| blob | 9a6e975dd93158ff77eae58ea870cee2054ce72a |
1 ;;; cc-engine.el --- core syntax guessing engine for CC mode -*- coding: utf-8 -*-
3 ;; Copyright (C) 1985, 1987, 1992-2014 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.
628 noerror comma-delim)
629 "Move to the start of the current statement or declaration, or to
630 the previous one if already at the beginning of one. Only
631 statements/declarations on the same level are considered, i.e. don't
632 move into or out of sexps (not even normal expression parentheses).
634 If point is already at the earliest statement within braces or parens,
635 this function doesn't move back into any whitespace preceding it; it
636 returns 'same in this case.
640 is within the continuation. If starting at such a token, move to the
641 corresponding statement start. If at the beginning of a statement,
642 move to the closest containing statement if there is any. This might
643 also stop at a continuation clause.
645 Labels are treated as part of the following statements if
646 IGNORE-LABELS is non-nil. (FIXME: Doesn't work if we stop at a known
647 statement start keyword.) Otherwise, each label is treated as a
648 separate statement.
650 Macros are ignored \(i.e. skipped over) unless point is within one, in
651 which case the content of the macro is treated as normal code. Aside
652 from any normal statement starts found in it, stop at the first token
655 macros before leaving them.
657 Return:
659 'same if stopped at the beginning of the current statement;
660 'up if stepped to a containing statement;
661 'previous if stepped to a preceding statement;
662 'beginning if stepped from a statement continuation clause to
663 its start clause; or
664 'macro if stepped to a macro start.
665 Note that 'same and not 'label is returned if stopped at the same
666 label without crossing the colon character.
668 LIM may be given to limit the search. If the search hits the limit,
669 point will be left at the closest following token, or at the start
670 position if that is less ('same is returned in this case).
672 NOERROR turns off error logging to `c-parsing-error'.
674 Normally only ';' and virtual semicolons are considered to delimit
675 statements, but if COMMA-DELIM is non-nil then ',' is treated
676 as a delimiter too.
678 Note that this function might do hidden buffer changes. See the
679 comment at the start of cc-engine.el for more info."
681 ;; The bulk of this function is a pushdown automaton that looks at statement
682 ;; boundaries and the tokens (such as "while") in c-opt-block-stmt-key. Its
683 ;; purpose is to keep track of nested statements, ensuring that such
684 ;; statements are skipped over in their entirety (somewhat akin to what C-M-p
685 ;; does with nested braces/brackets/parentheses).
686 ;;
687 ;; Note: The position of a boundary is the following token.
688 ;;
689 ;; Beginning with the current token (the one following point), move back one
690 ;; sexp at a time (where a sexp is, more or less, either a token or the
691 ;; entire contents of a brace/bracket/paren pair). Each time a statement
692 ;; boundary is crossed or a "while"-like token is found, update the state of
693 ;; the PDA. Stop at the beginning of a statement when the stack (holding
694 ;; nested statement info) is empty and the position has been moved.
695 ;;
696 ;; The following variables constitute the PDA:
697 ;;
698 ;; sym: This is either the "while"-like token (e.g. 'for) we've just
699 ;; scanned back over, 'boundary if we've just gone back over a
700 ;; statement boundary, or nil otherwise.
701 ;; state: takes one of the values (nil else else-boundary while
702 ;; while-boundary catch catch-boundary).
703 ;; nil means "no "while"-like token yet scanned".
704 ;; 'else, for example, means "just gone back over an else".
705 ;; 'else-boundary means "just gone back over a statement boundary
706 ;; immediately after having gone back over an else".
707 ;; saved-pos: A vector of either saved positions (tok ptok pptok, etc.) or
708 ;; of error reporting information.
709 ;; stack: The stack onto which the PDA pushes its state. Each entry
710 ;; consists of a saved value of state and saved-pos. An entry is
711 ;; pushed when we move back over a "continuation" token (e.g. else)
712 ;; and popped when we encounter the corresponding opening token
713 ;; (e.g. if).
714 ;;
715 ;;
716 ;; The following diagram briefly outlines the PDA.
717 ;;
718 ;; Common state:
719 ;; "else": Push state, goto state `else'.
720 ;; "while": Push state, goto state `while'.
721 ;; "catch" or "finally": Push state, goto state `catch'.
722 ;; boundary: Pop state.
723 ;; other: Do nothing special.
724 ;;
725 ;; State `else':
726 ;; boundary: Goto state `else-boundary'.
727 ;; other: Error, pop state, retry token.
728 ;;
729 ;; State `else-boundary':
730 ;; "if": Pop state.
731 ;; boundary: Error, pop state.
732 ;; other: See common state.
733 ;;
734 ;; State `while':
735 ;; boundary: Save position, goto state `while-boundary'.
736 ;; other: Pop state, retry token.
737 ;;
738 ;; State `while-boundary':
739 ;; "do": Pop state.
740 ;; boundary: Restore position if it's not at start, pop state. [*see below]
741 ;; other: See common state.
742 ;;
743 ;; State `catch':
744 ;; boundary: Goto state `catch-boundary'.
745 ;; other: Error, pop state, retry token.
746 ;;
747 ;; State `catch-boundary':
748 ;; "try": Pop state.
749 ;; "catch": Goto state `catch'.
750 ;; boundary: Error, pop state.
751 ;; other: See common state.
752 ;;
753 ;; [*] In the `while-boundary' state, we had pushed a 'while state, and were
754 ;; searching for a "do" which would have opened a do-while. If we didn't
755 ;; find it, we discard the analysis done since the "while", go back to this
756 ;; token in the buffer and restart the scanning there, this time WITHOUT
757 ;; pushing the 'while state onto the stack.
758 ;;
759 ;; In addition to the above there is some special handling of labels
760 ;; and macros.
764 macro-start
767 c-stmt-delim-chars-with-comma
768 c-stmt-delim-chars))
769 c-in-literal-cache c-maybe-labelp after-case:-pos saved
770 ;; Current position.
771 pos
772 ;; Position of last stmt boundary character (e.g. ;).
773 boundary-pos
774 ;; The position of the last sexp or bound that follows the
775 ;; first found colon, i.e. the start of the nonlabel part of
776 ;; the statement. It's `start' if a colon is found just after
777 ;; the start.
778 after-labels-pos
779 ;; Like `after-labels-pos', but the first such position inside
780 ;; a label, i.e. the start of the last label before the start
781 ;; of the nonlabel part of the statement.
782 last-label-pos
783 ;; The last position where a label is possible provided the
784 ;; statement started there. It's nil as long as no invalid
785 ;; label content has been found (according to
786 ;; `c-nonlabel-token-key'). It's `start' if no valid label
787 ;; content was found in the label. Note that we might still
788 ;; regard it a label if it starts with `c-label-kwds'.
789 label-good-pos
790 ;; Putative positions of the components of a bitfield declaration,
791 ;; e.g. "int foo : NUM_FOO_BITS ;"
792 bitfield-type-pos bitfield-id-pos bitfield-size-pos
793 ;; Symbol just scanned back over (e.g. 'while or 'boundary).
794 ;; See above.
795 sym
796 ;; Current state in the automaton. See above.
797 state
798 ;; Current saved positions. See above.
799 saved-pos
800 ;; Stack of conses (state . saved-pos).
801 stack
802 ;; Regexp which matches "for", "if", etc.
805 ;; Return value.
807 ;; Positions of the last three sexps or bounds we've stopped at.
808 tok ptok pptok)
818 ;; Try to skip back over unary operator characters, to register
819 ;; that we've moved.
823 ;; Protect post-++/-- operators just before a virtual semicolon.
827 ;; Skip back over any semicolon here. If it was a bare semicolon, we're
828 ;; done. Later on we ignore the boundaries for statements that don't
829 ;; contain any sexp. The only thing that is affected is that the error
830 ;; checking is a little less strict, and we really don't bother.
840 pos saved)
842 ;; Begin at start and not pos to detect macros if we stand
843 ;; directly after the #.
846 ;; Record this as the first token if not starting inside it.
849 ;; The following while loop goes back one sexp (balanced parens,
850 ;; etc. with contents, or symbol or suchlike) each iteration. This
851 ;; movement is accomplished with a call to c-backward-sexp approx 170
852 ;; lines below.
853 ;;
854 ;; The loop is exited only by throwing nil to the (catch 'loop ...):
855 ;; 1. On reaching the start of a macro;
856 ;; 2. On having passed a stmt boundary with the PDA stack empty;
857 ;; 3. On reaching the start of an Objective C method def;
858 ;; 4. From macro `c-bos-pop-state'; when the stack is empty;
859 ;; 5. From macro `c-bos-pop-state-and-retry' when the stack is empty.
863 ;; Are we in a macro, just after the opening #?
878 ;; Stop at the first token in the content of the macro.
882 ret 'macro
883 ignore-labels t))
886 ;; Do a round through the automaton if we've just passed a
887 ;; statement boundary or passed a "while"-like token.
895 ;; The PDA state handling.
896 ;;
897 ;; Refer to the description of the PDA in the opening
898 ;; comments. In the following OR form, the first leaf
899 ;; attempts to handles one of the specific actions detailed
900 ;; (e.g., finding token "if" whilst in state `else-boundary').
901 ;; We drop through to the second leaf (which handles common
902 ;; state) if no specific handler is found in the first cond.
903 ;; If a parsing error is detected (e.g. an "else" with no
904 ;; preceding "if"), we throw to the enclosing catch.
905 ;;
906 ;; Note that the (eq state 'else) means
907 ;; "we've just passed an else", NOT "we're looking for an
908 ;; else".
925 ;; Since this can cause backtracking we do a
926 ;; little more careful analysis to avoid it:
927 ;; If there's a label in front of the while
928 ;; it can't be part of a do-while.
957 ;; This is state common. We get here when the previous
958 ;; cond statement found no particular state handler.
960 ;; If we have a boundary at the start
961 ;; position we push a frame to go to the
962 ;; previous statement.
971 ;; Is this a real while, or a do-while?
972 ;; The next `when' triggers unless we are SURE that
973 ;; the `while' is not the tail end of a `do-while'.
976 ;; The following kludge is to prevent
977 ;; infinite recursion when called from
978 ;; c-awk-after-if-for-while-condition-p,
979 ;; or the like.
983 ;; Since this can cause backtracking we do a
984 ;; little more careful analysis to avoid it: If
985 ;; the while isn't followed by a (possibly
986 ;; virtual) semicolon it can't be a do-while.
995 ;; We're either past a statement boundary or at the
996 ;; start of a statement, so throw away any label data
997 ;; for the previous one.
999 last-label-pos nil
1000 c-maybe-labelp nil))))
1002 ;; Step to the previous sexp, but not if we crossed a
1003 ;; boundary, since that doesn't consume an sexp.
1007 ;; HERE IS THE SINGLE PLACE INSIDE THE PDA LOOP WHERE WE MOVE
1008 ;; BACKWARDS THROUGH THE SOURCE.
1012 ;; The end position of the area to search for statement
1013 ;; barriers in this round.
1016 ;; Go back over exactly one logical sexp, taking proper
1017 ;; account of macros and escaped EOLs.
1021 ;; Give up if we hit an unbalanced block. Since the
1022 ;; stack won't be empty the code below will report a
1023 ;; suitable error.
1026 ;; Have we moved into a macro?
1029 ;; Have we crossed a statement boundary? If not,
1030 ;; keep going back until we find one or a "real" sexp.
1037 ;; Have we just gone back over an escaped NL? This
1038 ;; doesn't count as a sexp.
1041 ;; Have we crossed a statement boundary?
1044 ;; Are we at a macro beginning?
1046 c-opt-cpp-prefix
1052 ;; Just gone back over a brace block?
1061 ;; Just gone back over some paren block?
1065 before-sws-pos)))
1068 ;; Just gone back over an ordinary symbol of some sort?
1074 ptok tok
1075 tok boundary-pos
1077 ;; Like a C "continue". Analyze the next sexp.
1080 ;; ObjC method def?
1087 ;; Might we have a bitfield declaration, "<type> <id> : <size>"?
1090 ;; The : <size> and <id> fields?
1100 ;; The <type> field?
1108 bitfield-id-pos nil)))))
1110 ;; Handle labels.
1113 ;; `c-crosses-statement-barrier-p' has found a colon, so we
1114 ;; might be in a label now. Have we got a real label
1115 ;; (including a case label) or something like C++'s "public:"?
1116 ;; A case label might use an expression rather than a token.
1119 ;; Catch C++'s inheritance construct "class foo : bar".
1133 ; been found.
1135 ;; We're in a potential label and it's the first
1136 ;; time we've found something that isn't allowed in
1137 ;; one.
1140 ;; We've moved back by a sexp, so update the token positions.
1142 pptok ptok
1143 ptok tok
1149 ;; If the stack isn't empty there might be errors to report.
1158 ;; Need to investigate closer whether we've crossed
1159 ;; between a substatement and its containing statement.
1163 pptok)
1165 ptok)
1173 after-labels-pos
1180 last-label-pos
1181 pos))
1183 ;; We're in a label. Maybe we should step to the statement
1184 ;; after it.
1189 ;; Might have jumped over several labels. Go to the last one.
1192 ;; Have we got "case <expression>:"?
1202 ;; Skip over the unary operators that can start the statement.
1205 ;; protect AWK post-inc/decrement operators, etc.
1210 ret)))
1213 "Return non-nil if there is a non-comment non-macro punctuation character
1214 between FROM and TO. FROM must not be in a string or comment. The returned
1215 value is the position of the first such character."
1226 "Return non-nil if buffer positions FROM to TO cross one or more
1227 statement or declaration boundaries. The returned value is actually
1228 the position of the earliest boundary char. FROM must not be within
1229 a string or comment.
1231 The variable `c-maybe-labelp' is set to the position of the first `:' that
1232 might start a label (i.e. not part of `::' and not preceded by `?'). If a
1233 single `?' is found, then `c-maybe-labelp' is cleared.
1235 For AWK, a statement which is terminated by an EOL (not a \; or a }) is
1237 the line. If this virtual semicolon is _at_ from, the function recognizes it.
1239 Note that this function might do hidden buffer changes. See the
1240 comment at the start of cc-engine.el for more info."
1242 ;; If the current language has CPP macros, insert # into skip-chars.
1245 c-opt-cpp-symbol ; usually "#"
1247 c-stmt-delim-chars))
1250 lit-range vsemi-pos)
1257 skip-chars
1261 ;; Virtual semicolon?
1271 ;; In a string/comment?
1278 ;; Ignore scope operators.
1282 ;; A question mark. Can't be a label, so stop
1283 ;; looking for more : and ?.
1286 ;; At a CPP construct or a "#" or "##" operator?
1297 ;; In trailing space after an as yet undetected virtual semicolon?
1300 ; unterminated string/regexp.
1305 nil))))))
1308 "Return non-nil if the point is at the first token in a statement
1309 or somewhere in the syntactic whitespace before it.
1312 Any kind of declaration-like construct that occur outside function
1315 Note that this function might do hidden buffer changes. See the
1316 comment at the start of cc-engine.el for more info."
1320 c-maybe-labelp)
1331 "Return non-nil if the point is at the first token in an expression or
1332 statement, or somewhere in the syntactic whitespace before it.
1335 grammar sense: It's any sequence of expression tokens except commas,
1336 unless they are enclosed inside parentheses of some kind. Also, an
1337 expression never continues past an enclosing parenthesis, but it might
1338 contain parenthesis pairs of any sort except braces.
1340 Since expressions never cross statement boundaries, this function also
1341 recognizes statement beginnings, just like `c-at-statement-start-p'.
1343 Note that this function might do hidden buffer changes. See the
1344 comment at the start of cc-engine.el for more info."
1349 c-maybe-labelp)
1357 \f
1358 ;; A set of functions that covers various idiosyncrasies in
1359 ;; implementations of `forward-comment'.
1361 ;; Note: Some emacsen considers incorrectly that any line comment
1362 ;; ending with a backslash continues to the next line. I can't think
1363 ;; of any way to work around that in a reliable way without changing
1364 ;; the buffer, though. Suggestions welcome. ;) (No, temporarily
1365 ;; changing the syntax for backslash doesn't work since we must treat
1366 ;; escapes in string literals correctly.)
1369 "Move forward past whitespace and the closest following comment, if any.
1370 Return t if a comment was found, nil otherwise. In either case, the
1371 point is moved past the following whitespace. Line continuations,
1372 i.e. a backslashes followed by line breaks, are treated as whitespace.
1373 The line breaks that end line comments are considered to be the
1374 comment enders, so the point will be put on the beginning of the next
1375 line if it moved past a line comment.
1377 This function does not do any hidden buffer changes."
1385 ;; Some emacsen (e.g. XEmacs 21) return t when moving
1386 ;; forwards at eob.
1387 nil
1389 ;; Emacs includes the ending newline in a b-style (c++)
1390 ;; comment, but XEmacs doesn't. We depend on the Emacs
1391 ;; behavior (which also is symmetric).
1395 t))))
1398 "Move forward past all following whitespace and comments.
1399 Line continuations, i.e. a backslashes followed by line breaks, are
1400 treated as whitespace.
1402 Note that this function might do hidden buffer changes. See the
1403 comment at the start of cc-engine.el for more info."
1406 ;; If forward-comment in at least XEmacs 21 is given a large
1407 ;; positive value, it'll loop all the way through if it hits
1408 ;; eob.
1410 ;; Some emacsen (e.g. XEmacs 21) return t when moving
1411 ;; forwards at eob.
1416 t))))
1419 "Move backward past whitespace and the closest preceding comment, if any.
1420 Return t if a comment was found, nil otherwise. In either case, the
1421 point is moved past the preceding whitespace. Line continuations,
1422 i.e. a backslashes followed by line breaks, are treated as whitespace.
1423 The line breaks that end line comments are considered to be the
1424 comment enders, so the point cannot be at the end of the same line to
1425 move over a line comment.
1427 This function does not do any hidden buffer changes."
1430 ;; When we got newline terminated comments, forward-comment in all
1431 ;; supported emacsen so far will stop at eol of each line not
1432 ;; ending with a comment when moving backwards. This corrects for
1433 ;; that, and at the same time handles line continuations.
1441 ;; Some emacsen (e.g. Emacs 19.34) return t when moving
1442 ;; backwards at bob.
1443 nil
1445 ;; Leave point after the closest following newline if we've
1446 ;; backed up over any above, since forward-comment won't move
1447 ;; backward over a line comment if point is at the end of the
1448 ;; same line.
1453 ;; If forward-comment above succeeded and we're at eol
1454 ;; then the newline we moved over above didn't end a
1455 ;; 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 doesn'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.
1722 ;; Find the last rung position to avoid setting properties in all
1723 ;; the cases when the marked rung is complete.
1724 ;; (`next-single-property-change' is certain to move at least one
1725 ;; step forward.)
1728 ;; Got no marked rung here. Since the simple ws might have started
1729 ;; inside a line comment or cpp directive we must set `rung-pos' as
1730 ;; high as possible.
1740 ;; The following search is the main reason that `c-in-sws'
1741 ;; and `c-is-sws' aren't combined to one property.
1745 ;; If the `c-in-sws' region extended past the last
1746 ;; `c-is-sws' char we have to go back a bit.
1753 "c-forward-sws cached move %s -> %s (max %s)"
1760 ;; We'll loop here if there is simple ws after the last rung.
1761 ;; That means that there's been some change in it and it's
1762 ;; possible that we've stepped into another ladder, so extend
1763 ;; the previous one to join with it if there is one, and try to
1764 ;; use the cache again.
1766 "c-forward-sws extending rung with [%s..%s] (max %s)"
1769 ;; Remove any `c-in-sws' property from the last char of
1770 ;; the rung before we mark it with `c-is-sws', so that we
1771 ;; won't connect with the remains of a broken "ladder".
1777 last-put-in-sws-pos rung-pos)))
1784 ;; Skipped over comments. Don't cache at eob in case the buffer
1785 ;; is narrowed.
