows
10145 containing-sexp
10159 nil
10162 res))))
10165 ;; Returns non-nil if we're looking at the end of an in-expression
10166 ;; block, otherwise the same as `c-looking-at-inexpr-block'.
10167 ;; PAREN-STATE is the paren state relevant at the current position.
10168 ;;
10169 ;; This function might do hidden buffer changes.
10171 ;; We currently only recognize a block.
10174 containing-sexp)
10183 paren-state)
10184 containing-sexp)))))
10187 ;; Return non-nil if we're at the opening "[" of the capture list of a C++
10188 ;; lambda function, nil otherwise.
10199 ;; Is there a "virtual semicolon" at POS or point?
10200 ;; (See cc-defs.el for full details of "virtual semicolons".)
10201 ;;
10202 ;; This is true when point is at the last non syntactic WS position on the
10203 ;; line, there is a macro call last on the line, and this particular macro's
10204 ;; name is defined by the regexp `c-vs-macro-regexp' as not needing a
10205 ;; semicolon.
10213 c-macro-with-semi-re
10216 ;; Check we've got nothing after this except comments and empty lines
10217 ;; joined by escaped EOLs.
10230 nil)
10238 ;; Check for one of the listed macros being before point.
10242 t))
10250 \f
10251 ;; `c-guess-basic-syntax' and the functions that precedes it below
10252 ;; implements the main decision tree for determining the syntactic
10253 ;; analysis of the current line of code.
10255 ;; Dynamically bound to t when `c-guess-basic-syntax' is called during
10256 ;; auto newline analysis.
10260 ;; BRACEPOS is the position of a brace in a construct like "namespace
10261 ;; Bar {". Return the anchor point in this construct; this is the
10262 ;; earliest symbol on the brace's line which isn't earlier than
10263 ;; "namespace".
10264 ;;
10265 ;; Currently (2007-08-17), "like namespace" means "matches
10266 ;; c-other-block-decl-kwds". It doesn't work with "class" or "struct"
10267 ;; or anything like that.
10277 ;; A simple function to prepend a new syntax element to
10278 ;; `c-syntactic-context'. Using `setq' on it is unsafe since it
10279 ;; should always be dynamically bound but since we read it first
10280 ;; we'll fail properly anyway if this function is misused.
10282 c-syntactic-context)))
10285 ;; Like `c-add-syntax' but appends to the end of the syntax list.
10286 ;; (Normally not necessary.)
10291 syntax-extra-args
10292 stop-at-boi-only
10293 containing-sexp
10294 paren-state)
10295 ;; Add the indicated SYNTAX-SYMBOL to `c-syntactic-context', extending it as
10296 ;; needed with further syntax elements of the types `substatement',
10297 ;; `inexpr-statement', `arglist-cont-nonempty', `statement-block-intro', and
10298 ;; `defun-block-intro'.
10299 ;;
10300 ;; Do the generic processing to anchor the given syntax symbol on
10301 ;; the preceding statement: Skip over any labels and containing
10302 ;; statements on the same line, and then search backward until we
10303 ;; find a statement or block start that begins at boi without a
10304 ;; label or comment.
10305 ;;
10306 ;; Point is assumed to be at the prospective anchor point for the
10307 ;; given SYNTAX-SYMBOL. More syntax entries are added if we need to
10308 ;; skip past open parens and containing statements. Most of the added
10309 ;; syntax elements will get the same anchor point - the exception is
10310 ;; for an anchor in a construct like "namespace"[*] - this is as early
10311 ;; as possible in the construct but on the same line as the {.
10312 ;;
10313 ;; [*] i.e. with a keyword matching c-other-block-decl-kwds.
10314 ;;
10315 ;; SYNTAX-EXTRA-ARGS are a list of the extra arguments for the
10316 ;; syntax symbol. They are appended after the anchor point.
10317 ;;
10318 ;; If STOP-AT-BOI-ONLY is nil, we can stop in the middle of the line
10319 ;; if the current statement starts there.
10320 ;;
10321 ;; Note: It's not a problem if PAREN-STATE "overshoots"
10322 ;; CONTAINING-SEXP, i.e. contains info about parens further down.
10323 ;;
10324 ;; This function might do hidden buffer changes.
10327 ;; This is by far the most common case, so let's give it special
10328 ;; treatment.
10333 ;; Set when we're on a label, so that we don't stop there.
