| blob | 76a0b201816743235973542e30ec6eb25b5e2aab |
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, and
87 ;; to "hide" obtrusive characters in preprocessor lines.
88 ;;
89 ;; This property is used on single characters and is therefore
90 ;; always treated as front and rear nonsticky (or start and end open
91 ;; in XEmacs vocabulary). It's therefore installed on
92 ;; `text-property-default-nonsticky' if that variable exists (Emacs
93 ;; >= 21).
94 ;;
95 ;; 'c-is-sws and 'c-in-sws
96 ;; Used by `c-forward-syntactic-ws' and `c-backward-syntactic-ws' to
97 ;; speed them up. See the comment blurb before `c-put-is-sws'
98 ;; below for further details.
99 ;;
100 ;; 'c-type
101 ;; This property is used on single characters to mark positions with
102 ;; special syntactic relevance of various sorts. Its primary use is
103 ;; to avoid glitches when multiline constructs are refontified
104 ;; interactively (on font lock decoration level 3). It's cleared in
105 ;; a region before it's fontified and is then put on relevant chars
106 ;; in that region as they are encountered during the fontification.
107 ;; The value specifies the kind of position:
108 ;;
109 ;; 'c-decl-arg-start
110 ;; Put on the last char of the token preceding each declaration
111 ;; inside a declaration style arglist (typically in a function
112 ;; prototype).
113 ;;
114 ;; 'c-decl-end
115 ;; Put on the last char of the token preceding a declaration.
116 ;; This is used in cases where declaration boundaries can't be
117 ;; recognized simply by looking for a token like ";" or "}".
118 ;; `c-type-decl-end-used' must be set if this is used (see also
119 ;; `c-find-decl-spots').
120 ;;
121 ;; 'c-<>-arg-sep
122 ;; Put on the commas that separate arguments in angle bracket
123 ;; arglists like C++ template arglists.
124 ;;
125 ;; 'c-decl-id-start and 'c-decl-type-start
126 ;; Put on the last char of the token preceding each declarator
127 ;; in the declarator list of a declaration. They are also used
128 ;; between the identifiers cases like enum declarations.
129 ;; 'c-decl-type-start is used when the declarators are types,
130 ;; 'c-decl-id-start otherwise.
131 ;;
132 ;; 'c-awk-NL-prop
133 ;; Used in AWK mode to mark the various kinds of newlines. See
134 ;; cc-awk.el.
136 ;;; Code:
143 load-path)))
152 \f
153 ;; Make declarations for all the `c-lang-defvar' variables in cc-langs.
165 \f
166 ;;; Internal state variables.
168 ;; Internal state of hungry delete key feature
172 ;; The electric flag (toggled by `c-toggle-electric-state').
173 ;; If t, electric actions (like automatic reindentation, and (if
174 ;; c-auto-newline is also set) auto newlining) will happen when an electric
175 ;; key like `{' is pressed (or an electric keyword like `else').
179 ;; Internal state of auto newline feature.
183 ;; Included in the mode line to indicate the active submodes.
184 ;; (defvar c-submode-indicators nil)
185 ;; (make-variable-buffer-local 'c-submode-indicators)
188 ;; Calculate the new state of PREVSTATE, t or nil, based on arg. If
189 ;; arg is nil or zero, toggle the state. If arg is negative, turn
190 ;; the state off, and if arg is positive, turn the state on
196 \f
197 ;; Basic handling of preprocessor directives.
199 ;; This is a dynamically bound cache used together with
200 ;; `c-query-macro-start' and `c-query-and-set-macro-start'. It only
201 ;; works as long as point doesn't cross a macro boundary.
210 c-macro-start))
218 c-macro-start))
220 ;; One element macro cache to cope with continual movement within very large
221 ;; CPP macros.
224 ;; Nil or cons of the bounds of the most recent CPP form probed by
225 ;; `c-beginning-of-macro', `c-end-of-macro' or `c-syntactic-end-of-macro'.
226 ;; The cdr will be nil if we know only the start of the CPP form.
229 ;; The starting position from where we determined `c-macro-cache'.
232 ;; non-nil iff `c-macro-cache' has both elements set AND the cdr is at a
233 ;; syntactic end of macro, not merely an apparent one.
236 ;; Called from a before-change function. If the change region is before or
237 ;; in the macro characterized by `c-macro-cache' etc., nullify it
238 ;; appropriately. BEG and END are the standard before-change-functions
239 ;; parameters. END isn't used.
244 c-macro-cache-start-pos nil
245 c-macro-cache-syntactic nil))
250 c-macro-cache-syntactic nil))))
253 ;; Check that the ostensible CPP construct at point is a real one. In
254 ;; particular, if point is on the first line of a narrowed buffer, make sure
255 ;; that the "#" isn't, say, the second character of a "##" operator. Return
256 ;; t when the macro is real, nil otherwise.
267 t)
271 "Go to the beginning of a preprocessor directive.
272 Leave point at the beginning of the directive and return t if in one,
273 otherwise return nil and leave point unchanged.
275 Note that this function might do hidden buffer changes. See the
276 comment at the start of cc-engine.el for more info."
288 t))
290 c-macro-cache-start-pos nil
291 c-macro-cache-syntactic nil)
304 c-macro-cache-start-pos here)
305 t)
307 nil))))))
310 "Go to the end of a preprocessor directive.
311 More accurately, move the point to the end of the closest following
312 line that doesn't end with a line continuation backslash - no check is
313 done that the point is inside a cpp directive to begin with.
315 Note that this function might do hidden buffer changes. See the
316 comment at the start of cc-engine.el for more info."
325 c-macro-cache-start-pos nil
326 c-macro-cache-syntactic nil))
332 t)))
337 ;; Go to the end of a CPP directive, or a "safe" pos just before.
338 ;;
339 ;; This is normally the end of the next non-escaped line. A "safe"
340 ;; position is one not within a string or comment. (The EOL on a line
341 ;; comment is NOT "safe").
342 ;;
343 ;; This function must only be called from the beginning of a CPP construct.
344 ;;
345 ;; Note that this function might do hidden buffer changes. See the comment
346 ;; at the start of cc-engine.el for more info.
349 s)
361 ;; Assuming point is at the "#" that introduces a preprocessor
362 ;; directive, it's moved forward to the end of the identifier which is
363 ;; "#define"d (or whatever c-opt-cpp-macro-define specifies). Non-nil
364 ;; is returned in this case, in all other cases nil is returned and
365 ;; point isn't moved.
366 ;;
367 ;; This function might do hidden buffer changes.
373 ;; Assuming point is at the "#" that introduces a preprocessor
374 ;; directive, it's moved forward to the start of the definition body
375 ;; if it's a "#define" (or whatever c-opt-cpp-macro-define
376 ;; specifies). Non-nil is returned in this case, in all other cases
377 ;; nil is returned and point isn't moved.
378 ;;
379 ;; This function might do hidden buffer changes.
385 \f
386 ;;; Basic utility functions.
389 ;; Return the given region as a string where all syntactic
390 ;; whitespace is removed or, where necessary, replaced with a single
391 ;; space. If PAREN-LEVEL is given then all parens in the region are
392 ;; collapsed to "()", "[]" etc.
393 ;;
394 ;; This function might do hidden buffer changes.
438 ;; Shift the indentation of the current line with the specified
439 ;; amount (positive inwards). The buffer is modified only if
440 ;; SHIFT-AMT isn't equal to zero.
443 tmp-char-inserted)
445 nil
446 ;; If we're on an empty line inside a macro, we take the point
447 ;; to be at the current indentation and shift it to the
448 ;; appropriate column. This way we don't treat the extra
449 ;; whitespace out to the line continuation as indentation.
465 ;; If initial point was within line's indentation and we're not on
466 ;; a line with a line continuation in a macro, position after the
467 ;; indentation. Else stay at same point in text.
475 ;; Return non-nil if the string KEYWORD is a known keyword. More
476 ;; precisely, the value is the symbol for the keyword in
477 ;; `c-keywords-obarray'.
481 ;; Return non-nil if the symbol KEYWORD-SYM, as returned by
482 ;; `c-keyword-sym', is a member of LANG-CONSTANT, which is the name
483 ;; of a language constant that ends with "-kwds". If KEYWORD-SYM is
484 ;; nil then the result is nil.
487 ;; String syntax chars, suitable for skip-syntax-(forward|backward).
492 ;; Regexp matching string limit syntax.
497 ;; Regexp matching WS followed by string limit syntax.
501 ;; Holds formatted error strings for the few cases where parse errors
502 ;; are reported.
509 c-parsing-error)
511 ;; Faces given to comments and string literals. This is used in some
512 ;; situations to speed up recognition; it isn't mandatory that font
513 ;; locking is in use. This variable is extended with the face in
514 ;; `c-doc-face-name' when fontification is activated in cc-fonts.el.
518 ;; New in Emacs 22.
522 ;; Put a c-type property with the given value at POS.
526 ;; Remove all occurrences of the c-type property that has the given
527 ;; value in the region between FROM and TO. VALUE is assumed to not
528 ;; be nil.
529 ;;
530 ;; Note: This assumes that c-type is put on single chars only; it's
531 ;; very inefficient if matching properties cover large regions.
540 \f
541 ;; Some debug tools to visualize various special positions. This
542 ;; debug code isn't as portable as the rest of CC Mode.
578 \f
579 ;; `c-beginning-of-statement-1' and accompanying stuff.
581 ;; KLUDGE ALERT: c-maybe-labelp is used to pass information between
582 ;; c-crosses-statement-barrier-p and c-beginning-of-statement-1. A
583 ;; better way should be implemented, but this will at least shut up
584 ;; the byte compiler.
587 ;; New awk-compatible version of c-beginning-of-statement-1, ACM 2002/6/22
589 ;; Macros used internally in c-beginning-of-statement-1 for the
590 ;; automaton actions.
593 stack)))
598 t
599 ,do-if-done
604 ;; Throw nil if stack is empty, else throw non-nil.
622 "No matching `%s' found for `%s' on line %d"
629 noerror comma-delim)
630 "Move to the start of the current statement or declaration, or to
631 the previous one if already at the beginning of one. Only
632 statements/declarations on the same level are considered, i.e. don't
633 move into or out of sexps (not even normal expression parentheses).
635 If point is already at the earliest statement within braces or parens,
636 this function doesn't move back into any whitespace preceding it; it
637 returns `same' in this case.
641 is within the continuation. If starting at such a token, move to the
642 corresponding statement start. If at the beginning of a statement,
643 move to the closest containing statement if there is any. This might
644 also stop at a continuation clause.
646 Labels are treated as part of the following statements if
647 IGNORE-LABELS is non-nil. (FIXME: Doesn't work if we stop at a known
648 statement start keyword.) Otherwise, each label is treated as a
649 separate statement.
651 Macros are ignored \(i.e. skipped over) unless point is within one, in
652 which case the content of the macro is treated as normal code. Aside
653 from any normal statement starts found in it, stop at the first token
656 macros before leaving them.
658 Return:
660 `same' if stopped at the beginning of the current statement;
661 `up' if stepped to a containing statement;
662 `previous' if stepped to a preceding statement;
663 `beginning' if stepped from a statement continuation clause to
664 its start clause; or
665 `macro' if stepped to a macro start.
666 Note that `same' and not `label' is returned if stopped at the same
667 label without crossing the colon character.
669 LIM may be given to limit the search. If the search hits the limit,
670 point will be left at the closest following token, or at the start
671 position if that is less (`same' is returned in this case).
673 NOERROR turns off error logging to `c-parsing-error'.
675 Normally only `;' and virtual semicolons are considered to delimit
676 statements, but if COMMA-DELIM is non-nil then `,' is treated
677 as a delimiter too.
679 Note that this function might do hidden buffer changes. See the
680 comment at the start of cc-engine.el for more info."
682 ;; The bulk of this function is a pushdown automaton that looks at statement
683 ;; boundaries and the tokens (such as "while") in c-opt-block-stmt-key. Its
684 ;; purpose is to keep track of nested statements, ensuring that such
685 ;; statements are skipped over in their entirety (somewhat akin to what C-M-p
686 ;; does with nested braces/brackets/parentheses).
687 ;;
688 ;; Note: The position of a boundary is the following token.
689 ;;
690 ;; Beginning with the current token (the one following point), move back one
691 ;; sexp at a time (where a sexp is, more or less, either a token or the
692 ;; entire contents of a brace/bracket/paren pair). Each time a statement
693 ;; boundary is crossed or a "while"-like token is found, update the state of
694 ;; the PDA. Stop at the beginning of a statement when the stack (holding
695 ;; nested statement info) is empty and the position has been moved.
696 ;;
697 ;; The following variables constitute the PDA:
698 ;;
699 ;; sym: This is either the "while"-like token (e.g. 'for) we've just
700 ;; scanned back over, 'boundary if we've just gone back over a
701 ;; statement boundary, or nil otherwise.
702 ;; state: takes one of the values (nil else else-boundary while
703 ;; while-boundary catch catch-boundary).
704 ;; nil means "no "while"-like token yet scanned".
705 ;; 'else, for example, means "just gone back over an else".
706 ;; 'else-boundary means "just gone back over a statement boundary
707 ;; immediately after having gone back over an else".
708 ;; saved-pos: A vector of either saved positions (tok ptok pptok, etc.) or
709 ;; of error reporting information.
710 ;; stack: The stack onto which the PDA pushes its state. Each entry
711 ;; consists of a saved value of state and saved-pos. An entry is
712 ;; pushed when we move back over a "continuation" token (e.g. else)
713 ;; and popped when we encounter the corresponding opening token
714 ;; (e.g. if).
715 ;;
716 ;;
717 ;; The following diagram briefly outlines the PDA.
718 ;;
719 ;; Common state:
720 ;; "else": Push state, goto state `else'.
721 ;; "while": Push state, goto state `while'.
722 ;; "catch" or "finally": Push state, goto state `catch'.
723 ;; boundary: Pop state.
724 ;; other: Do nothing special.
725 ;;
726 ;; State `else':
727 ;; boundary: Goto state `else-boundary'.
728 ;; other: Error, pop state, retry token.
729 ;;
730 ;; State `else-boundary':
731 ;; "if": Pop state.
732 ;; boundary: Error, pop state.
733 ;; other: See common state.
734 ;;
735 ;; State `while':
736 ;; boundary: Save position, goto state `while-boundary'.
737 ;; other: Pop state, retry token.
738 ;;
739 ;; State `while-boundary':
740 ;; "do": Pop state.
741 ;; boundary: Restore position if it's not at start, pop state. [*see below]
742 ;; other: See common state.
743 ;;
744 ;; State `catch':
745 ;; boundary: Goto state `catch-boundary'.
746 ;; other: Error, pop state, retry token.
747 ;;
748 ;; State `catch-boundary':
749 ;; "try": Pop state.
750 ;; "catch": Goto state `catch'.
751 ;; boundary: Error, pop state.
752 ;; other: See common state.
753 ;;
754 ;; [*] In the `while-boundary' state, we had pushed a 'while state, and were
755 ;; searching for a "do" which would have opened a do-while. If we didn't
756 ;; find it, we discard the analysis done since the "while", go back to this
757 ;; token in the buffer and restart the scanning there, this time WITHOUT
758 ;; pushing the 'while state onto the stack.
759 ;;
760 ;; In addition to the above there is some special handling of labels
761 ;; and macros.
765 macro-start
768 c-stmt-delim-chars-with-comma
769 c-stmt-delim-chars))
770 c-in-literal-cache c-maybe-labelp after-case:-pos saved
771 ;; Current position.
772 pos
773 ;; Position of last stmt boundary character (e.g. ;).
774 boundary-pos
775 ;; The position of the last sexp or bound that follows the
776 ;; first found colon, i.e. the start of the nonlabel part of
777 ;; the statement. It's `start' if a colon is found just after
778 ;; the start.
779 after-labels-pos
780 ;; Like `after-labels-pos', but the first such position inside
781 ;; a label, i.e. the start of the last label before the start
782 ;; of the nonlabel part of the statement.
783 last-label-pos
784 ;; The last position where a label is possible provided the
785 ;; statement started there. It's nil as long as no invalid
786 ;; label content has been found (according to
787 ;; `c-nonlabel-token-key'). It's `start' if no valid label
788 ;; content was found in the label. Note that we might still
789 ;; regard it a label if it starts with `c-label-kwds'.
790 label-good-pos
791 ;; Putative positions of the components of a bitfield declaration,
792 ;; e.g. "int foo : NUM_FOO_BITS ;"
793 bitfield-type-pos bitfield-id-pos bitfield-size-pos
794 ;; Symbol just scanned back over (e.g. 'while or 'boundary).
795 ;; See above.
796 sym
797 ;; Current state in the automaton. See above.
798 state
799 ;; Current saved positions. See above.
800 saved-pos
801 ;; Stack of conses (state . saved-pos).
802 stack
803 ;; Regexp which matches "for", "if", etc.
806 ;; Return value.
808 ;; Positions of the last three sexps or bounds we've stopped at.
809 tok ptok pptok)
819 ;; Try to skip back over unary operator characters, to register
820 ;; that we've moved.
824 ;; Protect post-++/-- operators just before a virtual semicolon.
828 ;; Skip back over any semicolon here. If it was a bare semicolon, we're
829 ;; done. Later on we ignore the boundaries for statements that don't
830 ;; contain any sexp. The only thing that is affected is that the error
831 ;; checking is a little less strict, and we really don't bother.
841 pos saved)
843 ;; Begin at start and not pos to detect macros if we stand
844 ;; directly after the #.
847 ;; Record this as the first token if not starting inside it.
850 ;; The following while loop goes back one sexp (balanced parens,
851 ;; etc. with contents, or symbol or suchlike) each iteration. This
852 ;; movement is accomplished with a call to c-backward-sexp approx 170
853 ;; lines below.
854 ;;
855 ;; The loop is exited only by throwing nil to the (catch 'loop ...):
856 ;; 1. On reaching the start of a macro;
857 ;; 2. On having passed a stmt boundary with the PDA stack empty;
858 ;; 3. On reaching the start of an Objective C method def;
859 ;; 4. From macro `c-bos-pop-state'; when the stack is empty;
860 ;; 5. From macro `c-bos-pop-state-and-retry' when the stack is empty.
864 ;; Are we in a macro, just after the opening #?
879 ;; Stop at the first token in the content of the macro.
883 ret 'macro
884 ignore-labels t))
887 ;; Do a round through the automaton if we've just passed a
888 ;; statement boundary or passed a "while"-like token.
896 ;; The PDA state handling.
897 ;;
898 ;; Refer to the description of the PDA in the opening
899 ;; comments. In the following OR form, the first leaf
900 ;; attempts to handles one of the specific actions detailed
901 ;; (e.g., finding token "if" whilst in state `else-boundary').
902 ;; We drop through to the second leaf (which handles common
903 ;; state) if no specific handler is found in the first cond.
904 ;; If a parsing error is detected (e.g. an "else" with no
905 ;; preceding "if"), we throw to the enclosing catch.
906 ;;
907 ;; Note that the (eq state 'else) means
908 ;; "we've just passed an else", NOT "we're looking for an
909 ;; else".
926 ;; Since this can cause backtracking we do a
927 ;; little more careful analysis to avoid it:
928 ;; If there's a label in front of the while
929 ;; it can't be part of a do-while.
958 ;; This is state common. We get here when the previous
959 ;; cond statement found no particular state handler.
961 ;; If we have a boundary at the start
962 ;; position we push a frame to go to the
963 ;; previous statement.
972 ;; Is this a real while, or a do-while?
973 ;; The next `when' triggers unless we are SURE that
974 ;; the `while' is not the tail end of a `do-while'.
977 ;; The following kludge is to prevent
978 ;; infinite recursion when called from
979 ;; c-awk-after-if-for-while-condition-p,
980 ;; or the like.
984 ;; Since this can cause backtracking we do a
985 ;; little more careful analysis to avoid it: If
986 ;; the while isn't followed by a (possibly
987 ;; virtual) semicolon it can't be a do-while.
996 ;; We're either past a statement boundary or at the
997 ;; start of a statement, so throw away any label data
998 ;; for the previous one.
1000 last-label-pos nil
1001 c-maybe-labelp nil))))
1003 ;; Step to the previous sexp, but not if we crossed a
1004 ;; boundary, since that doesn't consume an sexp.
1008 ;; HERE IS THE SINGLE PLACE INSIDE THE PDA LOOP WHERE WE MOVE
1009 ;; BACKWARDS THROUGH THE SOURCE.
1013 ;; The end position of the area to search for statement
1014 ;; barriers in this round.
1017 ;; Go back over exactly one logical sexp, taking proper
1018 ;; account of macros and escaped EOLs.
1022 ;; Give up if we hit an unbalanced block. Since the
1023 ;; stack won't be empty the code below will report a
1024 ;; suitable error.
1027 ;; Have we moved into a macro?
1030 ;; Have we crossed a statement boundary? If not,
1031 ;; keep going back until we find one or a "real" sexp.
1038 ;; Have we just gone back over an escaped NL? This
1039 ;; doesn't count as a sexp.
1042 ;; Have we crossed a statement boundary?
1045 ;; Are we at a macro beginning?
1047 c-opt-cpp-prefix
1053 ;; Just gone back over a brace block?
1062 ;; Just gone back over some paren block?
1066 before-sws-pos)))
1069 ;; Just gone back over an ordinary symbol of some sort?
1075 ptok tok
1076 tok boundary-pos
1078 ;; Like a C "continue". Analyze the next sexp.
1081 ;; ObjC method def?
1088 ;; Might we have a bitfield declaration, "<type> <id> : <size>"?
1091 ;; The : <size> and <id> fields?
1101 ;; The <type> field?
1109 bitfield-id-pos nil)))))
1111 ;; Handle labels.
1114 ;; `c-crosses-statement-barrier-p' has found a colon, so we
1115 ;; might be in a label now. Have we got a real label
1116 ;; (including a case label) or something like C++'s "public:"?
1117 ;; A case label might use an expression rather than a token.
1120 ;; Catch C++'s inheritance construct "class foo : bar".
1134 ; been found.
1136 ;; We're in a potential label and it's the first
1137 ;; time we've found something that isn't allowed in
1138 ;; one.
1141 ;; We've moved back by a sexp, so update the token positions.
1143 pptok ptok
1144 ptok tok
1150 ;; If the stack isn't empty there might be errors to report.
1159 ;; Need to investigate closer whether we've crossed
1160 ;; between a substatement and its containing statement.
1164 pptok)
1166 ptok)
1174 after-labels-pos
1181 last-label-pos
1182 pos))
1184 ;; We're in a label. Maybe we should step to the statement
1185 ;; after it.
1190 ;; Might have jumped over several labels. Go to the last one.
1193 ;; Have we got "case <expression>:"?
1203 ;; Skip over the unary operators that can start the statement.
1206 ;; protect AWK post-inc/decrement operators, etc.
1211 ret)))
1214 "Return non-nil if there is a non-comment non-macro punctuation character
1215 between FROM and TO. FROM must not be in a string or comment. The returned
1216 value is the position of the first such character."
1227 "Return non-nil if buffer positions FROM to TO cross one or more
1228 statement or declaration boundaries. The returned value is actually
1229 the position of the earliest boundary char. FROM must not be within
1230 a string or comment.
1232 The variable `c-maybe-labelp' is set to the position of the first `:' that
1233 might start a label (i.e. not part of `::' and not preceded by `?'). If a
1234 single `?' is found, then `c-maybe-labelp' is cleared.
1236 For AWK, a statement which is terminated by an EOL (not a ; or a }) is
1238 the line. If this virtual semicolon is _at_ from, the function recognizes it.
1240 Note that this function might do hidden buffer changes. See the
1241 comment at the start of cc-engine.el for more info."
1243 ;; If the current language has CPP macros, insert # into skip-chars.
1246 c-opt-cpp-symbol ; usually "#"
1248 c-stmt-delim-chars))
1251 lit-range vsemi-pos)
1258 skip-chars
1262 ;; Virtual semicolon?
1272 ;; In a string/comment?
1279 ;; Ignore scope operators.
1283 ;; A question mark. Can't be a label, so stop
1284 ;; looking for more : and ?.
1287 ;; At a CPP construct or a "#" or "##" operator?
1298 ;; In trailing space after an as yet undetected virtual semicolon?
1301 ; unterminated string/regexp.
1306 nil))))))
1309 "Return non-nil if the point is at the first token in a statement
1310 or somewhere in the syntactic whitespace before it.
1313 Any kind of declaration-like construct that occur outside function
1316 Note that this function might do hidden buffer changes. See the
1317 comment at the start of cc-engine.el for more info."
1321 c-maybe-labelp)
1332 "Return non-nil if the point is at the first token in an expression or
1333 statement, or somewhere in the syntactic whitespace before it.
1336 grammar sense: It's any sequence of expression tokens except commas,
1337 unless they are enclosed inside parentheses of some kind. Also, an
1338 expression never continues past an enclosing parenthesis, but it might
1339 contain parenthesis pairs of any sort except braces.
1341 Since expressions never cross statement boundaries, this function also
1342 recognizes statement beginnings, just like `c-at-statement-start-p'.
1344 Note that this function might do hidden buffer changes. See the
1345 comment at the start of cc-engine.el for more info."
1350 c-maybe-labelp)
1358 \f
1359 ;; A set of functions that covers various idiosyncrasies in
1360 ;; implementations of `forward-comment'.
1362 ;; Note: Some emacsen considers incorrectly that any line comment
1363 ;; ending with a backslash continues to the next line. I can't think
1364 ;; of any way to work around that in a reliable way without changing
1365 ;; the buffer, though. Suggestions welcome. ;) (No, temporarily
1366 ;; changing the syntax for backslash doesn't work since we must treat
1367 ;; escapes in string literals correctly.)
1370 "Move forward past whitespace and the closest following comment, if any.
1371 Return t if a comment was found, nil otherwise. In either case, the
1372 point is moved past the following whitespace. Line continuations,
1373 i.e. a backslashes followed by line breaks, are treated as whitespace.
1374 The line breaks that end line comments are considered to be the
1375 comment enders, so the point will be put on the beginning of the next
1376 line if it moved past a line comment.
1378 This function does not do any hidden buffer changes."
1386 ;; Some emacsen (e.g. XEmacs 21) return t when moving
1387 ;; forwards at eob.
1388 nil
1390 ;; Emacs includes the ending newline in a b-style (c++)
1391 ;; comment, but XEmacs doesn't. We depend on the Emacs
1392 ;; behavior (which also is symmetric).
1396 t))))
1399 "Move forward past all following whitespace and comments.
1400 Line continuations, i.e. a backslashes followed by line breaks, are
1401 treated as whitespace.
1403 Note that this function might do hidden buffer changes. See the
1404 comment at the start of cc-engine.el for more info."
1407 ;; If forward-comment in at least XEmacs 21 is given a large
1408 ;; positive value, it'll loop all the way through if it hits
1409 ;; eob.
1411 ;; Some emacsen (e.g. XEmacs 21) return t when moving
1412 ;; forwards at eob.
1417 t))))
1420 "Move backward past whitespace and the closest preceding comment, if any.
1421 Return t if a comment was found, nil otherwise. In either case, the
1422 point is moved past the preceding whitespace. Line continuations,
1423 i.e. a backslashes followed by line breaks, are treated as whitespace.
1424 The line breaks that end line comments are considered to be the
1425 comment enders, so the point cannot be at the end of the same line to
1426 move over a line comment.
1428 This function does not do any hidden buffer changes."
1431 ;; When we got newline terminated comments, forward-comment in all
1432 ;; supported emacsen so far will stop at eol of each line not
1433 ;; ending with a comment when moving backwards. This corrects for
1434 ;; that, and at the same time handles line continuations.
1442 ;; Some emacsen (e.g. Emacs 19.34) return t when moving
1443 ;; backwards at bob.
1444 nil
1446 ;; Leave point after the closest following newline if we've
1447 ;; backed up over any above, since forward-comment won't move
1448 ;; backward over a line comment if point is at the end of the
1449 ;; same line.
1454 ;; If forward-comment above succeeded and we're at eol
1455 ;; then the newline we moved over above didn't end a
1456 ;; line comment, so we give it another go.
1458 t))
1460 ;; Emacs <= 20 and XEmacs move back over the closer of a
1461 ;; block comment that lacks an opener.
1464 t)))))
1467 "Move backward past all preceding whitespace and comments.
1468 Line continuations, i.e. a backslashes followed by line breaks, are
1469 treated as whitespace. The line breaks that end line comments are
1470 considered to be the comment enders, so the point cannot be at the end
1471 of the same line to move over a line comment. Unlike
1472 c-backward-syntactic-ws, this function doesn't move back over
1473 preprocessor directives.
1475 Note that this function might do hidden buffer changes. See the
1476 comment at the start of cc-engine.el for more info."
1480 ;; `forward-comment' in some emacsen (e.g. XEmacs 21.4)
1481 ;; return t when moving backwards at bob.
1487 ;; Cope specifically with ^M^J here -
1488 ;; forward-comment sometimes gets stuck after ^Ms,
1489 ;; sometimes after ^M^J.
1493 t)
1497 t))))
1498 moved-comment)
1500 ;; Emacs <= 20 and XEmacs move back over the
1501 ;; closer of a block comment that lacks an opener.
1503 t)
1505 ;; XEmacs treats line continuations as whitespace but
1506 ;; only in the backward direction, which seems a bit
1507 ;; odd. Anyway, this is necessary for Emacs.
1512 t))))))
1514 \f
1515 ;; Tools for skipping over syntactic whitespace.
1517 ;; The following functions use text properties to cache searches over
1518 ;; large regions of syntactic whitespace. It works as follows:
1519 ;;
1520 ;; o If a syntactic whitespace region contains anything but simple
1521 ;; whitespace (i.e. space, tab and line breaks), the text property
1522 ;; `c-in-sws' is put over it. At places where we have stopped
1523 ;; within that region there's also a `c-is-sws' text property.