1796 ;; Skip a preprocessor directive.
1803 ;; Don't cache at eob in case the buffer is narrowed.
1806 ;; We've searched over a piece of non-white syntactic ws. See if this
1807 ;; can be cached.
1813 ;; Cache if we haven't skipped comments only, and if we started
1814 ;; either from a marked rung or from a completely uncached
1815 ;; position.
1820 ;; See if there's a marked rung in the encountered simple ws. If
1821 ;; so then we can cache, unless `safe-start' is nil. Even then
1822 ;; we need to do this to check if the cache can be used for the
1823 ;; next step.
1827 safe-start))
1831 "c-forward-sws caching [%s..%s] - [%s..%s] (max %s)"
1835 ;; Remove the properties for any nested ws that might be cached.
1836 ;; Only necessary for `c-is-sws' since `c-in-sws' will be set
1837 ;; anyway.
1845 last-put-in-sws-pos rung-pos))
1847 ;; Remove any `c-in-sws' property from the last char of
1848 ;; the rung before we mark it with `c-is-sws', so that we
1849 ;; won't connect with the remains of a broken "ladder".
1852 rung-end-pos))
1855 "c-forward-sws not caching [%s..%s] - [%s..%s] (max %s)"
1859 ;; Set `rung-pos' for the next rung. It's the same thing here as
1860 ;; initially, except that the rung position is set as early as
1861 ;; possible since we can't be in the ending ws of a line comment or
1862 ;; cpp directive now.
1869 ;; Make sure that the newly marked `c-in-sws' region doesn't connect to
1870 ;; another one after the point (which might occur when editing inside a
1871 ;; comment or macro).
1875 "c-forward-sws clearing at %s for cache separation"
1876 last-put-in-sws-pos)
1880 ;; If at eob we have to clear the last character before the end
1881 ;; instead since the buffer might be narrowed and there might
1882 ;; be a `c-in-sws' after (point-max). In this case it's
1883 ;; necessary to clear both properties.
1885 "c-forward-sws clearing thoroughly at %s for cache separation"
1888 last-put-in-sws-pos))))
1889 ))))
1892 ;; Used by `c-backward-syntactic-ws' to implement the unbounded search.
1893 ;;
1894 ;; This function might do hidden buffer changes.
1897 ;; part of the simple ws region.
1899 rung-is-marked simple-ws-beg cmt-skip-pos)
1901 ;; Skip simple horizontal ws and do a quick check on the preceding
1902 ;; character to see if it's anything that can't end syntactic ws, so we can
1903 ;; bail out early in the majority of cases when there just are a few ws
1904 ;; chars. Newlines are complicated in the backward direction, so we can't
1905 ;; skip over them.
1912 ;; Try to find a rung position in the simple ws preceding point, so that
1913 ;; we can get a cache hit even if the last bit of the simple ws has
1914 ;; changed recently.
1920 ;; `rung-pos' will be the earliest marked position, which means that
1921 ;; there might be later unmarked parts in the simple ws region.
1922 ;; It's not worth the effort to fix that; the last part of the
1923 ;; simple ws is also typically edited often, so it could be wasted.
1935 ;; The following search is the main reason that `c-in-sws'
1936 ;; and `c-is-sws' aren't combined to one property.
1940 ;; If the `c-in-sws' region extended past the first
1941 ;; `c-is-sws' char we have to go forward a bit.
1946 "c-backward-sws cached move %s <- %s (min %s)"
1958 t
1960 nil))
1962 ;; We'll loop here if there is simple ws before the first rung.
1963 ;; That means that there's been some change in it and it's
1964 ;; possible that we've stepped into another ladder, so extend
1965 ;; the previous one to join with it if there is one, and try to
1966 ;; use the cache again.
1968 "c-backward-sws extending rung with [%s..%s] (min %s)"
1971 ;; Remove any `c-in-sws' property from the last char of
1972 ;; the rung before we mark it with `c-is-sws', so that we
1973 ;; won't connect with the remains of a broken "ladder".
1976 rung-pos)
1980 last-put-in-sws-pos rung-pos))
1989 ;; Inside a cpp directive. See if it should be skipped over.
1992 ;; Move back over all line continuations in the region skipped
1993 ;; over by `c-backward-comments'. If we go past it then we
1994 ;; started inside the cpp directive.
2003 ;; Don't move past the cpp directive if we began inside
2004 ;; it. Note that the position at the end of the last line
2005 ;; of the macro is also considered to be within it.
2007 nil)
2009 ;; It's worthwhile to spend a little bit of effort on finding
2010 ;; the end of the macro, to get a good `simple-ws-beg'
2011 ;; position for the cache. Note that `c-backward-comments'
2012 ;; could have stepped over some comments before going into
2013 ;; the macro, and then `simple-ws-beg' must be kept on the
2014 ;; same side of those comments.
2021 ;; Might happen if comments after the macro were skipped
2022 ;; over.
2026 t)))
2031 simple-ws-beg)
2032 ;; Skipped over comments. Must put point at the end of
2033 ;; the simple ws at point since we might be after a line
2034 ;; comment or cpp directive that's been partially
2035 ;; narrowed out, and we can't risk marking the simple ws
2036 ;; at the end of it.
2038 t)))
2040 ;; We've searched over a piece of non-white syntactic ws. See if this
2041 ;; can be cached.
2046 ;; Cache if we started either from a marked rung or from a
2047 ;; completely uncached position.
2048 rung-is-marked
2051 ;; Cache if there's a marked rung in the encountered simple ws.
2059 "c-backward-sws caching [%s..%s] - [%s..%s] (min %s)"
2064 ;; Remove the properties for any nested ws that might be cached.
2065 ;; Only necessary for `c-is-sws' since `c-in-sws' will be set
2066 ;; anyway.
2071 ;; Remove any `c-in-sws' property from the last char of
2072 ;; the rung before we mark it with `c-is-sws', so that we
2073 ;; won't connect with the remains of a broken "ladder".
2076 rung-end-pos)
2079 last-put-in-sws-pos simple-ws-beg)
2080 rung-pos)
2085 "c-backward-sws not caching [%s..%s] - [%s..%s] (min %s)"
2091 ))
2093 ;; Make sure that the newly marked `c-in-sws' region doesn't connect to
2094 ;; another one before the point (which might occur when editing inside a
2095 ;; comment or macro).
2099 "c-backward-sws clearing at %s for cache separation"
2102 last-put-in-sws-pos))
2104 ;; If at bob and the buffer is narrowed, we have to clear the
2105 ;; character we're standing on instead since there might be a
2106 ;; `c-in-sws' before (point-min). In this case it's necessary
2107 ;; to clear both properties.
2109 "c-backward-sws clearing thoroughly at %s for cache separation"
2110 last-put-in-sws-pos)
2113 ))))
2115 \f
2116 ;; Other whitespace tools
2118 ;; Is the region (beg end) WS, and is there WS (or BOB/EOB) next to the
2119 ;; region? This is a "heuristic" function. .....
2120 ;;
2121 ;; The motivation for the second bit is to check whether removing this
2122 ;; region would coalesce two symbols.
2123 ;;
2124 ;; FIXME!!! This function doesn't check virtual semicolons in any way. Be
2125 ;; careful about using this function for, e.g. AWK. (2007/3/7)
2134 \f
2135 ;; A system for finding noteworthy parens before the point.
2138 ;; A maximum comfortable scanning distance, e.g. between
2139 ;; `c-state-cache-good-pos' and "HERE" (where we call c-parse-state). When
2140 ;; this distance is exceeded, we take "emergency measures", e.g. by clearing
2141 ;; the cache and starting again from point-min or a beginning of defun. This
2142 ;; value can be tuned for efficiency or set to a lower value for testing.
2146 ;; The state cache used by `c-parse-state' to cut down the amount of
2147 ;; searching. It's the result from some earlier `c-parse-state' call. See
2148 ;; `c-parse-state''s doc string for details of its structure.
2149 ;;
2150 ;; The use of the cached info is more effective if the next
2151 ;; `c-parse-state' call is on a line close by the one the cached state
2152 ;; was made at; the cache can actually slow down a little if the
2153 ;; cached state was made very far back in the buffer. The cache is
2154 ;; most effective if `c-parse-state' is used on each line while moving
2155 ;; forward.
2159 ;; This is a position where `c-state-cache' is known to be correct, or
2160 ;; nil (see below). It's a position inside one of the recorded unclosed
2161 ;; parens or the top level, but not further nested inside any literal or
2162 ;; subparen that is closed before the last recorded position.
2163 ;;
2164 ;; The exact position is chosen to try to be close to yet earlier than
2165 ;; the position where `c-state-cache' will be called next. Right now
2166 ;; the heuristic is to set it to the position after the last found
2167 ;; closing paren (of any type) before the line on which
2168 ;; `c-parse-state' was called. That is chosen primarily to work well
2169 ;; with refontification of the current line.
2170 ;;
2171 ;; 2009-07-28: When `c-state-point-min' and the last position where
2172 ;; `c-parse-state' or for which `c-invalidate-state-cache' was called, are
2173 ;; both in the same literal, there is no such "good position", and
2174 ;; c-state-cache-good-pos is then nil. This is the ONLY circumstance in which
2175 ;; it can be nil. In this case, `c-state-point-min-literal' will be non-nil.
2176 ;;
2177 ;; 2009-06-12: In a brace desert, c-state-cache-good-pos may also be in
2178 ;; the middle of the desert, as long as it is not within a brace pair
2179 ;; recorded in `c-state-cache' or a paren/bracket pair.
2181 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2182 ;; We maintain a simple cache of positions which aren't in a literal, so as to
2183 ;; speed up testing for non-literality.
2185 ;; The approximate interval between entries in `c-state-nonlit-pos-cache'.
2189 ;; A list of buffer positions which are known not to be in a literal or a cpp
2190 ;; construct. This is ordered with higher positions at the front of the list.
2191 ;; Only those which are less than `c-state-nonlit-pos-cache-limit' are valid.
2195 ;; An upper limit on valid entries in `c-state-nonlit-pos-cache'. This is
2196 ;; reduced by buffer changes, and increased by invocations of
2197 ;; `c-state-literal-at'.
2201 ;; A list of buffer positions which are known not to be in a literal. This is
2202 ;; ordered with higher positions at the front of the list. Only those which
2203 ;; are less than `c-state-semi-nonlit-pos-cache-limit' are valid.
2207 ;; An upper limit on valid entries in `c-state-semi-nonlit-pos-cache'. This is
2208 ;; reduced by buffer changes, and increased by invocations of
2209 ;; `c-state-literal-at'. FIXME!!!
2212 ;; Do a parse-partial-sexp from FROM to TO, returning either
2213 ;; (STATE TYPE (BEG . END)) if TO is in a literal; or
2214 ;; (STATE) otherwise,
2215 ;; where STATE is the parsing state at TO, TYPE is the type of the literal
2216 ;; (one of 'c, 'c++, 'string) and (BEG . END) is the boundaries of the literal.
2217 ;;
2218 ;; Unless NOT-IN-DELIMITER is non-nil, when TO is inside a two-character
2219 ;; comment opener, this is recognized as being in a comment literal.
2220 ;;
2221 ;; Only elements 3 (in a string), 4 (in a comment), 5 (following a quote),
2222 ;; 7 (comment type) and 8 (start of comment/string) (and possibly 9) of
2223 ;; STATE are valid.
2226 ty co-st)
2234 nil ; TARGETDEPTH
2235 nil ; STOPBEFORE
2236 s ; OLDSTATE
2253 ;; Return a buffer position before HERE which is "safe", i.e. outside any
2254 ;; string, comment, or macro.
2255 ;;
2256 ;; NOTE: This function manipulates `c-state-nonlit-pos-cache'. This cache
2257 ;; MAY NOT contain any positions within macros, since macros are frequently
2258 ;; turned into comments by use of the `c-cpp-delimiter' category properties.
2259 ;; We cannot rely on this mechanism whilst determining a cache pos since
2260 ;; this function is also called from outwith `c-parse-state'.
2265 pos npos high-pos lit macro-beg macro-end)
2266 ;; Trim the cache to take account of buffer changes.
2278 ;; Add an element to `c-state-nonlit-pos-cache' each iteration.
2282 ;; Test for being in a literal. If so, go to after it.
2289 ;; Test for being in a macro. If so, go to after it.
2304 ;; Add one extra element above HERE so as to to avoid the previous
2305 ;; expensive calculation when the next call is close to the current
2306 ;; one. This is especially useful when inside a large macro.
2311 pos))))
2314 ;; Return a buffer position before HERE which is "safe", i.e. outside any
2315 ;; string or comment. It may be in a macro.
2320 pos npos high-pos lit macro-beg macro-end)
2321 ;; Trim the cache to take account of buffer changes.
2333 ;; Add an element to `c-state-semi-nonlit-pos-cache' each iteration.
2337 ;; Test for being in a literal. If so, go to after it.
2350 pos))))
2353 ;; If position HERE is inside a literal, return (START . END), the
2354 ;; boundaries of the literal (which may be outside the accessible bit of the
2355 ;; buffer). Otherwise, return nil.
2356 ;;
2357 ;; This function is almost the same as `c-literal-limits'. Previously, it
2358 ;; differed in that it was a lower level function, and that it rigorously
2359 ;; followed the syntax from BOB. `c-literal-limits' is now (2011-12)
2360 ;; virtually identical to this function.
2368 ;; Return the start of the literal containing POS, or POS itself.
2370 pos))
2373 ;; Return a position outside of a string/comment/macro at or before POS.
2374 ;; STATE is the parse-partial-sexp state at POS.
2378 pos)))
2383 res))))
2385 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2386 ;; Stuff to do with point-min, and coping with any literal there.
2389 ;; This is (point-min) when `c-state-cache' was last calculated. A change of
2390 ;; narrowing is likely to affect the parens that are visible before the point.
2396 ;; These two variables define the literal, if any, containing point-min.
2397 ;; Their values are, respectively, 'string, c, or c++, and the start of the
2398 ;; literal. If there's no literal there, they're both nil.
2402 ;; This is the earliest buffer-pos from which scanning can be done. It is
2403 ;; either the end of the literal containing point-min, or point-min itself.
2404 ;; It becomes nil if the buffer is changed earlier than this point.
2406 ;; Return the lowest valid scanning pos. This will be the end of the
2407 ;; literal enclosing point-min, or point-min itself.
2419 ;; Determine the properties of any literal containing POINT-MIN, setting the
2420 ;; variables `c-state-point-min-lit-type', `c-state-point-min-lit-start',
2421 ;; and `c-state-min-scan-pos' accordingly. The return value is meaningless.
2423 lit)
2438 c-state-point-min-lit-start nil
2439 c-state-min-scan-pos p-min)))))
2442 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2443 ;; A variable which signals a brace dessert - helpful for reducing the number
2444 ;; of fruitless backward scans.
2447 ;; Used only in `c-append-lower-brace-pair-to-state-cache'. It is set when
2448 ;; that defun has searched backwards for a brace pair and not found one. Its
2449 ;; value is either nil or a cons (PA . FROM), where PA is the position of the
2450 ;; enclosing opening paren/brace/bracket which bounds the backwards search (or
2451 ;; nil when at top level) and FROM is where the backward search started. It
2452 ;; is reset to nil in `c-invalidate-state-cache'.
2455 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2456 ;; Lowish level functions/macros which work directly on `c-state-cache', or a
2457 ;; list of like structure.
2459 ;; Return the address of the top left brace/bracket/paren recorded in CACHE
2460 ;; (default `c-state-cache') (or nil).
2467 ;; Return the address of the latest brace/bracket/paren (whether left or
2468 ;; right) recorded in CACHE (default `c-state-cache') or nil.
2475 ;; Return the position just after the latest brace/bracket/paren (whether
2476 ;; left or right) recorded in CACHE (default `c-state-cache') or nil.
2484 ;; From the state-cache, determine the buffer position from which we might
2485 ;; scan forward to HERE to update this cache. This position will be just
2486 ;; after a paren/brace/bracket recorded in the cache, if possible, otherwise
2487 ;; return the earliest position in the accessible region which isn't within
2488 ;; a literal. If the visible portion of the buffer is entirely within a
2489 ;; literal, return NIL.
2491 ;(while (>= (or (c-state-cache-top-lparen c) 1) here)
2507 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2508 ;; Variables which keep track of preprocessor constructs.
2517 ;; These are the limits of the macro containing point at the previous call of
2518 ;; `c-parse-state', or nil.
2520 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2521 ;; Defuns which analyze the buffer, yet don't change `c-state-cache'.
2523 ;; Return the position of the opening paren/brace/bracket before HERE- which
2524 ;; matches the outermost close p/b/b between HERE+ and TOP. Except when
2525 ;; there's a macro, HERE- and HERE+ are the same. Like this:
2526 ;;
2527 ;; ............................................
2528 ;; | |
2529 ;; ( [ ( .........#macro.. ) ( ) ] )
2530 ;; ^ ^ ^ ^
2531 ;; | | | |
2532 ;; return HERE- HERE+ TOP
2533 ;;
2534 ;; If there aren't enough opening paren/brace/brackets, return the position
2535 ;; of the outermost one found, or HERE- if there are none. If there are no
2536 ;; closing p/b/bs between HERE+ and TOP, return HERE-. HERE-/+ and TOP
2537 ;; must not be inside literals. Only the accessible portion of the buffer
2538 ;; will be scanned.
2540 ;; PART 1: scan from `here+' up to `top', accumulating ")"s which enclose
2541 ;; `here'. Go round the next loop each time we pass over such a ")". These
2542 ;; probably match "("s before `here-'.
2554 ;; PART 2: Scan back before `here-' searching for the "("s
2555 ;; matching/mismatching the ")"s found above. We only need to direct the
2556 ;; caller to scan when we've encountered unmatched right parens.
2565 pos))
2568 ;; Determine the scanning strategy for adjusting `c-parse-state', attempting
2569 ;; to minimize the amount of scanning. HERE is the pertinent position in
2570 ;; the buffer, GOOD-POS is a position where `c-state-cache' (possibly with
2571 ;; its head trimmed) is known to be good, or nil if there is no such
2572 ;; position.
2573 ;;
2574 ;; The return value is a list, one of the following:
2575 ;;
2576 ;; o - ('forward START-POINT) - scan forward from START-POINT,
2577 ;; which is not less than the highest position in `c-state-cache' below HERE,
2578 ;; which is after GOOD-POS.
2579 ;; o - ('backward nil) - scan backwards (from HERE).
2580 ;; o - ('back-and-forward START-POINT) - like 'forward, but when HERE is earlier
2581 ;; than GOOD-POS.
2582 ;; o - ('IN-LIT nil) - point is inside the literal containing point-min.
2584 strategy ; 'forward, 'backward, or 'IN-LIT.
2585 start-point)
2597 start-point cache-pos)))
2601 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2602 ;; Routines which change `c-state-cache' and associated values.
2604 ;; The region (more precisely, point-min) has changed since we
2605 ;; calculated `c-state-cache'. Amend `c-state-cache' accordingly.
2607 ;; If point-min has MOVED BACKWARDS then we drop the state completely.
2608 ;; It would be possible to do a better job here and recalculate the top
2609 ;; only.
2613 c-state-cache-good-pos c-state-min-scan-pos
2614 c-state-brace-pair-desert nil))
2616 ;; point-min has MOVED FORWARD.
2618 ;; Is the new point-min inside a (different) literal?
2623 ;; Cut off a bit of the tail from `c-state-cache'.
2625 pa)
2633 c-state-cache-good-pos c-state-min-scan-pos)
2636 )))
2641 ;; If there is a brace pair preceding FROM in the buffer, at the same level
2642 ;; of nesting (not necessarily immediately preceding), push a cons onto
2643 ;; `c-state-cache' to represent it. FROM must not be inside a literal. If
2644 ;; UPPER-LIM is non-nil, we append the highest brace pair whose "}" is below
2645 ;; UPPER-LIM.