10334 ;; FIXME: To be complete we should check if we're on a label
10335 ;; now at the start.
10336 on-label)
10338 ;; Use point as the anchor point for "namespace", "extern", etc.
10342 syntax-extra-args)
10344 ;; Loop while we have to back out of containing blocks.
10349 ;; Loop while we have to back up statements.
10351 on-label
10354 ;; Skip past any comments that stands between the
10355 ;; statement start and boi.
10363 ;; Skip to the beginning of this statement or backward
10364 ;; another one.
10372 ;; If we didn't move we're at the start of a block and
10373 ;; have to continue outside it.
10382 ;; Special case to avoid deeper and deeper indentation
10383 ;; of "else if" clauses.
10384 )
10389 ;; If stop-at-boi-only is nil, we shouldn't back up
10390 ;; over previous or containing statements to try to
10391 ;; reach boi, so go back to the last position and
10392 ;; exit.
10399 ;; If we've moved into another statement then we
10400 ;; should no longer try to stop in the middle of a
10401 ;; line.
10404 ;; Record this as a substatement if we skipped up one
10405 ;; level.
10408 )))
10410 containing-sexp)
10412 ;; Now we have to go out of this block.
10415 ;; Don't stop in the middle of a special brace list opener
10416 ;; like "({".
10428 ;; Analyze the construct in front of the block we've stepped out
10429 ;; from and add the right syntactic element for it.
10432 step-type)
10435 ;; Stepped out of a parenthesis block, so we're in an
10436 ;; expression now.
10448 ;; Stepped out of an in-expression statement. This
10449 ;; syntactic element won't get an anchor pos.
10452 ;; A parenthesis normally belongs to an arglist.
10460 on-label nil))
10462 ;; Stepped out of a brace block.
10474 containing-sexp)))
10476 'defun-block-intro
10478 nil))
10482 c-other-decl-block-key-in-symbols-alist))
10489 ;; Always done if the open brace was at boi. The
10490 ;; c-beginning-of-statement-1 call above is necessary
10491 ;; anyway, to decide the type of block-intro to add.
10494 ))
10496 ;; Fill in the current point as the anchor for all the symbols
10497 ;; added above.
10506 )))
10509 containing-decl-open
10510 containing-decl-start
10511 containing-decl-kwd
10512 paren-state)
10513 ;; The inclass and class-close syntactic symbols are added in
10514 ;; several places and some work is needed to fix everything.
10515 ;; Therefore it's collected here.
10516 ;;
10517 ;; This function might do hidden buffer changes.
10522 containing-decl-open)
10524 ;; Ought to use `c-add-stmt-syntax' instead of backing up to boi
10525 ;; here, but we have to do like this for compatibility.
10535 char-after-ip
10536 beg-of-same-or-containing-stmt
10537 containing-sexp
10538 paren-state)
10539 ;; This function contains the decision tree reached through both
10540 ;; cases 18 and 10. It's a continued statement or top level
10541 ;; construct of some kind.
10542 ;;
10543 ;; This function might do hidden buffer changes.
10550 ;; (CASE A removed.)
10551 ;; CASE B: open braces for class or brace-lists
10558 ;; CASE B.1: class-open
10565 ;; CASE B.2: brace-list-open
10570 indent-point t t t)))
10571 ;; The most semantically accurate symbol here is
10572 ;; brace-list-open, but we normally report it simply as a
10573 ;; statement-cont. The reason is that one normally adjusts
10574 ;; brace-list-open for brace lists as top-level constructs,
10575 ;; and brace lists inside statements is a completely different
10576 ;; context. C.f. case 5A.3.
10579 ;; Turn off the dwim above when we're
10580 ;; analyzing the nature of the brace
10581 ;; for the auto newline feature.
10582 'brace-list-open
10584 nil nil
10585 containing-sexp paren-state))
10587 ;; CASE B.3: The body of a function declared inside a normal
10588 ;; block. Can occur e.g. in Pike and when using gcc
10589 ;; extensions, but watch out for macros followed by blocks.
10590 ;; C.f. cases E, 16F and 17G.
10596 ;; Turn off recognition of constructs that lacks a
10597 ;; type in this case, since that's more likely to be
10598 ;; a macro followed by a block.