1524 ;; That since there typically are nested whitespace inside that
1525 ;; must be handled separately, e.g. whitespace inside a comment or
1526 ;; cpp directive. Thus, from one point with `c-is-sws' it's safe
1527 ;; to jump to another point with that property within the same
1528 ;; `c-in-sws' region. It can be likened to a ladder where
1529 ;; `c-in-sws' marks the bars and `c-is-sws' the rungs.
1530 ;;
1531 ;; o The `c-is-sws' property is put on the simple whitespace chars at
1532 ;; a "rung position" and also maybe on the first following char.
1533 ;; As many characters as can be conveniently found in this range
1534 ;; are marked, but no assumption can be made that the whole range
1535 ;; is marked (it could be clobbered by later changes, for
1536 ;; instance).
1537 ;;
1538 ;; Note that some part of the beginning of a sequence of simple
1539 ;; whitespace might be part of the end of a preceding line comment
1540 ;; or cpp directive and must not be considered part of the "rung".
1541 ;; Such whitespace is some amount of horizontal whitespace followed
1542 ;; by a newline. In the case of cpp directives it could also be
1543 ;; two newlines with horizontal whitespace between them.
1544 ;;
1545 ;; The reason to include the first following char is to cope with
1546 ;; "rung positions" that don't have any ordinary whitespace. If
1547 ;; `c-is-sws' is put on a token character it does not have
1548 ;; `c-in-sws' set simultaneously. That's the only case when that
1549 ;; can occur, and the reason for not extending the `c-in-sws'
1550 ;; region to cover it is that the `c-in-sws' region could then be
1551 ;; accidentally merged with a following one if the token is only
1552 ;; one character long.
1553 ;;
1554 ;; o On buffer changes the `c-in-sws' and `c-is-sws' properties are
1555 ;; removed in the changed region. If the change was inside
1556 ;; syntactic whitespace that means that the "ladder" is broken, but
1557 ;; a later call to `c-forward-sws' or `c-backward-sws' will use the
1558 ;; parts on either side and use an ordinary search only to "repair"
1559 ;; the gap.
1560 ;;
1561 ;; Special care needs to be taken if a region is removed: If there
1562 ;; are `c-in-sws' on both sides of it which do not connect inside
1563 ;; the region then they can't be joined. If e.g. a marked macro is
1564 ;; broken, syntactic whitespace inside the new text might be
1565 ;; marked. If those marks would become connected with the old
1566 ;; `c-in-sws' range around the macro then we could get a ladder
1567 ;; with one end outside the macro and the other at some whitespace
1568 ;; within it.
1569 ;;
1570 ;; The main motivation for this system is to increase the speed in
1571 ;; skipping over the large whitespace regions that can occur at the
1572 ;; top level in e.g. header files that contain a lot of comments and
1573 ;; cpp directives. For small comments inside code it's probably
1574 ;; slower than using `forward-comment' straightforwardly, but speed is
1575 ;; not a significant factor there anyway.
1577 ; (defface c-debug-is-sws-face
1578 ; '((t (:background "GreenYellow")))
1579 ; "Debug face to mark the `c-is-sws' property.")
1580 ; (defface c-debug-in-sws-face
1581 ; '((t (:underline t)))
1582 ; "Debug face to mark the `c-in-sws' property.")
1584 ; (defun c-debug-put-sws-faces ()
1585 ; ;; Put the sws debug faces on all the `c-is-sws' and `c-in-sws'
1586 ; ;; properties in the buffer.
1587 ; (interactive)
1588 ; (save-excursion
1589 ; (c-save-buffer-state (in-face)
1590 ; (goto-char (point-min))
1591 ; (setq in-face (if (get-text-property (point) 'c-is-sws)
1592 ; (point)))
1593 ; (while (progn
1594 ; (goto-char (next-single-property-change
1595 ; (point) 'c-is-sws nil (point-max)))
1596 ; (if in-face
1597 ; (progn
1598 ; (c-debug-add-face in-face (point) 'c-debug-is-sws-face)
1599 ; (setq in-face nil))
1600 ; (setq in-face (point)))
1601 ; (not (eobp))))
1602 ; (goto-char (point-min))
1603 ; (setq in-face (if (get-text-property (point) 'c-in-sws)
1604 ; (point)))
1605 ; (while (progn
1606 ; (goto-char (next-single-property-change
1607 ; (point) 'c-in-sws nil (point-max)))
1608 ; (if in-face
1609 ; (progn
1610 ; (c-debug-add-face in-face (point) 'c-debug-in-sws-face)
1611 ; (setq in-face nil))
1612 ; (setq in-face (point)))
1613 ; (not (eobp)))))))
1616 ;;`(message ,@args)
1617 )
1620 ;; This macro does a hidden buffer change.
1627 ;; This macro does a hidden buffer change.
1634 ;; This macro does a hidden buffer change.
1641 ;; This macro does a hidden buffer change.
1648 ;; This macro does a hidden buffer change.
1656 ;; Called from `after-change-functions'. Note that if
1657 ;; `c-forward-sws' or `c-backward-sws' are used outside
1658 ;; `c-save-buffer-state' or similar then this will remove the cache
1659 ;; properties right after they're added.
1660 ;;
1661 ;; This function does hidden buffer changes.
1664 ;; Adjust the end to remove the properties in any following simple
1665 ;; ws up to and including the next line break, if there is any
1666 ;; after the changed region. This is necessary e.g. when a rung
1667 ;; marked empty line is converted to a line comment by inserting
1668 ;; "//" before the line break. In that case the line break would
1669 ;; keep the rung mark which could make a later `c-backward-sws'
1670 ;; move into the line comment instead of over it.
1680 ;; Ensure that an `c-in-sws' range gets broken. Note that it isn't
1681 ;; safe to keep a range that was continuous before the change. E.g:
1682 ;;
1683 ;; #define foo
1684 ;; \
1685 ;; bar
1686 ;;
1687 ;; There can be a "ladder" between "#" and "b". Now, if the newline
1688 ;; after "foo" is removed then "bar" will become part of the cpp
1689 ;; directive instead of a syntactically relevant token. In that
1690 ;; case there's no longer syntactic ws from "#" to "b".
1697 ;; Used by `c-forward-syntactic-ws' to implement the unbounded search.
1698 ;;
1699 ;; This function might do hidden buffer changes.
1702 ;; unmarked part of the simple ws region.
1704 rung-is-marked next-rung-is-marked simple-ws-end
1705 ;; `safe-start' is set when it's safe to cache the start position.
1706 ;; It's not set if we've initially skipped over comments and line
1707 ;; continuations since we might have gone out through the end of a
1708 ;; macro then. This provision makes `c-forward-sws' not populate the
1709 ;; cache in the majority of cases, but otoh is `c-backward-sws' by far
1710 ;; more common.
1711 safe-start)
1713 ;; Skip simple ws and do a quick check on the following character to see
1714 ;; if it's anything that can't start syntactic ws, so we can bail out
1715 ;; early in the majority of cases when there just are a few ws chars.
1724 ;; Find the last rung position to avoid setting properties in all
1725 ;; the cases when the marked rung is complete.
1726 ;; (`next-single-property-change' is certain to move at least one
1727 ;; step forward.)
1730 ;; Got no marked rung here. Since the simple ws might have started
1731 ;; inside a line comment or cpp directive we must set `rung-pos' as
1732 ;; high as possible.
1738 ;; In the following while form, we move over a "ladder" and
1739 ;; following simple WS each time round the loop, appending the WS
1740 ;; onto the ladder, joining adjacent ladders, and terminating when
1741 ;; there is no more WS or we reach EOB.
1746 ;; The following search is the main reason that `c-in-sws'
1747 ;; and `c-is-sws' aren't combined to one property.
1751 ;; If the `c-in-sws' region extended past the last
1752 ;; `c-is-sws' char we have to go back a bit.
1759 "c-forward-sws cached move %s -> %s (max %s)"
1766 ;; We'll loop here if there is simple ws after the last rung.
1767 ;; That means that there's been some change in it and it's
1768 ;; possible that we've stepped into another ladder, so extend
1769 ;; the previous one to join with it if there is one, and try to
1770 ;; use the cache again.
1772 "c-forward-sws extending rung with [%s..%s] (max %s)"
1775 ;; Remove any `c-in-sws' property from the last char of
1776 ;; the rung before we mark it with `c-is-sws', so that we
1777 ;; won't connect with the remains of a broken "ladder".
1783 last-put-in-sws-pos rung-pos)))
1785 ;; Now move over any comments (x)or a CPP construct.
1791 ;; Skipped over comments. Don't cache at eob in case the buffer
1792 ;; is narrowed.
1803 ;; Skip a preprocessor directive.
1810 ;; Don't cache at eob in case the buffer is narrowed.
1815 ;; Skip over a noise macro.
1820 ;; We've searched over a piece of non-white syntactic ws. See if this
1821 ;; can be cached.
1827 ;; Cache if we haven't skipped comments only, and if we started
1828 ;; either from a marked rung or from a completely uncached
1829 ;; position.
1834 ;; See if there's a marked rung in the encountered simple ws. If
1835 ;; so then we can cache, unless `safe-start' is nil. Even then
1836 ;; we need to do this to check if the cache can be used for the
1837 ;; next step.
1841 safe-start))
1845 "c-forward-sws caching [%s..%s] - [%s..%s] (max %s)"
1849 ;; Remove the properties for any nested ws that might be cached.
1850 ;; Only necessary for `c-is-sws' since `c-in-sws' will be set
1851 ;; anyway.
1859 last-put-in-sws-pos rung-pos))
1861 ;; Remove any `c-in-sws' property from the last char of
1862 ;; the rung before we mark it with `c-is-sws', so that we
1863 ;; won't connect with the remains of a broken "ladder".
1866 rung-end-pos))
1869 "c-forward-sws not caching [%s..%s] - [%s..%s] (max %s)"
1873 ;; Set `rung-pos' for the next rung. It's the same thing here as
1874 ;; initially, except that the rung position is set as early as
1875 ;; possible since we can't be in the ending ws of a line comment or
1876 ;; cpp directive now.
1883 ;; Make sure that the newly marked `c-in-sws' region doesn't connect to
1884 ;; another one after the point (which might occur when editing inside a
1885 ;; comment or macro).
1889 "c-forward-sws clearing at %s for cache separation"
1890 last-put-in-sws-pos)
1894 ;; If at eob we have to clear the last character before the end
1895 ;; instead since the buffer might be narrowed and there might
1896 ;; be a `c-in-sws' after (point-max). In this case it's
1897 ;; necessary to clear both properties.
1899 "c-forward-sws clearing thoroughly at %s for cache separation"
1902 last-put-in-sws-pos))))
1903 ))))
1906 ;; Used by `c-backward-syntactic-ws' to implement the unbounded search.
1907 ;;
1908 ;; This function might do hidden buffer changes.
1911 ;; part of the simple ws region.
1913 rung-is-marked simple-ws-beg cmt-skip-pos)
1915 ;; Skip simple horizontal ws and do a quick check on the preceding
1916 ;; character to see if it's anything that can't end syntactic ws, so we can
1917 ;; bail out early in the majority of cases when there just are a few ws
1918 ;; chars. Newlines are complicated in the backward direction, so we can't
1919 ;; skip over them.
1929 ;; Try to find a rung position in the simple ws preceding point, so that
1930 ;; we can get a cache hit even if the last bit of the simple ws has
1931 ;; changed recently.
1937 ;; `rung-pos' will be the earliest marked position, which means that
1938 ;; there might be later unmarked parts in the simple ws region.
1939 ;; It's not worth the effort to fix that; the last part of the
1940 ;; simple ws is also typically edited often, so it could be wasted.
1947 ;; Each time round the next while form, we move back over a ladder
1948 ;; and append any simple WS preceding it, if possible joining with
1949 ;; the previous ladder.
1955 ;; The following search is the main reason that `c-in-sws'
1956 ;; and `c-is-sws' aren't combined to one property.
1960 ;; If the `c-in-sws' region extended past the first
1961 ;; `c-is-sws' char we have to go forward a bit.
1966 "c-backward-sws cached move %s <- %s (min %s)"
1978 t
1980 nil))
1982 ;; We'll loop here if there is simple ws before the first rung.
1983 ;; That means that there's been some change in it and it's
1984 ;; possible that we've stepped into another ladder, so extend
1985 ;; the previous one to join with it if there is one, and try to
1986 ;; use the cache again.
1988 "c-backward-sws extending rung with [%s..%s] (min %s)"
1991 ;; Remove any `c-in-sws' property from the last char of
1992 ;; the rung before we mark it with `c-is-sws', so that we
1993 ;; won't connect with the remains of a broken "ladder".
1996 rung-pos)
2000 last-put-in-sws-pos rung-pos))
2009 ;; Inside a cpp directive. See if it should be skipped over.
2012 ;; Move back over all line continuations in the region skipped
2013 ;; over by `c-backward-comments'. If we go past it then we
2014 ;; started inside the cpp directive.
2023 ;; Don't move past the cpp directive if we began inside
2024 ;; it. Note that the position at the end of the last line
2025 ;; of the macro is also considered to be within it.
2027 nil)
2029 ;; It's worthwhile to spend a little bit of effort on finding
2030 ;; the end of the macro, to get a good `simple-ws-beg'
2031 ;; position for the cache. Note that `c-backward-comments'
2032 ;; could have stepped over some comments before going into
2033 ;; the macro, and then `simple-ws-beg' must be kept on the
2034 ;; same side of those comments.
2041 ;; Might happen if comments after the macro were skipped
2042 ;; over.
2046 t)))
2051 simple-ws-beg)
2052 ;; Skipped over comments. Must put point at the end of
2053 ;; the simple ws at point since we might be after a line
2054 ;; comment or cpp directive that's been partially
2055 ;; narrowed out, and we can't risk marking the simple ws
2056 ;; at the end of it.
2058 t)
2065 ;; Skipped over a noise macro
2067 t)))
2069 ;; We've searched over a piece of non-white syntactic ws. See if this
2070 ;; can be cached.
2075 ;; Cache if we started either from a marked rung or from a
2076 ;; completely uncached position.
2077 rung-is-marked
2080 ;; Cache if there's a marked rung in the encountered simple ws.
2088 "c-backward-sws caching [%s..%s] - [%s..%s] (min %s)"
2093 ;; Remove the properties for any nested ws that might be cached.
2094 ;; Only necessary for `c-is-sws' since `c-in-sws' will be set
2095 ;; anyway.
2100 ;; Remove any `c-in-sws' property from the last char of
2101 ;; the rung before we mark it with `c-is-sws', so that we
2102 ;; won't connect with the remains of a broken "ladder".
2105 rung-end-pos)
2108 last-put-in-sws-pos simple-ws-beg)
2109 rung-pos)
2114 "c-backward-sws not caching [%s..%s] - [%s..%s] (min %s)"
2120 ))
2122 ;; Make sure that the newly marked `c-in-sws' region doesn't connect to
2123 ;; another one before the point (which might occur when editing inside a
2124 ;; comment or macro).
2128 "c-backward-sws clearing at %s for cache separation"
2131 last-put-in-sws-pos))
2133 ;; If at bob and the buffer is narrowed, we have to clear the
2134 ;; character we're standing on instead since there might be a
2135 ;; `c-in-sws' before (point-min). In this case it's necessary
2136 ;; to clear both properties.
2138 "c-backward-sws clearing thoroughly at %s for cache separation"
2139 last-put-in-sws-pos)
2142 ))))
2144 \f
2145 ;; Other whitespace tools
2147 ;; Is the region (beg end) WS, and is there WS (or BOB/EOB) next to the
2148 ;; region? This is a "heuristic" function. .....
2149 ;;
2150 ;; The motivation for the second bit is to check whether removing this
2151 ;; region would coalesce two symbols.
2152 ;;
2153 ;; FIXME!!! This function doesn't check virtual semicolons in any way. Be
2154 ;; careful about using this function for, e.g. AWK. (2007/3/7)
2163 \f
2164 ;; A system for finding noteworthy parens before the point.
2167 ;; A maximum comfortable scanning distance, e.g. between
2168 ;; `c-state-cache-good-pos' and "HERE" (where we call c-parse-state). When
2169 ;; this distance is exceeded, we take "emergency measures", e.g. by clearing
2170 ;; the cache and starting again from point-min or a beginning of defun. This
2171 ;; value can be tuned for efficiency or set to a lower value for testing.
2175 ;; The state cache used by `c-parse-state' to cut down the amount of
2176 ;; searching. It's the result from some earlier `c-parse-state' call. See
2177 ;; `c-parse-state''s doc string for details of its structure.
2178 ;;
2179 ;; The use of the cached info is more effective if the next
2180 ;; `c-parse-state' call is on a line close by the one the cached state
2181 ;; was made at; the cache can actually slow down a little if the
2182 ;; cached state was made very far back in the buffer. The cache is
2183 ;; most effective if `c-parse-state' is used on each line while moving
2184 ;; forward.
2188 ;; This is a position where `c-state-cache' is known to be correct, or
2189 ;; nil (see below). It's a position inside one of the recorded unclosed
2190 ;; parens or the top level, but not further nested inside any literal or
2191 ;; subparen that is closed before the last recorded position.
2192 ;;
2193 ;; The exact position is chosen to try to be close to yet earlier than
2194 ;; the position where `c-state-cache' will be called next. Right now
2195 ;; the heuristic is to set it to the position after the last found
2196 ;; closing paren (of any type) before the line on which
2197 ;; `c-parse-state' was called. That is chosen primarily to work well
2198 ;; with refontification of the current line.
2199 ;;
2200 ;; 2009-07-28: When `c-state-point-min' and the last position where
2201 ;; `c-parse-state' or for which `c-invalidate-state-cache' was called, are
2202 ;; both in the same literal, there is no such "good position", and
2203 ;; c-state-cache-good-pos is then nil. This is the ONLY circumstance in which
2204 ;; it can be nil. In this case, `c-state-point-min-literal' will be non-nil.
2205 ;;
2206 ;; 2009-06-12: In a brace desert, c-state-cache-good-pos may also be in
2207 ;; the middle of the desert, as long as it is not within a brace pair
2208 ;; recorded in `c-state-cache' or a paren/bracket pair.
2210 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2211 ;; We maintain a simple cache of positions which aren't in a literal, so as to
2212 ;; speed up testing for non-literality.
2214 ;; The approximate interval between entries in `c-state-nonlit-pos-cache'.
2218 ;; A list of buffer positions which are known not to be in a literal or a cpp
2219 ;; construct. This is ordered with higher positions at the front of the list.
2220 ;; Only those which are less than `c-state-nonlit-pos-cache-limit' are valid.
2224 ;; An upper limit on valid entries in `c-state-nonlit-pos-cache'. This is
2225 ;; reduced by buffer changes, and increased by invocations of
2226 ;; `c-state-literal-at'.
2230 ;; A list of buffer positions which are known not to be in a literal. This is
2231 ;; ordered with higher positions at the front of the list. Only those which
2232 ;; are less than `c-state-semi-nonlit-pos-cache-limit' are valid.
2236 ;; An upper limit on valid entries in `c-state-semi-nonlit-pos-cache'. This is
2237 ;; reduced by buffer changes, and increased by invocations of
2238 ;; `c-state-literal-at'. FIXME!!!
2241 ;; Do a parse-partial-sexp from FROM to TO, returning either
2242 ;; (STATE TYPE (BEG . END)) if TO is in a literal; or
2243 ;; (STATE) otherwise,
2244 ;; where STATE is the parsing state at TO, TYPE is the type of the literal
2245 ;; (one of 'c, 'c++, 'string) and (BEG . END) is the boundaries of the literal.
2246 ;;
2247 ;; Unless NOT-IN-DELIMITER is non-nil, when TO is inside a two-character
2248 ;; comment opener, this is recognized as being in a comment literal.
2249 ;;
2250 ;; Only elements 3 (in a string), 4 (in a comment), 5 (following a quote),
2251 ;; 7 (comment type) and 8 (start of comment/string) (and possibly 9) of
2252 ;; STATE are valid.
2255 ty co-st)
2263 nil ; TARGETDEPTH
2264 nil ; STOPBEFORE
2265 s ; OLDSTATE
2282 ;; Return a buffer position before HERE which is "safe", i.e. outside any
2283 ;; string, comment, or macro.
2284 ;;
2285 ;; NOTE: This function manipulates `c-state-nonlit-pos-cache'. This cache
2286 ;; MAY NOT contain any positions within macros, since macros are frequently
2287 ;; turned into comments by use of the `c-cpp-delimiter' category properties.
2288 ;; We cannot rely on this mechanism whilst determining a cache pos since
2289 ;; this function is also called from outwith `c-parse-state'.
2294 pos npos high-pos lit macro-beg macro-end)
2295 ;; Trim the cache to take account of buffer changes.
2307 ;; Add an element to `c-state-nonlit-pos-cache' each iteration.
2313 ;; Test for being in a literal. If so, go to after it.
2320 ;; Test for being in a macro. If so, go to after it.
2335 ;; Add one extra element above HERE so as to to avoid the previous
2336 ;; expensive calculation when the next call is close to the current
2337 ;; one. This is especially useful when inside a large macro.
2344 pos))))
2347 ;; Return a buffer position before HERE which is "safe", i.e. outside any
2348 ;; string or comment. It may be in a macro.
2353 pos npos high-pos lit macro-beg macro-end)
2354 ;; Trim the cache to take account of buffer changes.
2366 ;; Add an element to `c-state-semi-nonlit-pos-cache' each iteration.
2370 ;; Test for being in a literal. If so, go to after it.
2383 pos))))
2386 ;; If position HERE is inside a literal, return (START . END), the
2387 ;; boundaries of the literal (which may be outside the accessible bit of the
2388 ;; buffer). Otherwise, return nil.
2389 ;;
2390 ;; This function is almost the same as `c-literal-limits'. Previously, it
2391 ;; differed in that it was a lower level function, and that it rigorously
2392 ;; followed the syntax from BOB. `c-literal-limits' is now (2011-12)
2393 ;; virtually identical to this function.
2401 ;; Return the start of the literal containing POS, or POS itself.
2403 pos))
2406 ;; Return a position outside of a string/comment/macro at or before POS.
2407 ;; STATE is the parse-partial-sexp state at POS.
2411 pos)))
2416 res))))
2418 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2419 ;; Stuff to do with point-min, and coping with any literal there.
2422 ;; This is (point-min) when `c-state-cache' was last calculated. A change of
2423 ;; narrowing is likely to affect the parens that are visible before the point.
2429 ;; These two variables define the literal, if any, containing point-min.
2430 ;; Their values are, respectively, 'string, c, or c++, and the start of the
2431 ;; literal. If there's no literal there, they're both nil.
2435 ;; This is the earliest buffer-pos from which scanning can be done. It is
2436 ;; either the end of the literal containing point-min, or point-min itself.
2437 ;; It becomes nil if the buffer is changed earlier than this point.
2439 ;; Return the lowest valid scanning pos. This will be the end of the
2440 ;; literal enclosing point-min, or point-min itself.
2452 ;; Determine the properties of any literal containing POINT-MIN, setting the
2453 ;; variables `c-state-point-min-lit-type', `c-state-point-min-lit-start',
2454 ;; and `c-state-min-scan-pos' accordingly. The return value is meaningless.
2456 lit)
2471 c-state-point-min-lit-start nil
2472 c-state-min-scan-pos p-min)))))
2475 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2476 ;; A variable which signals a brace dessert - helpful for reducing the number
2477 ;; of fruitless backward scans.
2480 ;; Used only in `c-append-lower-brace-pair-to-state-cache'. It is set when
2481 ;; that defun has searched backwards for a brace pair and not found one. Its
2482 ;; value is either nil or a cons (PA . FROM), where PA is the position of the
2483 ;; enclosing opening paren/brace/bracket which bounds the backwards search (or
2484 ;; nil when at top level) and FROM is where the backward search started. It
2485 ;; is reset to nil in `c-invalidate-state-cache'.
2488 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2489 ;; Lowish level functions/macros which work directly on `c-state-cache', or a
2490 ;; list of like structure.
2492 ;; Return the address of the top left brace/bracket/paren recorded in CACHE
2493 ;; (default `c-state-cache') (or nil).
2500 ;; Return the address of the latest brace/bracket/paren (whether left or
2501 ;; right) recorded in CACHE (default `c-state-cache') or nil.
2508 ;; Return the position just after the latest brace/bracket/paren (whether
2509 ;; left or right) recorded in CACHE (default `c-state-cache') or nil.
2517 ;; From the state-cache, determine the buffer position from which we might
2518 ;; scan forward to HERE to update this cache. This position will be just
2519 ;; after a paren/brace/bracket recorded in the cache, if possible, otherwise
2520 ;; return the earliest position in the accessible region which isn't within
2521 ;; a literal. If the visible portion of the buffer is entirely within a
2522 ;; literal, return NIL.
2524 ;(while (>= (or (c-state-cache-top-lparen c) 1) here)
2540 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2541 ;; Variables which keep track of preprocessor constructs.
2550 ;; These are the limits of the macro containing point at the previous call of
2551 ;; `c-parse-state', or nil.
2553 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2554 ;; Defuns which analyze the buffer, yet don't change `c-state-cache'.
2556 ;; Return a start position for building `c-state-cache' from
2557 ;; scratch. This will be at the top level, 2 defuns back.
2559 ;; Go back 2 bods, but ignore any bogus positions returned by
2560 ;; beginning-of-defun (i.e. open paren in column zero).
2570 ;; Return the position of the opening paren/brace/bracket before HERE- which
2571 ;; matches the outermost close p/b/b between HERE+ and TOP. Except when
2572 ;; there's a macro, HERE- and HERE+ are the same. Like this:
2573 ;;
2574 ;; ............................................
2575 ;; | |
2576 ;; ( [ ( .........#macro.. ) ( ) ] )
2577 ;; ^ ^ ^ ^
2578 ;; | | | |
2579 ;; return HERE- HERE+ TOP
2580 ;;
2581 ;; If there aren't enough opening paren/brace/brackets, return the position
2582 ;; of the outermost one found, or HERE- if there are none. If there are no
2583 ;; closing p/b/bs between HERE+ and TOP, return HERE-. HERE-/+ and TOP
2584 ;; must not be inside literals. Only the accessible portion of the buffer
2585 ;; will be scanned.
2587 ;; PART 1: scan from `here+' up to `top', accumulating ")"s which enclose
2588 ;; `here'. Go round the next loop each time we pass over such a ")". These
2589 ;; probably match "("s before `here-'.
2601 ;; PART 2: Scan back before `here-' searching for the "("s
2602 ;; matching/mismatching the ")"s found above. We only need to direct the
2603 ;; caller to scan when we've encountered unmatched right parens.
2612 pos))
2615 ;; Determine the scanning strategy for adjusting `c-parse-state', attempting
2616 ;; to minimize the amount of scanning. HERE is the pertinent position in
2617 ;; the buffer, GOOD-POS is a position where `c-state-cache' (possibly with
2618 ;; its head trimmed) is known to be good, or nil if there is no such
2619 ;; position.
2620 ;;
2621 ;; The return value is a list, one of the following:
2622 ;;
2623 ;; o - ('forward START-POINT) - scan forward from START-POINT,
2624 ;; which is not less than the highest position in `c-state-cache' below HERE,
2625 ;; which is after GOOD-POS.
2626 ;; o - ('backward nil) - scan backwards (from HERE).
2627 ;; o - ('back-and-forward START-POINT) - like 'forward, but when HERE is earlier
2628 ;; than GOOD-POS.
2629 ;; o - ('BOD START-POINT) - scan forwards from START-POINT, which is at the
2630 ;; top level.
2631 ;; o - ('IN-LIT nil) - point is inside the literal containing point-min.
2633 BOD-pos ; position of 2nd BOD before HERE.
2634 strategy ; 'forward, 'backward, 'BOD, or 'IN-LIT.
2635 start-point
2641 start-point nil
2642 cache-pos nil
2653 start-point cache-pos
2656 ;; Might we be better off starting from the top level, two defuns back,
2657 ;; instead? This heuristic no longer works well in C++, where
2658 ;; declarations inside namespace brace blocks are frequently placed at
2659 ;; column zero. (2015-11-10): Remove the condition on C++ Mode.
2661 open-paren-in-column-0-is-defun-start)
2662 ;; (not (c-major-mode-is 'c++-mode))
2667 start-point BOD-pos)))
2672 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2673 ;; Routines which change `c-state-cache' and associated values.
2675 ;; The region (more precisely, point-min) has changed since we
2676 ;; calculated `c-state-cache'. Amend `c-state-cache' accordingly.
2678 ;; If point-min has MOVED BACKWARDS then we drop the state completely.
2679 ;; It would be possible to do a better job here and recalculate the top
2680 ;; only.
2684 c-state-cache-good-pos c-state-min-scan-pos
2685 c-state-brace-pair-desert nil))
2687 ;; point-min has MOVED FORWARD.
2689 ;; Is the new point-min inside a (different) literal?
2694 ;; Cut off a bit of the tail from `c-state-cache'.
2696 pa)
2704 c-state-cache-good-pos c-state-min-scan-pos)
2707 )))
2712 ;; If there is a brace pair preceding FROM in the buffer, at the same level
2713 ;; of nesting (not necessarily immediately preceding), push a cons onto
2714 ;; `c-state-cache' to represent it. FROM must not be inside a literal. If
2715 ;; UPPER-LIM is non-nil, we append the highest brace pair whose "}" is below
2716 ;; UPPER-LIM.
2717 ;;
2718 ;; Return non-nil when this has been done.
2719 ;;
2720 ;; The situation it copes with is this transformation:
2721 ;;
2722 ;; OLD: { (.) {...........}
2723 ;; ^ ^
2724 ;; FROM HERE
2725 ;;
2726 ;; NEW: { {....} (.) {.........
2727 ;; ^ ^ ^
2728 ;; LOWER BRACE PAIR HERE or HERE
2729 ;;
2730 ;; This routine should be fast. Since it can get called a LOT, we maintain
2731 ;; `c-state-brace-pair-desert', a small cache of "failures", such that we
2732 ;; reduce the time wasted in repeated fruitless searches in brace deserts.
2742 ;; Don't start scanning in the middle of a CPP construct unless
2743 ;; it contains HERE - these constructs, in Emacs, are "commented
2744 ;; out" with category properties.