2646 ;;
2647 ;; Return non-nil when this has been done.
2648 ;;
2649 ;; The situation it copes with is this transformation:
2650 ;;
2651 ;; OLD: { (.) {...........}
2652 ;; ^ ^
2653 ;; FROM HERE
2654 ;;
2655 ;; NEW: { {....} (.) {.........
2656 ;; ^ ^ ^
2657 ;; LOWER BRACE PAIR HERE or HERE
2658 ;;
2659 ;; This routine should be fast. Since it can get called a LOT, we maintain
2660 ;; `c-state-brace-pair-desert', a small cache of "failures", such that we
2661 ;; reduce the time wasted in repeated fruitless searches in brace deserts.
2671 ;; Don't start scanning in the middle of a CPP construct unless
2672 ;; it contains HERE - these constructs, in Emacs, are "commented
2673 ;; out" with category properties.
2676 macro-start-or-from
2677 from))
2681 ;; If we're essentially repeating a fruitless search, just give up.
2687 ;; DESERT-LIM. Avoid repeated searching through the cached desert.
2693 ;; CACHE-LIM. This limit will be necessary when an opening
2694 ;; paren at `cache-pos' has just had its matching close paren
2695 ;; inserted into the buffer. `cache-pos' continues to be a
2696 ;; search bound, even though the algorithm below would skip
2697 ;; over the new paren pair.
2706 ;; The top limit is EOB to ensure that `bra' is inside the
2707 ;; accessible part of the buffer at the next scan operation.
2710 ;; In the next pair of nested loops, the inner one moves back past a
2711 ;; pair of (mis-)matching parens or brackets; the outer one moves
2712 ;; back over a sequence of unmatched close brace/paren/bracket each
2713 ;; time round.
2729 ; brace, ignore it.
2732 ;; We've found the desired brace-pair.
2744 ;; We haven't found a brace pair. Record this in the cache.
2747 bra
2752 ;; If BRA+1 is nil, do nothing. Otherwise, BRA+1 is the buffer position
2753 ;; following a {, and that brace has a (mis-)matching } (or ]), and we
2754 ;; "push" "a" brace pair onto `c-state-cache'.
2755 ;;
2756 ;; Here "push" means overwrite the top element if it's itself a brace-pair,
2757 ;; otherwise push it normally.
2758 ;;
2759 ;; The brace pair we push is normally the one surrounding BRA+1, but if the
2760 ;; latter is inside a macro, not being a macro containing
2761 ;; MACRO-START-OR-HERE, we scan backwards through the buffer for a non-macro
2762 ;; base pair. This latter case is assumed to be rare.
2763 ;;
2764 ;; Note: POINT is not preserved in this routine.
2774 c-state-cache)))
2775 ;; N.B. This defsubst codes one method for the simple, normal case,
2776 ;; and a more sophisticated, slower way for the general case. Don't
2777 ;; eliminate this defsubst - it's a speed optimization.
2781 ;; Scan the buffer from FROM to HERE, adding elements into `c-state-cache'
2782 ;; for braces etc. Return a candidate for `c-state-cache-good-pos'.
2783 ;;
2784 ;; FROM must be after the latest brace/paren/bracket in `c-state-cache', if
2785 ;; any. Typically, it is immediately after it. It must not be inside a
2786 ;; literal.
2792 here)))
2795 ; Is actually the pos. to scan for a (/{/[ from,
2796 ; which sometimes is after a silly )/}/].
2799 ; good-pos.
2807 ;; Each time round the following loop, we enter a successively deeper
2808 ;; level of brace/paren nesting. (Except sometimes we "continue at
2809 ;; the existing level".) `pa+1' is a pos inside an opening
2810 ;; brace/paren/bracket, usually just after it.
2813 ;; Each time round the next loop moves forward over an opening then
2814 ;; a closing brace/bracket/paren. This loop is white hot, so it
2815 ;; plays ugly tricks to go fast. DON'T PUT ANYTHING INTO THIS
2816 ;; LOOP WHICH ISN'T ABSOLUTELY NECESSARY!!! It terminates when a
2817 ;; call of `scan-lists' signals an error, which happens when there
2818 ;; are no more b/b/p's to scan.
2829 ;; We've just entered a deeper nesting level.
2831 ;; Insert the brace pair (if present) and the single open
2832 ;; paren/brace/bracket into `c-state-cache' It cannot be
2833 ;; inside a macro, except one around point, because of what
2834 ;; `c-neutralize-syntax-in-CPP' has done.
2836 ;; Insert the opening brace/bracket/paren position.
2838 ;; Clear admin stuff for the next more nested part of the scan.
2842 ;; All open p/b/b's at this nesting level, if any, have probably
2843 ;; been closed by matching/mismatching ones. We're probably
2844 ;; finished - we just need to check for having found an
2845 ;; unmatched )/}/], which we ignore. Such a )/}/] can't be in a
2846 ;; macro, due the action of `c-neutralize-syntax-in-CPP'.
2849 ;; Record the final, innermost, brace-pair if there is one.
2852 ;; Determine a good pos
2860 here-bol))
2869 ;; Remove stale entries from the `c-cache-state', i.e. those which will
2870 ;; not be in it when it is amended for position HERE. This may involve
2871 ;; replacing a CONS element for a brace pair containing HERE with its car.
2872 ;; Additionally, the "outermost" open-brace entry before HERE will be
2873 ;; converted to a cons if the matching close-brace is below HERE.
2874 ;;
2875 ;; START-POINT is a "maximal" "safe position" - there must be no open
2876 ;; parens/braces/brackets between START-POINT and HERE.
2877 ;;
2878 ;; As a second thing, calculate the result of parse-partial-sexp at
2879 ;; PPS-POINT, w.r.t. START-POINT. The motivation here is that
2880 ;; `c-state-cache-good-pos' may become PPS-POINT, but the caller may need to
2881 ;; adjust it to get outside a string/comment. (Sorry about this! The code
2882 ;; needs to be FAST).
2883 ;;
2884 ;; Return a list (GOOD-POS SCAN-BACK-POS CONS-SEPARATED PPS-STATE), where
2885 ;; o - GOOD-POS is a position where the new value `c-state-cache' is known
2886 ;; to be good (we aim for this to be as high as possible);
2887 ;; o - SCAN-BACK-POS, if not nil, indicates there may be a brace pair
2888 ;; preceding POS which needs to be recorded in `c-state-cache'. It is a
2889 ;; position to scan backwards from. It is the position of the "{" of the
2890 ;; last element to be removed from `c-state-cache', when that elt is a
2891 ;; cons, otherwise nil.
2892 ;; o - CONS-SEPARATED is t when a cons element in `c-state-cache' has been
2893 ;; replaced by its car because HERE lies inside the brace pair represented
2894 ;; by the cons.
2895 ;; o - PPS-STATE is the parse-partial-sexp state at PPS-POINT.
2905 ; start-point or nil
2912 ; HERE), or nil.
2918 pps-state ; Will be 9 or 10 elements long.
2919 pos
2920 upper-lim ; ,beyond which `c-state-cache' entries are removed
2921 scan-back-pos
2922 cons-separated
2923 pair-beg pps-point-state target-depth)
2925 ;; Remove entries beyond HERE. Also remove any entries inside
2926 ;; a macro, unless HERE is in the same macro.
2931 here
2937 ;; If `upper-lim' is inside the last recorded brace pair, remove its
2938 ;; RBrace and indicate we'll need to search backwards for a previous
2939 ;; brace pair.
2945 cons-separated t))
2947 ;; The next loop jumps forward out of a nested level of parens each
2948 ;; time round; the corresponding elements in `c-state-cache' are
2949 ;; removed. `pos' is just after the brace-pair or the open paren at
2950 ;; (car c-state-cache). There can be no open parens/braces/brackets
2951 ;; between `start-point'/`start-point-actual-macro-start' and HERE,
2952 ;; due to the interface spec to this function.
2955 in-macro-start)))
2957 ; marked by a `category' text property.
2958 start-point))
2969 ;; Do nothing when we've merely reached pps-point.
2970 )
2972 ;; Scan!
2976 target-depth
2977 nil pps-state))
2987 ;; We've just moved out of the paren pair containing the brace-pair
2988 ;; at (car c-state-cache). `pair-beg' is where the open paren is,
2989 ;; and is potentially where the open brace of a cons in
2990 ;; c-state-cache will be.
2996 ; containing Lparen
3000 ; together with enclosed brace pair.
3001 ;; (t nil) ; Ignore an unmated Rparen.
3002 )))
3007 nil nil ; TARGETDEPTH, STOPBEFORE
3008 pps-state)))
3010 ;; If the last paren pair we moved out of was actually a brace pair,
3011 ;; insert it into `c-state-cache'.
3016 c-state-cache)))
3021 ;; Strip stale elements of `c-state-cache' by moving backwards through the
3022 ;; buffer, and inform the caller of the scenario detected.
3023 ;;
3024 ;; HERE is the position we're setting `c-state-cache' for.
3025 ;; CACHE-POS (a locally bound variable) is just after the latest recorded
3026 ;; position in `c-state-cache' before HERE, or a position at or near
3027 ;; point-min which isn't in a literal.
3028 ;;
3029 ;; This function must only be called only when (> `c-state-cache-good-pos'
3030 ;; HERE). Usually the gap between CACHE-POS and HERE is large. It is thus
3031 ;; optimized to eliminate (or minimize) scanning between these two
3032 ;; positions.
3033 ;;
3034 ;; Return a three element list (GOOD-POS SCAN-BACK-POS FWD-FLAG), where:
3035 ;; o - GOOD-POS is a "good position", where `c-state-cache' is valid, or
3036 ;; could become so after missing elements are inserted into
3037 ;; `c-state-cache'. This is JUST AFTER an opening or closing
3038 ;; brace/paren/bracket which is already in `c-state-cache' or just before
3039 ;; one otherwise. exceptionally (when there's no such b/p/b handy) the BOL
3040 ;; before `here''s line, or the start of the literal containing it.
3041 ;; o - SCAN-BACK-POS, if non-nil, indicates there may be a brace pair
3042 ;; preceding POS which isn't recorded in `c-state-cache'. It is a position
3043 ;; to scan backwards from.
3044 ;; o - FWD-FLAG, if non-nil, indicates there may be parens/braces between
3045 ;; POS and HERE which aren't recorded in `c-state-cache'.
3046 ;;
3047 ;; The comments in this defun use "paren" to mean parenthesis or square
3048 ;; bracket (as contrasted with a brace), and "(" and ")" likewise.
3049 ;;
3050 ;; . {..} (..) (..) ( .. { } ) (...) ( .... . ..)
3051 ;; | | | | | |
3052 ;; CP E here D C good
3055 pa ren ; positions of "(" and ")"
3056 dropped-cons ; whether the last element dropped from `c-state-cache'
3057 ; was a cons (representing a brace-pair)
3058 good-pos ; see above.
3059 lit ; (START . END) of a literal containing some point.
3060 here-lit-start here-lit-end ; bounds of literal containing `here'
3061 ; or `here' itself.
3066 ;; Remove completely irrelevant entries from `c-state-cache'.
3072 ;; At this stage, (> pos here);
3073 ;; (< (c-state-cache-top-lparen) here) (or is nil).
3078 ;; CASE 1: The top of the cache is a brace pair which now encloses
3079 ;; `here'. As good-pos, return the address. of the "{". Since we've no
3080 ;; knowledge of what's inside these braces, we have no alternative but
3081 ;; to direct the caller to scan the buffer from the opening brace.
3085 ; entry into a { entry, so the caller needs to
3086 ; search for a brace pair before the {.
3088 ;; `here' might be inside a literal. Check for this.
3093 ;; Has `here' just "newly entered" a macro?
3106 ;; `here' might be nested inside any depth of parens (or brackets but
3107 ;; not braces). Scan backwards to find the outermost such opening
3108 ;; paren, if there is one. This will be the scan position to return.
3115 ;; CASE 2: Address of outermost ( or [ which now encloses `here', but
3116 ;; didn't enclose the (previous) `c-state-cache-good-pos'. If there is
3117 ;; a brace pair preceding this, it will already be in `c-state-cache',
3118 ;; unless there was a brace pair after it, i.e. there'll only be one to
3119 ;; scan for if we've just deleted one.
3122 ;; `here' isn't enclosed in a (previously unrecorded) bracket/paren.
3123 ;; Further forward scanning isn't needed, but we still need to find a
3124 ;; GOOD-POS. Step out of all enclosing "("s on HERE's line.
3137 ;; CASE 3: After a }/)/] before `here''s BOL.
3141 ;; CASE 4; Best of a bad job: BOL before `here-bol', or beginning of
3142 ;; literal containing it.
3147 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3148 ;; Externally visible routines.
3152 c-state-cache-good-pos 1
3153 c-state-nonlit-pos-cache nil
3154 c-state-nonlit-pos-cache-limit 1
3155 c-state-semi-nonlit-pos-cache nil
3156 c-state-semi-nonlit-pos-cache-limit 1
3157 c-state-brace-pair-desert nil
3158 c-state-point-min 1
3159 c-state-point-min-lit-type nil
3160 c-state-point-min-lit-start nil
3161 c-state-min-scan-pos 1
3162 c-state-old-cpp-beg nil
3163 c-state-old-cpp-end nil)
3166 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3167 ;; Debugging routines to dump `c-state-cache' in a "replayable" form.
3168 ;; (defmacro c-sc-de (elt) ; "c-state-cache-dump-element"
3169 ;; `(format ,(concat "(setq " (symbol-name elt) " %s) ") ,elt))
3170 ;; (defmacro c-sc-qde (elt) ; "c-state-cache-quote-dump-element"
3171 ;; `(format ,(concat "(setq " (symbol-name elt) " '%s) ") ,elt))
3172 ;; (defun c-state-dump ()
3173 ;; ;; For debugging.
3174 ;; ;(message
3175 ;; (concat
3176 ;; (c-sc-qde c-state-cache)
3177 ;; (c-sc-de c-state-cache-good-pos)
3178 ;; (c-sc-qde c-state-nonlit-pos-cache)
3179 ;; (c-sc-de c-state-nonlit-pos-cache-limit)
3180 ;; (c-sc-qde c-state-brace-pair-desert)
3181 ;; (c-sc-de c-state-point-min)
3182 ;; (c-sc-de c-state-point-min-lit-type)
3183 ;; (c-sc-de c-state-point-min-lit-start)
3184 ;; (c-sc-de c-state-min-scan-pos)
3185 ;; (c-sc-de c-state-old-cpp-beg)
3186 ;; (c-sc-de c-state-old-cpp-end)))
3187 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3190 ;; Invalidate all info on `c-state-cache' that applies to the buffer at HERE
3191 ;; or higher and set `c-state-cache-good-pos' accordingly. The cache is
3192 ;; left in a consistent state.
3193 ;;
3194 ;; This is much like `c-whack-state-after', but it never changes a paren
3195 ;; pair element into an open paren element. Doing that would mean that the
3196 ;; new open paren wouldn't have the required preceding paren pair element.
3197 ;;
3198 ;; This function is called from c-after-change.
3200 ;; The caches of non-literals:
3201 ;; Note that we use "<=" for the possibility of the second char of a two-char
3202 ;; comment opener being typed; this would invalidate any cache position at
3203 ;; HERE.
3209 ;; `c-state-cache':
3210 ;; Case 1: if `here' is in a literal containing point-min, everything
3211 ;; becomes (or is already) nil.
3215 c-state-cache-good-pos nil
3216 c-state-min-scan-pos nil)
3218 ;; Truncate `c-state-cache' and set `c-state-cache-good-pos' to a value
3219 ;; below `here'. To maintain its consistency, we may need to insert a new
3220 ;; brace pair.
3223 too-high-pa ; recorded {/(/[ next above here, or nil.
3224 dropped-cons ; was the last removed element a brace pair?
3225 pa)
3226 ;; The easy bit - knock over-the-top bits off `c-state-cache'.
3233 ;; Do we need to add in an earlier brace pair, having lopped one off?
3240 ;; The brace-pair desert marker:
3248 ;; Find and record all noteworthy parens between some good point earlier in
3249 ;; the file and point. That good point is at least the beginning of the
3250 ;; top-level construct we are in, or the beginning of the preceding
3251 ;; top-level construct if we aren't in one.
3252 ;;
3253 ;; The returned value is a list of the noteworthy parens with the last one
3254 ;; first. If an element in the list is an integer, it's the position of an
3255 ;; open paren (of any type) which has not been closed before the point. If
3256 ;; an element is a cons, it gives the position of a closed BRACE paren
3257 ;; pair[*]; the car is the start brace position and the cdr is the position
3258 ;; following the closing brace. Only the last closed brace paren pair
3259 ;; before each open paren and before the point is recorded, and thus the
3260 ;; state never contains two cons elements in succession. When a close brace
3261 ;; has no matching open brace (e.g., the matching brace is outside the
3262 ;; visible region), it is not represented in the returned value.
3263 ;;
3264 ;; [*] N.B. The close "brace" might be a mismatching close bracket or paren.
3265 ;; This defun explicitly treats mismatching parens/braces/brackets as
3266 ;; matching. It is the open brace which makes it a "brace" pair.
3267 ;;
3268 ;; If POINT is within a macro, open parens and brace pairs within
3269 ;; THIS macro MIGHT be recorded. This depends on whether their
3270 ;; syntactic properties have been suppressed by
3271 ;; `c-neutralize-syntax-in-CPP'. This might need fixing (2008-12-11).
3272 ;;
3273 ;; Currently no characters which are given paren syntax with the
3274 ;; syntax-table property are recorded, i.e. angle bracket arglist
3275 ;; parens are never present here. Note that this might change.
3276 ;;
3277 ;; BUG: This function doesn't cope entirely well with unbalanced
3278 ;; parens in macros. (2008-12-11: this has probably been resolved
3279 ;; by the function `c-neutralize-syntax-in-CPP'.) E.g. in the
3280 ;; following case the brace before the macro isn't balanced with the
3281 ;; one after it:
3282 ;;
3283 ;; {
3284 ;; #define X {
3285 ;; }
3286 ;;
3287 ;; Note to maintainers: this function DOES get called with point
3288 ;; within comments and strings, so don't assume it doesn't!
3289 ;;
3290 ;; This function might do hidden buffer changes.
3293 open-paren-in-column-0-is-defun-start
3294 strategy ; 'forward, 'backward etc..
3295 ;; Candidate positions to start scanning from:
3296 cache-pos ; highest position below HERE already existing in
3297 ; cache (or 1).
3298 good-pos
3299 start-point ; (when scanning forward) a place below HERE where there
3300 ; are no open parens/braces between it and HERE.
3301 bopl-state
3302 res
3303 cons-separated
3305 ;; If POINT-MIN has changed, adjust the cache
3309 ;; Strategy?
3314 ;; SCAN!
3322 ; start-point)
3332 good-pos)))
3344 good-pos)))
3348 c-state-cache-good-pos nil))))
3350 c-state-cache)
3353 ;; This is a wrapper over `c-invalidate-state-cache-1'.
3354 ;;
3355 ;; It suppresses the syntactic effect of the < and > (template) brackets and
3356 ;; of all parens in preprocessor constructs, except for any such construct
3357 ;; containing point. We can then call `c-invalidate-state-cache-1' without
3358 ;; worrying further about macros and template delimiters.
3360 ;; Emacs
3365 c-state-old-cpp-beg
3370 ;; XEmacs
3374 ;; If PLACE is non-nil, return a marker marking it, otherwise nil.
3375 ;; We (re)use MARKER.
3381 ;; This is a wrapper over `c-parse-state-1'. See that function for a
3382 ;; description of the functionality and return value.
3383 ;;
3384 ;; It suppresses the syntactic effect of the < and > (template) brackets and
3385 ;; of all parens in preprocessor constructs, except for any such construct
3386 ;; containing point. We can then call `c-parse-state-1' without worrying
3387 ;; further about macros and template delimiters.