10601 containing-sexp paren-state))
10603 ;; CASE B.4: Continued statement with block open. The most
10604 ;; accurate analysis is perhaps `statement-cont' together with
10605 ;; `block-open' but we play DWIM and use `substatement-open'
10606 ;; instead. The rationale is that this typically is a macro
10607 ;; followed by a block which makes it very similar to a
10608 ;; statement with a substatement block.
10611 containing-sexp paren-state))
10612 ))
10614 ;; CASE C: iostream insertion or extraction operator
10618 ;; If there is no preceding streamop in the statement
10619 ;; then indent this line as a normal statement-cont.
10623 t))))
10625 ;; CASE E: In the "K&R region" of a function declared inside a
10626 ;; normal block. C.f. case B.3.
10628 ;; Check that the next token is a '{'. This works as
10629 ;; long as no language that allows nested function
10630 ;; definitions allows stuff like member init lists, K&R
10631 ;; declarations or throws clauses there.
10632 ;;
10633 ;; Note that we do a forward search for something ahead
10634 ;; of the indentation line here. That's not good since
10635 ;; the user might not have typed it yet. Unfortunately
10636 ;; it's exceedingly tricky to recognize a function
10637 ;; prototype in a code block without resorting to this.
10645 ;; Turn off recognition of constructs that lacks a
10646 ;; type in this case, since that's more likely to be
10647 ;; a macro followed by a block.
10650 containing-sexp paren-state))
10652 ;;CASE F: continued statement and the only preceding items are
10653 ;;annotations.
10661 t)
10668 ;; CASE G: a template list continuation?
10669 ;; Mostly a duplication of case 5D.3 to fix templates-19:
10687 ;; In a nested template arglist.
10693 ;; FIXME: Should use c-add-stmt-syntax, but it's not yet
10694 ;; template aware.
10697 ;; CASE D: continued statement.
10701 containing-sexp paren-state))
10702 )))
10704 ;; The next autoload was added by RMS on 2005/8/9 - don't know why (ACM,
10705 ;; 2005/11/29).
10706 ;;;###autoload
10708 "Return the syntactic context of the current line."
10714 ;; A whole ugly bunch of various temporary variables. Have
10715 ;; to declare them here since it's not possible to declare
10716 ;; a variable with only the scope of a cond test and the
10717 ;; following result clauses, and most of this function is a
10718 ;; single gigantic cond. :P
10719 literal char-before-ip before-ws-ip char-after-ip macro-start
10720 in-macro-expr c-syntactic-context placeholder c-in-literal-cache
10721 step-type tmpsymbol keyword injava-inher special-brace-list tmp-pos
10722 containing-<
10723 ;; The following record some positions for the containing
10724 ;; declaration block if we're directly within one:
10725 ;; `containing-decl-open' is the position of the open
10726 ;; brace. `containing-decl-start' is the start of the
10727 ;; declaration. `containing-decl-kwd' is the keyword
10728 ;; symbol of the keyword that tells what kind of block it
10729 ;; is.
10730 containing-decl-open
10731 containing-decl-start
10732 containing-decl-kwd
10733 ;; The open paren of the closest surrounding sexp or nil if
10734 ;; there is none.
10735 containing-sexp
10736 ;; The position after the closest preceding brace sexp
10737 ;; (nested sexps are ignored), or the position after
10738 ;; `containing-sexp' if there is none, or (point-min) if
10739 ;; `containing-sexp' is nil.
10740 lim
10741 ;; The paren state outside `containing-sexp', or at
10742 ;; `indent-point' if `containing-sexp' is nil.
10744 ;; There's always at most one syntactic element which got
10745 ;; an anchor pos. It's stored in syntactic-relpos.
10746 syntactic-relpos
10749 ;; Check if we're directly inside an enclosing declaration
10750 ;; level block.
10759 containing-sexp)
10760 t)))
10763 containing-sexp nil)
10768 ;; Init some position variables.
10781 ;; Ignore balanced paren. The next entry
10782 ;; can't be another one.
10785 ;; If there is no surrounding open paren then
10786 ;; put the last balanced pair back on paren-state.
10788 containing-sexp nil)))
10792 ;; If we're in a parenthesis list then ',' delimits the
10793 ;; "statements" rather than being an operator (with the
10794 ;; exception of the "for" clause). This difference is
10795 ;; typically only noticeable when statements are used in macro
10796 ;; arglists.
10800 ;; cache char before and after indent point, and move point to
10801 ;; the most likely position to perform the majority of tests
10810 ;; are we in a literal?