2747 macro-start-or-from
2748 from))
2752 ;; If we're essentially repeating a fruitless search, just give up.
2758 ;; DESERT-LIM. Avoid repeated searching through the cached desert.
2764 ;; CACHE-LIM. This limit will be necessary when an opening
2765 ;; paren at `cache-pos' has just had its matching close paren
2766 ;; inserted into the buffer. `cache-pos' continues to be a
2767 ;; search bound, even though the algorithm below would skip
2768 ;; over the new paren pair.
2777 ;; The top limit is EOB to ensure that `bra' is inside the
2778 ;; accessible part of the buffer at the next scan operation.
2781 ;; In the next pair of nested loops, the inner one moves back past a
2782 ;; pair of (mis-)matching parens or brackets; the outer one moves
2783 ;; back over a sequence of unmatched close brace/paren/bracket each
2784 ;; time round.
2800 ; brace, ignore it.
2803 ;; We've found the desired brace-pair.
2815 ;; We haven't found a brace pair. Record this in the cache.
2818 bra
2823 ;; If BRA+1 is nil, do nothing. Otherwise, BRA+1 is the buffer position
2824 ;; following a {, and that brace has a (mis-)matching } (or ]), and we
2825 ;; "push" "a" brace pair onto `c-state-cache'.
2826 ;;
2827 ;; Here "push" means overwrite the top element if it's itself a brace-pair,
2828 ;; otherwise push it normally.
2829 ;;
2830 ;; The brace pair we push is normally the one surrounding BRA+1, but if the
2831 ;; latter is inside a macro, not being a macro containing
2832 ;; MACRO-START-OR-HERE, we scan backwards through the buffer for a non-macro
2833 ;; base pair. This latter case is assumed to be rare.
2834 ;;
2835 ;; Note: POINT is not preserved in this routine.
2845 c-state-cache)))
2846 ;; N.B. This defsubst codes one method for the simple, normal case,
2847 ;; and a more sophisticated, slower way for the general case. Don't
2848 ;; eliminate this defsubst - it's a speed optimization.
2852 ;; Scan the buffer from FROM to HERE, adding elements into `c-state-cache'
2853 ;; for braces etc. Return a candidate for `c-state-cache-good-pos'.
2854 ;;
2855 ;; FROM must be after the latest brace/paren/bracket in `c-state-cache', if
2856 ;; any. Typically, it is immediately after it. It must not be inside a
2857 ;; literal.
2863 here)))
2866 ; Is actually the pos. to scan for a (/{/[ from,
2867 ; which sometimes is after a silly )/}/].
2870 ; good-pos.
2878 ;; Each time round the following loop, we enter a successively deeper
2879 ;; level of brace/paren nesting. (Except sometimes we "continue at
2880 ;; the existing level".) `pa+1' is a pos inside an opening
2881 ;; brace/paren/bracket, usually just after it.
2884 ;; Each time round the next loop moves forward over an opening then
2885 ;; a closing brace/bracket/paren. This loop is white hot, so it
2886 ;; plays ugly tricks to go fast. DON'T PUT ANYTHING INTO THIS
2887 ;; LOOP WHICH ISN'T ABSOLUTELY NECESSARY!!! It terminates when a
2888 ;; call of `scan-lists' signals an error, which happens when there
2889 ;; are no more b/b/p's to scan.
2900 ;; We've just entered a deeper nesting level.
2902 ;; Insert the brace pair (if present) and the single open
2903 ;; paren/brace/bracket into `c-state-cache' It cannot be
2904 ;; inside a macro, except one around point, because of what
2905 ;; `c-neutralize-syntax-in-CPP' has done.
2907 ;; Insert the opening brace/bracket/paren position.
2909 ;; Clear admin stuff for the next more nested part of the scan.
2913 ;; All open p/b/b's at this nesting level, if any, have probably
2914 ;; been closed by matching/mismatching ones. We're probably
2915 ;; finished - we just need to check for having found an
2916 ;; unmatched )/}/], which we ignore. Such a )/}/] can't be in a
2917 ;; macro, due the action of `c-neutralize-syntax-in-CPP'.
2920 ;; Record the final, innermost, brace-pair if there is one.
2923 ;; Determine a good pos
2931 here-bol))
2940 ;; Remove stale entries from the `c-cache-state', i.e. those which will
2941 ;; not be in it when it is amended for position HERE. This may involve
2942 ;; replacing a CONS element for a brace pair containing HERE with its car.
2943 ;; Additionally, the "outermost" open-brace entry before HERE will be
2944 ;; converted to a cons if the matching close-brace is below HERE.
2945 ;;
2946 ;; START-POINT is a "maximal" "safe position" - there must be no open
2947 ;; parens/braces/brackets between START-POINT and HERE.
2948 ;;
2949 ;; As a second thing, calculate the result of parse-partial-sexp at
2950 ;; PPS-POINT, w.r.t. START-POINT. The motivation here is that
2951 ;; `c-state-cache-good-pos' may become PPS-POINT, but the caller may need to
2952 ;; adjust it to get outside a string/comment. (Sorry about this! The code
2953 ;; needs to be FAST).
2954 ;;
2955 ;; Return a list (GOOD-POS SCAN-BACK-POS CONS-SEPARATED PPS-STATE), where
2956 ;; o - GOOD-POS is a position where the new value `c-state-cache' is known
2957 ;; to be good (we aim for this to be as high as possible);
2958 ;; o - SCAN-BACK-POS, if not nil, indicates there may be a brace pair
2959 ;; preceding POS which needs to be recorded in `c-state-cache'. It is a
2960 ;; position to scan backwards from. It is the position of the "{" of the
2961 ;; last element to be removed from `c-state-cache', when that elt is a
2962 ;; cons, otherwise nil.
2963 ;; o - CONS-SEPARATED is t when a cons element in `c-state-cache' has been
2964 ;; replaced by its car because HERE lies inside the brace pair represented
2965 ;; by the cons.
2966 ;; o - PPS-STATE is the parse-partial-sexp state at PPS-POINT.
2976 ; start-point or nil
2983 ; HERE), or nil.
2989 pps-state ; Will be 9 or 10 elements long.
2990 pos
2991 upper-lim ; ,beyond which `c-state-cache' entries are removed
2992 scan-back-pos
2993 cons-separated
2994 pair-beg pps-point-state target-depth)
2996 ;; Remove entries beyond HERE. Also remove any entries inside
2997 ;; a macro, unless HERE is in the same macro.
3002 here
3008 ;; If `upper-lim' is inside the last recorded brace pair, remove its
3009 ;; RBrace and indicate we'll need to search backwards for a previous
3010 ;; brace pair.
3016 cons-separated t))
3018 ;; The next loop jumps forward out of a nested level of parens each
3019 ;; time round; the corresponding elements in `c-state-cache' are
3020 ;; removed. `pos' is just after the brace-pair or the open paren at
3021 ;; (car c-state-cache). There can be no open parens/braces/brackets
3022 ;; between `start-point'/`start-point-actual-macro-start' and HERE,
3023 ;; due to the interface spec to this function.
3026 in-macro-start)))
3028 ; marked by a `category' text property.
3029 start-point))
3040 ;; Do nothing when we've merely reached pps-point.
3041 )
3043 ;; Scan!
3047 target-depth
3048 nil pps-state))
3058 ;; We've just moved out of the paren pair containing the brace-pair
3059 ;; at (car c-state-cache). `pair-beg' is where the open paren is,
3060 ;; and is potentially where the open brace of a cons in
3061 ;; c-state-cache will be.
3067 ; containing Lparen
3071 ; together with enclosed brace pair.
3072 ;; (t nil) ; Ignore an unmated Rparen.
3073 )))
3078 nil nil ; TARGETDEPTH, STOPBEFORE
3079 pps-state)))
3081 ;; If the last paren pair we moved out of was actually a brace pair,
3082 ;; insert it into `c-state-cache'.
3087 c-state-cache)))
3092 ;; Strip stale elements of `c-state-cache' by moving backwards through the
3093 ;; buffer, and inform the caller of the scenario detected.
3094 ;;
3095 ;; HERE is the position we're setting `c-state-cache' for.
3096 ;; CACHE-POS (a locally bound variable) is just after the latest recorded
3097 ;; position in `c-state-cache' before HERE, or a position at or near
3098 ;; point-min which isn't in a literal.
3099 ;;
3100 ;; This function must only be called only when (> `c-state-cache-good-pos'
3101 ;; HERE). Usually the gap between CACHE-POS and HERE is large. It is thus
3102 ;; optimized to eliminate (or minimize) scanning between these two
3103 ;; positions.
3104 ;;
3105 ;; Return a three element list (GOOD-POS SCAN-BACK-POS FWD-FLAG), where:
3106 ;; o - GOOD-POS is a "good position", where `c-state-cache' is valid, or
3107 ;; could become so after missing elements are inserted into
3108 ;; `c-state-cache'. This is JUST AFTER an opening or closing
3109 ;; brace/paren/bracket which is already in `c-state-cache' or just before
3110 ;; one otherwise. exceptionally (when there's no such b/p/b handy) the BOL
3111 ;; before `here''s line, or the start of the literal containing it.
3112 ;; o - SCAN-BACK-POS, if non-nil, indicates there may be a brace pair
3113 ;; preceding POS which isn't recorded in `c-state-cache'. It is a position
3114 ;; to scan backwards from.
3115 ;; o - FWD-FLAG, if non-nil, indicates there may be parens/braces between
3116 ;; POS and HERE which aren't recorded in `c-state-cache'.
3117 ;;
3118 ;; The comments in this defun use "paren" to mean parenthesis or square
3119 ;; bracket (as contrasted with a brace), and "(" and ")" likewise.
3120 ;;
3121 ;; . {..} (..) (..) ( .. { } ) (...) ( .... . ..)
3122 ;; | | | | | |
3123 ;; CP E here D C good
3126 pa ren ; positions of "(" and ")"
3127 dropped-cons ; whether the last element dropped from `c-state-cache'
3128 ; was a cons (representing a brace-pair)
3129 good-pos ; see above.
3130 lit ; (START . END) of a literal containing some point.
3131 here-lit-start here-lit-end ; bounds of literal containing `here'
3132 ; or `here' itself.
3137 ;; Remove completely irrelevant entries from `c-state-cache'.
3143 ;; At this stage, (>= pos here);
3144 ;; (< (c-state-cache-top-lparen) here) (or is nil).
3149 ;; CASE 1: The top of the cache is a brace pair which now encloses
3150 ;; `here'. As good-pos, return the address. of the "{". Since we've no
3151 ;; knowledge of what's inside these braces, we have no alternative but
3152 ;; to direct the caller to scan the buffer from the opening brace.
3156 ; entry into a { entry, so the caller needs to
3157 ; search for a brace pair before the {.
3159 ;; `here' might be inside a literal. Check for this.
3164 ;; Has `here' just "newly entered" a macro?
3177 ;; `here' might be nested inside any depth of parens (or brackets but
3178 ;; not braces). Scan backwards to find the outermost such opening
3179 ;; paren, if there is one. This will be the scan position to return.
3186 ;; CASE 2: Address of outermost ( or [ which now encloses `here', but
3187 ;; didn't enclose the (previous) `c-state-cache-good-pos'. If there is
3188 ;; a brace pair preceding this, it will already be in `c-state-cache',
3189 ;; unless there was a brace pair after it, i.e. there'll only be one to
3190 ;; scan for if we've just deleted one.
3193 ;; `here' isn't enclosed in a (previously unrecorded) bracket/paren.
3194 ;; Further forward scanning isn't needed, but we still need to find a
3195 ;; GOOD-POS. Step out of all enclosing "("s on HERE's line.
3207 ;; CASE 3: After a }/)/] before `here''s BOL.
3212 ;; CASE 3.5: Just after an existing entry in `c-state-cache' on `here''s
3213 ;; line or the previous line.
3217 ;; CASE 4; Best of a bad job: BOL before `here-bol', or beginning of
3218 ;; literal containing it.
3222 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3223 ;; Externally visible routines.
3227 c-state-cache-good-pos 1
3228 c-state-nonlit-pos-cache nil
3229 c-state-nonlit-pos-cache-limit 1
3230 c-state-semi-nonlit-pos-cache nil
3231 c-state-semi-nonlit-pos-cache-limit 1
3232 c-state-brace-pair-desert nil
3233 c-state-point-min 1
3234 c-state-point-min-lit-type nil
3235 c-state-point-min-lit-start nil
3236 c-state-min-scan-pos 1
3237 c-state-old-cpp-beg nil
3238 c-state-old-cpp-end nil)
3241 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3242 ;; Debugging routines to dump `c-state-cache' in a "replayable" form.
3243 ;; (defmacro c-sc-de (elt) ; "c-state-cache-dump-element"
3244 ;; `(format ,(concat "(setq " (symbol-name elt) " %s) ") ,elt))
3245 ;; (defmacro c-sc-qde (elt) ; "c-state-cache-quote-dump-element"
3246 ;; `(format ,(concat "(setq " (symbol-name elt) " '%s) ") ,elt))
3247 ;; (defun c-state-dump ()
3248 ;; ;; For debugging.
3249 ;; ;(message
3250 ;; (concat
3251 ;; (c-sc-qde c-state-cache)
3252 ;; (c-sc-de c-state-cache-good-pos)
3253 ;; (c-sc-qde c-state-nonlit-pos-cache)
3254 ;; (c-sc-de c-state-nonlit-pos-cache-limit)
3255 ;; (c-sc-qde c-state-brace-pair-desert)
3256 ;; (c-sc-de c-state-point-min)
3257 ;; (c-sc-de c-state-point-min-lit-type)
3258 ;; (c-sc-de c-state-point-min-lit-start)
3259 ;; (c-sc-de c-state-min-scan-pos)
3260 ;; (c-sc-de c-state-old-cpp-beg)
3261 ;; (c-sc-de c-state-old-cpp-end)))
3262 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3265 ;; Invalidate all info on `c-state-cache' that applies to the buffer at HERE
3266 ;; or higher and set `c-state-cache-good-pos' accordingly. The cache is
3267 ;; left in a consistent state.
3268 ;;
3269 ;; This is much like `c-whack-state-after', but it never changes a paren
3270 ;; pair element into an open paren element. Doing that would mean that the
3271 ;; new open paren wouldn't have the required preceding paren pair element.
3272 ;;
3273 ;; This function is called from c-before-change.
3275 ;; The caches of non-literals:
3276 ;; Note that we use "<=" for the possibility of the second char of a two-char
3277 ;; comment opener being typed; this would invalidate any cache position at
3278 ;; HERE.
3284 ;; `c-state-cache':
3285 ;; Case 1: if `here' is in a literal containing point-min, everything
3286 ;; becomes (or is already) nil.
3290 c-state-cache-good-pos nil
3291 c-state-min-scan-pos nil)
3293 ;; Truncate `c-state-cache' and set `c-state-cache-good-pos' to a value
3294 ;; below `here'. To maintain its consistency, we may need to insert a new
3295 ;; brace pair.
3297 too-high-pa ; recorded {/(/[ next above or just below here, or nil.
3298 dropped-cons ; was the last removed element a brace pair?
3299 pa)
3300 ;; The easy bit - knock over-the-top bits off `c-state-cache'.
3307 ;; Do we need to add in an earlier brace pair, having lopped one off?
3314 ;; The brace-pair desert marker:
3322 ;; Find and record all noteworthy parens between some good point earlier in
3323 ;; the file and point. That good point is at least the beginning of the
3324 ;; top-level construct we are in, or the beginning of the preceding
3325 ;; top-level construct if we aren't in one.
3326 ;;
3327 ;; The returned value is a list of the noteworthy parens with the last one
3328 ;; first. If an element in the list is an integer, it's the position of an
3329 ;; open paren (of any type) which has not been closed before the point. If
3330 ;; an element is a cons, it gives the position of a closed BRACE paren
3331 ;; pair[*]; the car is the start brace position and the cdr is the position
3332 ;; following the closing brace. Only the last closed brace paren pair
3333 ;; before each open paren and before the point is recorded, and thus the
3334 ;; state never contains two cons elements in succession. When a close brace
3335 ;; has no matching open brace (e.g., the matching brace is outside the
3336 ;; visible region), it is not represented in the returned value.
3337 ;;
3338 ;; [*] N.B. The close "brace" might be a mismatching close bracket or paren.
3339 ;; This defun explicitly treats mismatching parens/braces/brackets as
3340 ;; matching. It is the open brace which makes it a "brace" pair.
3341 ;;
3342 ;; If POINT is within a macro, open parens and brace pairs within
3343 ;; THIS macro MIGHT be recorded. This depends on whether their
3344 ;; syntactic properties have been suppressed by
3345 ;; `c-neutralize-syntax-in-CPP'. This might need fixing (2008-12-11).
3346 ;;
3347 ;; Currently no characters which are given paren syntax with the
3348 ;; syntax-table property are recorded, i.e. angle bracket arglist
3349 ;; parens are never present here. Note that this might change.
3350 ;;
3351 ;; BUG: This function doesn't cope entirely well with unbalanced
3352 ;; parens in macros. (2008-12-11: this has probably been resolved
3353 ;; by the function `c-neutralize-syntax-in-CPP'.) E.g. in the
3354 ;; following case the brace before the macro isn't balanced with the
3355 ;; one after it:
3356 ;;
3357 ;; {
3358 ;; #define X {
3359 ;; }
3360 ;;
3361 ;; Note to maintainers: this function DOES get called with point
3362 ;; within comments and strings, so don't assume it doesn't!
3363 ;;
3364 ;; This function might do hidden buffer changes.
3367 strategy ; 'forward, 'backward etc..
3368 ;; Candidate positions to start scanning from:
3369 cache-pos ; highest position below HERE already existing in
3370 ; cache (or 1).
3371 good-pos
3372 start-point ; (when scanning forward) a place below HERE where there
3373 ; are no open parens/braces between it and HERE.
3374 bopl-state
3375 res
3376 cons-separated
3378 ;; If POINT-MIN has changed, adjust the cache
3382 ;; Strategy?
3389 c-state-cache-good-pos start-point))
3391 ;; SCAN!
3399 ; start-point)
3409 good-pos)))
3421 good-pos)))
3425 c-state-cache-good-pos nil))))
3427 c-state-cache)
3430 ;; This is a wrapper over `c-invalidate-state-cache-1'.
3431 ;;
3432 ;; It suppresses the syntactic effect of the < and > (template) brackets and
3433 ;; of all parens in preprocessor constructs, except for any such construct
3434 ;; containing point. We can then call `c-invalidate-state-cache-1' without
3435 ;; worrying further about macros and template delimiters.
3437 ;; Emacs
3442 c-state-old-cpp-beg
3443 c-state-old-cpp-end
3447 ;; XEmacs
3451 ;; If PLACE is non-nil, return a marker marking it, otherwise nil.
3452 ;; We (re)use MARKER.
3458 ;; This is a wrapper over `c-parse-state-1'. See that function for a
3459 ;; description of the functionality and return value.
3460 ;;
3461 ;; It suppresses the syntactic effect of the < and > (template) brackets and
3462 ;; of all parens in preprocessor constructs, except for any such construct
3463 ;; containing point. We can then call `c-parse-state-1' without worrying
3464 ;; further about macros and template delimiters.
3471 here-cpp-beg)
3473 ;; FIXME!!! Put in a `condition-case' here to protect the integrity of the
3474 ;; subsystem.
3477 ;; Emacs
3481 here-cpp-beg here-cpp-end
3485 ;; XEmacs
3489 c-state-old-cpp-end
3492 ;; Debug tool to catch cache inconsistencies. This is called from
3493 ;; 000tests.el.
3516 c-state-cache-good-pos
3517 c-state-nonlit-pos-cache
3518 c-state-nonlit-pos-cache-limit
3519 c-state-semi-nonlit-pos-cache
3520 c-state-semi-nonlit-pos-cache-limit
3521 c-state-brace-pair-desert
3522 c-state-point-min
3523 c-state-point-min-lit-type
3524 c-state-point-min-lit-start
3525 c-state-min-scan-pos
3526 c-state-old-cpp-beg
3527 c-state-old-cpp-end
3528 c-parse-state-point))))
3550 conses-not-ok))
3567 ;; The cache can actually go further back due to the ad-hoc way
3568 ;; the first paren is found, so try to whack off a bit of its
3569 ;; start before complaining.
3570 ;; (save-excursion
3571 ;; (goto-char (or (c-least-enclosing-brace res2) (point)))
3572 ;; (c-beginning-of-defun-1)
3573 ;; (while (not (or (bobp) (eq (char-after) ?{)))
3574 ;; (c-beginning-of-defun-1))
3575 ;; (unless (equal (c-whack-state-before (point) res1) res2)
3576 ;; (message (concat "c-parse-state inconsistency at %s: "
3577 ;; "using cache: %s, from scratch: %s")
3578 ;; here res1 res2)))
3581 here res1 res2)
3587 here res1)
3592 res2 ; res1 correct a cascading series of errors ASAP
3593 ))
3599 'c-debug-parse-state
3607 \f
3609 ;; Whack off any state information from PAREN-STATE which lies
3610 ;; before BUFPOS. Not destructive on PAREN-STATE.
3613 car)
3624 ;; Whack off any state information from PAREN-STATE which lies at or
3625 ;; after BUFPOS. Not destructive on PAREN-STATE.
3630 ;; just check the car, because in a balanced brace
3631 ;; expression, it must be impossible for the corresponding
3632 ;; close brace to be before point, but the open brace to
3633 ;; be after.
3635 nil ; whack it off
3637 ;; its possible that the open brace is before
3638 ;; bufpos, but the close brace is after. In that
3639 ;; case, convert this to a non-cons element. The
3640 ;; rest of the state is before bufpos, so we're
3641 ;; done.
3643 ;; we know that both the open and close braces are
3644 ;; before bufpos, so we also know that everything else
3645 ;; on state is before bufpos.
3648 nil ; whack it off
3649 ;; it's before bufpos, so everything else should too.
3652 nil)))
3655 ;; Return the bufpos of the innermost enclosing open paren before
3656 ;; bufpos, or nil if none was found.
3666 enclosingp))
3669 ;; Return the bufpos of the outermost enclosing open paren, or nil
3670 ;; if none was found.
3677 pos))
3680 ;; Return the closest "safe" position recorded on PAREN-STATE that
3681 ;; is higher up than BUFPOS. Return nil if PAREN-STATE doesn't
3682 ;; contain any. Return nil if BUFPOS is nil, which is useful to
3683 ;; find the closest limit before a given limit that might be nil.
3684 ;;
3685 ;; A "safe" position is a position at or after a recorded open
3686 ;; paren, or after a recorded close paren. The returned position is
3687 ;; thus either the first position after a close brace, or the first
3688 ;; position after an enclosing paren, or at the enclosing paren in
3689 ;; case BUFPOS is immediately after it.
3699 ;; See below.
3702 ;; elem is the position at and not after the opening paren, so
3703 ;; we can go forward one more step unless it's equal to
3704 ;; bufpos. This is useful in some cases avoid an extra paren
3705 ;; level between the safe position and bufpos.
3710 ;; This is used for `font-lock-beginning-of-syntax-function'. It
3711 ;; goes to the closest previous point that is known to be outside
3712 ;; any string literal or comment. `c-state-cache' is used if it has
3713 ;; a position in the vicinity.
3715 elem
3718 ;; Note: Similar code in `c-safe-position'. The
3719 ;; difference is that we accept a safe position at
3720 ;; the point and don't bother to go forward past open
3721 ;; parens.
3736 ;; The position is far back. Try `c-beginning-of-defun-1'
3737 ;; (although we can't be entirely sure it will go to a position
3738 ;; outside a comment or string in current emacsen). FIXME:
3739 ;; Consult `syntax-ppss' here.
3744 \f
3745 ;; Tools for scanning identifiers and other tokens.
3748 "Return non-nil if the point is on or directly after an identifier.
3749 Keywords are recognized and not considered identifiers. If an
3750 identifier is detected, the returned value is its starting position.
3751 If an identifier ends at the point and another begins at it \(can only
3752 happen in Pike) then the point for the preceding one is returned.
3754 Note that this function might do hidden buffer changes. See the
3755 comment at the start of cc-engine.el for more info."
3757 ;; FIXME: Shouldn't this function handle "operator" in C++?
3764 ;; Check for a normal (non-keyword) identifier.
3770 ;; Handle the `<operator> syntax in Pike.
3774 t
3781 ;; Handle the "operator +" syntax in C++.
3798 )))
3801 ;; If the point is at the end of a symbol then skip backward to the
3802 ;; beginning of it. Don't move otherwise. Return non-nil if point
3803 ;; moved.
3804 ;;
3805 ;; This function might do hidden buffer changes.
3808 ;; Handle the `<operator> syntax in Pike.
3814 t
3816 nil)))))
3819 ;; Move to the beginning of the current token. Do not move if not
3820 ;; in the middle of one. BACK-LIMIT may be used to bound the
3821 ;; backward search; if given it's assumed to be at the boundary
3822 ;; between two tokens. Return non-nil if the point is moved, nil
3823 ;; otherwise.
3824 ;;
3825 ;; This function might do hidden buffer changes.
3832 ;; `c-nonsymbol-token-regexp' should always match
3833 ;; since we've skipped backward over punctuation
3834 ;; or paren syntax, but consume one char in case
3835 ;; it doesn't so that we don't leave point before
3836 ;; some earlier incorrect token.
3843 ;; Move to the end of the current token. Do not move if not in the
3844 ;; middle of one. BACK-LIMIT may be used to bound the backward
3845 ;; search; if given it's assumed to be at the boundary between two
3846 ;; tokens. Return non-nil if the point is moved, nil otherwise.
3847 ;;
3848 ;; This function might do hidden buffer changes.
3856 ;; `c-nonsymbol-token-regexp' should always match since
3857 ;; we've skipped backward over punctuation or paren
3858 ;; syntax, but move forward in case it doesn't so that
3859 ;; we don't leave point earlier than we started with.
3875 "Move forward by tokens.
3876 A token is defined as all symbols and identifiers which aren't
3878 treated properly). Point is always either left at the beginning of a
3879 token or not moved at all. COUNT specifies the number of tokens to
3880 move; a negative COUNT moves in the opposite direction. A COUNT of 0
3881 moves to the next token beginning only if not already at one. If
3882 BALANCED is true, move over balanced parens, otherwise move into them.
3883 Also, if BALANCED is true, never move out of an enclosing paren.
3885 LIMIT sets the limit for the movement and defaults to the point limit.
3886 The case when LIMIT is set in the middle of a token, comment or macro
3887 is handled correctly, i.e. the point won't be left there.
3889 Return the number of tokens left to move \(positive or negative). If
3890 BALANCED is true, a move over a balanced paren counts as one. Note
3891 that if COUNT is 0 and no appropriate token beginning is found, 1 will
3892 be returned. Thus, a return value of 0 guarantees that point is at
3893 the requested position and a return value less \(without signs) than
3894 COUNT guarantees that point is at the beginning of some token.
3896 Note that this function might do hidden buffer changes. See the
3897 comment at the start of cc-engine.el for more info."
3904 c-jump-syntax-balanced
3905 c-jump-syntax-unbalanced))
3910 ;; If count is zero we should jump if in the middle of a token.
3917 ;; Skip whitespace. Count this as a move if we did in
3918 ;; fact move.
3922 ;; Moved out of bounds. Make sure the returned count isn't zero.
3927 ;; Use `condition-case' to avoid having the limit tests
3928 ;; inside the loop.
3936 t)
3939 t)
3940 ;; `c-nonsymbol-token-regexp' above should always
3941 ;; match if there are correct tokens. Try to
3942 ;; widen to see if the limit was set in the
3943 ;; middle of one, else fall back to treating
3944 ;; the offending thing as a one character token.
3949 nil)
3952 t))))
3962 count)))
3965 "Move backward by tokens.
3966 See `c-forward-token-2' for details."
3974 c-jump-syntax-balanced
3975 c-jump-syntax-unbalanced))
3979 ;; The count is zero so try to skip to the beginning of the
3980 ;; current token.
3984 ;; The limit is inside the same token, so return 1.
3987 ;; We're not in the middle of a token. If there's
3988 ;; whitespace after the point then we must move backward,
3989 ;; so set count to 1 in that case.
3991 ;; If we're looking at a '#' that might start a cpp
3992 ;; directive then we have to do a more elaborate check.
4005 ;; Use `condition-case' to avoid having to check for buffer
4006 ;; limits in `backward-char', `scan-sexps' and `goto-char' below.
4015 ;; This can be very inefficient if there's a long
4016 ;; sequence of operator tokens without any separation.
4017 ;; That doesn't happen in practice, anyway.
4027 count)))
4030 "Like `c-forward-token-2' but doesn't treat multicharacter operator
4032 characters are jumped over character by character. This function is
4033 for compatibility only; it's only a wrapper over `c-forward-token-2'."
4038 "Like `c-backward-token-2' but doesn't treat multicharacter operator
4040 characters are jumped over character by character. This function is
4041 for compatibility only; it's only a wrapper over `c-backward-token-2'."
4045 \f
4046 ;; Tools for doing searches restricted to syntactically relevant text.
4049 paren-level not-inside-token
4050 lookbehind-submatch)
4051 "Like `re-search-forward', but only report matches that are found
4052 in syntactically significant text. I.e. matches in comments, macros
4053 or string literals are ignored. The start point is assumed to be
4054 outside any comment, macro or string literal, or else the content of
4055 that region is taken as syntactically significant text.
4057 If PAREN-LEVEL is non-nil, an additional restriction is added to
4058 ignore matches in nested paren sexps. The search will also not go
4059 outside the current list sexp, which has the effect that if the point
4060 should be moved to BOUND when no match is found \(i.e. NOERROR is
4061 neither nil nor t), then it will be at the closing paren if the end of
4062 the current list sexp is encountered first.
4064 If NOT-INSIDE-TOKEN is non-nil, matches in the middle of tokens are
4065 ignored. Things like multicharacter operators and special symbols
4067 constants.