3394 here-cpp-beg)
3396 ;; FIXME!!! Put in a `condition-case' here to protect the integrity of the
3397 ;; subsystem.
3400 ;; Emacs
3404 here-cpp-beg here-cpp-end
3408 ;; XEmacs
3412 c-state-old-cpp-end
3415 ;; Debug tool to catch cache inconsistencies. This is called from
3416 ;; 000tests.el.
3436 c-state-cache-good-pos
3437 c-state-nonlit-pos-cache
3438 c-state-nonlit-pos-cache-limit
3439 c-state-semi-nonlit-pos-cache
3440 c-state-semi-nonlit-pos-cache-limit
3441 c-state-brace-pair-desert
3442 c-state-point-min
3443 c-state-point-min-lit-type
3444 c-state-point-min-lit-start
3445 c-state-min-scan-pos
3446 c-state-old-cpp-beg
3447 c-state-old-cpp-end
3448 c-parse-state-point))))
3470 conses-not-ok))
3487 ;; The cache can actually go further back due to the ad-hoc way
3488 ;; the first paren is found, so try to whack off a bit of its
3489 ;; start before complaining.
3490 ;; (save-excursion
3491 ;; (goto-char (or (c-least-enclosing-brace res2) (point)))
3492 ;; (c-beginning-of-defun-1)
3493 ;; (while (not (or (bobp) (eq (char-after) ?{)))
3494 ;; (c-beginning-of-defun-1))
3495 ;; (unless (equal (c-whack-state-before (point) res1) res2)
3496 ;; (message (concat "c-parse-state inconsistency at %s: "
3497 ;; "using cache: %s, from scratch: %s")
3498 ;; here res1 res2)))
3501 here res1 res2)
3507 here res1)
3512 res2 ; res1 correct a cascading series of errors ASAP
3513 ))
3519 'c-debug-parse-state
3527 \f
3529 ;; Whack off any state information from PAREN-STATE which lies
3530 ;; before BUFPOS. Not destructive on PAREN-STATE.
3533 car)
3544 ;; Whack off any state information from PAREN-STATE which lies at or
3545 ;; after BUFPOS. Not destructive on PAREN-STATE.
3550 ;; just check the car, because in a balanced brace
3551 ;; expression, it must be impossible for the corresponding
3552 ;; close brace to be before point, but the open brace to
3553 ;; be after.
3555 nil ; whack it off
3557 ;; its possible that the open brace is before
3558 ;; bufpos, but the close brace is after. In that
3559 ;; case, convert this to a non-cons element. The
3560 ;; rest of the state is before bufpos, so we're
3561 ;; done.
3563 ;; we know that both the open and close braces are
3564 ;; before bufpos, so we also know that everything else
3565 ;; on state is before bufpos.
3568 nil ; whack it off
3569 ;; it's before bufpos, so everything else should too.
3572 nil)))
3575 ;; Return the bufpos of the innermost enclosing open paren before
3576 ;; bufpos, or nil if none was found.
3586 enclosingp))
3589 ;; Return the bufpos of the outermost enclosing open paren, or nil
3590 ;; if none was found.
3597 pos))
3600 ;; Return the closest "safe" position recorded on PAREN-STATE that
3601 ;; is higher up than BUFPOS. Return nil if PAREN-STATE doesn't
3602 ;; contain any. Return nil if BUFPOS is nil, which is useful to
3603 ;; find the closest limit before a given limit that might be nil.
3604 ;;
3605 ;; A "safe" position is a position at or after a recorded open
3606 ;; paren, or after a recorded close paren. The returned position is
3607 ;; thus either the first position after a close brace, or the first
3608 ;; position after an enclosing paren, or at the enclosing paren in
3609 ;; case BUFPOS is immediately after it.
3619 ;; See below.
3622 ;; elem is the position at and not after the opening paren, so
3623 ;; we can go forward one more step unless it's equal to
3624 ;; bufpos. This is useful in some cases avoid an extra paren
3625 ;; level between the safe position and bufpos.
3630 ;; This is used for `font-lock-beginning-of-syntax-function'. It
3631 ;; goes to the closest previous point that is known to be outside
3632 ;; any string literal or comment. `c-state-cache' is used if it has
3633 ;; a position in the vicinity.
3635 elem
3638 ;; Note: Similar code in `c-safe-position'. The
3639 ;; difference is that we accept a safe position at
3640 ;; the point and don't bother to go forward past open
3641 ;; parens.
3656 ;; The position is far back. Try `c-beginning-of-defun-1'
3657 ;; (although we can't be entirely sure it will go to a position
3658 ;; outside a comment or string in current emacsen). FIXME:
3659 ;; Consult `syntax-ppss' here.
3664 \f
3665 ;; Tools for scanning identifiers and other tokens.
3668 "Return non-nil if the point is on or directly after an identifier.
3669 Keywords are recognized and not considered identifiers. If an
3670 identifier is detected, the returned value is its starting position.
3671 If an identifier ends at the point and another begins at it \(can only
3672 happen in Pike) then the point for the preceding one is returned.
3674 Note that this function might do hidden buffer changes. See the
3675 comment at the start of cc-engine.el for more info."
3677 ;; FIXME: Shouldn't this function handle "operator" in C++?
3684 ;; Check for a normal (non-keyword) identifier.
3690 ;; Handle the `<operator> syntax in Pike.
3694 t
3701 ;; Handle the "operator +" syntax in C++.
3718 )))
3721 ;; If the point is at the end of a symbol then skip backward to the
3722 ;; beginning of it. Don't move otherwise. Return non-nil if point
3723 ;; moved.
3724 ;;
3725 ;; This function might do hidden buffer changes.
3728 ;; Handle the `<operator> syntax in Pike.
3734 t
3736 nil)))))
3739 ;; Move to the beginning of the current token. Do not move if not
3740 ;; in the middle of one. BACK-LIMIT may be used to bound the
3741 ;; backward search; if given it's assumed to be at the boundary
3742 ;; between two tokens. Return non-nil if the point is moved, nil
3743 ;; otherwise.
3744 ;;
3745 ;; This function might do hidden buffer changes.
3752 ;; `c-nonsymbol-token-regexp' should always match
3753 ;; since we've skipped backward over punctuation
3754 ;; or paren syntax, but consume one char in case
3755 ;; it doesn't so that we don't leave point before
3756 ;; some earlier incorrect token.
3763 ;; Move to the end of the current token. Do not move if not in the
3764 ;; middle of one. BACK-LIMIT may be used to bound the backward
3765 ;; search; if given it's assumed to be at the boundary between two
3766 ;; tokens. Return non-nil if the point is moved, nil otherwise.
3767 ;;
3768 ;; This function might do hidden buffer changes.
3776 ;; `c-nonsymbol-token-regexp' should always match since
3777 ;; we've skipped backward over punctuation or paren
3778 ;; syntax, but move forward in case it doesn't so that
3779 ;; we don't leave point earlier than we started with.
3795 "Move forward by tokens.
3796 A token is defined as all symbols and identifiers which aren't
3798 treated properly). Point is always either left at the beginning of a
3799 token or not moved at all. COUNT specifies the number of tokens to
3800 move; a negative COUNT moves in the opposite direction. A COUNT of 0
3801 moves to the next token beginning only if not already at one. If
3802 BALANCED is true, move over balanced parens, otherwise move into them.
3803 Also, if BALANCED is true, never move out of an enclosing paren.
3805 LIMIT sets the limit for the movement and defaults to the point limit.
3806 The case when LIMIT is set in the middle of a token, comment or macro
3807 is handled correctly, i.e. the point won't be left there.
3809 Return the number of tokens left to move \(positive or negative). If
3810 BALANCED is true, a move over a balanced paren counts as one. Note
3811 that if COUNT is 0 and no appropriate token beginning is found, 1 will
3812 be returned. Thus, a return value of 0 guarantees that point is at
3813 the requested position and a return value less \(without signs) than
3814 COUNT guarantees that point is at the beginning of some token.
3816 Note that this function might do hidden buffer changes. See the
3817 comment at the start of cc-engine.el for more info."
3824 c-jump-syntax-balanced
3825 c-jump-syntax-unbalanced))
3830 ;; If count is zero we should jump if in the middle of a token.
3837 ;; Skip whitespace. Count this as a move if we did in
3838 ;; fact move.
3842 ;; Moved out of bounds. Make sure the returned count isn't zero.
3847 ;; Use `condition-case' to avoid having the limit tests
3848 ;; inside the loop.
3856 t)
3859 t)
3860 ;; `c-nonsymbol-token-regexp' above should always
3861 ;; match if there are correct tokens. Try to
3862 ;; widen to see if the limit was set in the
3863 ;; middle of one, else fall back to treating
3864 ;; the offending thing as a one character token.
3869 nil)
3872 t))))
3882 count)))
3885 "Move backward by tokens.
3886 See `c-forward-token-2' for details."
3894 c-jump-syntax-balanced
3895 c-jump-syntax-unbalanced))
3899 ;; The count is zero so try to skip to the beginning of the
3900 ;; current token.
3904 ;; The limit is inside the same token, so return 1.
3907 ;; We're not in the middle of a token. If there's
3908 ;; whitespace after the point then we must move backward,
3909 ;; so set count to 1 in that case.
3911 ;; If we're looking at a '#' that might start a cpp
3912 ;; directive then we have to do a more elaborate check.
3925 ;; Use `condition-case' to avoid having to check for buffer
3926 ;; limits in `backward-char', `scan-sexps' and `goto-char' below.
3935 ;; This can be very inefficient if there's a long
3936 ;; sequence of operator tokens without any separation.
3937 ;; That doesn't happen in practice, anyway.
3947 count)))
3950 "Like `c-forward-token-2' but doesn't treat multicharacter operator
3952 characters are jumped over character by character. This function is
3953 for compatibility only; it's only a wrapper over `c-forward-token-2'."
3958 "Like `c-backward-token-2' but doesn't treat multicharacter operator
3960 characters are jumped over character by character. This function is
3961 for compatibility only; it's only a wrapper over `c-backward-token-2'."
3965 \f
3966 ;; Tools for doing searches restricted to syntactically relevant text.
3969 paren-level not-inside-token
3970 lookbehind-submatch)
3971 "Like `re-search-forward', but only report matches that are found
3972 in syntactically significant text. I.e. matches in comments, macros
3973 or string literals are ignored. The start point is assumed to be
3974 outside any comment, macro or string literal, or else the content of
3975 that region is taken as syntactically significant text.
3977 If PAREN-LEVEL is non-nil, an additional restriction is added to
3978 ignore matches in nested paren sexps. The search will also not go
3979 outside the current list sexp, which has the effect that if the point
3980 should be moved to BOUND when no match is found \(i.e. NOERROR is
3981 neither nil nor t), then it will be at the closing paren if the end of
3982 the current list sexp is encountered first.
3984 If NOT-INSIDE-TOKEN is non-nil, matches in the middle of tokens are
3985 ignored. Things like multicharacter operators and special symbols
3987 constants.
3989 If LOOKBEHIND-SUBMATCH is non-nil, it's taken as a number of a
3990 subexpression in REGEXP. The end of that submatch is used as the
3991 position to check for syntactic significance. If LOOKBEHIND-SUBMATCH
3992 isn't used or if that subexpression didn't match then the start
3994 subexpression is never tested before the starting position, so it
3998 subexpression can match the end of nonwhite syntactic whitespace,
3999 i.e. the end of comments or cpp directives. This since the function
4000 skips over such things before resuming the search. It's on the other
4002 matches syntactic whitespace.
4004 Bug: Unbalanced parens inside cpp directives are currently not handled
4005 correctly \(i.e. they don't get ignored as they should) when
4006 PAREN-LEVEL is set.
4008 Note that this function might do hidden buffer changes. See the
4009 comment at the start of cc-engine.el for more info."
4014 ;;(message "c-syntactic-re-search-forward %s %s %S" (point) bound regexp)
4017 tmp
4018 ;; Start position for the last search.
4019 search-pos
4020 ;; The `parse-partial-sexp' state between the start position
4021 ;; and the point.
4022 state
4023 ;; The current position after the last state update. The next
4024 ;; `parse-partial-sexp' continues from here.
4026 ;; The position at which to check the state and the state
4027 ;; there. This is separate from `state-pos' since we might
4028 ;; need to back up before doing the next search round.
4029 check-pos check-state
4030 ;; Last position known to end a token.
4032 ;; Set when a valid match is found.
4033 found)
4051 state-pos check-pos paren-level nil state))
4053 check-state state))
4055 ;; NOTE: If we got a look behind subexpression and get
4056 ;; an insignificant match in something that isn't
4057 ;; syntactic whitespace (i.e. strings or in nested
4058 ;; parentheses), then we can never skip more than a
4059 ;; single character from the match start position
4060 ;; (i.e. `state-pos' here) before continuing the
4061 ;; search. That since the look behind subexpression
4062 ;; might match the end of the insignificant region in
4063 ;; the next search.
4067 ;; Match inside a line comment. Skip to eol. Use
4068 ;; `re-search-forward' instead of `skip-chars-forward' to get
4069 ;; the right bound behavior.
4073 ;; Match inside a block comment. Skip to the '*/'.
4080 ;; Match in the middle of the opener of a block or line
4081 ;; comment.
4086 ;; The last `parse-partial-sexp' above might have
4087 ;; stopped short of the real check position if the end
4088 ;; of the current sexp was encountered in paren-level
4089 ;; mode. The checks above are always false in that
4090 ;; case, and since they can do better skipping in
4091 ;; lookbehind-submatch mode, we do them before
4092 ;; checking the paren level.
4096 ;; Check the paren level first since we're short of the
4097 ;; syntactic checking position if the end of the
4098 ;; current sexp was encountered by `parse-partial-sexp'.
4101 ;; Inside a nested paren sexp.
4103 ;; See the NOTE above.
4105 ;; Skip out of the paren quickly.
4109 ;; Have exited the current paren sexp.
4112 ;; The last `parse-partial-sexp' call above
4113 ;; has left us just after the closing paren
4114 ;; in this case, so we can modify the bound
4115 ;; to leave the point at the right position
4116 ;; upon return.
4118 nil)
4122 ;; Match inside a string.
4125 ;; See the NOTE above.
4127 ;; Skip to the end of the string before continuing.
4136 continue)))
4141 ;; Match inside a macro. Skip to the end of it.
4149 (< check-pos
4155 ;; Inside a token.
4157 ;; See the NOTE above.
4162 ;; A real match.
4164 nil)))
4166 ;; Should loop to search again, but take care to avoid
4167 ;; looping on the same spot.
4171 nil
4174 t))))
4180 ;;(message "c-syntactic-re-search-forward done %s" (or (match-end 0) (point)))
4187 ;; Search failed. Set point as appropriate.
4191 nil)))
4196 ;; Return the start of the literal point is in, or nil.
4197 ;; We read and write the variables `safe-pos', `safe-pos-list', `state'
4198 ;; bound in the caller.
4200 ;; Use `parse-partial-sexp' from a safe position down to the point to check
4201 ;; if it's outside comments and strings.
4204 ;; Pick a safe position as close to the point as possible.
4205 ;;
4206 ;; FIXME: Consult `syntax-ppss' here if our cache doesn't give a good
4207 ;; position.
4220 ;; Cache positions along the way to use if we have to back up more. We
4221 ;; cache every closing paren on the same level. If the paren cache is
4222 ;; relevant in this region then we're typically already on the same
4223 ;; level as the target position. Note that we might cache positions
4224 ;; after opening parens in case safe-pos is in a nested list. That's
4225 ;; both uncommon and harmless.
4233 ;; If the state contains the start of the containing sexp we cache that
4234 ;; position too, so that parse-partial-sexp in the next run has a bigger
4235 ;; chance of starting at the same level as the target position and thus
4236 ;; will get more good safe positions into the list.
4242 ;; Inside string or comment. Continue search at the
4243 ;; beginning of it.
4247 "Like `skip-chars-backward' but only look at syntactically relevant chars,
4248 i.e. don't stop at positions inside syntactic whitespace or string
4249 literals. Preprocessor directives are also ignored, with the exception
4250 of the one that the point starts within, if any. If LIMIT is given,
4251 it's assumed to be at a syntactically relevant position.
4253 If PAREN-LEVEL is non-nil, the function won't stop in nested paren
4254 sexps, and the search will also not go outside the current paren sexp.
4255 However, if LIMIT or the buffer limit is reached inside a nested paren
4256 then the point will be left at the limit.
4258 Non-nil is returned if the point moved, nil otherwise.
4260 Note that this function might do hidden buffer changes. See the
4261 comment at the start of cc-engine.el for more info."
4264 state-2
4265 ;; A list of syntactically relevant positions in descending
4266 ;; order. It's used to avoid scanning repeatedly over
4267 ;; potentially large regions with `parse-partial-sexp' to verify
4268 ;; each position. Used in `c-ssb-lit-begin'
4269 safe-pos-list
4270 ;; The result from `c-beginning-of-macro' at the start position or the
4271 ;; start position itself if it isn't within a macro. Evaluated on
4272 ;; demand.
4273 start-macro-beg
4274 ;; The earliest position after the current one with the same paren
4275 ;; level. Used only when `paren-level' is set.
4276 lit-beg
4281 ;; The next loop "tries" to find the end point each time round,
4282 ;; loops when it hasn't succeeded.
4288 ;; Don't stop inside a literal.
4291 t)))
4303 ;; Not at the right level.
4306 ;; We stop above if we go out of a paren.
4307 ;; Now check whether it precedes or is
4308 ;; nested in the starting sexp.
4312 pps-end-pos paren-level-pos
4313 nil nil state-2))
4316 ;; We've stopped short of the starting position
4317 ;; so the hit was inside a nested list. Go up
4318 ;; until we are at the right level.
4327 nil)
4328 t))
4331 nil))
4333 ;; The hit was outside the list at the start
4334 ;; position. Go to the start of the list and exit.
4336 nil))
4339 ;; Inside a macro.
4347 t
4349 ;; It's inside the same macro we started in so it's
4350 ;; a relevant match.
4352 nil))))))
4356 ;; Skip syntactic ws afterwards so that we don't stop at the
4357 ;; end of a comment if `skip-chars' is something like "^/".
4361 ;; We might want to extend this with more useful return values in
4362 ;; the future.
4365 ;; The following is an alternative implementation of
4366 ;; `c-syntactic-skip-backward' that uses backward movement to keep
4367 ;; track of the syntactic context. It turned out to be generally
4368 ;; slower than the one above which uses forward checks from earlier
4369 ;; safe positions.
4370 ;;
4371 ;;(defconst c-ssb-stop-re
4372 ;; ;; The regexp matching chars `c-syntactic-skip-backward' needs to
4373 ;; ;; stop at to avoid going into comments and literals.
4374 ;; (concat
4375 ;; ;; Match comment end syntax and string literal syntax. Also match
4376 ;; ;; '/' for block comment endings (not covered by comment end
4377 ;; ;; syntax).
4378 ;; "\\s>\\|/\\|\\s\""
4379 ;; (if (memq 'gen-string-delim c-emacs-features)
4380 ;; "\\|\\s|"
4381 ;; "")
4382 ;; (if (memq 'gen-comment-delim c-emacs-features)
4383 ;; "\\|\\s!"
4384 ;; "")))
4385 ;;
4386 ;;(defconst c-ssb-stop-paren-re
4387 ;; ;; Like `c-ssb-stop-re' but also stops at paren chars.
4388 ;; (concat c-ssb-stop-re "\\|\\s(\\|\\s)"))
4389 ;;
4390 ;;(defconst c-ssb-sexp-end-re
4391 ;; ;; Regexp matching the ending syntax of a complex sexp.
4392 ;; (concat c-string-limit-regexp "\\|\\s)"))
4393 ;;
4394 ;;(defun c-syntactic-skip-backward (skip-chars &optional limit paren-level)
4395 ;; "Like `skip-chars-backward' but only look at syntactically relevant chars,
4396 ;;i.e. don't stop at positions inside syntactic whitespace or string
4397 ;;literals. Preprocessor directives are also ignored. However, if the
4398 ;;point is within a comment, string literal or preprocessor directory to
4399 ;;begin with, its contents is treated as syntactically relevant chars.