10813 ;; now figure out syntactic qualities of the current line
10816 ;; CASE 1: in a string.
10820 ;; CASE 2: in a C or C++ style comment.
10822 ;; This is a kludge for XEmacs where we use
10823 ;; `buffer-syntactic-context', which doesn't correctly
10824 ;; recognize "\*/" to end a block comment.
10825 ;; `parse-partial-sexp' which is used by
10826 ;; `c-literal-limits' will however do that in most
10827 ;; versions, which results in that we get nil from
10828 ;; `c-literal-limits' even when `c-in-literal' claims
10829 ;; we're inside a comment.
10833 ;; CASE 3: in a cpp preprocessor macro continuation.
10843 ;; If at the beginning of the body of a #define
10844 ;; directive then analyze as cpp-define-intro
10845 ;; only. Go on with the syntactic analysis
10846 ;; otherwise. in-macro-expr is set if we're in a
10847 ;; cpp expression, i.e. before the #define body
10848 ;; or anywhere in a non-#define directive.
10855 nil)))))
10859 ;; CASE 11: an else clause?
10863 containing-sexp paren-state))
10865 ;; CASE 12: while closure of a do/while construct?
10873 containing-sexp paren-state))
10875 ;; CASE 13: A catch or finally clause? This case is simpler
10876 ;; than if-else and do-while, because a block is required
10877 ;; after every try, catch and finally.
10885 t)
10889 t)
10892 t))
10897 containing-sexp paren-state))
10899 ;; CASE 18: A substatement we can recognize by keyword.
10909 ;; Ought to cache the result from the
10910 ;; c-beginning-of-statement-1 calls here.
10920 ;; Step up to the containing statement if we
10921 ;; stayed in the same one.
10929 ;; There was no containing statement after all.
10931 placeholder))
10933 ;; Require a parenthesis after these keywords.
10934 ;; Necessary to catch e.g. synchronized in Java,
10935 ;; which can be used both as statement and
10936 ;; modifier.
10942 ;; CASE 18A: Simple substatement.
10948 containing-sexp paren-state))
10954 containing-sexp paren-state))
10957 containing-sexp paren-state))))
10959 ;; CASE 18B: Some other substatement. This is shared
10960 ;; with case 10.
10962 char-after-ip
10963 placeholder
10964 lim
10965 paren-state)))
10967 ;; CASE 14: A case or default label
10972 containing-sexp
10985 containing-sexp))
10988 ;; Got a bogus label at the top level. In lack of better
10989 ;; alternatives, anchor it on (point-min).
10992 ;; CASE 15: any other label
11000 containing-decl-open
11001 containing-decl-start
11002 containing-decl-kwd
11003 paren-state))
11004 ;; Append access-label with the same anchor point as
11005 ;; inclass gets.
11011 containing-sexp))
11016 ;; If the surrounding statement is a switch then
11017 ;; let's analyze all labels as switch labels, so
11018 ;; that they get lined up consistently.
11019 'case-label
11025 ;; A label on the top level. Treat it as a class
11026 ;; context. (point-min) is the closest we get to the
11027 ;; class open brace.
11030 ;; CASE 4: In-expression statement. C.f. cases 7B, 16A and
11031 ;; 17E.
11034 containing-sexp
11035 ;; Have to turn on the heuristics after
11036 ;; the point even though it doesn't work
11037 ;; very well. C.f. test case class-16.pike.
11038 t))
11043 ;; It's a statement block or an anonymous class.
11045 ;; It's a Pike lambda. Check whether we are between the
11046 ;; lambda keyword and the argument list or at the defun
11047 ;; opener.
11049 'inline-open
11055 paren-state)
11059 ;; CASE 5: Line is inside a declaration level block or at top level.
11063 ;; CASE 5A: we are looking at a defun, brace list, class,
11064 ;; or inline-inclass method opening brace
11071 ;; CASE 5A.1: Non-class declaration block open.
11086 ;; Special case for extern-lang-open.
11087 'extern-lang-open
11089 nil t containing-sexp paren-state))
11091 ;; CASE 5A.2: we are looking at a class opening brace
11100 ;; CASE 5A.3: brace list open
11128 ))))
11132 ;; We're in Java and have found that the open brace
11133 ;; belongs to a "new Foo[]" initialization list,
11134 ;; which means the brace list is part of an
11135 ;; expression and not a top level definition. We
11136 ;; therefore treat it as any topmost continuation
11137 ;; even though the semantically correct symbol still
11138 ;; is brace-list-open, on the same grounds as in
11139 ;; case B.2.