4069 If LOOKBEHIND-SUBMATCH is non-nil, it's taken as a number of a
4070 subexpression in REGEXP. The end of that submatch is used as the
4071 position to check for syntactic significance. If LOOKBEHIND-SUBMATCH
4072 isn't used or if that subexpression didn't match then the start
4074 subexpression is never tested before the starting position, so it
4078 subexpression can match the end of nonwhite syntactic whitespace,
4079 i.e. the end of comments or cpp directives. This since the function
4080 skips over such things before resuming the search. It's on the other
4082 matches syntactic whitespace.
4084 Bug: Unbalanced parens inside cpp directives are currently not handled
4085 correctly \(i.e. they don't get ignored as they should) when
4086 PAREN-LEVEL is set.
4088 Note that this function might do hidden buffer changes. See the
4089 comment at the start of cc-engine.el for more info."
4094 ;;(message "c-syntactic-re-search-forward %s %s %S" (point) bound regexp)
4097 tmp
4098 ;; Start position for the last search.
4099 search-pos
4100 ;; The `parse-partial-sexp' state between the start position
4101 ;; and the point.
4102 state
4103 ;; The current position after the last state update. The next
4104 ;; `parse-partial-sexp' continues from here.
4106 ;; The position at which to check the state and the state
4107 ;; there. This is separate from `state-pos' since we might
4108 ;; need to back up before doing the next search round.
4109 check-pos check-state
4110 ;; Last position known to end a token.
4112 ;; Set when a valid match is found.
4113 found)
4131 state-pos check-pos paren-level nil state))
4133 check-state state))
4135 ;; NOTE: If we got a look behind subexpression and get
4136 ;; an insignificant match in something that isn't
4137 ;; syntactic whitespace (i.e. strings or in nested
4138 ;; parentheses), then we can never skip more than a
4139 ;; single character from the match start position
4140 ;; (i.e. `state-pos' here) before continuing the
4141 ;; search. That since the look behind subexpression
4142 ;; might match the end of the insignificant region in
4143 ;; the next search.
4147 ;; Match inside a line comment. Skip to eol. Use
4148 ;; `re-search-forward' instead of `skip-chars-forward' to get
4149 ;; the right bound behavior.
4153 ;; Match inside a block comment. Skip to the '*/'.
4160 ;; Match in the middle of the opener of a block or line
4161 ;; comment.
4166 ;; The last `parse-partial-sexp' above might have
4167 ;; stopped short of the real check position if the end
4168 ;; of the current sexp was encountered in paren-level
4169 ;; mode. The checks above are always false in that
4170 ;; case, and since they can do better skipping in
4171 ;; lookbehind-submatch mode, we do them before
4172 ;; checking the paren level.
4176 ;; Check the paren level first since we're short of the
4177 ;; syntactic checking position if the end of the
4178 ;; current sexp was encountered by `parse-partial-sexp'.
4181 ;; Inside a nested paren sexp.
4183 ;; See the NOTE above.
4185 ;; Skip out of the paren quickly.
4189 ;; Have exited the current paren sexp.
4192 ;; The last `parse-partial-sexp' call above
4193 ;; has left us just after the closing paren
4194 ;; in this case, so we can modify the bound
4195 ;; to leave the point at the right position
4196 ;; upon return.
4198 nil)
4202 ;; Match inside a string.
4205 ;; See the NOTE above.
4207 ;; Skip to the end of the string before continuing.
4216 continue)))
4221 ;; Match inside a macro. Skip to the end of it.
4229 (< check-pos
4235 ;; Inside a token.
4237 ;; See the NOTE above.
4242 ;; A real match.
4244 nil)))
4246 ;; Should loop to search again, but take care to avoid
4247 ;; looping on the same spot.
4251 nil
4254 t))))
4260 ;;(message "c-syntactic-re-search-forward done %s" (or (match-end 0) (point)))
4267 ;; Search failed. Set point as appropriate.
4271 nil)))
4276 ;; Return the start of the literal point is in, or nil.
4277 ;; We read and write the variables `safe-pos', `safe-pos-list', `state'
4278 ;; bound in the caller.
4280 ;; Use `parse-partial-sexp' from a safe position down to the point to check
4281 ;; if it's outside comments and strings.
4284 ;; Pick a safe position as close to the point as possible.
4285 ;;
4286 ;; FIXME: Consult `syntax-ppss' here if our cache doesn't give a good
4287 ;; position.
4299 ;; Cache positions along the way to use if we have to back up more. We
4300 ;; cache every closing paren on the same level. If the paren cache is
4301 ;; relevant in this region then we're typically already on the same
4302 ;; level as the target position. Note that we might cache positions
4303 ;; after opening parens in case safe-pos is in a nested list. That's
4304 ;; both uncommon and harmless.
4312 ;; If the state contains the start of the containing sexp we cache that
4313 ;; position too, so that parse-partial-sexp in the next run has a bigger
4314 ;; chance of starting at the same level as the target position and thus
4315 ;; will get more good safe positions into the list.
4321 ;; Inside string or comment. Continue search at the
4322 ;; beginning of it.
4326 "Like `skip-chars-backward' but only look at syntactically relevant chars,
4327 i.e. don't stop at positions inside syntactic whitespace or string
4328 literals. Preprocessor directives are also ignored, with the exception
4329 of the one that the point starts within, if any. If LIMIT is given,
4330 it's assumed to be at a syntactically relevant position.
4332 If PAREN-LEVEL is non-nil, the function won't stop in nested paren
4333 sexps, and the search will also not go outside the current paren sexp.
4334 However, if LIMIT or the buffer limit is reached inside a nested paren
4335 then the point will be left at the limit.
4337 Non-nil is returned if the point moved, nil otherwise.
4339 Note that this function might do hidden buffer changes. See the
4340 comment at the start of cc-engine.el for more info."
4344 state-2
4345 ;; A list of syntactically relevant positions in descending
4346 ;; order. It's used to avoid scanning repeatedly over
4347 ;; potentially large regions with `parse-partial-sexp' to verify
4348 ;; each position. Used in `c-ssb-lit-begin'
4349 safe-pos-list
4350 ;; The result from `c-beginning-of-macro' at the start position or the
4351 ;; start position itself if it isn't within a macro. Evaluated on
4352 ;; demand.
4353 start-macro-beg
4354 ;; The earliest position after the current one with the same paren
4355 ;; level. Used only when `paren-level' is set.
4356 lit-beg
4361 ;; The next loop "tries" to find the end point each time round,
4362 ;; loops when it hasn't succeeded.
4368 ;; Don't stop inside a literal.
4371 t)))
4383 ;; Not at the right level.
4386 ;; We stop above if we go out of a paren.
4387 ;; Now check whether it precedes or is
4388 ;; nested in the starting sexp.
4392 pps-end-pos paren-level-pos
4393 nil nil state-2))
4396 ;; We've stopped short of the starting position
4397 ;; so the hit was inside a nested list. Go up
4398 ;; until we are at the right level.
4407 nil)
4408 t))
4411 nil))
4413 ;; The hit was outside the list at the start
4414 ;; position. Go to the start of the list and exit.
4416 nil))
4419 ;; Inside a macro.
4427 t
4429 ;; It's inside the same macro we started in so it's
4430 ;; a relevant match.
4432 nil))))))
4436 ;; Skip syntactic ws afterwards so that we don't stop at the
4437 ;; end of a comment if `skip-chars' is something like "^/".
4441 ;; We might want to extend this with more useful return values in
4442 ;; the future.
4445 ;; The following is an alternative implementation of
4446 ;; `c-syntactic-skip-backward' that uses backward movement to keep
4447 ;; track of the syntactic context. It turned out to be generally
4448 ;; slower than the one above which uses forward checks from earlier
4449 ;; safe positions.
4450 ;;
4451 ;;(defconst c-ssb-stop-re
4452 ;; ;; The regexp matching chars `c-syntactic-skip-backward' needs to
4453 ;; ;; stop at to avoid going into comments and literals.
4454 ;; (concat
4455 ;; ;; Match comment end syntax and string literal syntax. Also match
4456 ;; ;; '/' for block comment endings (not covered by comment end
4457 ;; ;; syntax).
4458 ;; "\\s>\\|/\\|\\s\""
4459 ;; (if (memq 'gen-string-delim c-emacs-features)
4460 ;; "\\|\\s|"
4461 ;; "")
4462 ;; (if (memq 'gen-comment-delim c-emacs-features)
4463 ;; "\\|\\s!"
4464 ;; "")))
4465 ;;
4466 ;;(defconst c-ssb-stop-paren-re
4467 ;; ;; Like `c-ssb-stop-re' but also stops at paren chars.
4468 ;; (concat c-ssb-stop-re "\\|\\s(\\|\\s)"))
4469 ;;
4470 ;;(defconst c-ssb-sexp-end-re
4471 ;; ;; Regexp matching the ending syntax of a complex sexp.
4472 ;; (concat c-string-limit-regexp "\\|\\s)"))
4473 ;;
4474 ;;(defun c-syntactic-skip-backward (skip-chars &optional limit paren-level)
4475 ;; "Like `skip-chars-backward' but only look at syntactically relevant chars,
4476 ;;i.e. don't stop at positions inside syntactic whitespace or string
4477 ;;literals. Preprocessor directives are also ignored. However, if the
4478 ;;point is within a comment, string literal or preprocessor directory to
4479 ;;begin with, its contents is treated as syntactically relevant chars.
4480 ;;If LIMIT is given, it limits the backward search and the point will be
4481 ;;left there if no earlier position is found.
4482 ;;
4483 ;;If PAREN-LEVEL is non-nil, the function won't stop in nested paren
4484 ;;sexps, and the search will also not go outside the current paren sexp.
4485 ;;However, if LIMIT or the buffer limit is reached inside a nested paren
4486 ;;then the point will be left at the limit.
4487 ;;
4488 ;;Non-nil is returned if the point moved, nil otherwise.
4489 ;;
4490 ;;Note that this function might do hidden buffer changes. See the
4491 ;;comment at the start of cc-engine.el for more info."
4492 ;;
4493 ;; (save-restriction
4494 ;; (when limit
4495 ;; (narrow-to-region limit (point-max)))
4496 ;;
4497 ;; (let ((start (point)))
4498 ;; (catch 'done
4499 ;; (while (let ((last-pos (point))
4500 ;; (stop-pos (progn
4501 ;; (skip-chars-backward skip-chars)
4502 ;; (point))))
4503 ;;
4504 ;; ;; Skip back over the same region as
4505 ;; ;; `skip-chars-backward' above, but keep to
4506 ;; ;; syntactically relevant positions.
4507 ;; (goto-char last-pos)
4508 ;; (while (and
4509 ;; ;; `re-search-backward' with a single char regexp
4510 ;; ;; should be fast.
4511 ;; (re-search-backward
4512 ;; (if paren-level c-ssb-stop-paren-re c-ssb-stop-re)
4513 ;; stop-pos 'move)
4514 ;;
4515 ;; (progn
4516 ;; (cond
4517 ;; ((looking-at "\\s(")
4518 ;; ;; `paren-level' is set and we've found the
4519 ;; ;; start of the containing paren.
4520 ;; (forward-char)
4521 ;; (throw 'done t))
4522 ;;
4523 ;; ((looking-at c-ssb-sexp-end-re)
4524 ;; ;; We're at the end of a string literal or paren
4525 ;; ;; sexp (if `paren-level' is set).
4526 ;; (forward-char)
4527 ;; (condition-case nil
4528 ;; (c-backward-sexp)
4529 ;; (error
4530 ;; (goto-char limit)
4531 ;; (throw 'done t))))
4532 ;;
4533 ;; (t
4534 ;; (forward-char)
4535 ;; ;; At the end of some syntactic ws or possibly
4536 ;; ;; after a plain '/' operator.
4537 ;; (let ((pos (point)))
4538 ;; (c-backward-syntactic-ws)
4539 ;; (if (= pos (point))
4540 ;; ;; Was a plain '/' operator. Go past it.
4541 ;; (backward-char)))))
4542 ;;
4543 ;; (> (point) stop-pos))))
4544 ;;
4545 ;; ;; Now the point is either at `stop-pos' or at some
4546 ;; ;; position further back if `stop-pos' was at a
4547 ;; ;; syntactically irrelevant place.
4548 ;;
4549 ;; ;; Skip additional syntactic ws so that we don't stop
4550 ;; ;; at the end of a comment if `skip-chars' is
4551 ;; ;; something like "^/".
4552 ;; (c-backward-syntactic-ws)
4553 ;;
4554 ;; (< (point) stop-pos))))
4555 ;;
4556 ;; ;; We might want to extend this with more useful return values
4557 ;; ;; in the future.
4558 ;; (/= (point) start))))
4560 \f
4561 ;; Tools for handling comments and string literals.
4564 "Return the type of literal point is in, if any.
4565 The return value is `c' if in a C-style comment, `c++' if in a C++
4566 style comment, `string' if in a string literal, `pound' if DETECT-CPP
4567 is non-nil and in a preprocessor line, or nil if somewhere else.
4568 Optional LIM is used as the backward limit of the search. If omitted,
4569 or nil, `c-beginning-of-defun' is used.
4571 The last point calculated is cached if the cache is enabled, i.e. if
4572 `c-in-literal-cache' is bound to a two element vector.
4574 Note that this function might do hidden buffer changes. See the
4575 comment at the start of cc-engine.el for more info."
4586 "Return a cons of the beginning and end positions of the comment or
4587 string surrounding point (including both delimiters), or nil if point
4589 to start parsing from. If NEAR is non-nil, then the limits of any
4591 spaces and tabs between point and the literal. The search for such a
4592 literal is done first in forward direction. If NOT-IN-DELIMITER is
4593 non-nil, the case when point is inside a starting delimiter won't be
4594 recognized. This only has effect for comments which have starting
4595 delimiters with more than one character.
4597 Note that this function might do hidden buffer changes. See the
4598 comment at the start of cc-engine.el for more info."
4614 ;; Search forward for a literal.
4625 ;; Search backward.
4636 ;; Block comment. Due to the nature of line
4637 ;; comments, they will always be covered by the
4638 ;; normal case above.
4641 ;; If LIM is bogus, beg will be bogus.
4645 ))))
4647 ;; In case external callers use this; it did have a docstring.
4651 "If the argument is a cons of two buffer positions (such as returned by
4652 `c-literal-limits'), and that range contains a C++ style line comment,
4653 then an extended range is returned that contains all adjacent line
4654 comments (i.e. all comments that starts in the same column with no
4655 empty lines or non-whitespace characters between them). Otherwise the
4656 argument is returned.
4658 Note that this function might do hidden buffer changes. See the
4659 comment at the start of cc-engine.el for more info."
4670 ;; Got to take care in the backward direction to handle
4671 ;; comments which are preceded by code.
4685 range)
4689 "Convenience function that given the result of `c-literal-limits',
4690 returns nil or the type of literal that the range surrounds, one
4691 of the symbols `c', `c++' or `string'. It's much faster than using
4692 `c-in-literal' and is intended to be used when you need both the
4693 type of a literal and its limits.
4695 Note that this function might do hidden buffer changes. See the
4696 comment at the start of cc-engine.el for more info."
4707 range))
4710 ;; Get a "safe place" approximately TRY-SIZE characters before START.
4711 ;; This doesn't preserve point.
4720 ;; Return a buffer position HOW-FAR-BACK non-literal characters from START
4721 ;; (default point). This is done by going back further in the buffer then
4722 ;; searching forward for literals. The position found won't be in a
4723 ;; literal. We start searching for the sought position TRY-SIZE (default
4724 ;; twice HOW-FAR-BACK) bytes back from START. This function must be fast.
4725 ;; :-)
4733 stack elt size
4736 ;; Move forward one literal each time round this loop.
4737 ;; Move forward to the start of a comment or string.
4739 pos
4740 start
4741 nil ; target-depth
4742 nil ; stop-before
4743 s ; state
4746 ;; Gather details of the non-literal-bit - starting pos and size.
4750 pos))
4754 ;; Move forward to the end of the comment/string.
4758 start
4759 nil ; target-depth
4760 nil ; stop-before
4761 s ; state
4765 ;; Now try and find enough non-literal characters recorded on the stack.
4766 ;; Go back one recorded literal each time round this loop.
4768 stack)
4773 ;; Have we found enough yet?
4783 ;; Return a buffer position about HOW-FAR non-literal characters forward
4784 ;; from START-POS (default point), which must not be inside a literal.
4791 ;; Scan over counted characters.
4793 pos
4795 nil ; target-depth
4796 nil ; stop-before
4797 s ; state
4801 ;; Scan over literal characters.
4804 pos
4806 nil ; target-depth
4807 nil ; stop-before
4808 s ; state
4813 \f
4814 ;; `c-find-decl-spots' and accompanying stuff.
4816 ;; Variables used in `c-find-decl-spots' to cache the search done for
4817 ;; the first declaration in the last call. When that function starts,
4818 ;; it needs to back up over syntactic whitespace to look at the last
4819 ;; token before the region being searched. That can sometimes cause
4820 ;; moves back and forth over a quite large region of comments and
4821 ;; macros, which would be repeated for each changed character when
4822 ;; we're called during fontification, since font-lock refontifies the
4823 ;; current line for each change. Thus it's worthwhile to cache the
4824 ;; first match.
4825 ;;
4826 ;; `c-find-decl-syntactic-pos' is a syntactically relevant position in
4827 ;; the syntactic whitespace less or equal to some start position.
4828 ;; There's no cached value if it's nil.
4829 ;;
4830 ;; `c-find-decl-match-pos' is the match position if
4831 ;; `c-find-decl-prefix-search' matched before the syntactic whitespace
4832 ;; at `c-find-decl-syntactic-pos', or nil if there's no such match.
4843 ; (defface c-debug-decl-spot-face
4844 ; '((t (:background "Turquoise")))
4845 ; "Debug face to mark the spots where `c-find-decl-spots' stopped.")
4846 ; (defface c-debug-decl-sws-face
4847 ; '((t (:background "Khaki")))
4848 ; "Debug face to mark the syntactic whitespace between the declaration
4849 ; spots and the preceding token end.")
4865 ;; Macro used inside `c-find-decl-spots'. It ought to be a defun,
4866 ;; but it contains lots of free variables that refer to things
4867 ;; inside `c-find-decl-spots'. The point is left at `cfd-match-pos'
4868 ;; if there is a match, otherwise at `cfd-limit'.
4869 ;;
4870 ;; The macro moves point forward to the next putative start of a declaration
4871 ;; or cfd-limit. This decl start is the next token after a "declaration
4872 ;; prefix". The declaration prefix is the earlier of `cfd-prop-match' and
4873 ;; `cfd-re-match'. `cfd-match-pos' is set to the decl prefix.
4874 ;;
4875 ;; This macro might do hidden buffer changes.
4878 ;; Find the next property match position if we haven't got one already.
4889 ;; Find the next `c-decl-prefix-or-start-re' match if we haven't
4890 ;; got one already.
4896 ;; Each time round, the next `while' moves forward over a pseudo match
4897 ;; of `c-decl-prefix-or-start-re' which is either inside a literal, or
4898 ;; is a ":" not preceded by "public", etc.. `cfd-re-match' and
4899 ;; `cfd-re-match-end' get set.
4906 ;; No match. Finish up and exit the loop.
4908 nil)
4911 ;; Matched the end of a token preceding a decl spot.
4915 ;; Matched a token that start a decl spot.
4918 c-literal-faces)
4919 ;; Pseudo match inside a comment or string literal. Skip out
4920 ;; of comments and string literals.
4932 ;; Found a ":" which isn't part of "public:", etc.
4933 t)
4936 ;; If our match was at the decl start, we have to back up over the
4937 ;; preceding syntactic ws to set `cfd-match-pos' and to catch
4938 ;; any decl spots in the syntactic ws.
4943 ;; Choose whichever match is closer to the start.
4946 cfd-re-match nil)
4948 cfd-prop-match nil))
4953 ;; Skip forward past comments only so we don't skip macros.
4955 ;; Set the position to continue at. We can avoid going over
4956 ;; the comments skipped above a second time, but it's possible
4957 ;; that the comment skipping has taken us past `cfd-prop-match'
4958 ;; since the property might be used inside comments.
4964 ;; Call CFD-FUN for each possible spot for a declaration, cast or
4965 ;; label from the point to CFD-LIMIT.
4966 ;;
4967 ;; CFD-FUN is called with point at the start of the spot. It's passed two
4968 ;; arguments: The first is the end position of the token preceding the spot,
4969 ;; or 0 for the implicit match at bob. The second is a flag that is t when
4970 ;; the match is inside a macro. Point should be moved forward by at least
4971 ;; one token.
4972 ;;
4973 ;; If CFD-FUN adds `c-decl-end' properties somewhere below the current spot,
4974 ;; it should return non-nil to ensure that the next search will find them.
4975 ;;
4976 ;; Such a spot is:
4977 ;; o The first token after bob.
4978 ;; o The first token after the end of submatch 1 in
4979 ;; `c-decl-prefix-or-start-re' when that submatch matches. This
4980 ;; submatch is typically a (L or R) brace or paren, a ;, or a ,.
4981 ;; o The start of each `c-decl-prefix-or-start-re' match when
4982 ;; submatch 1 doesn't match. This is, for example, the keyword
4983 ;; "class" in Pike.
4984 ;; o The start of a previously recognized declaration; "recognized"
4985 ;; means that the last char of the previous token has a `c-type'
4986 ;; text property with the value `c-decl-end'; this only holds
4987 ;; when `c-type-decl-end-used' is set.
4988 ;;
4989 ;; Only a spot that match CFD-DECL-RE and whose face is in the
4990 ;; CFD-FACE-CHECKLIST list causes CFD-FUN to be called. The face
4991 ;; check is disabled if CFD-FACE-CHECKLIST is nil.
4992 ;;
4993 ;; If the match is inside a macro then the buffer is narrowed to the
4994 ;; end of it, so that CFD-FUN can investigate the following tokens
4995 ;; without matching something that begins inside a macro and ends
4996 ;; outside it. It's to avoid this work that the CFD-DECL-RE and
4997 ;; CFD-FACE-CHECKLIST checks exist.
4998 ;;
4999 ;; The spots are visited approximately in order from top to bottom.
5000 ;; It's however the positions where `c-decl-prefix-or-start-re'
5001 ;; matches and where `c-decl-end' properties are found that are in
5002 ;; order. Since the spots often are at the following token, they
5003 ;; might be visited out of order insofar as more spots are reported
5004 ;; later on within the syntactic whitespace between the match
5005 ;; positions and their spots.
5006 ;;
5007 ;; It's assumed that comments and strings are fontified in the
5008 ;; searched range.
5009 ;;
5010 ;; This is mainly used in fontification, and so has an elaborate
5011 ;; cache to handle repeated calls from the same start position; see
5012 ;; the variables above.
5013 ;;
5014 ;; All variables in this function begin with `cfd-' to avoid name
5015 ;; collision with the (dynamically bound) variables used in CFD-FUN.
5016 ;;
5017 ;; This function might do hidden buffer changes.
5021 ;; The end of the token preceding the decl spot last found
5022 ;; with `c-decl-prefix-or-start-re'. `cfd-limit' if there's
5023 ;; no match.
5024 cfd-re-match
5025 ;; The end position of the last `c-decl-prefix-or-start-re'
5026 ;; match. If this is greater than `cfd-continue-pos', the
5027 ;; next regexp search is started here instead.
5029 ;; The end of the last `c-decl-end' found by
5030 ;; `c-find-decl-prefix-search'. `cfd-limit' if there's no
5031 ;; match. If searching for the property isn't needed then we
5032 ;; disable it by setting it to `cfd-limit' directly.
5034 ;; The end of the token preceding the decl spot last found by
5035 ;; `c-find-decl-prefix-search'. 0 for the implicit match at
5036 ;; bob. `cfd-limit' if there's no match. In other words,
5037 ;; this is the minimum of `cfd-re-match' and `cfd-prop-match'.
5039 ;; The position to continue searching at.
5040 cfd-continue-pos
5041 ;; The position of the last "real" token we've stopped at.
5042 ;; This can be greater than `cfd-continue-pos' when we get
5043 ;; hits inside macros or at `c-decl-end' positions inside
5044 ;; comments.
5046 ;; The end position of the last entered macro.
5049 ;; Initialize by finding a syntactically relevant start position
5050 ;; before the point, and do the first `c-decl-prefix-or-start-re'
5051 ;; search unless we're at bob.
5054 ;; Must back up a bit since we look for the end of the previous
5055 ;; statement or declaration, which is earlier than the first
5056 ;; returned match.
5058 ;; This `cond' moves back over any literals or macros. It has special
5059 ;; handling for when the region being searched is entirely within a
5060 ;; macro. It sets `cfd-continue-pos' (unless we've reached
5061 ;; `cfd-limit').
5063 ;; First we need to move to a syntactically relevant position.
5064 ;; Begin by backing out of comment or string literals.
5065 ;;
5066 ;; This arm of the cond actually triggers if we're in a literal,
5067 ;; and cfd-limit is at most at BONL.
5069 ;; This arm of the `and' moves backwards out of a literal when
5070 ;; the face at point is a literal face. In this case, its value
5071 ;; is always non-nil.
5073 ;; Try to use the faces to back up to the start of the
5074 ;; literal. FIXME: What if the point is on a declaration
5075 ;; inside a comment?
5081 ;; XEmacs doesn't fontify the quotes surrounding string
5082 ;; literals.
5091 ;; Don't trust the literal to contain only literal faces
5092 ;; (the font lock package might not have fontified the
5093 ;; start of it at all, for instance) so check that we have
5094 ;; arrived at something that looks like a start or else
5095 ;; resort to `c-literal-limits'.
5102 ;; The start is in a literal. If the limit is in the same
5103 ;; one we don't have to find a syntactic position etc. We
5104 ;; only check that if the limit is at or before bonl to save
5105 ;; time; it covers the by far most common case when font-lock
5106 ;; refontifies the current line only.
5117 ;; Completely inside a literal. Set up variables to trig the
5118 ;; (< cfd-continue-pos cfd-start-pos) case below and it'll
5119 ;; find a suitable start position.
5122 ;; Check if the region might be completely inside a macro, to
5123 ;; optimize that like the completely-inside-literal above.
5130 ;; (Maybe) completely inside a macro. Only need to trig the
5131 ;; (< cfd-continue-pos cfd-start-pos) case below to make it
5132 ;; set things up.
5134 start-in-macro t))
5136 ;; The default arm of the `cond' moves back over any macro we're in
5137 ;; and over any syntactic WS. It sets `c-find-decl-syntactic-pos'.
5139 ;; Back out of any macro so we don't miss any declaration
5140 ;; that could follow after it.
5144 ;; Now we're at a proper syntactically relevant position so we
5145 ;; can use the cache. But first clear it if it applied
5146 ;; further down.
5151 ;; Don't have to do this if the cache is relevant here,
5152 ;; typically if the same line is refontified again. If
5153 ;; we're just some syntactic whitespace further down we can
5154 ;; still use the cache to limit the skipping.
5157 ;; If we hit `c-find-decl-syntactic-pos' and
5158 ;; `c-find-decl-match-pos' is set then we install the cached
5159 ;; values. If we hit `c-find-decl-syntactic-pos' and
5160 ;; `c-find-decl-match-pos' is nil then we know there's no decl
5161 ;; prefix in the whitespace before `c-find-decl-syntactic-pos'
5162 ;; and so we can continue the search from this point. If we
5163 ;; didn't hit `c-find-decl-syntactic-pos' then we're now in
5164 ;; the right spot to begin searching anyway.
5166 c-find-decl-match-pos)
5168 cfd-continue-pos syntactic-pos)
5173 ;; Always consider bob a match to get the first
5174 ;; declaration in the file. Do this separately instead of
5175 ;; letting `c-decl-prefix-or-start-re' match bob, so that
5176 ;; regexp always can consume at least one character to
5177 ;; ensure that we won't get stuck in an infinite loop.
5182 ;; Do an initial search now. In the bob case above it's
5183 ;; only done to search for a `c-decl-end' spot.
5189 ;; Advance `cfd-continue-pos' if it's before the start position.
5190 ;; The closest continue position that might have effect at or
5191 ;; after the start depends on what we started in. This also
5192 ;; finds a suitable start position in the special cases when the
5193 ;; region is completely within a literal or macro.
5198 ;; If we're in a macro then it's the closest preceding token
5199 ;; in the macro. Check this before `start-in-literal',
5200 ;; since if we're inside a literal in a macro, the preceding
5201 ;; token is earlier than any `c-decl-end' spot inside the
5202 ;; literal (comment).
5204 ;; The only syntactic ws in macros are comments.
5210 ;; If we're in a comment it can only be the closest
5211 ;; preceding `c-decl-end' position within that comment, if
5212 ;; any. Go back to the beginning of such a property so that
5213 ;; `c-find-decl-prefix-search' will find the end of it.
5214 ;; (Can't stop at the end and install it directly on
5215 ;; `cfd-prop-match' since that variable might be cleared
5216 ;; after `cfd-fun' below.)
5217 ;;
5218 ;; Note that if the literal is a string then the property
5219 ;; search will simply skip to the beginning of it right
5220 ;; away.
5232 ;; Didn't find any property inside the comment, so we can
5233 ;; skip it entirely. (This won't skip past a string, but
5234 ;; that'll be handled quickly by the next
5235 ;; `c-find-decl-prefix-search' anyway.)
5241 ;; If we started in normal code, the only match that might
5242 ;; apply before the start is what we already got in
5243 ;; `cfd-match-pos' so we can continue at the start position.
5244 ;; (Note that we don't get here if the first match is below
5245 ;; it.)
5248 ;; Delete found matches if they are before our new continue
5249 ;; position, so that `c-find-decl-prefix-search' won't back up
5250 ;; to them later on.
5258 ;; This is the normal case and we got a proper syntactic
5259 ;; position. If there's a match then it's always outside
5260 ;; macros and comments, so advance to the next token and set
5261 ;; `cfd-token-pos'. The loop below will later go back using
5262 ;; `cfd-continue-pos' to fix declarations inside the
5263 ;; syntactic ws.
5271 ;; Have one of the special cases when the region is completely
5272 ;; within a literal or macro. `cfd-continue-pos' is set to a
5273 ;; good start position for the search, so do it.