4400 ;;If LIMIT is given, it limits the backward search and the point will be
4401 ;;left there if no earlier position is found.
4402 ;;
4403 ;;If PAREN-LEVEL is non-nil, the function won't stop in nested paren
4404 ;;sexps, and the search will also not go outside the current paren sexp.
4405 ;;However, if LIMIT or the buffer limit is reached inside a nested paren
4406 ;;then the point will be left at the limit.
4407 ;;
4408 ;;Non-nil is returned if the point moved, nil otherwise.
4409 ;;
4410 ;;Note that this function might do hidden buffer changes. See the
4411 ;;comment at the start of cc-engine.el for more info."
4412 ;;
4413 ;; (save-restriction
4414 ;; (when limit
4415 ;; (narrow-to-region limit (point-max)))
4416 ;;
4417 ;; (let ((start (point)))
4418 ;; (catch 'done
4419 ;; (while (let ((last-pos (point))
4420 ;; (stop-pos (progn
4421 ;; (skip-chars-backward skip-chars)
4422 ;; (point))))
4423 ;;
4424 ;; ;; Skip back over the same region as
4425 ;; ;; `skip-chars-backward' above, but keep to
4426 ;; ;; syntactically relevant positions.
4427 ;; (goto-char last-pos)
4428 ;; (while (and
4429 ;; ;; `re-search-backward' with a single char regexp
4430 ;; ;; should be fast.
4431 ;; (re-search-backward
4432 ;; (if paren-level c-ssb-stop-paren-re c-ssb-stop-re)
4433 ;; stop-pos 'move)
4434 ;;
4435 ;; (progn
4436 ;; (cond
4437 ;; ((looking-at "\\s(")
4438 ;; ;; `paren-level' is set and we've found the
4439 ;; ;; start of the containing paren.
4440 ;; (forward-char)
4441 ;; (throw 'done t))
4442 ;;
4443 ;; ((looking-at c-ssb-sexp-end-re)
4444 ;; ;; We're at the end of a string literal or paren
4445 ;; ;; sexp (if `paren-level' is set).
4446 ;; (forward-char)
4447 ;; (condition-case nil
4448 ;; (c-backward-sexp)
4449 ;; (error
4450 ;; (goto-char limit)
4451 ;; (throw 'done t))))
4452 ;;
4453 ;; (t
4454 ;; (forward-char)
4455 ;; ;; At the end of some syntactic ws or possibly
4456 ;; ;; after a plain '/' operator.
4457 ;; (let ((pos (point)))
4458 ;; (c-backward-syntactic-ws)
4459 ;; (if (= pos (point))
4460 ;; ;; Was a plain '/' operator. Go past it.
4461 ;; (backward-char)))))
4462 ;;
4463 ;; (> (point) stop-pos))))
4464 ;;
4465 ;; ;; Now the point is either at `stop-pos' or at some
4466 ;; ;; position further back if `stop-pos' was at a
4467 ;; ;; syntactically irrelevant place.
4468 ;;
4469 ;; ;; Skip additional syntactic ws so that we don't stop
4470 ;; ;; at the end of a comment if `skip-chars' is
4471 ;; ;; something like "^/".
4472 ;; (c-backward-syntactic-ws)
4473 ;;
4474 ;; (< (point) stop-pos))))
4475 ;;
4476 ;; ;; We might want to extend this with more useful return values
4477 ;; ;; in the future.
4478 ;; (/= (point) start))))
4480 \f
4481 ;; Tools for handling comments and string literals.
4484 "Return the type of literal point is in, if any.
4485 The return value is `c' if in a C-style comment, `c++' if in a C++
4486 style comment, `string' if in a string literal, `pound' if DETECT-CPP
4487 is non-nil and in a preprocessor line, or nil if somewhere else.
4488 Optional LIM is used as the backward limit of the search. If omitted,
4489 or nil, `c-beginning-of-defun' is used.
4491 The last point calculated is cached if the cache is enabled, i.e. if
4492 `c-in-literal-cache' is bound to a two element vector.
4494 Note that this function might do hidden buffer changes. See the
4495 comment at the start of cc-engine.el for more info."
4506 "Return a cons of the beginning and end positions of the comment or
4507 string surrounding point (including both delimiters), or nil if point
4509 to start parsing from. If NEAR is non-nil, then the limits of any
4511 spaces and tabs between point and the literal. The search for such a
4512 literal is done first in forward direction. If NOT-IN-DELIMITER is
4513 non-nil, the case when point is inside a starting delimiter won't be
4514 recognized. This only has effect for comments which have starting
4515 delimiters with more than one character.
4517 Note that this function might do hidden buffer changes. See the
4518 comment at the start of cc-engine.el for more info."
4534 ;; Search forward for a literal.
4545 ;; Search backward.
4556 ;; Block comment. Due to the nature of line
4557 ;; comments, they will always be covered by the
4558 ;; normal case above.
4561 ;; If LIM is bogus, beg will be bogus.
4565 ))))
4567 ;; In case external callers use this; it did have a docstring.
4571 "If the argument is a cons of two buffer positions (such as returned by
4572 `c-literal-limits'), and that range contains a C++ style line comment,
4573 then an extended range is returned that contains all adjacent line
4574 comments (i.e. all comments that starts in the same column with no
4575 empty lines or non-whitespace characters between them). Otherwise the
4576 argument is returned.
4578 Note that this function might do hidden buffer changes. See the
4579 comment at the start of cc-engine.el for more info."
4590 ;; Got to take care in the backward direction to handle
4591 ;; comments which are preceded by code.
4605 range)
4609 "Convenience function that given the result of `c-literal-limits',
4610 returns nil or the type of literal that the range surrounds, one
4611 of the symbols 'c, 'c++ or 'string. It's much faster than using
4612 `c-in-literal' and is intended to be used when you need both the
4613 type of a literal and its limits.
4615 Note that this function might do hidden buffer changes. See the
4616 comment at the start of cc-engine.el for more info."
4627 range))
4630 ;; Get a "safe place" approximately TRY-SIZE characters before START.
4631 ;; This doesn't preserve point.
4640 ;; Return a buffer position HOW-FAR-BACK non-literal characters from START
4641 ;; (default point). This is done by going back further in the buffer then
4642 ;; searching forward for literals. The position found won't be in a
4643 ;; literal. We start searching for the sought position TRY-SIZE (default
4644 ;; twice HOW-FAR-BACK) bytes back from START. This function must be fast.
4645 ;; :-)
4653 stack elt size
4656 ;; Move forward one literal each time round this loop.
4657 ;; Move forward to the start of a comment or string.
4659 pos
4660 start
4661 nil ; target-depth
4662 nil ; stop-before
4663 s ; state
4666 ;; Gather details of the non-literal-bit - starting pos and size.
4670 pos))
4674 ;; Move forward to the end of the comment/string.
4678 start
4679 nil ; target-depth
4680 nil ; stop-before
4681 s ; state
4685 ;; Now try and find enough non-literal characters recorded on the stack.
4686 ;; Go back one recorded literal each time round this loop.
4688 stack)
4693 ;; Have we found enough yet?
4703 ;; Return a buffer position about HOW-FAR non-literal characters forward
4704 ;; from START-POS (default point), which must not be inside a literal.
4711 ;; Scan over counted characters.
4713 pos
4715 nil ; target-depth
4716 nil ; stop-before
4717 s ; state
4721 ;; Scan over literal characters.
4724 pos
4726 nil ; target-depth
4727 nil ; stop-before
4728 s ; state
4733 \f
4734 ;; `c-find-decl-spots' and accompanying stuff.
4736 ;; Variables used in `c-find-decl-spots' to cache the search done for
4737 ;; the first declaration in the last call. When that function starts,
4738 ;; it needs to back up over syntactic whitespace to look at the last
4739 ;; token before the region being searched. That can sometimes cause
4740 ;; moves back and forth over a quite large region of comments and
4741 ;; macros, which would be repeated for each changed character when
4742 ;; we're called during fontification, since font-lock refontifies the
4743 ;; current line for each change. Thus it's worthwhile to cache the
4744 ;; first match.
4745 ;;
4746 ;; `c-find-decl-syntactic-pos' is a syntactically relevant position in
4747 ;; the syntactic whitespace less or equal to some start position.
4748 ;; There's no cached value if it's nil.
4749 ;;
4750 ;; `c-find-decl-match-pos' is the match position if
4751 ;; `c-find-decl-prefix-search' matched before the syntactic whitespace
4752 ;; at `c-find-decl-syntactic-pos', or nil if there's no such match.
4763 ; (defface c-debug-decl-spot-face
4764 ; '((t (:background "Turquoise")))
4765 ; "Debug face to mark the spots where `c-find-decl-spots' stopped.")
4766 ; (defface c-debug-decl-sws-face
4767 ; '((t (:background "Khaki")))
4768 ; "Debug face to mark the syntactic whitespace between the declaration
4769 ; spots and the preceding token end.")
4785 ;; Macro used inside `c-find-decl-spots'. It ought to be a defun,
4786 ;; but it contains lots of free variables that refer to things
4787 ;; inside `c-find-decl-spots'. The point is left at `cfd-match-pos'
4788 ;; if there is a match, otherwise at `cfd-limit'.
4789 ;;
4790 ;; The macro moves point forward to the next putative start of a declaration
4791 ;; or cfd-limit. This decl start is the next token after a "declaration
4792 ;; prefix". The declaration prefix is the earlier of `cfd-prop-match' and
4793 ;; `cfd-re-match'. `cfd-match-pos' is set to the decl prefix.
4794 ;;
4795 ;; This macro might do hidden buffer changes.
4798 ;; Find the next property match position if we haven't got one already.
4809 ;; Find the next `c-decl-prefix-or-start-re' match if we haven't
4810 ;; got one already.
4816 ;; Each time round, the next `while' moves forward over a pseudo match
4817 ;; of `c-decl-prefix-or-start-re' which is either inside a literal, or
4818 ;; is a ":" not preceded by "public", etc.. `cfd-re-match' and
4819 ;; `cfd-re-match-end' get set.
4826 ;; No match. Finish up and exit the loop.
4828 nil)
4831 ;; Matched the end of a token preceding a decl spot.
4835 ;; Matched a token that start a decl spot.
4838 c-literal-faces)
4839 ;; Pseudo match inside a comment or string literal. Skip out
4840 ;; of comments and string literals.
4852 ;; Found a ":" which isn't part of "public:", etc.
4853 t)
4856 ;; If our match was at the decl start, we have to back up over the
4857 ;; preceding syntactic ws to set `cfd-match-pos' and to catch
4858 ;; any decl spots in the syntactic ws.
4863 ;; Choose whichever match is closer to the start.
4866 cfd-re-match nil)
4868 cfd-prop-match nil))
4873 ;; Skip forward past comments only so we don't skip macros.
4875 ;; Set the position to continue at. We can avoid going over
4876 ;; the comments skipped above a second time, but it's possible
4877 ;; that the comment skipping has taken us past `cfd-prop-match'
4878 ;; since the property might be used inside comments.
4884 ;; Call CFD-FUN for each possible spot for a declaration, cast or
4885 ;; label from the point to CFD-LIMIT.
4886 ;;
4887 ;; CFD-FUN is called with point at the start of the spot. It's passed two
4888 ;; arguments: The first is the end position of the token preceding the spot,
4889 ;; or 0 for the implicit match at bob. The second is a flag that is t when
4890 ;; the match is inside a macro. Point should be moved forward by at least
4891 ;; one token.
4892 ;;
4893 ;; If CFD-FUN adds `c-decl-end' properties somewhere below the current spot,
4894 ;; it should return non-nil to ensure that the next search will find them.
4895 ;;
4896 ;; Such a spot is:
4897 ;; o The first token after bob.
4898 ;; o The first token after the end of submatch 1 in
4899 ;; `c-decl-prefix-or-start-re' when that submatch matches. This
4900 ;; submatch is typically a (L or R) brace or paren, a ;, or a ,.
4901 ;; o The start of each `c-decl-prefix-or-start-re' match when
4902 ;; submatch 1 doesn't match. This is, for example, the keyword
4903 ;; "class" in Pike.
4904 ;; o The start of a previously recognized declaration; "recognized"
4905 ;; means that the last char of the previous token has a `c-type'
4906 ;; text property with the value `c-decl-end'; this only holds
4907 ;; when `c-type-decl-end-used' is set.
4908 ;;
4909 ;; Only a spot that match CFD-DECL-RE and whose face is in the
4910 ;; CFD-FACE-CHECKLIST list causes CFD-FUN to be called. The face
4911 ;; check is disabled if CFD-FACE-CHECKLIST is nil.
4912 ;;
4913 ;; If the match is inside a macro then the buffer is narrowed to the
4914 ;; end of it, so that CFD-FUN can investigate the following tokens
4915 ;; without matching something that begins inside a macro and ends
4916 ;; outside it. It's to avoid this work that the CFD-DECL-RE and
4917 ;; CFD-FACE-CHECKLIST checks exist.
4918 ;;
4919 ;; The spots are visited approximately in order from top to bottom.
4920 ;; It's however the positions where `c-decl-prefix-or-start-re'
4921 ;; matches and where `c-decl-end' properties are found that are in
4922 ;; order. Since the spots often are at the following token, they
4923 ;; might be visited out of order insofar as more spots are reported
4924 ;; later on within the syntactic whitespace between the match
4925 ;; positions and their spots.
4926 ;;
4927 ;; It's assumed that comments and strings are fontified in the
4928 ;; searched range.
4929 ;;
4930 ;; This is mainly used in fontification, and so has an elaborate
4931 ;; cache to handle repeated calls from the same start position; see
4932 ;; the variables above.
4933 ;;
4934 ;; All variables in this function begin with `cfd-' to avoid name
4935 ;; collision with the (dynamically bound) variables used in CFD-FUN.
4936 ;;
4937 ;; This function might do hidden buffer changes.
4941 ;; The end of the token preceding the decl spot last found
4942 ;; with `c-decl-prefix-or-start-re'. `cfd-limit' if there's
4943 ;; no match.
4944 cfd-re-match
4945 ;; The end position of the last `c-decl-prefix-or-start-re'
4946 ;; match. If this is greater than `cfd-continue-pos', the
4947 ;; next regexp search is started here instead.
4949 ;; The end of the last `c-decl-end' found by
4950 ;; `c-find-decl-prefix-search'. `cfd-limit' if there's no
4951 ;; match. If searching for the property isn't needed then we
4952 ;; disable it by setting it to `cfd-limit' directly.
4954 ;; The end of the token preceding the decl spot last found by
4955 ;; `c-find-decl-prefix-search'. 0 for the implicit match at
4956 ;; bob. `cfd-limit' if there's no match. In other words,
4957 ;; this is the minimum of `cfd-re-match' and `cfd-prop-match'.
4959 ;; The position to continue searching at.
4960 cfd-continue-pos
4961 ;; The position of the last "real" token we've stopped at.
4962 ;; This can be greater than `cfd-continue-pos' when we get
4963 ;; hits inside macros or at `c-decl-end' positions inside
4964 ;; comments.
4966 ;; The end position of the last entered macro.
4969 ;; Initialize by finding a syntactically relevant start position
4970 ;; before the point, and do the first `c-decl-prefix-or-start-re'
4971 ;; search unless we're at bob.
4974 ;; Must back up a bit since we look for the end of the previous
4975 ;; statement or declaration, which is earlier than the first
4976 ;; returned match.
4978 ;; This `cond' moves back over any literals or macros. It has special
4979 ;; handling for when the region being searched is entirely within a
4980 ;; macro. It sets `cfd-continue-pos' (unless we've reached
4981 ;; `cfd-limit').
4983 ;; First we need to move to a syntactically relevant position.
4984 ;; Begin by backing out of comment or string literals.
4985 ;;
4986 ;; This arm of the cond actually triggers if we're in a literal,
4987 ;; and cfd-limit is at most at BONL.
4989 ;; This arm of the `and' moves backwards out of a literal when
4990 ;; the face at point is a literal face. In this case, its value
4991 ;; is always non-nil.
4993 ;; Try to use the faces to back up to the start of the
4994 ;; literal. FIXME: What if the point is on a declaration
4995 ;; inside a comment?
5001 ;; XEmacs doesn't fontify the quotes surrounding string
5002 ;; literals.
5011 ;; Don't trust the literal to contain only literal faces
5012 ;; (the font lock package might not have fontified the
5013 ;; start of it at all, for instance) so check that we have
5014 ;; arrived at something that looks like a start or else
5015 ;; resort to `c-literal-limits'.
5022 ;; The start is in a literal. If the limit is in the same
5023 ;; one we don't have to find a syntactic position etc. We
5024 ;; only check that if the limit is at or before bonl to save
5025 ;; time; it covers the by far most common case when font-lock
5026 ;; refontifies the current line only.
5037 ;; Completely inside a literal. Set up variables to trig the
5038 ;; (< cfd-continue-pos cfd-start-pos) case below and it'll
5039 ;; find a suitable start position.
5042 ;; Check if the region might be completely inside a macro, to
5043 ;; optimize that like the completely-inside-literal above.
5050 ;; (Maybe) completely inside a macro. Only need to trig the
5051 ;; (< cfd-continue-pos cfd-start-pos) case below to make it
5052 ;; set things up.
5054 start-in-macro t))
5056 ;; The default arm of the `cond' moves back over any macro we're in
5057 ;; and over any syntactic WS. It sets `c-find-decl-syntactic-pos'.
5059 ;; Back out of any macro so we don't miss any declaration
5060 ;; that could follow after it.
5064 ;; Now we're at a proper syntactically relevant position so we
5065 ;; can use the cache. But first clear it if it applied
5066 ;; further down.
5071 ;; Don't have to do this if the cache is relevant here,
5072 ;; typically if the same line is refontified again. If
5073 ;; we're just some syntactic whitespace further down we can
5074 ;; still use the cache to limit the skipping.
5077 ;; If we hit `c-find-decl-syntactic-pos' and
5078 ;; `c-find-decl-match-pos' is set then we install the cached
5079 ;; values. If we hit `c-find-decl-syntactic-pos' and
5080 ;; `c-find-decl-match-pos' is nil then we know there's no decl
5081 ;; prefix in the whitespace before `c-find-decl-syntactic-pos'
5082 ;; and so we can continue the search from this point. If we
5083 ;; didn't hit `c-find-decl-syntactic-pos' then we're now in
5084 ;; the right spot to begin searching anyway.
5086 c-find-decl-match-pos)
5088 cfd-continue-pos syntactic-pos)
5093 ;; Always consider bob a match to get the first
5094 ;; declaration in the file. Do this separately instead of
5095 ;; letting `c-decl-prefix-or-start-re' match bob, so that
5096 ;; regexp always can consume at least one character to
5097 ;; ensure that we won't get stuck in an infinite loop.
5102 ;; Do an initial search now. In the bob case above it's
5103 ;; only done to search for a `c-decl-end' spot.
5109 ;; Advance `cfd-continue-pos' if it's before the start position.
5110 ;; The closest continue position that might have effect at or
5111 ;; after the start depends on what we started in. This also
5112 ;; finds a suitable start position in the special cases when the
5113 ;; region is completely within a literal or macro.
5118 ;; If we're in a macro then it's the closest preceding token
5119 ;; in the macro. Check this before `start-in-literal',
5120 ;; since if we're inside a literal in a macro, the preceding
5121 ;; token is earlier than any `c-decl-end' spot inside the
5122 ;; literal (comment).
5124 ;; The only syntactic ws in macros are comments.
5130 ;; If we're in a comment it can only be the closest
5131 ;; preceding `c-decl-end' position within that comment, if
5132 ;; any. Go back to the beginning of such a property so that
5133 ;; `c-find-decl-prefix-search' will find the end of it.