11145 ;; CASE 5A.4: inline defun open
11151 containing-decl-open
11152 containing-decl-start
11153 containing-decl-kwd
11154 paren-state))
11156 ;; CASE 5A.5: ordinary defun open
11167 ;; Bogus to use bol here, but it's the legacy. (Resolved,
11168 ;; 2007-11-09)
11169 ))))
11171 ;; CASE 5R: Member init list. (Used to be part of CASE 5B.1)
11172 ;; Note there is no limit on the backward search here, since member
11173 ;; init lists can, in practice, be very large.
11180 ;; There is no preceding member init clause.
11181 ;; Indent relative to the beginning of indentation
11182 ;; for the topmost-intro line that contains the
11183 ;; prototype's open paren.
11186 ;; Indent relative to the first member init clause.
11191 ;; CASE 5B: After a function header but before the body (or
11192 ;; the ending semicolon if there's no body).
11199 ;; CASE 5B.1: Member init list.
11201 ;; There is no preceding member init clause.
11202 ;; Indent relative to the beginning of indentation
11203 ;; for the topmost-intro line that contains the
11204 ;; prototype's open paren.
11208 ;; CASE 5B.2: K&R arg decl intro
11215 containing-decl-open
11216 containing-decl-start
11217 containing-decl-kwd
11218 paren-state)))
11220 ;; CASE 5B.4: Nether region after a C++ or Java func
11221 ;; decl, which could include a `throws' declaration.
11225 )))
11227 ;; CASE 5C: inheritance line. could be first inheritance
11228 ;; line, or continuation of a multiple inheritance
11236 ;; watch out for scope operator
11241 )))
11250 ;; for Java
11255 cont done)
11260 done t))
11264 )
11266 injava-inher)
11269 ))
11272 ;; CASE 5C.1: non-hanging colon on an inher intro
11276 ;; don't add inclass symbol since relative point already
11277 ;; contains any class offset
11278 )
11280 ;; CASE 5C.2: hanging colon on an inher intro
11286 containing-decl-open
11287 containing-decl-start
11288 containing-decl-kwd
11289 paren-state)))
11291 ;; CASE 5C.3: in a Java implements/extends
11305 )))
11307 ;; CASE 5C.4: a continued inheritance line
11311 ;; don't add inclass symbol since relative point already
11312 ;; contains any class offset
11313 )))
11315 ;; CASE 5P: AWK pattern or function or continuation
11316 ;; thereof.
11322 'topmost-intro-cont
11324 nil nil
11325 containing-sexp paren-state))
11327 ;; CASE 5D: this could be a top-level initialization, a
11328 ;; member init list continuation, or a template argument
11329 ;; list continuation.
11331 ;; Note: We use the fact that lim is always after any
11332 ;; preceding brace sexp.
11342 t))
11347 t)
11353 nil)
11354 t))
11356 ;; NB: No c-after-special-operator-id stuff in this
11357 ;; clause - we assume only C++ needs it.
11362 ;; CASE 5D.3: perhaps a template list continuation?
11377 ;; In a nested template arglist.
11383 ;; FIXME: Should use c-add-stmt-syntax, but it's not yet
11384 ;; template aware.
11387 ;; CASE 5D.4: perhaps a multiple inheritance line?
11399 t)
11404 ;; CASE 5D.5: Continuation of the "expression part" of a
11405 ;; top level construct. Or, perhaps, an unrecognized construct.
11417 ;; A preceding comma at the top level means that a
11418 ;; new variable declaration starts here. Use
11419 ;; topmost-intro-cont for it, for consistency with
11420 ;; the first variable declaration. C.f. case 5N.
11423 nil nil containing-sexp paren-state))
11424 ))
11426 ;; CASE 5F: Close of a non-class declaration level block.
11430 ;; This is inconsistent: Should use `containing-decl-open'
11431 ;; here if it's at boi, like in case 5J.
11435 ;; Special case for compatibility with the
11436 ;; extern-lang syntactic symbols.