5276 ;; Now loop, one decl spot per iteration. We already have the first
5277 ;; match in `cfd-match-pos'.
5279 ;; Go forward over "false matches", one per iteration.
5284 ;; Kludge to filter out matches on the "<" that
5285 ;; aren't open parens, for the sake of languages
5286 ;; that got `c-recognize-<>-arglists' set.
5291 ;; If `cfd-continue-pos' is less or equal to
5292 ;; `cfd-token-pos', we've got a hit inside a macro
5293 ;; that's in the syntactic whitespace before the last
5294 ;; "real" declaration we've checked. If they're equal
5295 ;; we've arrived at the declaration a second time, so
5296 ;; there's nothing to do.
5300 ;; If `cfd-continue-pos' is less than `cfd-token-pos'
5301 ;; we're still searching for declarations embedded in
5302 ;; the syntactic whitespace. In that case we need
5303 ;; only to skip comments and not macros, since they
5304 ;; can't be nested, and that's already been done in
5305 ;; `c-find-decl-prefix-search'.
5310 ;; Continue if the following token fails the
5311 ;; CFD-DECL-RE and CFD-FACE-CHECKLIST checks.
5318 t)))
5329 ;; Narrow to the end of the macro if we got a hit inside
5330 ;; one, to avoid recognizing things that start inside the
5331 ;; macro and end outside it.
5333 ;; Not in the same macro as in the previous round.
5344 t
5347 t)
5348 ;; The matched token was the last thing in the macro,
5349 ;; so the whole match is bogus.
5358 ;; Restore limits if we did macro narrowing above.
5365 ; cfd-match-pos, etc.
5367 \f
5368 ;; A cache for found types.
5370 ;; Buffer local variable that contains an obarray with the types we've
5371 ;; found. If a declaration is recognized somewhere we record the
5372 ;; fully qualified identifier in it to recognize it as a type
5373 ;; elsewhere in the file too. This is not accurate since we do not
5374 ;; bother with the scoping rules of the languages, but in practice the
5375 ;; same name is seldom used as both a type and something else in a
5376 ;; file, and we only use this as a last resort in ambiguous cases (see
5377 ;; `c-forward-decl-or-cast-1').
5378 ;;
5379 ;; Not every type need be in this cache. However, things which have
5380 ;; ceased to be types must be removed from it.
5381 ;;
5382 ;; Template types in C++ are added here too but with the template
5383 ;; arglist replaced with "<>" in references or "<" for the one in the
5384 ;; primary type. E.g. the type "Foo<A,B>::Bar<C>" is stored as
5385 ;; "Foo<>::Bar<". This avoids storing very long strings (since C++
5386 ;; template specs can be fairly sized programs in themselves) and
5387 ;; improves the hit ratio (it's a type regardless of the template
5388 ;; args; it's just not the same type, but we're only interested in
5389 ;; recognizing types, not telling distinct types apart). Note that
5390 ;; template types in references are added here too; from the example
5391 ;; above there will also be an entry "Foo<".
5396 ;; Clears `c-found-types'.
5400 ;; Add the given region as a type in `c-found-types'. If the region
5401 ;; doesn't match an existing type but there is a type which is equal
5402 ;; to the given one except that the last character is missing, then
5403 ;; the shorter type is removed. That's done to avoid adding all
5404 ;; prefixes of a type as it's being entered and font locked. This
5405 ;; doesn't cover cases like when characters are removed from a type
5406 ;; or added in the middle. We'd need the position of point when the
5407 ;; font locking is invoked to solve this well.
5408 ;;
5409 ;; This function might do hidden buffer changes.
5416 ;; Remove the "NAME" from c-found-types, if present.
5420 ;; Return non-nil if the given region contains a type in
5421 ;; `c-found-types'.
5422 ;;
5423 ;; This function might do hidden buffer changes.
5425 c-found-types))
5428 ;; Return all the types in `c-found-types' as a sorted list of
5429 ;; strings.
5433 type-list)))
5434 c-found-types)
5437 ;; Shut up the byte compiler.
5441 ;; An after change function which, in conjunction with the info in
5442 ;; c-maybe-stale-found-type (set in c-before-change), removes a type
5443 ;; from `c-found-types', should this type have become stale. For
5444 ;; example, this happens to "foo" when "foo \n bar();" becomes
5445 ;; "foo(); \n bar();". Such stale types, if not removed, foul up
5446 ;; the fontification.
5447 ;;
5448 ;; Have we, perhaps, added non-ws characters to the front/back of a found
5449 ;; type?
5469 ;; Changing the amount of (already existing) whitespace - don't do anything.
5474 ;; The syntactic relationship which defined a "found type" has been
5475 ;; destroyed.
5478 ;; ((eq (car c-maybe-stale-found-type) 'c-decl-type-start) FIXME!!!
5479 )))
5481 \f
5482 ;; Setting and removing syntax properties on < and > in languages (C++
5483 ;; and Java) where they can be template/generic delimiters as well as
5484 ;; their normal meaning of "less/greater than".
5486 ;; Normally, < and > have syntax 'punctuation'. When they are found to
5487 ;; be delimiters, they are marked as such with the category properties
5488 ;; c-<-as-paren-syntax, c->-as-paren-syntax respectively.
5490 ;; STRATEGY:
5491 ;;
5492 ;; It is impossible to determine with certainty whether a <..> pair in
5493 ;; C++ is two comparison operators or is template delimiters, unless
5494 ;; one duplicates a lot of a C++ compiler. For example, the following
5495 ;; code fragment:
5496 ;;
5497 ;; foo (a < b, c > d) ;
5498 ;;
5499 ;; could be a function call with two integer parameters (each a
5500 ;; relational expression), or it could be a constructor for class foo
5501 ;; taking one parameter d of templated type "a < b, c >". They are
5502 ;; somewhat easier to distinguish in Java.
5503 ;;
5504 ;; The strategy now (2010-01) adopted is to mark and unmark < and
5505 ;; > IN MATCHING PAIRS ONLY. [Previously, they were marked
5506 ;; individually when their context so indicated. This gave rise to
5507 ;; intractable problems when one of a matching pair was deleted, or
5508 ;; pulled into a literal.]
5509 ;;
5510 ;; At each buffer change, the syntax-table properties are removed in a
5511 ;; before-change function and reapplied, when needed, in an
5512 ;; after-change function. It is far more important that the
5513 ;; properties get removed when they they are spurious than that they
5514 ;; be present when wanted.
5515 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
5517 ;; POS (default point) is at a < character. If it is marked with
5518 ;; open paren syntax-table text property, remove the property,
5519 ;; together with the close paren property on the matching > (if
5520 ;; any).
5535 ;; POS (default point) is at a > character. If it is marked with
5536 ;; close paren syntax-table property, remove the property, together
5537 ;; with the open paren property on the matching < (if any).
5552 ;; POS (default point) is at a < or > character. If it has an
5553 ;; open/close paren syntax-table property, remove this property both
5554 ;; from the current character and its partner (which will also be
5555 ;; thusly marked).
5565 ;; POS (default point) is at a < character. If it is both marked
5566 ;; with open/close paren syntax-table property, and has a matching >
5567 ;; (also marked) which is after LIM, remove the property both from
5568 ;; the current > and its partner. Return t when this happens, nil
5569 ;; when it doesn't.
5583 t)))
5586 ;; POS (default point) is at a > character. If it is both marked
5587 ;; with open/close paren syntax-table property, and has a matching <
5588 ;; (also marked) which is before LIM, remove the property both from
5589 ;; the current < and its partner. Return t when this happens, nil
5590 ;; when it doesn't.
5604 t)))
5606 ;; Set by c-common-init in cc-mode.el.
5611 ;; Unmark certain pairs of "< .... >" which are currently marked as
5612 ;; template/generic delimiters. (This marking is via syntax-table text
5613 ;; properties), and expand the (c-new-BEG c-new-END) region to include all
5614 ;; unmarked < and > operators within the certain bounds (see below).
5615 ;;
5616 ;; These pairs are those which are in the current "statement" (i.e.,
5617 ;; the region between the {, }, or ; before BEG and the one after
5618 ;; END), and which enclose any part of the interval (BEG END).
5619 ;;
5620 ;; Note that in C++ (?and Java), template/generic parens cannot
5621 ;; enclose a brace or semicolon, so we use these as bounds on the
5622 ;; region we must work on.
5623 ;;
5624 ;; This function is called from before-change-functions (via
5625 ;; c-get-state-before-change-functions). Thus the buffer is widened,
5626 ;; and point is undefined, both at entry and exit.
5627 ;;
5628 ;; FIXME!!! This routine ignores the possibility of macros entirely.
5629 ;; 2010-01-29.
5634 new-beg new-end beg-limit end-limit)
5635 ;; Locate the earliest < after the barrier before the changed region,
5636 ;; which isn't already marked as a paren.
5640 ;; Remove the syntax-table/category properties from each pertinent <...>
5641 ;; pair. Firstly, the ones with the < before beg and > after beg....
5650 ;; ...Then the ones with < before end and > after end.
5661 ;; Extend the fontification region, if needed.
5670 ;; This is called from `after-change-functions' when
5671 ;; c-recognize-<>-arglists' is set. It ensures that no "<" or ">"
5672 ;; chars with paren syntax become part of another operator like "<<"
5673 ;; or ">=".
5674 ;;
5675 ;; This function might do hidden buffer changes.
5708 ;; This function is called as an after-change function. It restores the
5709 ;; category/syntax-table properties on template/generic <..> pairs between
5710 ;; c-new-BEG and c-new-END. It may do hidden buffer changes.
5729 \f
5730 ;; Handling of small scale constructs like types and names.
5732 ;; Dynamically bound variable that instructs `c-forward-type' to also
5733 ;; treat possible types (i.e. those that it normally returns 'maybe or
5734 ;; 'found for) as actual types (and always return 'found for them).
5735 ;; This means that it records them in `c-record-type-identifiers' if
5736 ;; that is set, and that it adds them to `c-found-types'.
5739 ;; Dynamically bound variable that instructs `c-forward-<>-arglist' to
5740 ;; mark up successfully parsed arglists with paren syntax properties on
5741 ;; the surrounding angle brackets and with `c-<>-arg-sep' in the
5742 ;; `c-type' property of each argument separating comma.
5743 ;;
5744 ;; Setting this variable also makes `c-forward-<>-arglist' recurse into
5745 ;; all arglists for side effects (i.e. recording types), otherwise it
5746 ;; exploits any existing paren syntax properties to quickly jump to the
5747 ;; end of already parsed arglists.
5748 ;;
5749 ;; Marking up the arglists is not the default since doing that correctly
5750 ;; depends on a proper value for `c-restricted-<>-arglists'.
5753 ;; Dynamically bound variable that instructs `c-forward-<>-arglist' to
5754 ;; not accept arglists that contain binary operators.
5755 ;;
5756 ;; This is primarily used to handle C++ template arglists. C++
5757 ;; disambiguates them by checking whether the preceding name is a
5758 ;; template or not. We can't do that, so we assume it is a template
5759 ;; if it can be parsed as one. That usually works well since
5760 ;; comparison expressions on the forms "a < b > c" or "a < b, c > d"
5761 ;; in almost all cases would be pointless.
5762 ;;
5763 ;; However, in function arglists, e.g. in "foo (a < b, c > d)", we
5764 ;; should let the comma separate the function arguments instead. And
5765 ;; in a context where the value of the expression is taken, e.g. in
5766 ;; "if (a < b || c > d)", it's probably not a template.
5769 ;; Dynamically bound variables that instructs
5770 ;; `c-forward-keyword-clause', `c-forward-<>-arglist',
5771 ;; `c-forward-name', `c-forward-type', `c-forward-decl-or-cast-1', and
5772 ;; `c-forward-label' to record the ranges of all the type and
5773 ;; reference identifiers they encounter. They will build lists on
5774 ;; these variables where each element is a cons of the buffer
5775 ;; positions surrounding each identifier. This recording is only
5776 ;; activated when `c-record-type-identifiers' is non-nil.
5777 ;;
5778 ;; All known types that can't be identifiers are recorded, and also
5779 ;; other possible types if `c-promote-possible-types' is set.
5780 ;; Recording is however disabled inside angle bracket arglists that
5781 ;; are encountered inside names and other angle bracket arglists.
5782 ;; Such occurrences are taken care of by `c-font-lock-<>-arglists'
5783 ;; instead.
5784 ;;
5785 ;; Only the names in C++ template style references (e.g. "tmpl" in
5786 ;; "tmpl<a,b>::foo") are recorded as references, other references
5787 ;; aren't handled here.
5788 ;;
5789 ;; `c-forward-label' records the label identifier(s) on
5790 ;; `c-record-ref-identifiers'.
5794 ;; This variable will receive a cons cell of the range of the last
5795 ;; single identifier symbol stepped over by `c-forward-name' if it's
5796 ;; successful. This is the range that should be put on one of the
5797 ;; record lists above by the caller. It's assigned nil if there's no
5798 ;; such symbol in the name.
5803 ;; Always true.
5813 ;; Always true.
5821 ;; Dynamically bound variable that instructs `c-forward-type' to
5822 ;; record the ranges of types that only are found. Behaves otherwise
5823 ;; like `c-record-type-identifiers'.
5827 ;; Used internally in `c-forward-keyword-clause' to move forward
5828 ;; over a type (if TYPE is 'type) or a name (otherwise) which
5829 ;; possibly is prefixed by keywords and their associated clauses.
5830 ;; Try with a type/name first to not trip up on those that begin
5831 ;; with a keyword. Return t if a known or found type is moved
5832 ;; over. The point is clobbered if nil is returned. If range
5833 ;; recording is enabled, the identifier is recorded on as a type
5834 ;; if TYPE is 'type or as a reference if TYPE is 'ref.
5835 ;;
5836 ;; This macro might do hidden buffer changes.
5842 nil
5853 t)))
5856 ;; Used internally in `c-forward-keyword-clause' to move forward
5857 ;; over a comma separated list of types or names using
5858 ;; `c-forward-keyword-prefixed-id'.
5859 ;;
5860 ;; This macro might do hidden buffer changes.
5871 ;; Point is at a c-noise-macro-with-parens-names macro identifier. Go
5872 ;; forward over this name, any parenthesis expression which follows it, and
5873 ;; any syntactic WS, ending up at the next token. If there is an unbalanced
5874 ;; paren expression, leave point at it. Always Return t.
5879 t)
5882 ;; Submatch MATCH in the current match data is assumed to surround a
5883 ;; token. If it's a keyword, move over it and any immediately
5884 ;; following clauses associated with it, stopping at the start of
5885 ;; the next token. t is returned in that case, otherwise the point
5886 ;; stays and nil is returned. The kind of clauses that are
5887 ;; recognized are those specified by `c-type-list-kwds',
5888 ;; `c-ref-list-kwds', `c-colon-type-list-kwds',
5889 ;; `c-paren-nontype-kwds', `c-paren-type-kwds', `c-<>-type-kwds',
5890 ;; and `c-<>-arglist-kwds'.
5891 ;;
5892 ;; This function records identifier ranges on
5893 ;; `c-record-type-identifiers' and `c-record-ref-identifiers' if
5894 ;; `c-record-type-identifiers' is non-nil.
5895 ;;
5896 ;; Note that for `c-colon-type-list-kwds', which doesn't necessary
5897 ;; apply directly after the keyword, the type list is moved over
5898 ;; only when there is no unaccounted token before it (i.e. a token
5899 ;; that isn't moved over due to some other keyword list). The
5900 ;; identifier ranges in the list are still recorded if that should
5901 ;; be done, though.
5902 ;;
5903 ;; This function might do hidden buffer changes.
5906 ;; The call to `c-forward-<>-arglist' below is made after
5907 ;; `c-<>-sexp-kwds' keywords, so we're certain they actually
5908 ;; are angle bracket arglists and `c-restricted-<>-arglists'
5909 ;; should therefore be nil.
5921 ;; There's a type directly after a keyword in `c-type-list-kwds'.
5926 ;; There's a name directly after a keyword in `c-ref-list-kwds'.
5931 ;; There's an open paren after a keyword in `c-paren-any-kwds'.
5938 ;; Use `c-forward-type' on every identifier we can find
5939 ;; inside the paren, to record the types.
5964 ;; If we are at the colon already, we move over the type
5965 ;; list after it.
5971 ;; Not at the colon, so stop here. But the identifier
5972 ;; ranges in the type list later on should still be
5973 ;; recorded.
5976 ;; If a keyword matched both one of the types above and
5977 ;; this one, we match `c-colon-type-list-re' after the
5978 ;; clause matched above.
5985 ;; There's a type after the `c-colon-type-list-re' match
5986 ;; after a keyword in `c-colon-type-list-kwds'.
5990 t)))
5992 ;; cc-mode requires cc-fonts.
5996 ;; The point is assumed to be at a "<". Try to treat it as the open
5997 ;; paren of an angle bracket arglist and move forward to the
5998 ;; corresponding ">". If successful, the point is left after the
5999 ;; ">" and t is returned, otherwise the point isn't moved and nil is
6000 ;; returned. If ALL-TYPES is t then all encountered arguments in
6001 ;; the arglist that might be types are treated as found types.
6002 ;;
6003 ;; The variable `c-parse-and-markup-<>-arglists' controls how this
6004 ;; function handles text properties on the angle brackets and argument
6005 ;; separating commas.
6006 ;;
6007 ;; `c-restricted-<>-arglists' controls how lenient the template
6008 ;; arglist recognition should be.
6009 ;;
6010 ;; This function records identifier ranges on
6011 ;; `c-record-type-identifiers' and `c-record-ref-identifiers' if
6012 ;; `c-record-type-identifiers' is non-nil.
6013 ;;
6014 ;; This function might do hidden buffer changes.
6017 ;; If `c-record-type-identifiers' is set then activate
6018 ;; recording of any found types that constitute an argument in
6019 ;; the arglist.
6027 ;; `nconc' doesn't mind that the tail of
6028 ;; `c-record-found-types' is t.
6030 t)
6033 nil)))
6036 ;; Recursive part of `c-forward-<>-arglist'.
6037 ;;
6038 ;; This function might do hidden buffer changes.
6040 ;; Cover this so that any recorded found type ranges are
6041 ;; automatically lost if it turns out to not be an angle
6042 ;; bracket arglist. It's propagated through the return value
6043 ;; on successful completion.
6045 ;; List that collects the positions after the argument
6046 ;; separating ',' in the arglist.
6047 arg-start-pos)
6048 ;; If the '<' has paren open syntax then we've marked it as an angle
6049 ;; bracket arglist before, so skip to the end.
6057 t
6058 ;; Got unmatched paren angle brackets. We don't clear the paren
6059 ;; syntax properties and retry, on the basis that it's very
6060 ;; unlikely that paren angle brackets become operators by code
6061 ;; manipulation. It's far more likely that it doesn't match due
6062 ;; to narrowing or some temporary change.
6064 nil))
6069 ;; go forward one non-alphanumeric character (group) per iteration of
6070 ;; this loop.
6084 ;; All encountered identifiers are types, so set the
6085 ;; promote flag and parse the type.
6103 ;; Note: These regexps exploit the match order in \| so
6104 ;; that "<>" is matched by "<" rather than "[^>:-]>".
6106 ;; Stop on ',', '|', '&', '+' and '-' to catch
6107 ;; common binary operators that could be between
6108 ;; two comparison expressions "a<b" and "c>d".
6109 ;; 2016-02-11: C++11 templates can now contain arithmetic
6110 ;; expressions, so template detection in C++ is now less
6111 ;; robust than it was.
6112 c-<>-notable-chars-re
6113 nil t t))
6117 ;; Either an operator starting with '>' or the end of
6118 ;; the angle bracket arglist.
6127 ;; The angle bracket arglist is finished.
6139 ;; Either an operator starting with '<' or a nested arglist.
6143 ;; The '<' begins a multi-char operator.
6146 ;; We're at a nested <.....>
6151 ;; There's always an identifier before an angle
6152 ;; bracket arglist, or a keyword in `c-<>-type-kwds'
6153 ;; or `c-<>-arglist-kwds'.
6170 subres)
6171 ;; It was an angle bracket arglist.
6174 ;; Record the identifier before the template as a type
6175 ;; or reference depending on whether the arglist is last
6176 ;; in a qualified identifier.
6186 ;; At a "less than" operator.
6188 ;; (forward-char) ; NO! We've already gone over the <.
6189 )))
6201 ;; Just another argument. Record the position. The
6202 ;; type check stuff that made us stop at it is at
6203 ;; the top of the loop.
6207 ;; Got a character that can't be in an angle bracket
6208 ;; arglist argument. Abort using `throw', since
6209 ;; it's useless to try to find a surrounding arglist
6210 ;; if we're nested.
6216 ;; The point is assumed to be directly after a ">". Try to treat it
6217 ;; as the close paren of an angle bracket arglist and move back to
6218 ;; the corresponding "<". If successful, the point is left at
6219 ;; the "<" and t is returned, otherwise the point isn't moved and
6220 ;; nil is returned. ALL-TYPES is passed on to
6221 ;; `c-forward-<>-arglist'.
6222 ;;
6223 ;; If the optional LIMIT is given, it bounds the backward search.
6224 ;; It's then assumed to be at a syntactically relevant position.
6225 ;;
6226 ;; This is a wrapper around `c-forward-<>-arglist'. See that
6227 ;; function for more details.
6236 t
6237 ;; See corresponding note in `c-forward-<>-arglist'.
6239 nil)
6246 t
6247 ;; Stopped at bob or a char that isn't allowed in an
6248 ;; arglist, so we've failed.
6250 nil)
6255 ;; If we moved then the "<" was part of some
6256 ;; multicharacter token.
6257 t
6263 ;; Matched the arglist. Break the while.
6265 nil)
6267 ;; We started from a non-paren ">" inside an
6268 ;; arglist.
6270 nil)
6272 ;; Matched a shorter arglist. Can be a nested
6273 ;; one so continue looking.
6275 t))
6276 t))))))
6281 ;; Move forward over a complete name if at the beginning of one,
6282 ;; stopping at the next following token. A keyword, as such,
6283 ;; doesn't count as a name. If the point is not at something that
6284 ;; is recognized as a name then it stays put.
6285 ;;
6286 ;; A name could be something as simple as "foo" in C or something as
6287 ;; complex as "X<Y<class A<int>::B, BIT_MAX >> b>, ::operator<> ::
6288 ;; Z<(a>b)> :: operator const X<&foo>::T Q::G<unsigned short
6289 ;; int>::*volatile const" in C++ (this function is actually little
6290 ;; more than a `looking-at' call in all modes except those that,
6291 ;; like C++, have `c-recognize-<>-arglists' set).
6292 ;;
6293 ;; Return
6294 ;; o - nil if no name is found;
6295 ;; o - 'template if it's an identifier ending with an angle bracket
6296 ;; arglist;
6297 ;; o - 'operator of it's an operator identifier;
6298 ;; o - t if it's some other kind of name.
6299 ;;
6300 ;; This function records identifier ranges on
6301 ;; `c-record-type-identifiers' and `c-record-ref-identifiers' if
6302 ;; `c-record-type-identifiers' is non-nil.
6303 ;;
6304 ;; This function might do hidden buffer changes.
6307 ;; Turn off `c-promote-possible-types' here since we might
6308 ;; call `c-forward-<>-arglist' and we don't want it to promote
6309 ;; every suspect thing in the arglist to a type. We're
6310 ;; typically called from `c-forward-type' in this case, and
6311 ;; the caller only wants the top level type that it finds to
6312 ;; be promoted.
6313 c-promote-possible-types)
6319 ;; Check for keyword. We go to the last symbol in
6320 ;; `c-identifier-key' first.
6333 ;; "template" is only valid inside an
6334 ;; identifier if preceded by "::".
6339 t))
6341 ;; Handle a C++ operator or template identifier.
6345 ;; Got "... ::template".
6349 res subres))))
6352 ;; Got a cast operator.
6356 ;; Now we should match a sequence of either
6357 ;; '*', '&' or a name followed by ":: *",
6358 ;; where each can be followed by a sequence
6359 ;; of `c-opt-type-modifier-key'.
6362 t)
6372 t))))
6380 ;; Got some other operator.
6388 nil)
6390 ;; `id-start' is equal to `id-end' if we've jumped over
6391 ;; an identifier that doesn't end with a symbol token.
6392 ;; That can occur e.g. for Java import directives on the
6393 ;; form "foo.bar.*".
6400 res t)))
6410 ;; Got a concatenated identifier. This handles the
6411 ;; cases with tricky syntactic whitespace that aren't
6412 ;; covered in `c-identifier-key'.
6415 t)
6419 ;; Maybe an angle bracket arglist.
6432 ;; Continue if there's an identifier concatenation
6433 ;; operator after the template argument.
6439 t)
6445 nil)))
6446 )))))
6449 res))
6452 ;; Move forward over a type spec if at the beginning of one,
6453 ;; stopping at the next following token. The keyword "typedef"
6454 ;; isn't part of a type spec here.
6455 ;;
6456 ;; BRACE-BLOCK-TOO, when non-nil, means move over the brace block in
6457 ;; constructs like "struct foo {...} bar ;" or "struct {...} bar;".
6458 ;; The current (2009-03-10) intention is to convert all uses of
6459 ;; `c-forward-type' to call with this parameter set, then to
6460 ;; eliminate it.
6461 ;;
6462 ;; Return
6463 ;; o - t if it's a known type that can't be a name or other
6464 ;; expression;
6465 ;; o - 'known if it's an otherwise known type (according to
6466 ;; `*-font-lock-extra-types');
6467 ;; o - 'prefix if it's a known prefix of a type;
6468 ;; o - 'found if it's a type that matches one in `c-found-types';
6469 ;; o - 'maybe if it's an identifier that might be a type;
6470 ;; o - 'decltype if it's a decltype(variable) declaration; - or
6471 ;; o - nil if it can't be a type (the point isn't moved then).
6472 ;;
6473 ;; The point is assumed to be at the beginning of a token.
6474 ;;
6475 ;; Note that this function doesn't skip past the brace definition
6476 ;; that might be considered part of the type, e.g.
6477 ;; "enum {a, b, c} foo".
6478 ;;
6479 ;; This function records identifier ranges on
6480 ;; `c-record-type-identifiers' and `c-record-ref-identifiers' if
6481 ;; `c-record-type-identifiers' is non-nil.
6482 ;;
6483 ;; This function might do hidden buffer changes.
6491 ;; Skip leading type modifiers. If any are found we know it's a
6492 ;; prefix of a type.
6511 ; "typedef".
6527 ;; In many languages the name can be used without the
6528 ;; prefix, so we add it to `c-found-types'.
6531 c-last-identifier-range)
6553 id-range c-last-identifier-range))))
6565 id-end)
6567 ;; Looking at a primitive or known type identifier. We've
6568 ;; checked for a name first so that we don't go here if the
6569 ;; known type match only is a prefix of another name.
6580 ;; There might be more keywords for the type.
6608 ;; A normal identifier.
6618 ;; It's an identifier that has been used as
6619 ;; a type somewhere else.
6620 'found
6621 ;; It's an identifier that might be a type.
6624 ;; A template is sometimes a type.
6630 ;; It's an identifier that has been used as
6631 ;; a type somewhere else.
6632 'found
6633 ;; It's an identifier that might be a type.
6635 t)))
6637 ;; Otherwise it's an operator identifier, which is not a type.
6642 ;; Skip trailing type modifiers. If any are found we know it's
6643 ;; a type.
6650 ;; Step over any type suffix operator. Do not let the existence
6651 ;; of these alter the classification of the found type, since
6652 ;; these operators typically are allowed in normal expressions
6653 ;; too.
6660 ;; Look for a trailing operator that concatenates the type
6661 ;; with a following one, and if so step past that one through
6662 ;; a recursive call. Note that we don't record concatenated
6663 ;; types in `c-found-types' - it's the component types that
6664 ;; are recorded when appropriate.
6667 c-promote-possible-types))
6668 ;; If we can't promote then set `c-record-found-types' so that
6669 ;; we can merge in the types from the second part afterwards if
6670 ;; it turns out to be a known type there.
6673 subres)
6682 ;; If either operand certainly is a type then both are, but we
6683 ;; don't let the existence of the operator itself promote two
6684 ;; uncertain types to a certain one.
6703 ;; Merge in the ranges of any types found by the second
6704 ;; `c-forward-type'.
6706 ;; `nconc' doesn't mind that the tail of
6707 ;; `c-record-found-types' is t.
6709 c-record-type-identifiers))))
6717 ;;(message "c-forward-type %s -> %s: %s" start (point) res)
6719 res))
6722 ;; Used for Java code only at the moment. Assumes point is on the @, moves
6723 ;; forward an annotation and returns t. Leaves point unmoved and returns
6724 ;; nil if there is no annotation at point.
6733 t)
6736 t))
6740 ;; Pull the next open brace from PS (which has the form of paren-state),
6741 ;; skipping over any brace pairs. Returns NIL when PS is exhausted.
6749 ;; Point is just after a closing brace/parenthesis. Try to parse this as a
6750 ;; C++ member initializer list, going back to just after the introducing ":"
6751 ;; and returning t. Otherwise return nil, leaving point unchanged.
6776 res))
6779 ;; Go back over a list of elements, each looking like:
6780 ;; <symbol> (<expression>) ,
6781 ;; or <symbol> {<expression>} ,
6782 ;; when we are putatively immediately after a comma. Stop when we don't see
6783 ;; a comma. If either of <symbol> or bracketed <expression> is missing,
6784 ;; throw nil to 'level. If the terminating } or ) is unmatched, throw nil
6785 ;; to 'done. This is not a general purpose macro!
6799 ;; Test whether we are in a C++ member initializer list, and if so, go back
6800 ;; to the introducing ":", returning the position of the opening paren of
6801 ;; the function's arglist. Otherwise return nil, leaving point unchanged.
6804 pos level-plausible at-top-level res)
6805 ;; Assume tentatively that we're at the top level. Try to go back to the
6806 ;; colon we seek.
6844 res))
6846 \f
6847 ;; Handling of large scale constructs like statements and declarations.