5134 ;; (Can't stop at the end and install it directly on
5135 ;; `cfd-prop-match' since that variable might be cleared
5136 ;; after `cfd-fun' below.)
5137 ;;
5138 ;; Note that if the literal is a string then the property
5139 ;; search will simply skip to the beginning of it right
5140 ;; away.
5152 ;; Didn't find any property inside the comment, so we can
5153 ;; skip it entirely. (This won't skip past a string, but
5154 ;; that'll be handled quickly by the next
5155 ;; `c-find-decl-prefix-search' anyway.)
5161 ;; If we started in normal code, the only match that might
5162 ;; apply before the start is what we already got in
5163 ;; `cfd-match-pos' so we can continue at the start position.
5164 ;; (Note that we don't get here if the first match is below
5165 ;; it.)
5168 ;; Delete found matches if they are before our new continue
5169 ;; position, so that `c-find-decl-prefix-search' won't back up
5170 ;; to them later on.
5178 ;; This is the normal case and we got a proper syntactic
5179 ;; position. If there's a match then it's always outside
5180 ;; macros and comments, so advance to the next token and set
5181 ;; `cfd-token-pos'. The loop below will later go back using
5182 ;; `cfd-continue-pos' to fix declarations inside the
5183 ;; syntactic ws.
5191 ;; Have one of the special cases when the region is completely
5192 ;; within a literal or macro. `cfd-continue-pos' is set to a
5193 ;; good start position for the search, so do it.
5196 ;; Now loop, one decl spot per iteration. We already have the first
5197 ;; match in `cfd-match-pos'.
5199 ;; Go forward over "false matches", one per iteration.
5204 ;; Kludge to filter out matches on the "<" that
5205 ;; aren't open parens, for the sake of languages
5206 ;; that got `c-recognize-<>-arglists' set.
5211 ;; If `cfd-continue-pos' is less or equal to
5212 ;; `cfd-token-pos', we've got a hit inside a macro
5213 ;; that's in the syntactic whitespace before the last
5214 ;; "real" declaration we've checked. If they're equal
5215 ;; we've arrived at the declaration a second time, so
5216 ;; there's nothing to do.
5220 ;; If `cfd-continue-pos' is less than `cfd-token-pos'
5221 ;; we're still searching for declarations embedded in
5222 ;; the syntactic whitespace. In that case we need
5223 ;; only to skip comments and not macros, since they
5224 ;; can't be nested, and that's already been done in
5225 ;; `c-find-decl-prefix-search'.
5230 ;; Continue if the following token fails the
5231 ;; CFD-DECL-RE and CFD-FACE-CHECKLIST checks.
5238 t)))
5249 ;; Narrow to the end of the macro if we got a hit inside
5250 ;; one, to avoid recognizing things that start inside the
5251 ;; macro and end outside it.
5253 ;; Not in the same macro as in the previous round.
5264 t
5267 t)
5268 ;; The matched token was the last thing in the macro,
5269 ;; so the whole match is bogus.
5278 ;; Restore limits if we did macro narrowing above.
5285 ; cfd-match-pos, etc.
5287 \f
5288 ;; A cache for found types.
5290 ;; Buffer local variable that contains an obarray with the types we've
5291 ;; found. If a declaration is recognized somewhere we record the
5292 ;; fully qualified identifier in it to recognize it as a type
5293 ;; elsewhere in the file too. This is not accurate since we do not
5294 ;; bother with the scoping rules of the languages, but in practice the
5295 ;; same name is seldom used as both a type and something else in a
5296 ;; file, and we only use this as a last resort in ambiguous cases (see
5297 ;; `c-forward-decl-or-cast-1').
5298 ;;
5299 ;; Not every type need be in this cache. However, things which have
5300 ;; ceased to be types must be removed from it.
5301 ;;
5302 ;; Template types in C++ are added here too but with the template
5303 ;; arglist replaced with "<>" in references or "<" for the one in the
5304 ;; primary type. E.g. the type "Foo<A,B>::Bar<C>" is stored as
5305 ;; "Foo<>::Bar<". This avoids storing very long strings (since C++
5306 ;; template specs can be fairly sized programs in themselves) and
5307 ;; improves the hit ratio (it's a type regardless of the template
5308 ;; args; it's just not the same type, but we're only interested in
5309 ;; recognizing types, not telling distinct types apart). Note that
5310 ;; template types in references are added here too; from the example
5311 ;; above there will also be an entry "Foo<".
5316 ;; Clears `c-found-types'.
5320 ;; Add the given region as a type in `c-found-types'. If the region
5321 ;; doesn't match an existing type but there is a type which is equal
5322 ;; to the given one except that the last character is missing, then
5323 ;; the shorter type is removed. That's done to avoid adding all
5324 ;; prefixes of a type as it's being entered and font locked. This
5325 ;; doesn't cover cases like when characters are removed from a type
5326 ;; or added in the middle. We'd need the position of point when the
5327 ;; font locking is invoked to solve this well.
5328 ;;
5329 ;; This function might do hidden buffer changes.
5336 ;; Remove the "NAME" from c-found-types, if present.
5340 ;; Return non-nil if the given region contains a type in
5341 ;; `c-found-types'.
5342 ;;
5343 ;; This function might do hidden buffer changes.
5345 c-found-types))
5348 ;; Return all the types in `c-found-types' as a sorted list of
5349 ;; strings.
5353 type-list)))
5354 c-found-types)
5357 ;; Shut up the byte compiler.
5361 ;; An after change function which, in conjunction with the info in
5362 ;; c-maybe-stale-found-type (set in c-before-change), removes a type
5363 ;; from `c-found-types', should this type have become stale. For
5364 ;; example, this happens to "foo" when "foo \n bar();" becomes
5365 ;; "foo(); \n bar();". Such stale types, if not removed, foul up
5366 ;; the fontification.
5367 ;;
5368 ;; Have we, perhaps, added non-ws characters to the front/back of a found
5369 ;; type?
5389 ;; Changing the amount of (already existing) whitespace - don't do anything.
5394 ;; The syntactic relationship which defined a "found type" has been
5395 ;; destroyed.
5398 ;; ((eq (car c-maybe-stale-found-type) 'c-decl-type-start) FIXME!!!
5399 )))
5401 \f
5402 ;; Setting and removing syntax properties on < and > in languages (C++
5403 ;; and Java) where they can be template/generic delimiters as well as
5404 ;; their normal meaning of "less/greater than".
5406 ;; Normally, < and > have syntax 'punctuation'. When they are found to
5407 ;; be delimiters, they are marked as such with the category properties
5408 ;; c-<-as-paren-syntax, c->-as-paren-syntax respectively.
5410 ;; STRATEGY:
5411 ;;
5412 ;; It is impossible to determine with certainty whether a <..> pair in
5413 ;; C++ is two comparison operators or is template delimiters, unless
5414 ;; one duplicates a lot of a C++ compiler. For example, the following
5415 ;; code fragment:
5416 ;;
5417 ;; foo (a < b, c > d) ;
5418 ;;
5419 ;; could be a function call with two integer parameters (each a
5420 ;; relational expression), or it could be a constructor for class foo
5421 ;; taking one parameter d of templated type "a < b, c >". They are
5422 ;; somewhat easier to distinguish in Java.
5423 ;;
5424 ;; The strategy now (2010-01) adopted is to mark and unmark < and
5425 ;; > IN MATCHING PAIRS ONLY. [Previously, they were marked
5426 ;; individually when their context so indicated. This gave rise to
5427 ;; intractable problems when one of a matching pair was deleted, or
5428 ;; pulled into a literal.]
5429 ;;
5430 ;; At each buffer change, the syntax-table properties are removed in a
5431 ;; before-change function and reapplied, when needed, in an
5432 ;; after-change function. It is far more important that the
5433 ;; properties get removed when they they are spurious than that they
5434 ;; be present when wanted.
5435 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
5437 ;; POS (default point) is at a < character. If it is marked with
5438 ;; open paren syntax-table text property, remove the property,
5439 ;; together with the close paren property on the matching > (if
5440 ;; any).
5455 ;; POS (default point) is at a > character. If it is marked with
5456 ;; close paren syntax-table property, remove the property, together
5457 ;; with the open paren property on the matching < (if any).
5472 ;; POS (default point) is at a < or > character. If it has an
5473 ;; open/close paren syntax-table property, remove this property both
5474 ;; from the current character and its partner (which will also be
5475 ;; thusly marked).
5485 ;; POS (default point) is at a < character. If it is both marked
5486 ;; with open/close paren syntax-table property, and has a matching >
5487 ;; (also marked) which is after LIM, remove the property both from
5488 ;; the current > and its partner. Return t when this happens, nil
5489 ;; when it doesn't.
5503 t)))
5506 ;; POS (default point) is at a > character. If it is both marked
5507 ;; with open/close paren syntax-table property, and has a matching <
5508 ;; (also marked) which is before LIM, remove the property both from
5509 ;; the current < and its partner. Return t when this happens, nil
5510 ;; when it doesn't.
5524 t)))
5526 ;; Set by c-common-init in cc-mode.el.
5531 ;; Unmark certain pairs of "< .... >" which are currently marked as
5532 ;; template/generic delimiters. (This marking is via syntax-table
5533 ;; text properties).
5534 ;;
5535 ;; These pairs are those which are in the current "statement" (i.e.,
5536 ;; the region between the {, }, or ; before BEG and the one after
5537 ;; END), and which enclose any part of the interval (BEG END).
5538 ;;
5539 ;; Note that in C++ (?and Java), template/generic parens cannot
5540 ;; enclose a brace or semicolon, so we use these as bounds on the
5541 ;; region we must work on.
5542 ;;
5543 ;; This function is called from before-change-functions (via
5544 ;; c-get-state-before-change-functions). Thus the buffer is widened,
5545 ;; and point is undefined, both at entry and exit.
5546 ;;
5547 ;; FIXME!!! This routine ignores the possibility of macros entirely.
5548 ;; 2010-01-29.
5552 new-beg new-end need-new-beg need-new-end)
5553 ;; Locate the barrier before the changed region
5558 ;; Remove the syntax-table/category properties from each pertinent <...>
5559 ;; pair. Firsly, the ones with the < before beg and > after beg.
5565 ;; Locate the barrier after END.
5570 ;; Remove syntax-table properties from the remaining pertinent <...>
5571 ;; pairs, those with a > after end and < before end.
5576 ;; Extend the fontification region, if needed.
5586 ;; This is called from `after-change-functions' when
5587 ;; c-recognize-<>-arglists' is set. It ensures that no "<" or ">"
5588 ;; chars with paren syntax become part of another operator like "<<"
5589 ;; or ">=".
5590 ;;
5591 ;; This function might do hidden buffer changes.
5624 \f
5625 ;; Handling of small scale constructs like types and names.
5627 ;; Dynamically bound variable that instructs `c-forward-type' to also
5628 ;; treat possible types (i.e. those that it normally returns 'maybe or
5629 ;; 'found for) as actual types (and always return 'found for them).
5630 ;; This means that it records them in `c-record-type-identifiers' if
5631 ;; that is set, and that it adds them to `c-found-types'.
5634 ;; Dynamically bound variable that instructs `c-forward-<>-arglist' to
5635 ;; mark up successfully parsed arglists with paren syntax properties on
5636 ;; the surrounding angle brackets and with `c-<>-arg-sep' in the
5637 ;; `c-type' property of each argument separating comma.
5638 ;;
5639 ;; Setting this variable also makes `c-forward-<>-arglist' recurse into
5640 ;; all arglists for side effects (i.e. recording types), otherwise it
5641 ;; exploits any existing paren syntax properties to quickly jump to the
5642 ;; end of already parsed arglists.
5643 ;;
5644 ;; Marking up the arglists is not the default since doing that correctly
5645 ;; depends on a proper value for `c-restricted-<>-arglists'.
5648 ;; Dynamically bound variable that instructs `c-forward-<>-arglist' to
5649 ;; not accept arglists that contain binary operators.
5650 ;;
5651 ;; This is primarily used to handle C++ template arglists. C++
5652 ;; disambiguates them by checking whether the preceding name is a
5653 ;; template or not. We can't do that, so we assume it is a template
5654 ;; if it can be parsed as one. That usually works well since
5655 ;; comparison expressions on the forms "a < b > c" or "a < b, c > d"
5656 ;; in almost all cases would be pointless.
5657 ;;
5658 ;; However, in function arglists, e.g. in "foo (a < b, c > d)", we
5659 ;; should let the comma separate the function arguments instead. And
5660 ;; in a context where the value of the expression is taken, e.g. in
5661 ;; "if (a < b || c > d)", it's probably not a template.
5664 ;; Dynamically bound variables that instructs
5665 ;; `c-forward-keyword-clause', `c-forward-<>-arglist',
5666 ;; `c-forward-name', `c-forward-type', `c-forward-decl-or-cast-1', and
5667 ;; `c-forward-label' to record the ranges of all the type and
5668 ;; reference identifiers they encounter. They will build lists on
5669 ;; these variables where each element is a cons of the buffer
5670 ;; positions surrounding each identifier. This recording is only
5671 ;; activated when `c-record-type-identifiers' is non-nil.
5672 ;;
5673 ;; All known types that can't be identifiers are recorded, and also
5674 ;; other possible types if `c-promote-possible-types' is set.
5675 ;; Recording is however disabled inside angle bracket arglists that
5676 ;; are encountered inside names and other angle bracket arglists.
5677 ;; Such occurrences are taken care of by `c-font-lock-<>-arglists'
5678 ;; instead.
5679 ;;
5680 ;; Only the names in C++ template style references (e.g. "tmpl" in
5681 ;; "tmpl<a,b>::foo") are recorded as references, other references
5682 ;; aren't handled here.
5683 ;;
5684 ;; `c-forward-label' records the label identifier(s) on
5685 ;; `c-record-ref-identifiers'.
5689 ;; This variable will receive a cons cell of the range of the last
5690 ;; single identifier symbol stepped over by `c-forward-name' if it's
5691 ;; successful. This is the range that should be put on one of the
5692 ;; record lists above by the caller. It's assigned nil if there's no
5693 ;; such symbol in the name.
5698 ;; Always true.
5708 ;; Always true.
5716 ;; Dynamically bound variable that instructs `c-forward-type' to
5717 ;; record the ranges of types that only are found. Behaves otherwise
5718 ;; like `c-record-type-identifiers'.
5722 ;; Used internally in `c-forward-keyword-clause' to move forward
5723 ;; over a type (if TYPE is 'type) or a name (otherwise) which
5724 ;; possibly is prefixed by keywords and their associated clauses.
5725 ;; Try with a type/name first to not trip up on those that begin
5726 ;; with a keyword. Return t if a known or found type is moved
5727 ;; over. The point is clobbered if nil is returned. If range
5728 ;; recording is enabled, the identifier is recorded on as a type
5729 ;; if TYPE is 'type or as a reference if TYPE is 'ref.
5730 ;;
5731 ;; This macro might do hidden buffer changes.
5736 nil
5743 t)))
5746 ;; Used internally in `c-forward-keyword-clause' to move forward
5747 ;; over a comma separated list of types or names using
5748 ;; `c-forward-keyword-prefixed-id'.
5749 ;;
5750 ;; This macro might do hidden buffer changes.
5761 ;; Submatch MATCH in the current match data is assumed to surround a
5762 ;; token. If it's a keyword, move over it and any immediately
5763 ;; following clauses associated with it, stopping at the start of
5764 ;; the next token. t is returned in that case, otherwise the point
5765 ;; stays and nil is returned. The kind of clauses that are
5766 ;; recognized are those specified by `c-type-list-kwds',
5767 ;; `c-ref-list-kwds', `c-colon-type-list-kwds',
5768 ;; `c-paren-nontype-kwds', `c-paren-type-kwds', `c-<>-type-kwds',
5769 ;; and `c-<>-arglist-kwds'.
5770 ;;
5771 ;; This function records identifier ranges on
5772 ;; `c-record-type-identifiers' and `c-record-ref-identifiers' if
5773 ;; `c-record-type-identifiers' is non-nil.
5774 ;;
5775 ;; Note that for `c-colon-type-list-kwds', which doesn't necessary
5776 ;; apply directly after the keyword, the type list is moved over
5777 ;; only when there is no unaccounted token before it (i.e. a token
5778 ;; that isn't moved over due to some other keyword list). The
5779 ;; identifier ranges in the list are still recorded if that should
5780 ;; be done, though.
5781 ;;
5782 ;; This function might do hidden buffer changes.
5785 ;; The call to `c-forward-<>-arglist' below is made after
5786 ;; `c-<>-sexp-kwds' keywords, so we're certain they actually
5787 ;; are angle bracket arglists and `c-restricted-<>-arglists'
5788 ;; should therefore be nil.
5800 ;; There's a type directly after a keyword in `c-type-list-kwds'.
5805 ;; There's a name directly after a keyword in `c-ref-list-kwds'.
5810 ;; There's an open paren after a keyword in `c-paren-any-kwds'.
5817 ;; Use `c-forward-type' on every identifier we can find
5818 ;; inside the paren, to record the types.
5843 ;; If we are at the colon already, we move over the type
5844 ;; list after it.
5850 ;; Not at the colon, so stop here. But the identifier
5851 ;; ranges in the type list later on should still be
5852 ;; recorded.
5855 ;; If a keyword matched both one of the types above and
5856 ;; this one, we match `c-colon-type-list-re' after the
5857 ;; clause matched above.
5864 ;; There's a type after the `c-colon-type-list-re' match
5865 ;; after a keyword in `c-colon-type-list-kwds'.
5869 t)))
5871 ;; cc-mode requires cc-fonts.
5875 ;; The point is assumed to be at a "<". Try to treat it as the open
5876 ;; paren of an angle bracket arglist and move forward to the
5877 ;; corresponding ">". If successful, the point is left after the
5878 ;; ">" and t is returned, otherwise the point isn't moved and nil is
5879 ;; returned. If ALL-TYPES is t then all encountered arguments in
5880 ;; the arglist that might be types are treated as found types.
5881 ;;
5882 ;; The variable `c-parse-and-markup-<>-arglists' controls how this
5883 ;; function handles text properties on the angle brackets and argument
5884 ;; separating commas.
5885 ;;
5886 ;; `c-restricted-<>-arglists' controls how lenient the template
5887 ;; arglist recognition should be.
5888 ;;
5889 ;; This function records identifier ranges on
5890 ;; `c-record-type-identifiers' and `c-record-ref-identifiers' if
5891 ;; `c-record-type-identifiers' is non-nil.
5892 ;;
5893 ;; This function might do hidden buffer changes.
5896 ;; If `c-record-type-identifiers' is set then activate
5897 ;; recording of any found types that constitute an argument in
5898 ;; the arglist.
5906 ;; `nconc' doesn't mind that the tail of
5907 ;; `c-record-found-types' is t.
5910 t)
5913 nil)))
5916 ;; Recursive part of `c-forward-<>-arglist'.
5917 ;;
5918 ;; This function might do hidden buffer changes.
5920 ;; Cover this so that any recorded found type ranges are
5921 ;; automatically lost if it turns out to not be an angle
5922 ;; bracket arglist. It's propagated through the return value
5923 ;; on successful completion.
5925 ;; List that collects the positions after the argument
5926 ;; separating ',' in the arglist.
5927 arg-start-pos)
5928 ;; If the '<' has paren open syntax then we've marked it as an angle
5929 ;; bracket arglist before, so skip to the end.
5937 t
5938 ;; Got unmatched paren angle brackets. We don't clear the paren
5939 ;; syntax properties and retry, on the basis that it's very
5940 ;; unlikely that paren angle brackets become operators by code
5941 ;; manipulation. It's far more likely that it doesn't match due
5942 ;; to narrowing or some temporary change.
5944 nil))
5949 ;; go forward one non-alphanumeric character (group) per iteration of
5950 ;; this loop.
5956 ;; All encountered identifiers are types, so set the
5957 ;; promote flag and parse the type.