11437 'extern-lang-close
11440 nil t
11442 paren-state))
11444 ;; CASE 5G: we are looking at the brace which closes the
11445 ;; enclosing nested class decl
11450 containing-decl-open
11451 containing-decl-start
11452 containing-decl-kwd
11453 paren-state))
11455 ;; CASE 5H: we could be looking at subsequent knr-argdecls
11462 ;; Do an extra check to avoid tripping up on
11463 ;; statements that occur in invalid contexts
11464 ;; (e.g. in macro bodies where we don't really
11465 ;; know the context of what we're looking at).
11472 ;; CASE 5I: ObjC method definition.
11477 ;; Handle the case when it's the first (non-comment)
11478 ;; thing in the buffer. Can't look for a 'same return
11479 ;; value from cbos1 since ObjC directives currently
11480 ;; aren't recognized fully, so that we get 'same
11481 ;; instead of 'previous if it moved over a preceding
11482 ;; directive.
11486 ;; CASE 5N: At a variable declaration that follows a class
11487 ;; definition or some other block declaration that doesn't
11488 ;; end at the closing '}'. C.f. case 5D.5.
11497 ;; Speed up the backward search a bit.
11502 ;; The '}' is unbalanced.
11503 nil
11508 containing-sexp paren-state))
11510 ;; NOTE: The point is at the end of the previous token here.
11512 ;; CASE 5J: we are at the topmost level, make
11513 ;; sure we skip back past any access specifiers
11515 ;; A macro continuation line is never at top level.
11537 nil)))))
11538 ;; For historic reasons we anchor at bol of the last
11539 ;; line of the previous declaration. That's clearly
11540 ;; highly bogus and useless, and it makes our lives hard
11541 ;; to remain compatible. :P
11552 ;; Special case for compatibility with the
11553 ;; extern-lang syntactic symbols.
11554 'inextern-lang
11557 nil t
11559 paren-state))
11561 containing-decl-open
11562 containing-decl-start
11563 containing-decl-kwd
11564 paren-state)))
11566 macro-start
11571 ;; CASE 5K: we are at an ObjC method definition
11572 ;; continuation line.
11581 ;; CASE 5L: we are at the first argument of a template
11582 ;; arglist that begins on the previous line.
11590 ;; CASE 5Q: we are at a statement within a macro.
11595 ;;CASE 5N: We are at a topmost continuation line and the only
11596 ;;preceding items are annotations.
11603 t)
11609 ;; CASE 5M: we are at a topmost continuation line
11620 ))
11622 ;; (CASE 6 has been removed.)
11624 ;; CASE 7: line is an expression, not a statement. Most
11625 ;; likely we are either in a function prototype or a function
11626 ;; call argument list
11634 ;; CASE 7A: we are looking at the arglist closing paren.
11635 ;; C.f. case 7F.
11647 paren-state))
11649 ;; CASE 7B: Looking at the opening brace of an
11650 ;; in-expression block or brace list. C.f. cases 4, 16A
11651 ;; and 17E.
11655 paren-state))
11659 placeholder
11662 paren-state)
11663 containing-sexp)))
11664 ;; placeholder is nil if it's a block directly in
11665 ;; a function arglist. That makes us skip out of
11666 ;; this case.
11667 )))
11672 paren-state)
11676 ;; CASE 7C: we are looking at the first argument in an empty
11677 ;; argument list. Use arglist-close if we're actually
11678 ;; looking at a close paren or bracket.
11690 paren-state))
11692 ;; CASE 7D: we are inside a conditional test clause. treat
11693 ;; these things as statements
11703 ))
11705 ;; CASE 7E: maybe a continued ObjC method call. This is the
11706 ;; case when we are inside a [] bracketed exp, and what
11707 ;; precede the opening bracket is not an identifier.
11715 )))
11717 ;; CASE 7F: we are looking at an arglist continuation line,
11718 ;; but the preceding argument is on the same line as the
11719 ;; opening paren. This case includes multi-line
11720 ;; mathematical paren groupings, but we could be on a
11721 ;; for-list continuation line. C.f. case 7A.
11738 paren-state))
11740 ;; CASE 7G: we are looking at just a normal arglist
11741 ;; continuation line
11744 ))
11746 ;; CASE 8: func-local multi-inheritance line
11756 ;; CASE 8A: non-hanging colon on an inher intro
11761 ;; CASE 8B: hanging colon on an inher intro
11765 ;; CASE 8C: a continued inheritance line
11769 )))
11771 ;; CASE 9: we are inside a brace-list
11781 ;; CASE 9A: In the middle of a special brace list opener.