6849 ;; Macro used inside `c-forward-decl-or-cast-1'. It ought to be a
6850 ;; defsubst or perhaps even a defun, but it contains lots of free
6851 ;; variables that refer to things inside `c-forward-decl-or-cast-1'.
6853 ;; `c-forward-decl-or-cast-1' can consume an arbitrary length list
6854 ;; of types when parsing a declaration, which means that it
6855 ;; sometimes consumes the identifier in the declaration as a type.
6856 ;; This is used to "backtrack" and make the last type be treated as
6857 ;; an identifier instead.
6860 ;; These identifiers are bound only in the inner let.
6862 identifier-start type-start
6863 got-parens nil
6864 got-identifier t
6865 got-suffix t
6866 got-suffix-after-parens id-start
6870 t
6871 backup-at-type))
6873 id-start backup-id-start)
6875 id-start start-pos))
6877 ;; When these flags already are set we've found specifiers that
6878 ;; unconditionally signal these attributes - backtracking doesn't
6879 ;; change that. So keep them set in that case.
6886 ;; This identifier is bound only in the inner let.
6890 ;; Assuming point is at the start of a declarator, move forward over it,
6891 ;; leaving point at the next token after it (e.g. a ) or a ; or a ,).
6892 ;;
6893 ;; Return a list (ID-START ID-END BRACKETS-AFTER-ID GOT-INIT), where ID-START and
6894 ;; ID-END are the bounds of the declarator's identifier, and
6895 ;; BRACKETS-AFTER-ID is non-nil if a [...] pair is present after the id.
6896 ;; GOT-INIT is non-nil when the declarator is followed by "=" or "(".
6897 ;;
6898 ;; If ACCEPT-ANON is non-nil, move forward over any "anonymous declarator",
6899 ;; i.e. something like the (*) in int (*), such as might be found in a
6900 ;; declaration. In such a case ID-START and ID-END in the return value are
6901 ;; both set to nil. A "null" "anonymous declarator" gives a non-nil result.
6902 ;;
6903 ;; If no declarator is found, leave point unmoved and return nil. LIMIT is
6904 ;; an optional limit for forward searching.
6905 ;;
6906 ;; Note that the global variable `c-last-identifier-range' is written to, so
6907 ;; the caller should bind it if necessary.
6909 ;; Inside the following "condition form", we move forward over the
6910 ;; declarator's identifier up as far as any opening bracket (for array
6911 ;; size) or paren (for parameters of function-type) or brace (for
6912 ;; array/struct initialization) or "=" or terminating delimiter
6913 ;; (e.g. "," or ";" or "}").
6915 id-start id-end brackets-after-id paren-depth)
6920 ;; The following form moves forward over the declarator's
6921 ;; identifier (and what precedes it), returning t. If there
6922 ;; wasn't one, it returns nil.
6925 ;; Skip over type decl prefix operators, one for each iteration
6926 ;; of the while. These are, e.g. "*" in "int *foo" or "(" and
6927 ;; "*" in "int (*foo) (void)" (Note similar code in
6928 ;; `c-forward-decl-or-cast-1'.)
6939 ;; If the third submatch matches in C++ then
6940 ;; we're looking at an identifier that's a
6941 ;; prefix only if it specifies a member pointer.
6946 ;; We only check for a trailing "::" and
6947 ;; let the "*" that should follow be
6948 ;; matched in the next round.
6949 t
6950 ;; It turned out to be the real identifier,
6951 ;; so flag that and stop.
6953 nil))
6954 t))
6961 t)))
6963 ;; If we haven't passed the identifier already, do it now.
6974 t)
6977 ;; Skip out of the parens surrounding the identifier. If closing
6978 ;; parens are missing, this form returns nil.
6984 ;; Skip over any trailing bit, such as "__attribute__".
6994 ;; Search syntactically to the end of the declarator (";",
6995 ;; ",", a closing paren, eob etc) or to the beginning of an
6996 ;; initializer or function prototype ("=" or "\\s\(").
6997 ;; Note that square brackets are now not also treated as
6998 ;; initializers, since this broke when there were also
6999 ;; initializing brace lists.
7008 found))
7012 nil)))
7015 ;; Move forward over a declaration or a cast if at the start of one.
7016 ;; The point is assumed to be at the start of some token. Nil is
7017 ;; returned if no declaration or cast is recognized, and the point
7018 ;; is clobbered in that case.
7019 ;;
7020 ;; If a declaration is parsed:
7021 ;;
7022 ;; The point is left at the first token after the first complete
7023 ;; declarator, if there is one. The return value is a cons where
7024 ;; the car is the position of the first token in the declarator. (See
7025 ;; below for the cdr.)
7026 ;; Some examples:
7027 ;;
7028 ;; void foo (int a, char *b) stuff ...
7029 ;; car ^ ^ point
7030 ;; float (*a)[], b;
7031 ;; car ^ ^ point
7032 ;; unsigned int a = c_style_initializer, b;
7033 ;; car ^ ^ point
7034 ;; unsigned int a (cplusplus_style_initializer), b;
7035 ;; car ^ ^ point (might change)
7036 ;; class Foo : public Bar {}
7037 ;; car ^ ^ point
7038 ;; class PikeClass (int a, string b) stuff ...
7039 ;; car ^ ^ point
7040 ;; enum bool;
7041 ;; car ^ ^ point
7042 ;; enum bool flag;
7043 ;; car ^ ^ point
7044 ;; void cplusplus_function (int x) throw (Bad);
7045 ;; car ^ ^ point
7046 ;; Foo::Foo (int b) : Base (b) {}
7047 ;; car ^ ^ point
7048 ;;
7049 ;; auto foo = 5;
7050 ;; car ^ ^ point
7051 ;; auto cplusplus_11 (int a, char *b) -> decltype (bar):
7052 ;; car ^ ^ point
7053 ;;
7054 ;;
7055 ;;
7056 ;; The cdr of the return value is non-nil when a
7057 ;; `c-typedef-decl-kwds' specifier is found in the declaration.
7058 ;; Specifically it is a dotted pair (A . B) where B is t when a
7059 ;; `c-typedef-kwds' ("typedef") is present, and A is t when some
7060 ;; other `c-typedef-decl-kwds' (e.g. class, struct, enum)
7061 ;; specifier is present. I.e., (some of) the declared
7062 ;; identifier(s) are types.
7063 ;;
7064 ;; If a cast is parsed:
7065 ;;
7066 ;; The point is left at the first token after the closing paren of
7067 ;; the cast. The return value is `cast'. Note that the start
7068 ;; position must be at the first token inside the cast parenthesis
7069 ;; to recognize it.
7070 ;;
7071 ;; PRECEDING-TOKEN-END is the first position after the preceding
7072 ;; token, i.e. on the other side of the syntactic ws from the point.
7073 ;; Use a value less than or equal to (point-min) if the point is at
7074 ;; the first token in (the visible part of) the buffer.
7075 ;;
7076 ;; CONTEXT is a symbol that describes the context at the point:
7077 ;; 'decl In a comma-separated declaration context (typically
7078 ;; inside a function declaration arglist).
7079 ;; '<> In an angle bracket arglist.
7080 ;; 'arglist Some other type of arglist.
7081 ;; nil Some other context or unknown context. Includes
7082 ;; within the parens of an if, for, ... construct.
7083 ;;
7084 ;; LAST-CAST-END is the first token after the closing paren of a
7085 ;; preceding cast, or nil if none is known. If
7086 ;; `c-forward-decl-or-cast-1' is used in succession, it should be
7087 ;; the position after the closest preceding call where a cast was
7088 ;; matched. In that case it's used to discover chains of casts like
7089 ;; "(a) (b) c".
7090 ;;
7091 ;; This function records identifier ranges on
7092 ;; `c-record-type-identifiers' and `c-record-ref-identifiers' if
7093 ;; `c-record-type-identifiers' is non-nil.
7094 ;;
7095 ;; This function might do hidden buffer changes.
7098 ;; first type, i.e. after any leading specifiers. It might
7099 ;; also point at the beginning of the preceding syntactic
7100 ;; whitespace.
7102 ;; Set to the result of `c-forward-type'.
7103 at-type
7104 ;; The position of the first token in what we currently
7105 ;; believe is the type in the declaration or cast, after any
7106 ;; specifiers and their associated clauses.
7107 type-start
7108 ;; The position of the first token in what we currently
7109 ;; believe is the declarator for the first identifier. Set
7110 ;; when the type is found, and moved forward over any
7111 ;; `c-decl-hangon-kwds' and their associated clauses that
7112 ;; occurs after the type.
7113 id-start
7114 ;; These store `at-type', `type-start' and `id-start' of the
7115 ;; identifier before the one in those variables. The previous
7116 ;; identifier might turn out to be the real type in a
7117 ;; declaration if the last one has to be the declarator in it.
7118 ;; If `backup-at-type' is nil then the other variables have
7119 ;; undefined values.
7120 backup-at-type backup-type-start backup-id-start
7121 ;; This stores `kwd-sym' of the symbol before the current one.
7122 ;; This is needed to distinguish the C++11 version of "auto" from
7123 ;; the pre C++11 meaning.
7124 backup-kwd-sym
7125 ;; Set if we've found a specifier (apart from "typedef") that makes
7126 ;; the defined identifier(s) types.
7127 at-type-decl
7128 ;; Set if we've a "typedef" keyword.
7129 at-typedef
7130 ;; Set if we've found a specifier that can start a declaration
7131 ;; where there's no type.
7132 maybe-typeless
7133 ;; Save the value of kwd-sym between loops of the "Check for a
7134 ;; type" loop. Needed to distinguish a C++11 "auto" from a pre
7135 ;; C++11 one.
7136 prev-kwd-sym
7137 ;; If a specifier is found that also can be a type prefix,
7138 ;; these flags are set instead of those above. If we need to
7139 ;; back up an identifier, they are copied to the real flag
7140 ;; variables. Thus they only take effect if we fail to
7141 ;; interpret it as a type.
7142 backup-at-type-decl backup-maybe-typeless
7143 ;; Whether we've found a declaration or a cast. We might know
7144 ;; this before we've found the type in it. It's 'ids if we've
7145 ;; found two consecutive identifiers (usually a sure sign, but
7146 ;; we should allow that in labels too), and t if we've found a
7147 ;; specifier keyword (a 100% sure sign).
7148 at-decl-or-cast
7149 ;; Set when we need to back up to parse this as a declaration
7150 ;; but not as a cast.
7151 backup-if-not-cast
7152 ;; For casts, the return position.
7153 cast-end
7154 ;; Have we got a new-style C++11 "auto"?
7155 new-style-auto
7156 ;; Set when the symbol before `preceding-token-end' is known to
7157 ;; terminate the previous construct, or when we're at point-min.
7158 at-decl-start
7159 ;; Save `c-record-type-identifiers' and
7160 ;; `c-record-ref-identifiers' since ranges are recorded
7161 ;; speculatively and should be thrown away if it turns out
7162 ;; that it isn't a declaration or cast.
7173 c-pre-start-tokens)))))
7178 ;; Check for a type. Unknown symbols are treated as possible
7179 ;; types, but they could also be specifiers disguised through
7180 ;; macros like __INLINE__, so we recognize both types and known
7181 ;; specifiers after them too.
7186 ;; Look for a specifier keyword clause.
7204 ;; Found a known or possible type or a prefix of a known type.
7213 ;; Got two identifiers with nothing but whitespace
7214 ;; between them. That can only happen in declarations.
7218 ;; If the previous identifier is a found type we
7219 ;; record it as a real one; it might be some sort of
7220 ;; alias for a prefix like "unsigned".
7227 backup-type-start type-start
7228 backup-id-start id-start
7229 backup-kwd-sym kwd-sym
7230 at-type found-type
7231 type-start start
7233 ;; The previous ambiguous specifier/type turned out
7234 ;; to be a type since we've parsed another one after
7235 ;; it, so clear these backup flags.
7236 backup-at-type-decl nil
7237 backup-maybe-typeless nil))
7241 ;; Handle known specifier keywords and
7242 ;; `c-decl-hangon-kwds' which can occur after known
7243 ;; types.
7246 noise-start)
7247 ;; It's a hang-on keyword or noise clause that can occur
7248 ;; anywhere.
7251 ;; Move the identifier start position if
7252 ;; we've passed a type.
7254 ;; Otherwise treat this as a specifier and
7255 ;; move the fallback position.
7259 ;; It's an ordinary specifier so we know that
7260 ;; anything before this can't be the type.
7262 start-pos kwd-clause-end)
7265 ;; It's ambiguous whether this keyword is a
7266 ;; specifier or a type prefix, so set the backup
7267 ;; flags. (It's assumed that `c-forward-type'
7268 ;; moved further than `c-forward-keyword-clause'.)
7276 ;; This test only happens after we've scanned a type.
7277 ;; So, with valid syntax, kwd-sym can't be 'typedef.
7282 ;; Haven't matched a type so it's an unambiguous
7283 ;; specifier keyword and we know we're in a
7284 ;; declaration.
7290 ;; If the type isn't known we continue so that we'll jump
7291 ;; over all specifiers and type identifiers. The reason
7292 ;; to do this for a known type prefix is to make things
7293 ;; like "unsigned INT16" work.
7298 ;; If a known type was found, we still need to skip over any
7299 ;; hangon keyword clauses after it. Otherwise it has already
7300 ;; been done in the loop above.
7310 ;; A prefix type is itself a primitive type when it's not
7311 ;; followed by another type.
7315 ;; Got no type but set things up to continue anyway to handle
7316 ;; the various cases when a declaration doesn't start with a
7317 ;; type.
7322 ;; If it's C++ then check if the last "type" ends on the form
7323 ;; "foo::foo" or "foo::~foo", i.e. if it's the name of a
7324 ;; (con|de)structor.
7335 ;; Cheating in the handling of syntactic ws below.
7342 name))
7343 ;; It is a (con|de)structor name. In that case the
7344 ;; declaration is typeless so zap out any preceding
7345 ;; identifier(s) that we might have taken as types.
7348 backup-at-type nil
7349 id-start type-start))))))
7351 ;; Check for and step over a type decl expression after the thing
7352 ;; that is or might be a type. This can't be skipped since we
7353 ;; need the correct end position of the declarator for
7354 ;; `max-type-decl-end-*'.
7356 ;; True if there's a non-open-paren match of
7357 ;; `c-type-decl-prefix-key'.
7358 got-prefix
7359 ;; True if the declarator is surrounded by a parenthesis pair.
7360 got-parens
7361 ;; True if there is an identifier in the declarator.
7362 got-identifier
7363 ;; True if there's a non-close-paren match of
7364 ;; `c-type-decl-suffix-key'.
7365 got-suffix
7366 ;; True if there's a prefix match outside the outermost
7367 ;; paren pair that surrounds the declarator.
7368 got-prefix-before-parens
7369 ;; True if there's a suffix match outside the outermost
7370 ;; paren pair that surrounds the declarator. The value is
7371 ;; the position of the first suffix match.
7372 got-suffix-after-parens
7373 ;; True if we've parsed the type decl to a token that is
7374 ;; known to end declarations in this context.
7375 at-decl-end
7376 ;; The earlier values of `at-type' and `type-start' if we've
7377 ;; shifted the type backwards.
7378 identifier-type identifier-start
7379 ;; If `c-parse-and-markup-<>-arglists' is set we need to
7380 ;; turn it off during the name skipping below to avoid
7381 ;; getting `c-type' properties that might be bogus. That
7382 ;; can happen since we don't know if
7383 ;; `c-restricted-<>-arglists' will be correct inside the
7384 ;; arglist paren that gets entered.
7385 c-parse-and-markup-<>-arglists
7386 ;; Start of the identifier for which `got-identifier' was set.
7387 name-start)
7391 ;; Skip over type decl prefix operators. (Note similar code in
7392 ;; `c-forward-declarator'.)
7402 ;; If the third submatch matches in C++ then
7403 ;; we're looking at an identifier that's a
7404 ;; prefix only if it specifies a member pointer.
7409 ;; We only check for a trailing "::" and
7410 ;; let the "*" that should follow be
7411 ;; matched in the next round.
7413 ;; It turned out to be the real identifier,
7414 ;; so stop.
7415 nil))
7416 t))
7430 ;; Skip over an identifier.
7437 ;; Skip over type decl suffix operators and trailing noise macros.
7449 t)
7453 t)
7460 ;; No suffix matched. We might have matched the
7461 ;; identifier as a type and the open paren of a
7462 ;; function arglist as a type decl prefix. In that
7463 ;; case we should "backtrack": Reinterpret the last
7464 ;; type as the identifier, move out of the arglist and
7465 ;; continue searching for suffix operators.
7466 ;;
7467 ;; Do this even if there's no preceding type, to cope
7468 ;; with old style function declarations in K&R C,
7469 ;; (con|de)structors in C++ and `c-typeless-decl-kwds'
7470 ;; style declarations. That isn't applicable in an
7471 ;; arglist context, though.
7476 maybe-typeless
7477 backup-maybe-typeless
7484 t)))
7493 at-type))
7494 ;; Have found no identifier but `c-typeless-decl-kwds' has
7495 ;; matched so we know we're inside a declaration. The
7496 ;; preceding type must be the identifier instead.
7499 ;; Prepare the "-> type;" for fontification later on.
7509 at-decl-or-cast
7512 ;; CASE 1
7514 ;; Encountered something inside parens that isn't matched by
7515 ;; the `c-type-decl-*' regexps, so it's not a type decl
7516 ;; expression. Try to skip out to the same paren depth to
7517 ;; not confuse the cast check below.
7519 ;; If we've found a specifier keyword then it's a
7520 ;; declaration regardless.
7528 ;; Now we've collected info about various characteristics of
7529 ;; the construct we're looking at. Below follows a decision
7530 ;; tree based on that. It's ordered to check more certain
7531 ;; signs before less certain ones.
7536 ;; CASE 2
7539 ;; Got another identifier directly after the type, so it's a
7540 ;; declaration.
7545 ;; (not got-suffix-after-parens)
7547 maybe-typeless
7548 backup-maybe-typeless
7553 ;; Got a declaration of the form "foo bar (gnu);" or "bar
7554 ;; (gnu);" where we've recognized "bar" as the type and "gnu"
7555 ;; as the declarator. In this case it's however more likely
7556 ;; that "bar" is the declarator and "gnu" a function argument
7557 ;; or initializer (if `c-recognize-paren-inits' is set),
7558 ;; since the parens around "gnu" would be superfluous if it's
7559 ;; a declarator. Shift the type one step backward.
7562 ;; Found no identifier.
7567 ;; CASE 3
7569 ;; Got a plain list of identifiers. If a colon follows it's
7570 ;; a valid label, or maybe a bitfield. Otherwise the last
7571 ;; one probably is the declared identifier and we should
7572 ;; back up to the previous type, providing it isn't a cast.
7576 ;; If we've found a specifier keyword then it's a
7577 ;; declaration regardless.
7588 ;; CASE 4
7592 ;; Got a plain list of identifiers followed by some suffix.
7593 ;; If this isn't a cast then the last identifier probably is
7594 ;; the declared one and we should back up to the previous
7595 ;; type.
7599 ;; CASE 5
7601 ;; If the type is known we know that there can't be any
7602 ;; identifier somewhere else, and it's only in declarations in
7603 ;; e.g. function prototypes and in casts that the identifier may
7604 ;; be left out.
7608 ;; Only got a single identifier (parsed as a type so far).
7609 ;; CASE 6
7611 ;; Check that the identifier isn't at the start of an
7612 ;; expression.
7613 at-decl-end
7616 ;; Inside an arglist that contains declarations. If K&R
7617 ;; style declarations and parenthesis style initializers
7618 ;; aren't allowed then the single identifier must be a
7619 ;; type, else we require that it's known or found
7620 ;; (primitive types are handled above).
7625 ;; Inside a template arglist. Accept known and found
7626 ;; types; other identifiers could just as well be
7627 ;; constants in C++.
7630 ;; CASE 7
7631 ;; Can't be a valid declaration or cast, but if we've found a
7632 ;; specifier it can't be anything else either, so treat it as
7633 ;; an invalid/unfinished declaration or cast.
7641 maybe-typeless
7642 backup-maybe-typeless
7646 c-after-suffixed-type-maybe-decl-key))))))
7647 ;; Got an empty paren pair and a preceding type that probably
7648 ;; really is the identifier. Shift the type backwards to make
7649 ;; the last one the identifier. This is analogous to the
7650 ;; "backtracking" done inside the `c-type-decl-suffix-key' loop
7651 ;; above.
7652 ;;
7653 ;; Exception: In addition to the conditions in that
7654 ;; "backtracking" code, do not shift backward if we're not
7655 ;; looking at either `c-after-suffixed-type-decl-key' or "[;,]".
7656 ;; Since there's no preceding type, the shift would mean that
7657 ;; the declaration is typeless. But if the regexp doesn't match
7658 ;; then we will simply fall through in the tests below and not
7659 ;; recognize it at all, so it's better to try it as an abstract
7660 ;; declarator instead.
7663 ;; Still no identifier.
7664 ;; CASE 8
7666 ;; Require `got-prefix' together with either `got-parens' or
7667 ;; `got-suffix' to recognize it as an abstract declarator:
7668 ;; `got-parens' only is probably an empty function call.
7669 ;; `got-suffix' only can build an ordinary expression together
7670 ;; with the preceding identifier which we've taken as a type.
7671 ;; We could actually accept on `got-prefix' only, but that can
7672 ;; easily occur temporarily while writing an expression so we
7673 ;; avoid that case anyway. We could do a better job if we knew
7674 ;; the point when the fontification was invoked.
7677 ;; CASE 9
7681 got-suffix-after-parens
7683 ;; Got a type, no declarator but a paren suffix. I.e. it's a
7684 ;; normal function call after all (or perhaps a C++ style object
7685 ;; instantiation expression).
7688 ;; CASE 10
7690 ;; By now we've located the type in the declaration that we know
7691 ;; we're in.
7694 ;; CASE 11
7702 ;; Shift the type backward in the case that there's a
7703 ;; single identifier inside parens. That can only
7704 ;; occur in K&R style function declarations so it's
7705 ;; more likely that it really is a function call.
7706 ;; Therefore we only do this after
7707 ;; `c-after-suffixed-type-decl-key' has matched.
7709 got-suffix-after-parens))
7710 ;; A declaration according to `c-after-suffixed-type-decl-key'.
7713 ;; CASE 12
7716 ;; It's a declaration if a known type precedes it and it can't be a
7717 ;; function call.
7720 ;; If we get here we can't tell if this is a type decl or a normal
7721 ;; expression by looking at it alone. (That's under the assumption
7722 ;; that normal expressions always can look like type decl expressions,
7723 ;; which isn't really true but the cases where it doesn't hold are so
7724 ;; uncommon (e.g. some placements of "const" in C++) it's not worth
7725 ;; the effort to look for them.)
7727 ;;; 2008-04-16: commented out the next form, to allow the function to recognize
7728 ;;; "foo (int bar)" in CC (an implicit type (in class foo) without a semicolon)
7729 ;;; as a(n almost complete) declaration, enabling it to be fontified.
7730 ;; CASE 13
7731 ;; (unless (or at-decl-end (looking-at "=[^=]"))
7732 ;; If this is a declaration it should end here or its initializer(*)
7733 ;; should start here, so check for allowed separation tokens. Note
7734 ;; that this rule doesn't work e.g. with a K&R arglist after a
7735 ;; function header.
7736 ;;
7737 ;; *) Don't check for C++ style initializers using parens
7738 ;; since those already have been matched as suffixes.
7739 ;;
7740 ;; If `at-decl-or-cast' is then we've found some other sign that
7741 ;; it's a declaration or cast, so then it's probably an
7742 ;; invalid/unfinished one.
7743 ;; (throw 'at-decl-or-cast at-decl-or-cast))
7745 ;; Below are tests that only should be applied when we're certain to
7746 ;; not have parsed halfway through an expression.
7748 ;; CASE 14
7750 ;; The expression starts with a known type so treat it as a
7751 ;; declaration.
7754 ;; CASE 15
7756 ;; In C++ we check if the identifier is a known type, since
7757 ;; (con|de)structors use the class name as identifier.
7758 ;; We've always shifted over the identifier as a type and
7759 ;; then backed up again in this case.
7760 identifier-type
7763 ;; `at-type' probably won't be 'found for
7764 ;; destructors since the "~" is then part of the
7765 ;; type name being checked against the list of
7766 ;; known types, so do a check without that
7767 ;; operator.
7772 c-identifier-syntax-table
7776 ;; We have already parsed the type earlier,
7777 ;; so it'd be possible to cache the end
7778 ;; position instead of redoing it here, but
7779 ;; then we'd need to keep track of another
7780 ;; position everywhere.
7788 ;; CASE 16
7790 at-type
7794 at-decl-start))
7795 ;; Got something like "foo * bar;". Since we're not inside
7796 ;; an arglist it would be a meaningless expression because
7797 ;; the result isn't used. We therefore choose to recognize
7798 ;; it as a declaration. We only allow a suffix (which makes
7799 ;; the construct look like a function call) when
7800 ;; `at-decl-start' provides additional evidence that we do
7801 ;; have a declaration.
7804 ;; CASE 17
7809 ;; Got something like "a (*b) (c);" or "a (b) = c;". It could
7810 ;; be an odd expression or it could be a declaration. Treat
7811 ;; it as a declaration if "a" has been used as a type
7812 ;; somewhere else (if it's a known type we won't get here).
7815 ;; CASE 18
7821 ;; Got a type followed by an abstract declarator. If `got-prefix'
7822 ;; is set it's something like "a *" without anything after it. If
7823 ;; `got-parens' or `got-suffix' is set it's "a()", "a[]", "a()[]",
7824 ;; or similar, which we accept only if the context rules out
7825 ;; expressions.
7828 ;; If we had a complete symbol table here (which rules out
7829 ;; `c-found-types') we should return t due to the disambiguation rule
7830 ;; (in at least C++) that anything that can be parsed as a declaration
7831 ;; is a declaration. Now we're being more defensive and prefer to
7832 ;; highlight things like "foo (bar);" as a declaration only if we're
7833 ;; inside an arglist that contains declarations.
7834 ;; CASE 19
7837 ;; The point is now after the type decl expression.
7840 ;; Check for a cast.
7843 c-cast-parens
7845 ;; Should be the first type/identifier in a cast paren.
7849 ;; The closing paren should follow.
7854 ;; There should be a primary expression after it.
7863 ;; Check if the expression begins with a prefix keyword.
7866 ;; Expression begins with an ambiguous operator. Treat
7867 ;; it as a cast if it's a type decl or if we've
7868 ;; recognized the type somewhere else.
7871 ;; Unless it's a keyword, it's the beginning of a primary
7872 ;; expression.
7874 ;; If `c-primary-expr-regexp' matched a nonsymbol token, check
7875 ;; that it matched a whole one so that we don't e.g. confuse
7876 ;; the operator '-' with '->'. It's ok if it matches further,
7877 ;; though, since it e.g. can match the float '.5' while the
7878 ;; operator regexp only matches '.'.
7882 ;; There should either be a cast before it or something that isn't an
7883 ;; identifier or close paren.
7891 ;; It's a symbol. Accept it only if it's one of the
7892 ;; keywords that can precede an expression (without
7893 ;; surrounding parens).
7896 ;; Check that it isn't a close paren (block close is ok,
7897 ;; though).
7899 ;; Check that it isn't a nonsymbol identifier.
7902 ;; Handle the cast.
7912 ;; We're at a declaration. Highlight the type and the following
7913 ;; declarators.
7919 ;; Make sure to propagate the `c-decl-arg-start' property to
7920 ;; the next argument if it's set in this one, to cope with
7921 ;; interactive refontification.
7924 ;; Record the type's coordinates in `c-record-type-identifiers' for
7925 ;; later fontification.
7927 ;; There seems no reason to exclude a token from
7928 ;; fontification just because it's "a known type that can't
7929 ;; be a name or other expression". 2013-09-18.
7930 )
7941 ;; False alarm. Restore the recorded ranges.
7943 c-record-ref-identifiers save-rec-ref-ids)
7944 nil))))
7947 ;; Assuming that point is at the beginning of a token, check if it starts a
7948 ;; label and if so move over it and return non-nil (t in default situations,
7949 ;; specific symbols (see below) for interesting situations), otherwise don't
7950 ;; move and return nil. "Label" here means "most things with a colon".
7951 ;;
7952 ;; More precisely, a "label" is regarded as one of:
7953 ;; (i) a goto target like "foo:" - returns the symbol `goto-target';
7954 ;; (ii) A case label - either the entire construct "case FOO:", or just the
7955 ;; bare "case", should the colon be missing. We return t;
7956 ;; (iii) a keyword which needs a colon, like "default:" or "private:"; We
7957 ;; return t;
7958 ;; (iv) One of QT's "extended" C++ variants of
7959 ;; "private:"/"protected:"/"public:"/"more:" looking like "public slots:".
7960 ;; Returns the symbol `qt-2kwds-colon'.
7961 ;; (v) QT's construct "signals:". Returns the symbol `qt-1kwd-colon'.
7962 ;; (vi) One of the keywords matched by `c-opt-extra-label-key' (without any
7963 ;; colon). Currently (2006-03), this applies only to Objective C's
7964 ;; keywords "@private", "@protected", and "@public". Returns t.
7965 ;;
7966 ;; One of the things which will NOT be recognized as a label is a bit-field
7967 ;; element of a struct, something like "int foo:5".
7968 ;;
7969 ;; The end of the label is taken to be just after the colon, or the end of
7970 ;; the first submatch in `c-opt-extra-label-key'. The point is directly
7971 ;; after the end on return. The terminating char gets marked with
7972 ;; `c-decl-end' to improve recognition of the following declaration or
7973 ;; statement.
7974 ;;
7975 ;; If ASSUME-MARKUP is non-nil, it's assumed that the preceding
7976 ;; label, if any, has already been marked up like that.
7977 ;;
7978 ;; If PRECEDING-TOKEN-END is given, it should be the first position
7979 ;; after the preceding token, i.e. on the other side of the
7980 ;; syntactic ws from the point. Use a value less than or equal to
7981 ;; (point-min) if the point is at the first token in (the visible
7982 ;; part of) the buffer.