5980 ;; Note: These regexps exploit the match order in \| so
5981 ;; that "<>" is matched by "<" rather than "[^>:-]>".
5983 ;; Stop on ',', '|', '&', '+' and '-' to catch
5984 ;; common binary operators that could be between
5985 ;; two comparison expressions "a<b" and "c>d".
5987 nil t t))
5991 ;; Either an operator starting with '>' or the end of
5992 ;; the angle bracket arglist.
5999 ;; The angle bracket arglist is finished.
6011 ;; Either an operator starting with '<' or a nested arglist.
6015 ;; The '<' begins a multi-char operator.
6019 ;; We're at a nested <.....>
6025 ;; There's always an identifier before an angle
6026 ;; bracket arglist, or a keyword in `c-<>-type-kwds'
6027 ;; or `c-<>-arglist-kwds'.
6043 ;; It was an angle bracket arglist.
6046 ;; Record the identifier before the template as a type
6047 ;; or reference depending on whether the arglist is last
6048 ;; in a qualified identifier.
6058 ;; At a "less than" operator.
6061 )))
6068 ;; Just another argument. Record the position. The
6069 ;; type check stuff that made us stop at it is at
6070 ;; the top of the loop.
6074 ;; Got a character that can't be in an angle bracket
6075 ;; arglist argument. Abort using `throw', since
6076 ;; it's useless to try to find a surrounding arglist
6077 ;; if we're nested.
6083 ;; The point is assumed to be directly after a ">". Try to treat it
6084 ;; as the close paren of an angle bracket arglist and move back to
6085 ;; the corresponding "<". If successful, the point is left at
6086 ;; the "<" and t is returned, otherwise the point isn't moved and
6087 ;; nil is returned. ALL-TYPES is passed on to
6088 ;; `c-forward-<>-arglist'.
6089 ;;
6090 ;; If the optional LIMIT is given, it bounds the backward search.
6091 ;; It's then assumed to be at a syntactically relevant position.
6092 ;;
6093 ;; This is a wrapper around `c-forward-<>-arglist'. See that
6094 ;; function for more details.
6103 t
6104 ;; See corresponding note in `c-forward-<>-arglist'.
6106 nil)
6113 t
6114 ;; Stopped at bob or a char that isn't allowed in an
6115 ;; arglist, so we've failed.
6117 nil)
6122 ;; If we moved then the "<" was part of some
6123 ;; multicharacter token.
6124 t
6130 ;; Matched the arglist. Break the while.
6132 nil)
6134 ;; We started from a non-paren ">" inside an
6135 ;; arglist.
6137 nil)
6139 ;; Matched a shorter arglist. Can be a nested
6140 ;; one so continue looking.
6142 t))
6143 t))))))
6148 ;; Move forward over a complete name if at the beginning of one,
6149 ;; stopping at the next following token. A keyword, as such,
6150 ;; doesn't count as a name. If the point is not at something that
6151 ;; is recognized as a name then it stays put.
6152 ;;
6153 ;; A name could be something as simple as "foo" in C or something as
6154 ;; complex as "X<Y<class A<int>::B, BIT_MAX >> b>, ::operator<> ::
6155 ;; Z<(a>b)> :: operator const X<&foo>::T Q::G<unsigned short
6156 ;; int>::*volatile const" in C++ (this function is actually little
6157 ;; more than a `looking-at' call in all modes except those that,
6158 ;; like C++, have `c-recognize-<>-arglists' set).
6159 ;;
6160 ;; Return
6161 ;; o - nil if no name is found;
6162 ;; o - 'template if it's an identifier ending with an angle bracket
6163 ;; arglist;
6164 ;; o - 'operator of it's an operator identifier;
6165 ;; o - t if it's some other kind of name.
6166 ;;
6167 ;; This function records identifier ranges on
6168 ;; `c-record-type-identifiers' and `c-record-ref-identifiers' if
6169 ;; `c-record-type-identifiers' is non-nil.
6170 ;;
6171 ;; This function might do hidden buffer changes.
6174 ;; Turn off `c-promote-possible-types' here since we might
6175 ;; call `c-forward-<>-arglist' and we don't want it to promote
6176 ;; every suspect thing in the arglist to a type. We're
6177 ;; typically called from `c-forward-type' in this case, and
6178 ;; the caller only wants the top level type that it finds to
6179 ;; be promoted.
6180 c-promote-possible-types)
6186 ;; Check for keyword. We go to the last symbol in
6187 ;; `c-identifier-key' first.
6200 ;; "template" is only valid inside an
6201 ;; identifier if preceded by "::".
6206 t))
6208 ;; Handle a C++ operator or template identifier.
6212 ;; Got "... ::template".
6216 res subres))))
6219 ;; Got a cast operator.
6223 ;; Now we should match a sequence of either
6224 ;; '*', '&' or a name followed by ":: *",
6225 ;; where each can be followed by a sequence
6226 ;; of `c-opt-type-modifier-key'.
6229 t)
6239 t))))
6247 ;; Got some other operator.
6255 nil)
6257 ;; `id-start' is equal to `id-end' if we've jumped over
6258 ;; an identifier that doesn't end with a symbol token.
6259 ;; That can occur e.g. for Java import directives on the
6260 ;; form "foo.bar.*".
6267 res t)))
6277 ;; Got a concatenated identifier. This handles the
6278 ;; cases with tricky syntactic whitespace that aren't
6279 ;; covered in `c-identifier-key'.
6282 t)
6286 ;; Maybe an angle bracket arglist.
6294 c-last-identifier-range nil)
6299 ;; Continue if there's an identifier concatenation
6300 ;; operator after the template argument.
6306 t)
6311 nil)))
6312 )))))
6315 res))
6318 ;; Move forward over a type spec if at the beginning of one,
6319 ;; stopping at the next following token. The keyword "typedef"
6320 ;; isn't part of a type spec here.
6321 ;;
6322 ;; BRACE-BLOCK-TOO, when non-nil, means move over the brace block in
6323 ;; constructs like "struct foo {...} bar ;" or "struct {...} bar;".
6324 ;; The current (2009-03-10) intention is to convert all uses of
6325 ;; `c-forward-type' to call with this parameter set, then to
6326 ;; eliminate it.
6327 ;;
6328 ;; Return
6329 ;; o - t if it's a known type that can't be a name or other
6330 ;; expression;
6331 ;; o - 'known if it's an otherwise known type (according to
6332 ;; `*-font-lock-extra-types');
6333 ;; o - 'prefix if it's a known prefix of a type;
6334 ;; o - 'found if it's a type that matches one in `c-found-types';
6335 ;; o - 'maybe if it's an identifier that might be a type;
6336 ;; o - 'decltype if it's a decltype(variable) declaration; - or
6337 ;; o - nil if it can't be a type (the point isn't moved then).
6338 ;;
6339 ;; The point is assumed to be at the beginning of a token.
6340 ;;
6341 ;; Note that this function doesn't skip past the brace definition
6342 ;; that might be considered part of the type, e.g.
6343 ;; "enum {a, b, c} foo".
6344 ;;
6345 ;; This function records identifier ranges on
6346 ;; `c-record-type-identifiers' and `c-record-ref-identifiers' if
6347 ;; `c-record-type-identifiers' is non-nil.
6348 ;;
6349 ;; This function might do hidden buffer changes.
6357 ;; Skip leading type modifiers. If any are found we know it's a
6358 ;; prefix of a type.
6377 ; "typedef".
6385 ;; In many languages the name can be used without the
6386 ;; prefix, so we add it to `c-found-types'.
6389 c-last-identifier-range)
6411 id-range c-last-identifier-range))))
6423 id-end)
6425 ;; Looking at a primitive or known type identifier. We've
6426 ;; checked for a name first so that we don't go here if the
6427 ;; known type match only is a prefix of another name.
6438 ;; There might be more keywords for the type.
6466 ;; A normal identifier.
6476 ;; It's an identifier that has been used as
6477 ;; a type somewhere else.
6478 'found
6479 ;; It's an identifier that might be a type.
6482 ;; A template is a type.
6486 ;; Otherwise it's an operator identifier, which is not a type.
6491 ;; Skip trailing type modifiers. If any are found we know it's
6492 ;; a type.
6499 ;; Step over any type suffix operator. Do not let the existence
6500 ;; of these alter the classification of the found type, since
6501 ;; these operators typically are allowed in normal expressions
6502 ;; too.
6509 ;; Look for a trailing operator that concatenates the type
6510 ;; with a following one, and if so step past that one through
6511 ;; a recursive call. Note that we don't record concatenated
6512 ;; types in `c-found-types' - it's the component types that
6513 ;; are recorded when appropriate.
6516 c-promote-possible-types))
6517 ;; If we can't promote then set `c-record-found-types' so that
6518 ;; we can merge in the types from the second part afterwards if
6519 ;; it turns out to be a known type there.
6522 subres)
6531 ;; If either operand certainly is a type then both are, but we
6532 ;; don't let the existence of the operator itself promote two
6533 ;; uncertain types to a certain one.
6552 ;; Merge in the ranges of any types found by the second
6553 ;; `c-forward-type'.
6555 ;; `nconc' doesn't mind that the tail of
6556 ;; `c-record-found-types' is t.
6558 c-record-type-identifiers))))
6566 ;;(message "c-forward-type %s -> %s: %s" start (point) res)
6568 res))
6571 ;; Used for Java code only at the moment. Assumes point is on the
6572 ;; @, moves forward an annotation. returns nil if there is no
6573 ;; annotation at point.
6580 t)))
6583 ;; Pull the next open brace from PS (which has the form of paren-state),
6584 ;; skipping over any brace pairs. Returns NIL when PS is exhausted.
6592 ;; Point is just after a closing brace/parenthesis. Try to parse this as a
6593 ;; C++ member initializer list, going back to just after the introducing ":"
6594 ;; and returning t. Otherwise return nil, leaving point unchanged.
6619 res))
6622 ;; Test whether we are in a C++ member initializer list, and if so, go back
6623 ;; to the introducing ":", returning the position of the opening paren of
6624 ;; the function's arglist. Otherwise return nil, leaving point unchanged.
6627 res)
6665 res))
6667 \f
6668 ;; Handling of large scale constructs like statements and declarations.
6670 ;; Macro used inside `c-forward-decl-or-cast-1'. It ought to be a
6671 ;; defsubst or perhaps even a defun, but it contains lots of free
6672 ;; variables that refer to things inside `c-forward-decl-or-cast-1'.
6674 ;; `c-forward-decl-or-cast-1' can consume an arbitrary length list
6675 ;; of types when parsing a declaration, which means that it
6676 ;; sometimes consumes the identifier in the declaration as a type.
6677 ;; This is used to "backtrack" and make the last type be treated as
6678 ;; an identifier instead.
6681 ;; These identifiers are bound only in the inner let.
6683 identifier-start type-start
6684 got-parens nil
6685 got-identifier t
6686 got-suffix t
6687 got-suffix-after-parens id-start
6691 t
6692 backup-at-type))
6694 id-start backup-id-start)
6696 id-start start-pos))
6698 ;; When these flags already are set we've found specifiers that
6699 ;; unconditionally signal these attributes - backtracking doesn't
6700 ;; change that. So keep them set in that case.
6707 ;; This identifier is bound only in the inner let.
6711 ;; Move forward over a declaration or a cast if at the start of one.
6712 ;; The point is assumed to be at the start of some token. Nil is
6713 ;; returned if no declaration or cast is recognized, and the point
6714 ;; is clobbered in that case.
6715 ;;
6716 ;; If a declaration is parsed:
6717 ;;
6718 ;; The point is left at the first token after the first complete
6719 ;; declarator, if there is one. The return value is a cons where
6720 ;; the car is the position of the first token in the declarator. (See
6721 ;; below for the cdr.)
6722 ;; Some examples:
6723 ;;
6724 ;; void foo (int a, char *b) stuff ...
6725 ;; car ^ ^ point
6726 ;; float (*a)[], b;
6727 ;; car ^ ^ point
6728 ;; unsigned int a = c_style_initializer, b;
6729 ;; car ^ ^ point
6730 ;; unsigned int a (cplusplus_style_initializer), b;
6731 ;; car ^ ^ point (might change)
6732 ;; class Foo : public Bar {}
6733 ;; car ^ ^ point
6734 ;; class PikeClass (int a, string b) stuff ...
6735 ;; car ^ ^ point
6736 ;; enum bool;
6737 ;; car ^ ^ point
6738 ;; enum bool flag;
6739 ;; car ^ ^ point
6740 ;; void cplusplus_function (int x) throw (Bad);
6741 ;; car ^ ^ point
6742 ;; Foo::Foo (int b) : Base (b) {}
6743 ;; car ^ ^ point
6744 ;;
6745 ;; auto foo = 5;
6746 ;; car ^ ^ point
6747 ;; auto cplusplus_11 (int a, char *b) -> decltype (bar):
6748 ;; car ^ ^ point
6749 ;;
6750 ;;
6751 ;;
6752 ;; The cdr of the return value is non-nil when a
6753 ;; `c-typedef-decl-kwds' specifier is found in the declaration.
6754 ;; Specifically it is a dotted pair (A . B) where B is t when a
6755 ;; `c-typedef-kwds' ("typedef") is present, and A is t when some
6756 ;; other `c-typedef-decl-kwds' (e.g. class, struct, enum)
6757 ;; specifier is present. I.e., (some of) the declared
6758 ;; identifier(s) are types.
6759 ;;
6760 ;; If a cast is parsed:
6761 ;;
6762 ;; The point is left at the first token after the closing paren of
6763 ;; the cast. The return value is `cast'. Note that the start
6764 ;; position must be at the first token inside the cast parenthesis
6765 ;; to recognize it.
6766 ;;
6767 ;; PRECEDING-TOKEN-END is the first position after the preceding
6768 ;; token, i.e. on the other side of the syntactic ws from the point.
6769 ;; Use a value less than or equal to (point-min) if the point is at
6770 ;; the first token in (the visible part of) the buffer.
6771 ;;
6772 ;; CONTEXT is a symbol that describes the context at the point:
6773 ;; 'decl In a comma-separated declaration context (typically
6774 ;; inside a function declaration arglist).
6775 ;; '<> In an angle bracket arglist.
6776 ;; 'arglist Some other type of arglist.
6777 ;; nil Some other context or unknown context. Includes
6778 ;; within the parens of an if, for, ... construct.
6779 ;;
6780 ;; LAST-CAST-END is the first token after the closing paren of a
6781 ;; preceding cast, or nil if none is known. If
6782 ;; `c-forward-decl-or-cast-1' is used in succession, it should be
6783 ;; the position after the closest preceding call where a cast was
6784 ;; matched. In that case it's used to discover chains of casts like
6785 ;; "(a) (b) c".
6786 ;;
6787 ;; This function records identifier ranges on
6788 ;; `c-record-type-identifiers' and `c-record-ref-identifiers' if
6789 ;; `c-record-type-identifiers' is non-nil.
6790 ;;
6791 ;; This function might do hidden buffer changes.
6794 ;; first type, i.e. after any leading specifiers. It might
6795 ;; also point at the beginning of the preceding syntactic
6796 ;; whitespace.
6798 ;; Set to the result of `c-forward-type'.
6799 at-type
6800 ;; The position of the first token in what we currently
6801 ;; believe is the type in the declaration or cast, after any
6802 ;; specifiers and their associated clauses.
6803 type-start
6804 ;; The position of the first token in what we currently
6805 ;; believe is the declarator for the first identifier. Set
6806 ;; when the type is found, and moved forward over any
6807 ;; `c-decl-hangon-kwds' and their associated clauses that
6808 ;; occurs after the type.
6809 id-start
6810 ;; These store `at-type', `type-start' and `id-start' of the
6811 ;; identifier before the one in those variables. The previous
6812 ;; identifier might turn out to be the real type in a
6813 ;; declaration if the last one has to be the declarator in it.
6814 ;; If `backup-at-type' is nil then the other variables have
6815 ;; undefined values.
6816 backup-at-type backup-type-start backup-id-start
6817 ;; This stores `kwd-sym' of the symbol before the current one.
6818 ;; This is needed to distinguish the C++11 version of "auto" from
6819 ;; the pre C++11 meaning.
6820 backup-kwd-sym
6821 ;; Set if we've found a specifier (apart from "typedef") that makes
6822 ;; the defined identifier(s) types.
6823 at-type-decl
6824 ;; Set if we've a "typedef" keyword.
6825 at-typedef
6826 ;; Set if we've found a specifier that can start a declaration
6827 ;; where there's no type.
6828 maybe-typeless
6829 ;; Save the value of kwd-sym between loops of the "Check for a
6830 ;; type" loop. Needed to distinguish a C++11 "auto" from a pre
6831 ;; C++11 one.
6832 prev-kwd-sym
6833 ;; If a specifier is found that also can be a type prefix,
6834 ;; these flags are set instead of those above. If we need to
6835 ;; back up an identifier, they are copied to the real flag
6836 ;; variables. Thus they only take effect if we fail to
6837 ;; interpret it as a type.
6838 backup-at-type-decl backup-maybe-typeless
6839 ;; Whether we've found a declaration or a cast. We might know
6840 ;; this before we've found the type in it. It's 'ids if we've
6841 ;; found two consecutive identifiers (usually a sure sign, but
6842 ;; we should allow that in labels too), and t if we've found a
6843 ;; specifier keyword (a 100% sure sign).
6844 at-decl-or-cast
6845 ;; Set when we need to back up to parse this as a declaration
6846 ;; but not as a cast.
6847 backup-if-not-cast
6848 ;; For casts, the return position.
6849 cast-end
6850 ;; Have we got a new-style C++11 "auto"?
6851 new-style-auto
6852 ;; Save `c-record-type-identifiers' and
6853 ;; `c-record-ref-identifiers' since ranges are recorded
6854 ;; speculatively and should be thrown away if it turns out
6855 ;; that it isn't a declaration or cast.
6862 ;; Check for a type. Unknown symbols are treated as possible
6863 ;; types, but they could also be specifiers disguised through
6864 ;; macros like __INLINE__, so we recognize both types and known
6865 ;; specifiers after them too.
6869 ;; Look for a specifier keyword clause.
6882 ;; Found a known or possible type or a prefix of a known type.
6891 ;; Got two identifiers with nothing but whitespace
6892 ;; between them. That can only happen in declarations.
6896 ;; If the previous identifier is a found type we
6897 ;; record it as a real one; it might be some sort of
6898 ;; alias for a prefix like "unsigned".
6905 backup-type-start type-start
6906 backup-id-start id-start
6907 backup-kwd-sym kwd-sym
6908 at-type found-type
6909 type-start start
6911 ;; The previous ambiguous specifier/type turned out
6912 ;; to be a type since we've parsed another one after
6913 ;; it, so clear these backup flags.
6914 backup-at-type-decl nil
6915 backup-maybe-typeless nil))
6919 ;; Handle known specifier keywords and
6920 ;; `c-decl-hangon-kwds' which can occur after known
6921 ;; types.
6924 ;; It's a hang-on keyword that can occur anywhere.
6928 ;; Move the identifier start position if
6929 ;; we've passed a type.
6931 ;; Otherwise treat this as a specifier and
6932 ;; move the fallback position.
6936 ;; It's an ordinary specifier so we know that
6937 ;; anything before this can't be the type.
6939 start-pos kwd-clause-end)
6942 ;; It's ambiguous whether this keyword is a
6943 ;; specifier or a type prefix, so set the backup
6944 ;; flags. (It's assumed that `c-forward-type'
6945 ;; moved further than `c-forward-keyword-clause'.)
6953 ;; This test only happens after we've scanned a type.
6954 ;; So, with valid syntax, kwd-sym can't be 'typedef.
6959 ;; Haven't matched a type so it's an unambiguous
6960 ;; specifier keyword and we know we're in a
6961 ;; declaration.