11804 ;; CASE 9B: brace-list-close brace
11806 ;; Check special brace list closer.
11813 ;; We were between the special close char and the `)'.
11816 ;; We were before the special close char.
11820 ;; Normal brace list check.
11831 ;; Prepare for the rest of the cases below by going to the
11832 ;; token following the opening brace
11845 ;; CASE 9C: we're looking at the first line in a brace-list
11856 ;; CASE 9D: this is just a later brace-list-entry or
11857 ;; brace-entry-open
11866 ))
11867 ))))
11869 ;; CASE 10: A continued statement or top level construct.
11879 ;; This is shared with case 18.
11881 char-after-ip
11882 placeholder
11883 containing-sexp
11884 paren-state))
11886 ;; CASE 16: block close brace, possibly closing the defun or
11887 ;; the class
11889 ;; From here on we have the next containing sexp in lim.
11894 ;; CASE 16E: Closing a statement block? This catches
11895 ;; cases where it's preceded by a statement keyword,
11896 ;; which works even when used in an "invalid" context,
11897 ;; e.g. a macro argument.
11902 ;; CASE 16A: closing a lambda defun or an in-expression
11903 ;; block? C.f. cases 4, 7B and 17E.
11906 nil))
11908 'inline-close
11918 paren-state)
11922 ;; CASE 16B: does this close an inline or a function in
11923 ;; a non-class declaration level block?
11930 nil))
11940 ;; CASE 16F: Can be a defun-close of a function declared
11941 ;; in a statement block, e.g. in Pike or when using gcc
11942 ;; extensions, but watch out for macros followed by
11943 ;; blocks. Let it through to be handled below.
11944 ;; C.f. cases B.3 and 17G.
11950 ;; Turn off recognition of constructs that
11951 ;; lacks a type in this case, since that's more
11952 ;; likely to be a macro followed by a block.
11959 ;; CASE 16C: If there is an enclosing brace then this is
11960 ;; a block close since defun closes inside declaration
11961 ;; level blocks have been handled above.
11963 ;; If the block is preceded by a case/switch label on
11964 ;; the same line, we anchor at the first preceding label
11965 ;; at boi. The default handling in c-add-stmt-syntax
11966 ;; really fixes it better, but we do like this to keep
11967 ;; the indentation compatible with version 5.28 and
11968 ;; earlier. C.f. case 17H.
11975 ;; c-backward-to-block-anchor not necessary here; those
11976 ;; situations are handled in case 16E above.
11979 ;; CASE 16D: Only top level defun close left.
11985 paren-state))
11986 ))
11988 ;; CASE 19: line is an expression, not a statement, and is directly
11989 ;; contained by a template delimiter. Most likely, we are in a
11990 ;; template arglist within a statement. This case is based on CASE
11991 ;; 7. At some point in the future, we may wish to create more
11992 ;; syntactic symbols such as `template-intro',
11993 ;; `template-cont-nonempty', etc., and distinguish between them as we
11994 ;; do for `arglist-intro' etc. (2009-12-07).
12000 ; '<') before indent-point.
12008 paren-state))
12010 ;; CASE 17: Statement or defun catchall.
12013 ;; Back up statements until we find one that starts at boi.
12016 containing-sexp)))
12020 nil)
12025 ;; CASE 17B: continued statement
12029 containing-sexp paren-state))
12031 ;; CASE 17A: After a case/default label?
12039 'statement-case-open
12041 nil t containing-sexp paren-state))
12043 ;; CASE 17D: any old statement
12050 containing-sexp paren-state)
12054 ;; CASE 17I: Inside a substatement block.
12056 ;; The following tests are all based on containing-sexp.
12058 ;; From here on we have the next containing sexp in lim.
12063 lim paren-state)
12067 ;; CASE 17E: first statement in an in-expression block.
12068 ;; C.f. cases 4, 7B and 16A.
12071 nil))
12073 'defun-block-intro
12082 paren-state)
12088 ;; CASE 17F: first statement in an inline, or first
12089 ;; statement in a top-level defun. we can tell this is it
12090 ;; if there are no enclosing braces that haven't been
12091 ;; narrowed out by a class (i.e. don't use bod here).
12094 paren-state)))
12100 nil))))
12105 ;; CASE 17G: First statement in a function declared inside
12106 ;; a normal block. This can occur in Pike and with
12107 ;; e.g. the gcc extensions, but watch out for macros
12108 ;; followed by blocks. C.f. cases B.3 and 16F.