7983 ;;
7984 ;; The optional LIMIT limits the forward scan for the colon.
7985 ;;
7986 ;; This function records the ranges of the label symbols on
7987 ;; `c-record-ref-identifiers' if `c-record-type-identifiers' (!) is
7988 ;; non-nil.
7989 ;;
7990 ;; This function might do hidden buffer changes.
7993 label-end
7994 qt-symbol-idx
7995 macro-start ; if we're in one.
7996 label-type
7997 kwd)
7999 ;; "case" or "default" (Doesn't apply to AWK).
8002 ;; Record only the keyword itself for fontification, since in
8003 ;; case labels the following is a constant expression and not
8004 ;; a label.
8008 ;; Find the label end.
8012 ;; Stop on chars that aren't allowed in expressions,
8013 ;; and on operator chars that would be meaningless
8014 ;; there. FIXME: This doesn't cope with ?: operators.
8022 t)
8024 ;; It's an unfinished label. We consider the keyword enough
8025 ;; to recognize it as a label, so that it gets fontified.
8026 ;; Leave the point at the end of it, but don't put any
8027 ;; `c-decl-end' marker.
8029 t))))
8031 ;; @private, @protected, @public, in Objective C, or similar.
8034 ;; For a `c-opt-extra-label-key' match, we record the whole
8035 ;; thing for fontification. That's to get the leading '@' in
8036 ;; Objective-C protection labels fontified.
8043 ;; All other cases of labels.
8046 ;; A colon label must have something before the colon.
8049 ;; Check that we're not after a token that can't precede a label.
8051 ;; Trivially succeeds when there's no preceding token.
8052 ;; Succeeds when we're at a virtual semicolon.
8061 ;; Check if we're after a label, if we're after a closing
8062 ;; paren that belong to statement, and with
8063 ;; `c-label-prefix-re'. It's done in different order
8064 ;; depending on `assume-markup' since the checks have
8065 ;; different expensiveness.
8092 ;; Might be after another label, so check it recursively.
8096 ;; Essentially the same as the
8097 ;; `c-syntactic-re-search-forward' regexp below.
8103 ;; Note: the following should work instead of the
8104 ;; narrow-to-region above. Investigate why not,
8105 ;; sometime. ACM, 2006-03-31.
8106 ;; (c-syntactic-skip-backward "^-]:?;}=*/%&|,<>!@+"
8107 ;; macro-start t)
8109 ;; If the caller turned on recording for us,
8110 ;; it shouldn't apply when we check the
8111 ;; preceding label.
8112 c-record-type-identifiers)
8113 ;; A label can't start at a cpp directive. Check for
8114 ;; this, since c-forward-syntactic-ws would foul up on it.
8119 ;; Point is still at the beginning of the possible label construct.
8120 ;;
8121 ;; Check that the next nonsymbol token is ":", or that we're in one
8122 ;; of QT's "slots" declarations. Allow '(' for the sake of macro
8123 ;; arguments. FIXME: Should build this regexp from the language
8124 ;; constants.
8126 ;; public: protected: private:
8136 ;; QT double keyword like "protected slots:" or goto target.
8155 'qt-1kwd-colon
8167 ;; Check that `c-nonlabel-token-key' doesn't match anywhere.
8177 t)
8180 ;; Record the identifiers in the label for fontification, unless
8181 ;; it begins with `c-label-kwds' in which case the following
8182 ;; identifiers are part of a (constant) expression that
8183 ;; shouldn't be fontified.
8195 ;; Not a label.
8197 label-type))
8200 ;; Assuming the point is at the beginning of a token, try to move
8201 ;; forward to the end of the Objective-C directive that starts
8202 ;; there. Return t if a directive was fully recognized, otherwise
8203 ;; the point is moved as far as one could be successfully parsed and
8204 ;; nil is returned.
8205 ;;
8206 ;; This function records identifier ranges on
8207 ;; `c-record-type-identifiers' and `c-record-ref-identifiers' if
8208 ;; `c-record-type-identifiers' is non-nil.
8209 ;;
8210 ;; This function might do hidden buffer changes.
8213 start-char
8215 lim
8216 ;; Turn off recognition of angle bracket arglists while parsing
8217 ;; types here since the protocol reference list might then be
8218 ;; considered part of the preceding name or superclass-name.
8229 t)
8238 ;; Handle the name of the class itself.
8240 ;; (c-forward-token-2) ; 2006/1/13 This doesn't move if the token's
8241 ;; at EOB.
8248 ;; Look for ": superclass-name" or "( category-name )".
8259 ;; Look for a protocol reference list.
8265 t))))
8271 t)
8274 nil)))
8277 ;; Go to the first non-whitespace after the colon that starts a
8278 ;; multiple inheritance introduction. Optional LIM is the farthest
8279 ;; back we should search.
8280 ;;
8281 ;; This function might do hidden buffer changes.
8289 ;; Return nil if we aren't in a method definition, otherwise the
8290 ;; position of the initial [+-].
8291 ;;
8292 ;; This function might do hidden buffer changes.
8298 ))
8300 ;; Contributed by Kevin Ryde <user42@zip.com.au>.
8302 ;; Return non-nil if point is within a gcc \"asm\" block.
8303 ;;
8304 ;; This should be called with point inside an argument list.
8305 ;;
8306 ;; Only one level of enclosing parentheses is considered, so for
8307 ;; instance nil is returned when in a function call within an asm
8308 ;; operand.
8309 ;;
8310 ;; This function might do hidden buffer changes.
8322 a function.
8324 More precisely, being at the top-level means that point is either
8325 outside any enclosing block (such as a function definition), or
8326 directly inside a class, namespace or other block that contains
8327 another declaration level.
8329 If point is not at the top-level (e.g. it is inside a method
8330 definition), then nil is returned. Otherwise, if point is at a
8331 top-level not enclosed within a class definition, t is returned.
8332 Otherwise, a 2-vector is returned where the zeroth element is the
8333 buffer position of the start of the class declaration, and the first
8334 element is the buffer position of the enclosing class's opening
8335 brace.
8337 Note that this function might do hidden buffer changes. See the
8338 comment at the start of cc-engine.el for more info."
8339 ;; Note to maintainers: this function consumes a great mass of CPU cycles.
8340 ;; Its use should thus be minimized as far as possible.
8346 ;; Return non-nil if the point is in the region after the argument
8347 ;; list of a function and its opening brace (or semicolon in case it
8348 ;; got no body). If there are K&R style argument declarations in
8349 ;; that region, the point has to be inside the first one for this
8350 ;; function to recognize it.
8351 ;;
8352 ;; If successful, the point is moved to the first token after the
8353 ;; function header (see `c-forward-decl-or-cast-1' for details) and
8354 ;; the position of the opening paren of the function arglist is
8355 ;; returned.
8356 ;;
8357 ;; The point is clobbered if not successful.
8358 ;;
8359 ;; LIM is used as bound for backward buffer searches.
8360 ;;
8361 ;; This function might do hidden buffer changes.
8374 ;; There should not be a '=' or ',' between beg and the
8375 ;; start of the declaration since that means we were in the
8376 ;; "expression part" of the declaration.
8381 ;; Check that there's an arglist paren in the
8382 ;; declaration.
8385 ;; The declarator is a paren expression, so skip past it
8386 ;; so that we don't get stuck on that instead of the
8387 ;; function arglist.
8391 ;; Don't trip up on "operator ()".
8398 ;; Return the position of the first argument declaration if point is
8399 ;; inside a K&R style argument declaration list, nil otherwise.
8400 ;; `c-recognize-knr-p' is not checked. If LIM is non-nil, it's a
8401 ;; position that bounds the backward search for the argument list. This
8402 ;; function doesn't move point.
8403 ;;
8404 ;; Point must be within a possible K&R region, e.g. just before a top-level
8405 ;; "{". It must be outside of parens and brackets. The test can return
8406 ;; false positives otherwise.
8407 ;;
8408 ;; This function might do hidden buffer changes.
8411 ;; If we're in a macro, our search range is restricted to it. Narrow to
8412 ;; the searchable range.
8416 before-lparen after-rparen
8419 ; we give up.
8420 ids ; List of identifiers in the parenthesized list.
8421 id-start after-prec-token decl-or-cast decl-res
8422 c-last-identifier-range identifier-ok)
8425 ;; Search backwards for the defun's argument list. We give up if we
8426 ;; encounter a "}" (end of a previous defun) an "=" (which can't be in
8427 ;; a knr region) or BOB.
8428 ;;
8429 ;; The criterion for a paren structure being the arg list is:
8430 ;; o - there is non-WS stuff after it but before any "{"; AND
8431 ;; o - the token after it isn't a ";" AND
8432 ;; o - it is preceded by either an identifier (the function name) or
8433 ;; a macro expansion like "DEFUN (...)"; AND
8434 ;; o - its content is a non-empty comma-separated list of identifiers
8435 ;; (an empty arg list won't have a knr region).
8436 ;;
8437 ;; The following snippet illustrates these rules:
8438 ;; int foo (bar, baz, yuk)
8439 ;; int bar [] ;
8440 ;; int (*baz) (my_type) ;
8441 ;; int (*(* yuk) (void)) (void) ;
8442 ;; {
8443 ;;
8444 ;; Additionally, for a knr list to be recognized:
8445 ;; o - The identifier of each declarator up to and including the
8446 ;; one "near" point must be contained in the arg list.
8457 ; parens/brackets.
8461 ;; We're inside a paren. Could it be our argument list....?
8467 ;; FIXME!!! What about macros between the parens? 2007/01/20
8470 ;; It can't be the arg list if next token is ; or {
8475 ;; Is the thing preceding the list an identifier (the
8476 ;; function name), or a macro expansion?
8486 ;; Have we got a non-empty list of comma-separated
8487 ;; identifiers?
8500 ids)
8508 ;; Are all the identifiers in the k&r list up to the
8509 ;; current one also in the argument list?
8517 after-prec-token
8518 nil ; Or 'arglist ???
8519 nil)))
8524 nil))
8532 ids))
8538 identifier-ok))
8539 ;; ...Yes. We've identified the function's argument list.
8544 ;; ...No. The current parens aren't the function's arg list.
8551 ;; skip forward over conditional at point, including any predicate
8552 ;; statements in parentheses. No error checking is performed.
8553 ;;
8554 ;; This function might do hidden buffer changes.
8556 ;; else if()
8561 ;; do, else, try, finally
8566 ;; for, if, while, switch, catch, synchronized, foreach
8570 ;; If looking at the token after a conditional then return the
8571 ;; position of its start, otherwise return nil.
8572 ;;
8573 ;; This function might do hidden buffer changes.
8584 ;; If the point is after an operator identifier that isn't handled
8585 ;; like an ordinary symbol (i.e. like "operator =" in C++) then the
8586 ;; position of the start of that identifier is returned. nil is
8587 ;; returned otherwise. The point may be anywhere in the syntactic
8588 ;; whitespace after the last token of the operator identifier.
8589 ;;
8590 ;; This function might do hidden buffer changes.
8602 ;; Assuming point is at a brace that opens a statement block of some
8603 ;; kind, move to the proper anchor point for that block. It might
8604 ;; need to be adjusted further by c-add-stmt-syntax, but the
8605 ;; position at return is suitable as start position for that
8606 ;; function.
8607 ;;
8608 ;; This function might do hidden buffer changes.
8615 ;; Assuming point is at a brace that opens the block of a top level
8616 ;; declaration of some kind, move to the proper anchor point for
8617 ;; that block.
8618 ;;
8619 ;; This function might do hidden buffer changes.
8624 ;; Search forward for the end of the "header" of the current
8625 ;; declaration. That's the position where the definition body
8626 ;; starts, or the first variable initializer, or the ending
8627 ;; semicolon. I.e. search forward for the closest following
8628 ;; (syntactically relevant) '{', '=' or ';' token. Point is left
8629 ;; _after_ the first found token, or at point-max if none is found.
8630 ;;
8631 ;; This function might do hidden buffer changes.
8636 ;; In C++ we need to take special care to handle operator
8637 ;; tokens and those pesky template brackets.
8642 ;; Handle operator identifiers, i.e. ignore any
8643 ;; operator token preceded by "operator".
8650 t
8652 nil)))))
8661 ;; Go to the beginning of the current declaration, or the beginning
8662 ;; of the previous one if already at the start of it. Point won't
8663 ;; be moved out of any surrounding paren. Return a cons cell of the
8664 ;; form (MOVE . KNR-POS). MOVE is like the return value from
8665 ;; `c-beginning-of-statement-1'. If point skipped over some K&R
8666 ;; style argument declarations (and they are to be recognized) then
8667 ;; KNR-POS is set to the start of the first such argument
8668 ;; declaration, otherwise KNR-POS is nil. If LIM is non-nil, it's a
8669 ;; position that bounds the backward search.
8670 ;;
8671 ;; NB: Cases where the declaration continues after the block, as in
8672 ;; "struct foo { ... } bar;", are currently recognized as two
8673 ;; declarations, e.g. "struct foo { ... }" and "bar;" in this case.
8674 ;;
8675 ;; This function might do hidden buffer changes.
8681 ;; `c-beginning-of-statement-1' stops at a block start, but we
8682 ;; want to continue if the block doesn't begin a top level
8683 ;; construct, i.e. if it isn't preceded by ';', '}', ':', bob,
8684 ;; or an open paren.
8686 ;; Go back one "statement" each time round the loop until we're just
8687 ;; after a ;, }, or :, or at BOB or the start of a macro or start of
8688 ;; an ObjC method. This will move over a multiple declaration whose
8689 ;; components are comma separated.
8691 ;; Must check with c-opt-method-key in ObjC mode.
8701 ;; Check that we don't move from the first thing in a
8702 ;; macro to its header.
8708 move tentative-move))
8713 ;; Handle K&R argdecls. Back up after the "statement" jumped
8714 ;; over by `c-beginning-of-statement-1', unless it was the
8715 ;; function body, in which case we're sitting on the opening
8716 ;; brace now. Then test if we're in a K&R argdecl region and
8717 ;; that we started at the other side of the first argdecl in
8718 ;; it.
8728 'previous
8730 knr-argdecl-start))
8733 ;; `c-beginning-of-statement-1' counts each brace block as a separate
8734 ;; statement, so the result will be 'previous if we've moved over any.
8735 ;; So change our result back to 'same if necessary.
8736 ;;
8737 ;; If they were brace list initializers we might not have moved over a
8738 ;; declaration boundary though, so change it to 'same if we've moved
8739 ;; past a '=' before '{', but not ';'. (This ought to be integrated
8740 ;; into `c-beginning-of-statement-1', so we avoid this extra pass which
8741 ;; potentially can search over a large amount of text.). Take special
8742 ;; pains not to get mislead by C++'s "operator=", and the like.
8745 c++-template-syntax-table
8751 ; no more, or get to one which isn't an "operator ="
8754 c-overloadable-operators-regexp
8755 c-opt-op-identifier-prefix
8770 ;; Assuming point is at the start of a declaration (as detected by
8771 ;; e.g. `c-beginning-of-decl-1'), go to the end of it. Unlike
8772 ;; `c-beginning-of-decl-1', this function handles the case when a
8773 ;; block is followed by identifiers in e.g. struct declarations in C
8774 ;; or C++. If a proper end was found then t is returned, otherwise
8775 ;; point is moved as far as possible within the current sexp and nil
8776 ;; is returned. This function doesn't handle macros; use
8777 ;; `c-end-of-macro' instead in those cases.
8778 ;;
8779 ;; This function might do hidden buffer changes.
8782 c++-template-syntax-table
8790 ;; Stopped at the ';' in a K&R argdecl section which is
8791 ;; detected using the same criteria as in
8792 ;; `c-beginning-of-decl-1'. Move to the following block
8793 ;; start.
8797 ;; Encountered a block in the declaration. Jump over it.
8808 ;; Check for `c-opt-block-decls-with-vars-key'
8809 ;; before the first paren.
8812 c-opt-block-decls-with-vars-key
8814 lim t t t)
8817 ;; Check that the first following paren is
8818 ;; the block.
8820 lim t t t)
8822 ;; The declaration doesn't have any of the
8823 ;; `c-opt-block-decls-with-vars' keywords in the
8824 ;; beginning, so it ends here at the end of the block.
8832 nil)))
8835 ;; Assuming the point is at an open brace, check if it starts a
8836 ;; block that contains another declaration level, i.e. that isn't a
8837 ;; statement block or a brace list, and if so return non-nil.
8838 ;;
8839 ;; If the check is successful, the return value is the start of the
8840 ;; keyword that tells what kind of construct it is, i.e. typically
8841 ;; what `c-decl-block-key' matched. Also, if GOTO-START is set then
8842 ;; the point will be at the start of the construct, before any
8843 ;; leading specifiers, otherwise it's at the returned position.
8844 ;;
8845 ;; The point is clobbered if the check is unsuccessful.
8846 ;;
8847 ;; CONTAINING-SEXP is the position of the open of the surrounding
8848 ;; paren, or nil if none.
8849 ;;
8850 ;; The optional LIMIT limits the backward search for the start of
8851 ;; the construct. It's assumed to be at a syntactically relevant
8852 ;; position.
8853 ;;
8854 ;; If any template arglists are found in the searched region before
8855 ;; the open brace, they get marked with paren syntax.
8856 ;;
8857 ;; This function might do hidden buffer changes.
8864 ;; Could be at the end of a template arglist.
8872 ;; Note: Can't get bogus hits inside template arglists below since they
8873 ;; have gotten paren syntax above.
8875 ;; If `goto-start' is set we begin by searching for the
8876 ;; first possible position of a leading specifier list.
8877 ;; The `c-decl-block-key' search continues from there since
8878 ;; we know it can't match earlier.
8881 open-brace t t)
8883 t)
8884 t)
8890 ;; Exclude cases where we matched what would ordinarily
8891 ;; be a block declaration keyword, except where it's not
8892 ;; legal because it's part of a "compound keyword" like
8893 ;; "enum class". Of course, if c-after-brace-list-key
8894 ;; is nil, we can skip the test.
8904 ;; Found a keyword that can't be a type?
8907 ;; Can be a type too, in which case it's the return type of a
8908 ;; function (under the assumption that no declaration level
8909 ;; block construct starts with a type).
8912 ;; Jumped over a type, but it could be a declaration keyword
8913 ;; followed by the declared identifier that we've jumped over
8914 ;; instead (e.g. in "class Foo {"). If it indeed is a type
8915 ;; then we should be at the declarator now, so check for a
8916 ;; valid declarator start.
8917 ;;
8918 ;; Note: This doesn't cope with the case when a declared
8919 ;; identifier is followed by e.g. '(' in a language where '('
8920 ;; also might be part of a declarator expression. Currently
8921 ;; there's no such language.
8925 ;; In Pike a list of modifiers may be followed by a brace
8926 ;; to make them apply to many identifiers. Note that the
8927 ;; match data will be empty on return in this case.
8933 ;; Use this variant to avoid yet another special regexp.
8937 t)))
8939 ;; Got a match.
8942 ;; Back up over any preceding specifiers and their clauses
8943 ;; by going forward from `first-specifier-pos', which is the
8944 ;; earliest possible position where the specifier list can
8945 ;; start.
8951 ;; Accept any plain symbol token on the ground that
8952 ;; it's a specifier masked through a macro (just
8953 ;; like `c-forward-decl-or-cast-1' skip forward over
8954 ;; such tokens).
8955 ;;
8956 ;; Could be more restrictive wrt invalid keywords,
8957 ;; but that'd only occur in invalid code so there's
8958 ;; no use spending effort on it.
8964 ;; Can't parse a declaration preamble and is still
8965 ;; before `kwd-start'. That means `first-specifier-pos'
8966 ;; was in some earlier construct. Search again.
8970 ;; Got no preamble before the block declaration keyword.
8976 kwd-start)))
8979 ;; Check if the closest containing paren sexp is a declaration
8980 ;; block, returning a 2 element vector in that case. Aref 0
8981 ;; contains the bufpos at boi of the class key line, and aref 1
8982 ;; contains the bufpos of the open brace. This function is an
8983 ;; obsolete wrapper for `c-looking-at-decl-block'.
8984 ;;
8985 ;; This function might do hidden buffer changes.
8993 nil))
8998 ;; Return the buffer position of the most enclosing decl-block brace (in the
8999 ;; sense of c-looking-at-decl-block) in the PAREN-STATE structure, or nil if
9000 ;; none was found.
9009 open-brace))
9012 ;; Return the position of the L-brace if point is inside a brace list
9013 ;; initialization of an array, etc. This is an approximate function,
9014 ;; designed for speed over accuracy. It will not find every bracelist, but
9015 ;; a non-nil result is reliable. We simply search for "= {" (naturally with
9016 ;; syntactic whitespace allowed). PAREN-STATE is the normal thing that it
9017 ;; is everywhere else.
9026 b-pos)))
9029 ;; We're at a "{" which might be the opening brace of a enum which is
9030 ;; strongly typed (by a ":" followed by a type). If this is the case, leave
9031 ;; point before the colon and return t. Otherwise leave point unchanged and return nil.
9032 ;; Match data will be clobbered.
9052 nil)
9054 t)
9056 t)
9060 t)))
9063 ;; We're at a "{". Move back to the enum-like keyword that starts this
9064 ;; declaration and return t, otherwise don't move and return nil.
9066 up-sexp-pos before-identifier)
9082 t)
9094 t)
9100 ;; return the buffer position of the beginning of the brace list
9101 ;; statement if we're inside a brace list, otherwise return nil.
9102 ;; CONTAINING-SEXP is the buffer pos of the innermost containing
9103 ;; paren. PAREN-STATE is the remainder of the state of enclosing
9104 ;; braces
9105 ;;
9106 ;; N.B.: This algorithm can potentially get confused by cpp macros
9107 ;; placed in inconvenient locations. It's a trade-off we make for
9108 ;; speed.
9109 ;;
9110 ;; This function might do hidden buffer changes.
9112 ;; This will pick up brace list declarations.
9116 ;; this will pick up array/aggregate init lists, even if they are nested.
9119 ;; Pike can have class definitions anywhere, so we must
9120 ;; check for the class key here.
9122 c-decl-block-key))
9123 bufpos braceassignp lim next-containing macro-start)
9125 containing-sexp)
9136 ;; We're in an in-expression block of some kind. Do not
9137 ;; check nesting. We deliberately set the limit to the
9138 ;; containing sexp, so that c-looking-at-inexpr-block
9139 ;; doesn't check for an identifier before it.
9141 ;; see if the open brace is preceded by = or [...] in
9142 ;; this statement, but watch out for operator=
9145 ;; Checks to do only on the first sexp before the brace.
9148 ;; In Java, an initialization brace list may follow
9149 ;; directly after "new Foo[]", so check for a "new"
9150 ;; earlier.
9156 ;; Carry on looking if this is an
9157 ;; identifier (may contain "." in Java)
9158 ;; or another "[]" sexp.
9161 ;; Checks to do on all sexps before the brace, up to the
9162 ;; beginning of the statement.
9170 ;; We've seen a =, but must check earlier tokens so
9171 ;; that it isn't something that should be ignored.
9177 ;; Check for operator =
9180 nil)
9181 ;; Check for `<opchar>= in Pike.
9184 ;; Special case for Pikes
9185 ;; `[]=, since '[' is not in
9186 ;; the punctuation class.
9189 nil)
9191 ;; make sure we're not in a C++ template
9192 ;; argument assignment
9204 ))))
9205 nil)
9212 ;; We've hit the beginning of the aggregate list.
9217 ;; Brace lists can't contain a semicolon, so we're done.
9222 ;; We've a macro whose expansion starts with the '{'.
9223 ;; Heuristically, if we have a ';' in it we've not got a
9224 ;; brace list, otherwise we have.
9231 containing-sexp nil)
9234 ;; Go up one level
9236 lim nil
9237 next-containing nil)))))
9239 bufpos))
9240 ))
9243 ;; If we're looking at the start of a pike-style list, i.e., `({Â })',
9244 ;; `([Â ])', `(<Â >)', etc., a cons of a cons of its starting and ending
9245 ;; positions and its entry in c-special-brace-lists is returned, nil
9246 ;; otherwise. The ending position is nil if the list is still open.
9247 ;; LIM is the limit for forward search. The point may either be at
9248 ;; the `(' or at the following paren character. Tries to check the
9249 ;; matching closer, but assumes it's correct if no balanced paren is
9250 ;; found (i.e. the case `({ ... } ... )' is detected as _not_ being
9251 ;; a special brace list).
9252 ;;
9253 ;; This function might do hidden buffer changes.
9258 inner-beg end type)
9273 nil)))))
9283 ;; Check balancing of the inner paren
9284 ;; below.
9287 t)
9288 ;; If the inner char isn't a paren then
9289 ;; we can't check balancing, so just
9290 ;; check the char before the outer
9291 ;; closing paren.
9306 ;; Return non-nil if between two statements or declarations, assuming
9307 ;; point is not inside a literal or comment.
9308 ;;
9309 ;; Obsolete - `c-at-statement-start-p' or `c-at-expression-start-p'
9310 ;; are recommended instead.
9311 ;;
9312 ;; This function might do hidden buffer changes.
9317 ;; Return non-nil if we're looking at the beginning of a block
9318 ;; inside an expression. The value returned is actually a cons of
9319 ;; either 'inlambda, 'inexpr-statement or 'inexpr-class and the
9320 ;; position of the beginning of the construct.
9321 ;;
9322 ;; LIM limits the backward search. CONTAINING-SEXP is the start
9323 ;; position of the closest containing list. If it's nil, the
9324 ;; containing paren isn't used to decide whether we're inside an
9325 ;; expression or not. If both LIM and CONTAINING-SEXP are used, LIM
9326 ;; needs to be farther back.
9327 ;;
9328 ;; If CHECK-AT-END is non-nil then extra checks at the end of the
9329 ;; brace block might be done. It should only be used when the
9330 ;; construct can be assumed to be complete, i.e. when the original
9331 ;; starting position was further down than that.
9332 ;;
9333 ;; This function might do hidden buffer changes.
9338 ;; Look at the character after point only as a last resort
9339 ;; when we can't disambiguate.
9359 ;; If the class definition is at the start of
9360 ;; a statement, we don't consider it an
9361 ;; in-expression class.
9369 ;; Also, in Pike we treat it as an
9370 ;; in-expression class if it's used in an
9371 ;; object clone expression.
9387 nil)
9395 ;; Accept several square bracket sexps for
9396 ;; Java array initializations.
9404 block-follows
9405 containing-sexp
9419 nil
9422 res))))
9425 ;; Returns non-nil if we're looking at the end of an in-expression
9426 ;; block, otherwise the same as `c-looking-at-inexpr-block'.
9427 ;; PAREN-STATE is the paren state relevant at the current position.
9428 ;;
9429 ;; This function might do hidden buffer changes.
9431 ;; We currently only recognize a block.
9434 containing-sexp)
9443 paren-state)
9444 containing-sexp)))))
9447 ;; Is there a "virtual semicolon" at POS or point?
9448 ;; (See cc-defs.el for full details of "virtual semicolons".)
9449 ;;
9450 ;; This is true when point is at the last non syntactic WS position on the
9451 ;; line, there is a macro call last on the line, and this particular macro's
9452 ;; name is defined by the regexp `c-vs-macro-regexp' as not needing a
9453 ;; semicolon.
9461 c-macro-with-semi-re
9464 ;; Check we've got nothing after this except comments and empty lines
9465 ;; joined by escaped EOLs.
9478 nil)
9486 ;; Check for one of the listed macros being before point.
9490 t))
9498 \f
9499 ;; `c-guess-basic-syntax' and the functions that precedes it below
9500 ;; implements the main decision tree for determining the syntactic
9501 ;; analysis of the current line of code.
9503 ;; Dynamically bound to t when `c-guess-basic-syntax' is called during
9504 ;; auto newline analysis.
9508 ;; BRACEPOS is the position of a brace in a construct like "namespace
9509 ;; Bar {". Return the anchor point in this construct; this is the
9510 ;; earliest symbol on the brace's line which isn't earlier than
9511 ;; "namespace".
9512 ;;
9513 ;; Currently (2007-08-17), "like namespace" means "matches
9514 ;; c-other-block-decl-kwds". It doesn't work with "class" or "struct"
9515 ;; or anything like that.
9525 ;; A simple function to prepend a new syntax element to
9526 ;; `c-syntactic-context'. Using `setq' on it is unsafe since it
9527 ;; should always be dynamically bound but since we read it first
9528 ;; we'll fail properly anyway if this function is misused.
9530 c-syntactic-context)))
9533 ;; Like `c-add-syntax' but appends to the end of the syntax list.
9534 ;; (Normally not necessary.)
9539 syntax-extra-args
9540 stop-at-boi-only
9541 containing-sexp
9542 paren-state)
9543 ;; Add the indicated SYNTAX-SYMBOL to `c-syntactic-context', extending it as
9544 ;; needed with further syntax elements of the types `substatement',
9545 ;; `inexpr-statement', `arglist-cont-nonempty', `statement-block-intro', and
9546 ;; `defun-block-intro'.
9547 ;;
9548 ;; Do the generic processing to anchor the given syntax symbol on
9549 ;; the preceding statement: Skip over any labels and containing
9550 ;; statements on the same line, and then search backward until we
9551 ;; find a statement or block start that begins at boi without a
9552 ;; label or comment.
9553 ;;
9554 ;; Point is assumed to be at the prospective anchor point for the
9555 ;; given SYNTAX-SYMBOL. More syntax entries are added if we need to
9556 ;; skip past open parens and containing statements. Most of the added
9557 ;; syntax elements will get the same anchor point - the exception is
9558 ;; for an anchor in a construct like "namespace"[*] - this is as early
9559 ;; as possible in the construct but on the same line as the {.
9560 ;;
9561 ;; [*] i.e. with a keyword matching c-other-block-decl-kwds.
9562 ;;
9563 ;; SYNTAX-EXTRA-ARGS are a list of the extra arguments for the
9564 ;; syntax symbol. They are appended after the anchor point.
9565 ;;
9566 ;; If STOP-AT-BOI-ONLY is nil, we can stop in the middle of the line
9567 ;; if the current statement starts there.
9568 ;;
9569 ;; Note: It's not a problem if PAREN-STATE "overshoots"
9570 ;; CONTAINING-SEXP, i.e. contains info about parens further down.
9571 ;;
9572 ;; This function might do hidden buffer changes.
9575 ;; This is by far the most common case, so let's give it special
9576 ;; treatment.
9581 ;; Set when we're on a label, so that we don't stop there.
9582 ;; FIXME: To be complete we should check if we're on a label
9583 ;; now at the start.
9584 on-label)
9586 ;; Use point as the anchor point for "namespace", "extern", etc.
9590 syntax-extra-args)
9592 ;; Loop while we have to back out of containing blocks.
9597 ;; Loop while we have to back up statements.
9599 on-label
9602 ;; Skip past any comments that stands between the
9603 ;; statement start and boi.
9611 ;; Skip to the beginning of this statement or backward
9612 ;; another one.
9620 ;; If we didn't move we're at the start of a block and
9621 ;; have to continue outside it.
9630 ;; Special case to avoid deeper and deeper indentation
9631 ;; of "else if" clauses.
9632 )
9637 ;; If stop-at-boi-only is nil, we shouldn't back up
9638 ;; over previous or containing statements to try to
9639 ;; reach boi, so go back to the last position and
9640 ;; exit.
9647 ;; If we've moved into another statement then we
9648 ;; should no longer try to stop in the middle of a
9649 ;; line.
9652 ;; Record this as a substatement if we skipped up one
9653 ;; level.
9656 )))
9658 containing-sexp)
9660 ;; Now we have to go out of this block.
9663 ;; Don't stop in the middle of a special brace list opener
9664 ;; like "({".
9676 ;; Analyze the construct in front of the block we've stepped out
9677 ;; from and add the right syntactic element for it.
9680 step-type)
9683 ;; Stepped out of a parenthesis block, so we're in an
9684 ;; expression now.
9696 ;; Stepped out of an in-expression statement. This
9697 ;; syntactic element won't get an anchor pos.
9700 ;; A parenthesis normally belongs to an arglist.
9708 on-label nil))
9710 ;; Stepped out of a brace block.
9722 containing-sexp)))
9724 'defun-block-intro
9726 nil))
9730 c-other-decl-block-key-in-symbols-alist))
9737 ;; Always done if the open brace was at boi. The
9738 ;; c-beginning-of-statement-1 call above is necessary
9739 ;; anyway, to decide the type of block-intro to add.
9742 ))
9744 ;; Fill in the current point as the anchor for all the symbols
9745 ;; added above.
9754 )))
9757 containing-decl-open
9758 containing-decl-start
9759 containing-decl-kwd
9760 paren-state)
9761 ;; The inclass and class-close syntactic symbols are added in
9762 ;; several places and some work is needed to fix everything.
9763 ;; Therefore it's collected here.
9764 ;;
9765 ;; This function might do hidden buffer changes.
9770 containing-decl-open)
9772 ;; Ought to use `c-add-stmt-syntax' instead of backing up to boi
9773 ;; here, but we have to do like this for compatibility.
9783 char-after-ip
9784 beg-of-same-or-containing-stmt
9785 containing-sexp
9786 paren-state)
9787 ;; This function contains the decision tree reached through both
9788 ;; cases 18 and 10. It's a continued statement or top level
9789 ;; construct of some kind.
9790 ;;
9791 ;; This function might do hidden buffer changes.
9798 ;; (CASE A removed.)
9799 ;; CASE B: open braces for class or brace-lists
9806 ;; CASE B.1: class-open
9813 ;; CASE B.2: brace-list-open
9818 indent-point t t t)))
9819 ;; The most semantically accurate symbol here is
9820 ;; brace-list-open, but we normally report it simply as a
9821 ;; statement-cont. The reason is that one normally adjusts
9822 ;; brace-list-open for brace lists as top-level constructs,
9823 ;; and brace lists inside statements is a completely different
9824 ;; context. C.f. case 5A.3.
9827 ;; Turn off the dwim above when we're
9828 ;; analyzing the nature of the brace
9829 ;; for the auto newline feature.
9830 'brace-list-open
9832 nil nil
9833 containing-sexp paren-state))
9835 ;; CASE B.3: The body of a function declared inside a normal
9836 ;; block. Can occur e.g. in Pike and when using gcc
9837 ;; extensions, but watch out for macros followed by blocks.
9838 ;; C.f. cases E, 16F and 17G.
9844 ;; Turn off recognition of constructs that lacks a
9845 ;; type in this case, since that's more likely to be
9846 ;; a macro followed by a block.
9849 containing-sexp paren-state))
9851 ;; CASE B.4: Continued statement with block open. The most
9852 ;; accurate analysis is perhaps `statement-cont' together with
9853 ;; `block-open' but we play DWIM and use `substatement-open'
9854 ;; instead. The rationale is that this typically is a macro
9855 ;; followed by a block which makes it very similar to a
9856 ;; statement with a substatement block.
9859 containing-sexp paren-state))
9860 ))
9862 ;; CASE C: iostream insertion or extraction operator
9866 ;; If there is no preceding streamop in the statement
9867 ;; then indent this line as a normal statement-cont.
9871 t))))
9873 ;; CASE E: In the "K&R region" of a function declared inside a
9874 ;; normal block. C.f. case B.3.
9876 ;; Check that the next token is a '{'. This works as
9877 ;; long as no language that allows nested function
9878 ;; definitions allows stuff like member init lists, K&R
9879 ;; declarations or throws clauses there.
9880 ;;
9881 ;; Note that we do a forward search for something ahead
9882 ;; of the indentation line here. That's not good since
9883 ;; the user might not have typed it yet. Unfortunately
9884 ;; it's exceedingly tricky to recognize a function
9885 ;; prototype in a code block without resorting to this.
9893 ;; Turn off recognition of constructs that lacks a
9894 ;; type in this case, since that's more likely to be
9895 ;; a macro followed by a block.
9898 containing-sexp paren-state))
9900 ;;CASE F: continued statement and the only preceding items are
9901 ;;annotations.
9909 t)
9916 ;; CASE G: a template list continuation?
9917 ;; Mostly a duplication of case 5D.3 to fix templates-19:
9935 ;; In a nested template arglist.
9941 ;; FIXME: Should use c-add-stmt-syntax, but it's not yet
9942 ;; template aware.
9945 ;; CASE D: continued statement.
9949 containing-sexp paren-state))
9950 )))
9952 ;; The next autoload was added by RMS on 2005/8/9 - don't know why (ACM,
9953 ;; 2005/11/29).
9954 ;;;###autoload
9956 "Return the syntactic context of the current line."
9962 ;; A whole ugly bunch of various temporary variables. Have
9963 ;; to declare them here since it's not possible to declare
9964 ;; a variable with only the scope of a cond test and the
9965 ;; following result clauses, and most of this function is a
9966 ;; single gigantic cond. :P
9967 literal char-before-ip before-ws-ip char-after-ip macro-start
9968 in-macro-expr c-syntactic-context placeholder c-in-literal-cache
9969 step-type tmpsymbol keyword injava-inher special-brace-list tmp-pos
9970 containing-<
9971 ;; The following record some positions for the containing
9972 ;; declaration block if we're directly within one:
9973 ;; `containing-decl-open' is the position of the open
9974 ;; brace. `containing-decl-start' is the start of the
9975 ;; declaration. `containing-decl-kwd' is the keyword
9976 ;; symbol of the keyword that tells what kind of block it
9977 ;; is.
9978 containing-decl-open
9979 containing-decl-start
9980 containing-decl-kwd
9981 ;; The open paren of the closest surrounding sexp or nil if
9982 ;; there is none.
9983 containing-sexp
9984 ;; The position after the closest preceding brace sexp
9985 ;; (nested sexps are ignored), or the position after
9986 ;; `containing-sexp' if there is none, or (point-min) if
9987 ;; `containing-sexp' is nil.
9988 lim
9989 ;; The paren state outside `containing-sexp', or at
9990 ;; `indent-point' if `containing-sexp' is nil.
9992 ;; There's always at most one syntactic element which got
9993 ;; an anchor pos. It's stored in syntactic-relpos.
9994 syntactic-relpos
9997 ;; Check if we're directly inside an enclosing declaration
9998 ;; level block.
10007 containing-sexp)
10008 t)))
10011 containing-sexp nil)
10016 ;; Init some position variables.
10029 ;; Ignore balanced paren. The next entry
10030 ;; can't be another one.
10033 ;; If there is no surrounding open paren then
10034 ;; put the last balanced pair back on paren-state.
10036 containing-sexp nil)))
10040 ;; If we're in a parenthesis list then ',' delimits the
10041 ;; "statements" rather than being an operator (with the
10042 ;; exception of the "for" clause). This difference is
10043 ;; typically only noticeable when statements are used in macro
10044 ;; arglists.
10048 ;; cache char before and after indent point, and move point to
10049 ;; the most likely position to perform the majority of tests
10058 ;; are we in a literal?
10061 ;; now figure out syntactic qualities of the current line
10064 ;; CASE 1: in a string.
10068 ;; CASE 2: in a C or C++ style comment.
10070 ;; This is a kludge for XEmacs where we use
10071 ;; `buffer-syntactic-context', which doesn't correctly
10072 ;; recognize "\*/" to end a block comment.
10073 ;; `parse-partial-sexp' which is used by
10074 ;; `c-literal-limits' will however do that in most
10075 ;; versions, which results in that we get nil from
10076 ;; `c-literal-limits' even when `c-in-literal' claims
10077 ;; we're inside a comment.
10081 ;; CASE 3: in a cpp preprocessor macro continuation.
10091 ;; If at the beginning of the body of a #define
10092 ;; directive then analyze as cpp-define-intro
10093 ;; only. Go on with the syntactic analysis
10094 ;; otherwise. in-macro-expr is set if we're in a
10095 ;; cpp expression, i.e. before the #define body
10096 ;; or anywhere in a non-#define directive.
10103 nil)))))
10107 ;; CASE 11: an else clause?
10111 containing-sexp paren-state))
10113 ;; CASE 12: while closure of a do/while construct?
10121 containing-sexp paren-state))
10123 ;; CASE 13: A catch or finally clause? This case is simpler
10124 ;; than if-else and do-while, because a block is required
10125 ;; after every try, catch and finally.
10133 t)
10137 t)
10140 t))
10145 containing-sexp paren-state))
10147 ;; CASE 18: A substatement we can recognize by keyword.
10157 ;; Ought to cache the result from the
10158 ;; c-beginning-of-statement-1 calls here.
10168 ;; Step up to the containing statement if we
10169 ;; stayed in the same one.
10177 ;; There was no containing statement after all.
10179 placeholder))
10181 ;; Require a parenthesis after these keywords.
10182 ;; Necessary to catch e.g. synchronized in Java,
10183 ;; which can be used both as statement and
10184 ;; modifier.
10190 ;; CASE 18A: Simple substatement.
10196 containing-sexp paren-state))
10202 containing-sexp paren-state))
10205 containing-sexp paren-state))))
10207 ;; CASE 18B: Some other substatement. This is shared
10208 ;; with case 10.
10210 char-after-ip
10211 placeholder
10212 lim
10213 paren-state)))
10215 ;; CASE 14: A case or default label
10220 containing-sexp
10233 containing-sexp))
10236 ;; Got a bogus label at the top level. In lack of better
10237 ;; alternatives, anchor it on (point-min).
10240 ;; CASE 15: any other label
10248 containing-decl-open
10249 containing-decl-start
10250 containing-decl-kwd
10251 paren-state))
10252 ;; Append access-label with the same anchor point as
10253 ;; inclass gets.
10259 containing-sexp))
10264 ;; If the surrounding statement is a switch then
10265 ;; let's analyze all labels as switch labels, so
10266 ;; that they get lined up consistently.
10267 'case-label
10273 ;; A label on the top level. Treat it as a class
10274 ;; context. (point-min) is the closest we get to the
10275 ;; class open brace.
10278 ;; CASE 4: In-expression statement. C.f. cases 7B, 16A and
10279 ;; 17E.
10282 containing-sexp
10283 ;; Have to turn on the heuristics after
10284 ;; the point even though it doesn't work
10285 ;; very well. C.f. test case class-16.pike.
10286 t))
10291 ;; It's a statement block or an anonymous class.
10293 ;; It's a Pike lambda. Check whether we are between the
10294 ;; lambda keyword and the argument list or at the defun
10295 ;; opener.
10297 'inline-open
10303 paren-state)
10307 ;; CASE 5: Line is inside a declaration level block or at top level.
10311 ;; CASE 5A: we are looking at a defun, brace list, class,
10312 ;; or inline-inclass method opening brace
10319 ;; CASE 5A.1: Non-class declaration block open.
10334 ;; Special case for extern-lang-open.
10335 'extern-lang-open
10337 nil t containing-sexp paren-state))
10339 ;; CASE 5A.2: we are looking at a class opening brace
10348 ;; CASE 5A.3: brace list open
10376 ))))
10380 ;; We're in Java and have found that the open brace
10381 ;; belongs to a "new Foo[]" initialization list,
10382 ;; which means the brace list is part of an
10383 ;; expression and not a top level definition. We
10384 ;; therefore treat it as any topmost continuation
10385 ;; even though the semantically correct symbol still
10386 ;; is brace-list-open, on the same grounds as in
10387 ;; case B.2.
10393 ;; CASE 5A.4: inline defun open
10399 containing-decl-open
10400 containing-decl-start
10401 containing-decl-kwd
10402 paren-state))
10404 ;; CASE 5A.5: ordinary defun open
10415 ;; Bogus to use bol here, but it's the legacy. (Resolved,
10416 ;; 2007-11-09)
10417 ))))
10419 ;; CASE 5R: Member init list. (Used to be part of CASE 5B.1)
10420 ;; Note there is no limit on the backward search here, since member
10421 ;; init lists can, in practice, be very large.
10428 ;; There is no preceding member init clause.
10429 ;; Indent relative to the beginning of indentation
10430 ;; for the topmost-intro line that contains the
10431 ;; prototype's open paren.
10434 ;; Indent relative to the first member init clause.
10439 ;; CASE 5B: After a function header but before the body (or
10440 ;; the ending semicolon if there's no body).
10447 ;; CASE 5B.1: Member init list.
10449 ;; There is no preceding member init clause.
10450 ;; Indent relative to the beginning of indentation
10451 ;; for the topmost-intro line that contains the
10452 ;; prototype's open paren.
10456 ;; CASE 5B.2: K&R arg decl intro
10463 containing-decl-open
10464 containing-decl-start
10465 containing-decl-kwd
10466 paren-state)))
10468 ;; CASE 5B.4: Nether region after a C++ or Java func
10469 ;; decl, which could include a `throws' declaration.
10473 )))
10475 ;; CASE 5C: inheritance line. could be first inheritance
10476 ;; line, or continuation of a multiple inheritance
10484 ;; watch out for scope operator
10489 )))
10498 ;; for Java
10503 cont done)
10508 done t))
10512 )
10514 injava-inher)
10517 ))
10520 ;; CASE 5C.1: non-hanging colon on an inher intro
10524 ;; don't add inclass symbol since relative point already
10525 ;; contains any class offset
10526 )
10528 ;; CASE 5C.2: hanging colon on an inher intro
10534 containing-decl-open
10535 containing-decl-start
10536 containing-decl-kwd
10537 paren-state)))
10539 ;; CASE 5C.3: in a Java implements/extends
10553 )))
10555 ;; CASE 5C.4: a continued inheritance line
10559 ;; don't add inclass symbol since relative point already
10560 ;; contains any class offset
10561 )))
10563 ;; CASE 5P: AWK pattern or function or continuation
10564 ;; thereof.
10570 'topmost-intro-cont
10572 nil nil
10573 containing-sexp paren-state))
10575 ;; CASE 5D: this could be a top-level initialization, a
10576 ;; member init list continuation, or a template argument
10577 ;; list continuation.
10579 ;; Note: We use the fact that lim is always after any
10580 ;; preceding brace sexp.
10590 t))
10595 t)
10601 nil)
10602 t))
10604 ;; NB: No c-after-special-operator-id stuff in this
10605 ;; clause - we assume only C++ needs it.
10610 ;; CASE 5D.3: perhaps a template list continuation?
10625 ;; In a nested template arglist.
10631 ;; FIXME: Should use c-add-stmt-syntax, but it's not yet
10632 ;; template aware.
10635 ;; CASE 5D.4: perhaps a multiple inheritance line?
10647 t)
10652 ;; CASE 5D.5: Continuation of the "expression part" of a
10653 ;; top level construct. Or, perhaps, an unrecognized construct.
10665 ;; A preceding comma at the top level means that a
10666 ;; new variable declaration starts here. Use
10667 ;; topmost-intro-cont for it, for consistency with
10668 ;; the first variable declaration. C.f. case 5N.
10671 nil nil containing-sexp paren-state))
10672 ))
10674 ;; CASE 5F: Close of a non-class declaration level block.
10678 ;; This is inconsistent: Should use `containing-decl-open'
10679 ;; here if it's at boi, like in case 5J.
10683 ;; Special case for compatibility with the
10684 ;; extern-lang syntactic symbols.
10685 'extern-lang-close
10688 nil t
10690 paren-state))
10692 ;; CASE 5G: we are looking at the brace which closes the
10693 ;; enclosing nested class decl
10698 containing-decl-open
10699 containing-decl-start
10700 containing-decl-kwd
10701 paren-state))
10703 ;; CASE 5H: we could be looking at subsequent knr-argdecls
10710 ;; Do an extra check to avoid tripping up on
10711 ;; statements that occur in invalid contexts
10712 ;; (e.g. in macro bodies where we don't really
10713 ;; know the context of what we're looking at).
10720 ;; CASE 5I: ObjC method definition.
10725 ;; Handle the case when it's the first (non-comment)
10726 ;; thing in the buffer. Can't look for a 'same return
10727 ;; value from cbos1 since ObjC directives currently
10728 ;; aren't recognized fully, so that we get 'same
10729 ;; instead of 'previous if it moved over a preceding
10730 ;; directive.
10734 ;; CASE 5N: At a variable declaration that follows a class
10735 ;; definition or some other block declaration that doesn't
10736 ;; end at the closing '}'. C.f. case 5D.5.
10745 ;; Speed up the backward search a bit.
10750 ;; The '}' is unbalanced.
10751 nil
10756 containing-sexp paren-state))
10758 ;; NOTE: The point is at the end of the previous token here.
10760 ;; CASE 5J: we are at the topmost level, make
10761 ;; sure we skip back past any access specifiers
10763 ;; A macro continuation line is never at top level.
10785 nil)))))
10786 ;; For historic reasons we anchor at bol of the last
10787 ;; line of the previous declaration. That's clearly
10788 ;; highly bogus and useless, and it makes our lives hard
10789 ;; to remain compatible. :P
10800 ;; Special case for compatibility with the
10801 ;; extern-lang syntactic symbols.
10802 'inextern-lang
10805 nil t
10807 paren-state))
10809 containing-decl-open
10810 containing-decl-start
10811 containing-decl-kwd
10812 paren-state)))
10814 macro-start
10819 ;; CASE 5K: we are at an ObjC method definition
10820 ;; continuation line.
10829 ;; CASE 5L: we are at the first argument of a template
10830 ;; arglist that begins on the previous line.
10838 ;; CASE 5Q: we are at a statement within a macro.
10843 ;;CASE 5N: We are at a topmost continuation line and the only
10844 ;;preceding items are annotations.
10851 t)
10857 ;; CASE 5M: we are at a topmost continuation line
10868 ))
10870 ;; (CASE 6 has been removed.)
10872 ;; CASE 7: line is an expression, not a statement. Most
10873 ;; likely we are either in a function prototype or a function
10874 ;; call argument list
10882 ;; CASE 7A: we are looking at the arglist closing paren.
10883 ;; C.f. case 7F.
10895 paren-state))
10897 ;; CASE 7B: Looking at the opening brace of an
10898 ;; in-expression block or brace list. C.f. cases 4, 16A
10899 ;; and 17E.
10903 paren-state))
10907 placeholder
10910 paren-state)
10911 containing-sexp)))
10912 ;; placeholder is nil if it's a block directly in
10913 ;; a function arglist. That makes us skip out of
10914 ;; this case.
10915 )))
10920 paren-state)
10924 ;; CASE 7C: we are looking at the first argument in an empty
10925 ;; argument list. Use arglist-close if we're actually
10926 ;; looking at a close paren or bracket.
10938 paren-state))
10940 ;; CASE 7D: we are inside a conditional test clause. treat
10941 ;; these things as statements
10951 ))
10953 ;; CASE 7E: maybe a continued ObjC method call. This is the
10954 ;; case when we are inside a [] bracketed exp, and what
10955 ;; precede the opening bracket is not an identifier.
10963 )))
10965 ;; CASE 7F: we are looking at an arglist continuation line,
10966 ;; but the preceding argument is on the same line as the
10967 ;; opening paren. This case includes multi-line
10968 ;; mathematical paren groupings, but we could be on a
10969 ;; for-list continuation line. C.f. case 7A.
10986 paren-state))
10988 ;; CASE 7G: we are looking at just a normal arglist
10989 ;; continuation line
10992 ))
10994 ;; CASE 8: func-local multi-inheritance line
11004 ;; CASE 8A: non-hanging colon on an inher intro
11009 ;; CASE 8B: hanging colon on an inher intro
11013 ;; CASE 8C: a continued inheritance line
11017 )))
11019 ;; CASE 9: we are inside a brace-list
11029 ;; CASE 9A: In the middle of a special brace list opener.
11052 ;; CASE 9B: brace-list-close brace
11054 ;; Check special brace list closer.
11061 ;; We were between the special close char and the `)'.
11064 ;; We were before the special close char.
11068 ;; Normal brace list check.
11079 ;; Prepare for the rest of the cases below by going to the
11080 ;; token following the opening brace
11093 ;; CASE 9C: we're looking at the first line in a brace-list
11104 ;; CASE 9D: this is just a later brace-list-entry or
11105 ;; brace-entry-open
11114 ))
11115 ))))
11117 ;; CASE 10: A continued statement or top level construct.
11127 ;; This is shared with case 18.
11129 char-after-ip
11130 placeholder
11131 containing-sexp
11132 paren-state))
11134 ;; CASE 16: block close brace, possibly closing the defun or
11135 ;; the class
11137 ;; From here on we have the next containing sexp in lim.
11142 ;; CASE 16E: Closing a statement block? This catches
11143 ;; cases where it's preceded by a statement keyword,
11144 ;; which works even when used in an "invalid" context,
11145 ;; e.g. a macro argument.
11150 ;; CASE 16A: closing a lambda defun or an in-expression
11151 ;; block? C.f. cases 4, 7B and 17E.
11154 nil))
11156 'inline-close
11166 paren-state)
11170 ;; CASE 16B: does this close an inline or a function in
11171 ;; a non-class declaration level block?
11178 nil))
11188 ;; CASE 16F: Can be a defun-close of a function declared
11189 ;; in a statement block, e.g. in Pike or when using gcc
11190 ;; extensions, but watch out for macros followed by
11191 ;; blocks. Let it through to be handled below.
11192 ;; C.f. cases B.3 and 17G.
11198 ;; Turn off recognition of constructs that
11199 ;; lacks a type in this case, since that's more
11200 ;; likely to be a macro followed by a block.
11207 ;; CASE 16C: If there is an enclosing brace then this is
11208 ;; a block close since defun closes inside declaration
11209 ;; level blocks have been handled above.
11211 ;; If the block is preceded by a case/switch label on
11212 ;; the same line, we anchor at the first preceding label
11213 ;; at boi. The default handling in c-add-stmt-syntax
11214 ;; really fixes it better, but we do like this to keep
11215 ;; the indentation compatible with version 5.28 and
11216 ;; earlier. C.f. case 17H.
11223 ;; c-backward-to-block-anchor not necessary here; those
11224 ;; situations are handled in case 16E above.
11227 ;; CASE 16D: Only top level defun close left.
11233 paren-state))
11234 ))
11236 ;; CASE 19: line is an expression, not a statement, and is directly
11237 ;; contained by a template delimiter. Most likely, we are in a
11238 ;; template arglist within a statement. This case is based on CASE
11239 ;; 7. At some point in the future, we may wish to create more
11240 ;; syntactic symbols such as `template-intro',
11241 ;; `template-cont-nonempty', etc., and distinguish between them as we
11242 ;; do for `arglist-intro' etc. (2009-12-07).
11248 ; '<') before indent-point.
11256 paren-state))
11258 ;; CASE 17: Statement or defun catchall.
11261 ;; Back up statements until we find one that starts at boi.
11264 containing-sexp)))
11268 nil)
11273 ;; CASE 17B: continued statement
11277 containing-sexp paren-state))
11279 ;; CASE 17A: After a case/default label?
11287 'statement-case-open
11289 nil t containing-sexp paren-state))
11291 ;; CASE 17D: any old statement
11298 containing-sexp paren-state)
11302 ;; CASE 17I: Inside a substatement block.
11304 ;; The following tests are all based on containing-sexp.
11306 ;; From here on we have the next containing sexp in lim.
11311 lim paren-state)
11315 ;; CASE 17E: first statement in an in-expression block.
11316 ;; C.f. cases 4, 7B and 16A.
11319 nil))
11321 'defun-block-intro
11330 paren-state)
11336 ;; CASE 17F: first statement in an inline, or first
11337 ;; statement in a top-level defun. we can tell this is it
11338 ;; if there are no enclosing braces that haven't been
11339 ;; narrowed out by a class (i.e. don't use bod here).
11342 paren-state)))
11348 nil))))
11353 ;; CASE 17G: First statement in a function declared inside
11354 ;; a normal block. This can occur in Pike and with
11355 ;; e.g. the gcc extensions, but watch out for macros
11356 ;; followed by blocks. C.f. cases B.3 and 16F.
11362 ;; Turn off recognition of constructs that lacks
11363 ;; a type in this case, since that's more likely
11364 ;; to be a macro followed by a block.
11370 lim paren-state))
11372 ;; CASE 17H: First statement in a block.
11374 ;; If the block is preceded by a case/switch label on the
11375 ;; same line, we anchor at the first preceding label at
11376 ;; boi. The default handling in c-add-stmt-syntax is
11377 ;; really fixes it better, but we do like this to keep the
11378 ;; indentation compatible with version 5.28 and earlier.
11379 ;; C.f. case 16C.
11386 ;; c-backward-to-block-anchor not necessary here; those
11387 ;; situations are handled in case 17I above.
11389 lim paren-state))
11392 ))
11393 )
11395 ;; now we need to look at any modifiers
11399 ;; are we looking at a comment only line?
11404 ;; we might want to give additional offset to friends (in C++).
11409 ;; Set syntactic-relpos.
11415 nil)
11416 t))
11419 ;; Start of or a continuation of a preprocessor directive?
11434 ;; If inside a cpp expression, i.e. anywhere in a
11435 ;; cpp directive except a #define body, we only let
11436 ;; through the syntactic analysis that is internal
11437 ;; in the expression. That means the arglist
11438 ;; elements, if they are anchored inside the cpp
11439 ;; expression.
11448 ;; Inside a #define body and the syntactic analysis is
11449 ;; anchored on the start of the #define. In this case
11450 ;; we add cpp-define-intro to get the extra
11451 ;; indentation of the #define body.
11454 ;; return the syntax
11455 c-syntactic-context)))
11457 \f
11458 ;; Indentation calculation.
11461 ;; offset can be a number, a function, a variable, a list, or one of
11462 ;; the symbols + or -
11463 ;;
11464 ;; This function might do hidden buffer changes.
11483 langelem symbol))
11489 offset symbol)
11490 nil)
11504 Error evaluating offset %S for %s: \
11505 Cannot combine absolute offset %S with relative %S in `%s' method"
11511 Error evaluating offset %S for %s: \
11512 Cannot combine relative offset %S with absolute %S in `%s' method"
11517 res))
11535 Error evaluating offset %S for %s: \
11536 Cannot combine absolute offsets %S and %S in `add' method"
11540 res))
11549 res))))
11556 nil))))
11560 res
11562 offset symbol res)
11563 nil)))
11566 ;; Get offset from LANGELEM which is a list beginning with the
11567 ;; syntactic symbol and followed by any analysis data it provides.
11568 ;; That data may be zero or more elements, but if at least one is
11569 ;; given then the first is the anchor position (or nil). The symbol
11570 ;; is matched against `c-offsets-alist' and the offset calculated
11571 ;; from that is returned.
11572 ;;
11573 ;; This function might do hidden buffer changes.
11583 offset
11587 ))
11590 ;; This is a compatibility wrapper for `c-calc-offset' in case
11591 ;; someone is calling it directly. It takes an old style syntactic
11592 ;; element on the form (SYMBOL . ANCHOR-POS) and converts it to the
11593 ;; new list form.
11594 ;;
11595 ;; This function might do hidden buffer changes.
11602 ;; Calculate the syntactic indentation from a syntactic description
11603 ;; as returned by `c-guess-syntax'.
11604 ;;
11605 ;; Note that topmost-intro always has an anchor position at bol, for
11606 ;; historical reasons. It's often used together with other symbols
11607 ;; that has more sane positions. Since we always use the first
11608 ;; found anchor position, we rely on that these other symbols always
11609 ;; precede topmost-intro in the LANGELEMS list.
11610 ;;
11611 ;; This function might do hidden buffer changes.
11619 ;; Got an absolute column that overrides any indentation
11620 ;; we've collected so far, but not the relative
11621 ;; indentation we might get for the nested structures
11622 ;; further down the langelems list.
11626 ;; Got a relative change of the current calculated
11627 ;; indentation.
11630 ;; Use the anchor position from the first syntactic
11631 ;; element with one.
11641 indent)))
11643 \f
11646 ;; Local Variables:
11647 ;; indent-tabs-mode: t
11648 ;; tab-width: 8
11649 ;; End:
11650 ;;; cc-engine.el ends here