12114 ;; Turn off recognition of constructs that lacks
12115 ;; a type in this case, since that's more likely
12116 ;; to be a macro followed by a block.
12122 lim paren-state))
12124 ;; CASE 17H: First statement in a block.
12126 ;; If the block is preceded by a case/switch label on the
12127 ;; same line, we anchor at the first preceding label at
12128 ;; boi. The default handling in c-add-stmt-syntax is
12129 ;; really fixes it better, but we do like this to keep the
12130 ;; indentation compatible with version 5.28 and earlier.
12131 ;; C.f. case 16C.
12138 ;; c-backward-to-block-anchor not necessary here; those
12139 ;; situations are handled in case 17I above.
12141 lim paren-state))
12144 ))
12145 )
12147 ;; now we need to look at any modifiers
12151 ;; are we looking at a comment only line?
12156 ;; we might want to give additional offset to friends (in C++).
12161 ;; Set syntactic-relpos.
12167 nil)
12168 t))
12171 ;; Start of or a continuation of a preprocessor directive?
12186 ;; If inside a cpp expression, i.e. anywhere in a
12187 ;; cpp directive except a #define body, we only let
12188 ;; through the syntactic analysis that is internal
12189 ;; in the expression. That means the arglist
12190 ;; elements, if they are anchored inside the cpp
12191 ;; expression.
12200 ;; Inside a #define body and the syntactic analysis is
12201 ;; anchored on the start of the #define. In this case
12202 ;; we add cpp-define-intro to get the extra
12203 ;; indentation of the #define body.
12206 ;; return the syntax
12207 c-syntactic-context)))
12209 \f
12210 ;; Indentation calculation.
12213 ;; offset can be a number, a function, a variable, a list, or one of
12214 ;; the symbols + or -
12215 ;;
12216 ;; This function might do hidden buffer changes.
12235 langelem symbol))
12241 offset symbol)
12242 nil)
12256 Error evaluating offset %S for %s: \
12257 Cannot combine absolute offset %S with relative %S in `%s' method"
12263 Error evaluating offset %S for %s: \
12264 Cannot combine relative offset %S with absolute %S in `%s' method"
12269 res))
12287 Error evaluating offset %S for %s: \
12288 Cannot combine absolute offsets %S and %S in `add' method"
12292 res))
12301 res))))
12308 nil))))
12312 res
12314 offset symbol res)
12315 nil)))
12318 ;; Get offset from LANGELEM which is a list beginning with the
12319 ;; syntactic symbol and followed by any analysis data it provides.
12320 ;; That data may be zero or more elements, but if at least one is
12321 ;; given then the first is the anchor position (or nil). The symbol
12322 ;; is matched against `c-offsets-alist' and the offset calculated
12323 ;; from that is returned.
12324 ;;
12325 ;; This function might do hidden buffer changes.
12335 offset
12339 ))
12342 ;; This is a compatibility wrapper for `c-calc-offset' in case
12343 ;; someone is calling it directly. It takes an old style syntactic
12344 ;; element on the form (SYMBOL . ANCHOR-POS) and converts it to the
12345 ;; new list form.
12346 ;;
12347 ;; This function might do hidden buffer changes.
12354 ;; Calculate the syntactic indentation from a syntactic description
12355 ;; as returned by `c-guess-syntax'.
12356 ;;
12357 ;; Note that topmost-intro always has an anchor position at bol, for
12358 ;; historical reasons. It's often used together with other symbols
12359 ;; that has more sane positions. Since we always use the first
12360 ;; found anchor position, we rely on that these other symbols always
12361 ;; precede topmost-intro in the LANGELEMS list.
12362 ;;
12363 ;; This function might do hidden buffer changes.
12371 ;; Got an absolute column that overrides any indentation
12372 ;; we've collected so far, but not the relative
12373 ;; indentation we might get for the nested structures
12374 ;; further down the langelems list.
12378 ;; Got a relative change of the current calculated
12379 ;; indentation.
12382 ;; Use the anchor position from the first syntactic
12383 ;; element with one.
12393 indent)))
12395 \f
12398 ;; Local Variables:
12399 ;; indent-tabs-mode: t
12400 ;; tab-width: 8
12401 ;; End:
12402 ;;; cc-engine.el ends here