6967 ;; If the type isn't known we continue so that we'll jump
6968 ;; over all specifiers and type identifiers. The reason
6969 ;; to do this for a known type prefix is to make things
6970 ;; like "unsigned INT16" work.
6975 ;; If a known type was found, we still need to skip over any
6976 ;; hangon keyword clauses after it. Otherwise it has already
6977 ;; been done in the loop above.
6983 ;; A prefix type is itself a primitive type when it's not
6984 ;; followed by another type.
6988 ;; Got no type but set things up to continue anyway to handle
6989 ;; the various cases when a declaration doesn't start with a
6990 ;; type.
6995 ;; If it's C++ then check if the last "type" ends on the form
6996 ;; "foo::foo" or "foo::~foo", i.e. if it's the name of a
6997 ;; (con|de)structor.
7008 ;; Cheating in the handling of syntactic ws below.
7015 name))
7016 ;; It is a (con|de)structor name. In that case the
7017 ;; declaration is typeless so zap out any preceding
7018 ;; identifier(s) that we might have taken as types.
7021 backup-at-type nil
7022 id-start type-start))))))
7024 ;; Check for and step over a type decl expression after the thing
7025 ;; that is or might be a type. This can't be skipped since we
7026 ;; need the correct end position of the declarator for
7027 ;; `max-type-decl-end-*'.
7029 ;; True if there's a non-open-paren match of
7030 ;; `c-type-decl-prefix-key'.
7031 got-prefix
7032 ;; True if the declarator is surrounded by a parenthesis pair.
7033 got-parens
7034 ;; True if there is an identifier in the declarator.
7035 got-identifier
7036 ;; True if there's a non-close-paren match of
7037 ;; `c-type-decl-suffix-key'.
7038 got-suffix
7039 ;; True if there's a prefix match outside the outermost
7040 ;; paren pair that surrounds the declarator.
7041 got-prefix-before-parens
7042 ;; True if there's a suffix match outside the outermost
7043 ;; paren pair that surrounds the declarator. The value is
7044 ;; the position of the first suffix match.
7045 got-suffix-after-parens
7046 ;; True if we've parsed the type decl to a token that is
7047 ;; known to end declarations in this context.
7048 at-decl-end
7049 ;; The earlier values of `at-type' and `type-start' if we've
7050 ;; shifted the type backwards.
7051 identifier-type identifier-start
7052 ;; If `c-parse-and-markup-<>-arglists' is set we need to
7053 ;; turn it off during the name skipping below to avoid
7054 ;; getting `c-type' properties that might be bogus. That
7055 ;; can happen since we don't know if
7056 ;; `c-restricted-<>-arglists' will be correct inside the
7057 ;; arglist paren that gets entered.
7058 c-parse-and-markup-<>-arglists
7059 ;; Start of the identifier for which `got-identifier' was set.
7060 name-start)
7064 ;; Skip over type decl prefix operators. (Note similar code in
7065 ;; `c-font-lock-declarators'.)
7075 ;; If the third submatch matches in C++ then
7076 ;; we're looking at an identifier that's a
7077 ;; prefix only if it specifies a member pointer.
7082 ;; We only check for a trailing "::" and
7083 ;; let the "*" that should follow be
7084 ;; matched in the next round.
7086 ;; It turned out to be the real identifier,
7087 ;; so stop.
7088 nil))
7089 t))
7103 ;; Skip over an identifier.
7110 ;; Skip over type decl suffix operators.
7117 t)
7121 t)
7127 ;; No suffix matched. We might have matched the
7128 ;; identifier as a type and the open paren of a
7129 ;; function arglist as a type decl prefix. In that
7130 ;; case we should "backtrack": Reinterpret the last
7131 ;; type as the identifier, move out of the arglist and
7132 ;; continue searching for suffix operators.
7133 ;;
7134 ;; Do this even if there's no preceding type, to cope
7135 ;; with old style function declarations in K&R C,
7136 ;; (con|de)structors in C++ and `c-typeless-decl-kwds'
7137 ;; style declarations. That isn't applicable in an
7138 ;; arglist context, though.
7143 maybe-typeless
7144 backup-maybe-typeless
7151 t))
7160 at-type))
7161 ;; Have found no identifier but `c-typeless-decl-kwds' has
7162 ;; matched so we know we're inside a declaration. The
7163 ;; preceding type must be the identifier instead.
7166 ;; Prepare the "-> type;" for fontification later on.
7176 at-decl-or-cast
7179 ;; CASE 1
7181 ;; Encountered something inside parens that isn't matched by
7182 ;; the `c-type-decl-*' regexps, so it's not a type decl
7183 ;; expression. Try to skip out to the same paren depth to
7184 ;; not confuse the cast check below.
7186 ;; If we've found a specifier keyword then it's a
7187 ;; declaration regardless.
7195 ;; Now we've collected info about various characteristics of
7196 ;; the construct we're looking at. Below follows a decision
7197 ;; tree based on that. It's ordered to check more certain
7198 ;; signs before less certain ones.
7203 ;; CASE 2
7206 ;; Got another identifier directly after the type, so it's a
7207 ;; declaration.
7212 ;; (not got-suffix-after-parens)
7214 maybe-typeless
7215 backup-maybe-typeless
7220 ;; Got a declaration of the form "foo bar (gnu);" or "bar
7221 ;; (gnu);" where we've recognized "bar" as the type and "gnu"
7222 ;; as the declarator. In this case it's however more likely
7223 ;; that "bar" is the declarator and "gnu" a function argument
7224 ;; or initializer (if `c-recognize-paren-inits' is set),
7225 ;; since the parens around "gnu" would be superfluous if it's
7226 ;; a declarator. Shift the type one step backward.
7229 ;; Found no identifier.
7234 ;; CASE 3
7236 ;; Got a plain list of identifiers. If a colon follows it's
7237 ;; a valid label, or maybe a bitfield. Otherwise the last
7238 ;; one probably is the declared identifier and we should
7239 ;; back up to the previous type, providing it isn't a cast.
7243 ;; If we've found a specifier keyword then it's a
7244 ;; declaration regardless.
7255 ;; CASE 4
7259 ;; Got a plain list of identifiers followed by some suffix.
7260 ;; If this isn't a cast then the last identifier probably is
7261 ;; the declared one and we should back up to the previous
7262 ;; type.
7266 ;; CASE 5
7268 ;; If the type is known we know that there can't be any
7269 ;; identifier somewhere else, and it's only in declarations in
7270 ;; e.g. function prototypes and in casts that the identifier may
7271 ;; be left out.
7275 ;; Only got a single identifier (parsed as a type so far).
7276 ;; CASE 6
7278 ;; Check that the identifier isn't at the start of an
7279 ;; expression.
7280 at-decl-end
7283 ;; Inside an arglist that contains declarations. If K&R
7284 ;; style declarations and parenthesis style initializers
7285 ;; aren't allowed then the single identifier must be a
7286 ;; type, else we require that it's known or found
7287 ;; (primitive types are handled above).
7292 ;; Inside a template arglist. Accept known and found
7293 ;; types; other identifiers could just as well be
7294 ;; constants in C++.
7297 ;; CASE 7
7298 ;; Can't be a valid declaration or cast, but if we've found a
7299 ;; specifier it can't be anything else either, so treat it as
7300 ;; an invalid/unfinished declaration or cast.
7308 maybe-typeless
7309 backup-maybe-typeless
7313 c-after-suffixed-type-maybe-decl-key))))))
7314 ;; Got an empty paren pair and a preceding type that probably
7315 ;; really is the identifier. Shift the type backwards to make
7316 ;; the last one the identifier. This is analogous to the
7317 ;; "backtracking" done inside the `c-type-decl-suffix-key' loop
7318 ;; above.
7319 ;;
7320 ;; Exception: In addition to the conditions in that
7321 ;; "backtracking" code, do not shift backward if we're not
7322 ;; looking at either `c-after-suffixed-type-decl-key' or "[;,]".
7323 ;; Since there's no preceding type, the shift would mean that
7324 ;; the declaration is typeless. But if the regexp doesn't match
7325 ;; then we will simply fall through in the tests below and not
7326 ;; recognize it at all, so it's better to try it as an abstract
7327 ;; declarator instead.
7330 ;; Still no identifier.
7331 ;; CASE 8
7333 ;; Require `got-prefix' together with either `got-parens' or
7334 ;; `got-suffix' to recognize it as an abstract declarator:
7335 ;; `got-parens' only is probably an empty function call.
7336 ;; `got-suffix' only can build an ordinary expression together
7337 ;; with the preceding identifier which we've taken as a type.
7338 ;; We could actually accept on `got-prefix' only, but that can
7339 ;; easily occur temporarily while writing an expression so we
7340 ;; avoid that case anyway. We could do a better job if we knew
7341 ;; the point when the fontification was invoked.
7344 ;; CASE 9
7348 got-suffix-after-parens
7350 ;; Got a type, no declarator but a paren suffix. I.e. it's a
7351 ;; normal function call after all (or perhaps a C++ style object
7352 ;; instantiation expression).
7355 ;; CASE 10
7357 ;; By now we've located the type in the declaration that we know
7358 ;; we're in.
7361 ;; CASE 11
7369 ;; Shift the type backward in the case that there's a
7370 ;; single identifier inside parens. That can only
7371 ;; occur in K&R style function declarations so it's
7372 ;; more likely that it really is a function call.
7373 ;; Therefore we only do this after
7374 ;; `c-after-suffixed-type-decl-key' has matched.
7376 got-suffix-after-parens))
7377 ;; A declaration according to `c-after-suffixed-type-decl-key'.
7380 ;; CASE 12
7383 ;; It's a declaration if a known type precedes it and it can't be a
7384 ;; function call.
7387 ;; If we get here we can't tell if this is a type decl or a normal
7388 ;; expression by looking at it alone. (That's under the assumption
7389 ;; that normal expressions always can look like type decl expressions,
7390 ;; which isn't really true but the cases where it doesn't hold are so
7391 ;; uncommon (e.g. some placements of "const" in C++) it's not worth
7392 ;; the effort to look for them.)
7394 ;;; 2008-04-16: commented out the next form, to allow the function to recognize
7395 ;;; "foo (int bar)" in CC (an implicit type (in class foo) without a semicolon)
7396 ;;; as a(n almost complete) declaration, enabling it to be fontified.
7397 ;; CASE 13
7398 ;; (unless (or at-decl-end (looking-at "=[^=]"))
7399 ;; If this is a declaration it should end here or its initializer(*)
7400 ;; should start here, so check for allowed separation tokens. Note
7401 ;; that this rule doesn't work e.g. with a K&R arglist after a
7402 ;; function header.
7403 ;;
7404 ;; *) Don't check for C++ style initializers using parens
7405 ;; since those already have been matched as suffixes.
7406 ;;
7407 ;; If `at-decl-or-cast' is then we've found some other sign that
7408 ;; it's a declaration or cast, so then it's probably an
7409 ;; invalid/unfinished one.
7410 ;; (throw 'at-decl-or-cast at-decl-or-cast))
7412 ;; Below are tests that only should be applied when we're certain to
7413 ;; not have parsed halfway through an expression.
7415 ;; CASE 14
7417 ;; The expression starts with a known type so treat it as a
7418 ;; declaration.
7421 ;; CASE 15
7423 ;; In C++ we check if the identifier is a known type, since
7424 ;; (con|de)structors use the class name as identifier.
7425 ;; We've always shifted over the identifier as a type and
7426 ;; then backed up again in this case.
7427 identifier-type
7430 ;; `at-type' probably won't be 'found for
7431 ;; destructors since the "~" is then part of the
7432 ;; type name being checked against the list of
7433 ;; known types, so do a check without that
7434 ;; operator.
7439 c-identifier-syntax-table
7443 ;; We have already parsed the type earlier,
7444 ;; so it'd be possible to cache the end
7445 ;; position instead of redoing it here, but
7446 ;; then we'd need to keep track of another
7447 ;; position everywhere.
7455 ;; CASE 16
7457 at-type
7461 ;; Got something like "foo * bar;". Since we're not inside an
7462 ;; arglist it would be a meaningless expression because the
7463 ;; result isn't used. We therefore choose to recognize it as
7464 ;; a declaration. Do not allow a suffix since it could then
7465 ;; be a function call.
7468 ;; CASE 17
7473 ;; Got something like "a (*b) (c);" or "a (b) = c;". It could
7474 ;; be an odd expression or it could be a declaration. Treat
7475 ;; it as a declaration if "a" has been used as a type
7476 ;; somewhere else (if it's a known type we won't get here).
7479 ;; CASE 18
7485 ;; Got a type followed by an abstract declarator. If `got-prefix'
7486 ;; is set it's something like "a *" without anything after it. If
7487 ;; `got-parens' or `got-suffix' is set it's "a()", "a[]", "a()[]",
7488 ;; or similar, which we accept only if the context rules out
7489 ;; expressions.
7492 ;; If we had a complete symbol table here (which rules out
7493 ;; `c-found-types') we should return t due to the disambiguation rule
7494 ;; (in at least C++) that anything that can be parsed as a declaration
7495 ;; is a declaration. Now we're being more defensive and prefer to
7496 ;; highlight things like "foo (bar);" as a declaration only if we're
7497 ;; inside an arglist that contains declarations.
7498 ;; CASE 19
7501 ;; The point is now after the type decl expression.
7504 ;; Check for a cast.
7507 c-cast-parens
7509 ;; Should be the first type/identifier in a cast paren.
7513 ;; The closing paren should follow.
7518 ;; There should be a primary expression after it.
7527 ;; Check if the expression begins with a prefix keyword.
7530 ;; Expression begins with an ambiguous operator. Treat
7531 ;; it as a cast if it's a type decl or if we've
7532 ;; recognized the type somewhere else.
7535 ;; Unless it's a keyword, it's the beginning of a primary
7536 ;; expression.
7538 ;; If `c-primary-expr-regexp' matched a nonsymbol token, check
7539 ;; that it matched a whole one so that we don't e.g. confuse
7540 ;; the operator '-' with '->'. It's ok if it matches further,
7541 ;; though, since it e.g. can match the float '.5' while the
7542 ;; operator regexp only matches '.'.
7546 ;; There should either be a cast before it or something that isn't an
7547 ;; identifier or close paren.
7555 ;; It's a symbol. Accept it only if it's one of the
7556 ;; keywords that can precede an expression (without
7557 ;; surrounding parens).
7560 ;; Check that it isn't a close paren (block close is ok,
7561 ;; though).
7563 ;; Check that it isn't a nonsymbol identifier.
7566 ;; Handle the cast.
7576 ;; We're at a declaration. Highlight the type and the following
7577 ;; declarators.
7583 ;; Make sure to propagate the `c-decl-arg-start' property to
7584 ;; the next argument if it's set in this one, to cope with
7585 ;; interactive refontification.
7588 ;; Record the type's coordinates in `c-record-type-identifiers' for
7589 ;; later fontification.
7591 ;; There seems no reason to exclude a token from
7592 ;; fontification just because it's "a known type that can't
7593 ;; be a name or other expression". 2013-09-18.
7594 )
7605 ;; False alarm. Restore the recorded ranges.
7607 c-record-ref-identifiers save-rec-ref-ids)
7608 nil))))
7611 ;; Assuming that point is at the beginning of a token, check if it starts a
7612 ;; label and if so move over it and return non-nil (t in default situations,
7613 ;; specific symbols (see below) for interesting situations), otherwise don't
7614 ;; move and return nil. "Label" here means "most things with a colon".
7615 ;;
7616 ;; More precisely, a "label" is regarded as one of:
7617 ;; (i) a goto target like "foo:" - returns the symbol `goto-target';
7618 ;; (ii) A case label - either the entire construct "case FOO:", or just the
7619 ;; bare "case", should the colon be missing. We return t;
7620 ;; (iii) a keyword which needs a colon, like "default:" or "private:"; We
7621 ;; return t;
7622 ;; (iv) One of QT's "extended" C++ variants of
7623 ;; "private:"/"protected:"/"public:"/"more:" looking like "public slots:".
7624 ;; Returns the symbol `qt-2kwds-colon'.
7625 ;; (v) QT's construct "signals:". Returns the symbol `qt-1kwd-colon'.
7626 ;; (vi) One of the keywords matched by `c-opt-extra-label-key' (without any
7627 ;; colon). Currently (2006-03), this applies only to Objective C's
7628 ;; keywords "@private", "@protected", and "@public". Returns t.
7629 ;;
7630 ;; One of the things which will NOT be recognized as a label is a bit-field
7631 ;; element of a struct, something like "int foo:5".
7632 ;;
7633 ;; The end of the label is taken to be just after the colon, or the end of
7634 ;; the first submatch in `c-opt-extra-label-key'. The point is directly
7635 ;; after the end on return. The terminating char gets marked with
7636 ;; `c-decl-end' to improve recognition of the following declaration or
7637 ;; statement.
7638 ;;
7639 ;; If ASSUME-MARKUP is non-nil, it's assumed that the preceding
7640 ;; label, if any, has already been marked up like that.
7641 ;;
7642 ;; If PRECEDING-TOKEN-END is given, it should be the first position
7643 ;; after the preceding token, i.e. on the other side of the
7644 ;; syntactic ws from the point. Use a value less than or equal to
7645 ;; (point-min) if the point is at the first token in (the visible
7646 ;; part of) the buffer.
7647 ;;
7648 ;; The optional LIMIT limits the forward scan for the colon.
7649 ;;
7650 ;; This function records the ranges of the label symbols on
7651 ;; `c-record-ref-identifiers' if `c-record-type-identifiers' (!) is
7652 ;; non-nil.
7653 ;;
7654 ;; This function might do hidden buffer changes.
7657 label-end
7658 qt-symbol-idx
7659 macro-start ; if we're in one.
7660 label-type
7661 kwd)
7663 ;; "case" or "default" (Doesn't apply to AWK).
7666 ;; Record only the keyword itself for fontification, since in
7667 ;; case labels the following is a constant expression and not
7668 ;; a label.
7672 ;; Find the label end.
7676 ;; Stop on chars that aren't allowed in expressions,
7677 ;; and on operator chars that would be meaningless
7678 ;; there. FIXME: This doesn't cope with ?: operators.
7686 t)
7688 ;; It's an unfinished label. We consider the keyword enough
7689 ;; to recognize it as a label, so that it gets fontified.
7690 ;; Leave the point at the end of it, but don't put any
7691 ;; `c-decl-end' marker.
7693 t))))
7695 ;; @private, @protected, @public, in Objective C, or similar.
7698 ;; For a `c-opt-extra-label-key' match, we record the whole
7699 ;; thing for fontification. That's to get the leading '@' in
7700 ;; Objective-C protection labels fontified.
7707 ;; All other cases of labels.
7710 ;; A colon label must have something before the colon.
7713 ;; Check that we're not after a token that can't precede a label.
7715 ;; Trivially succeeds when there's no preceding token.
7716 ;; Succeeds when we're at a virtual semicolon.
7725 ;; Check if we're after a label, if we're after a closing
7726 ;; paren that belong to statement, and with
7727 ;; `c-label-prefix-re'. It's done in different order
7728 ;; depending on `assume-markup' since the checks have
7729 ;; different expensiveness.
7756 ;; Might be after another label, so check it recursively.
7760 ;; Essentially the same as the
7761 ;; `c-syntactic-re-search-forward' regexp below.
7767 ;; Note: the following should work instead of the
7768 ;; narrow-to-region above. Investigate why not,
7769 ;; sometime. ACM, 2006-03-31.
7770 ;; (c-syntactic-skip-backward "^-]:?;}=*/%&|,<>!@+"
7771 ;; macro-start t)
7773 ;; If the caller turned on recording for us,
7774 ;; it shouldn't apply when we check the
7775 ;; preceding label.
7776 c-record-type-identifiers)
7777 ;; A label can't start at a cpp directive. Check for
7778 ;; this, since c-forward-syntactic-ws would foul up on it.
7783 ;; Point is still at the beginning of the possible label construct.
7784 ;;
7785 ;; Check that the next nonsymbol token is ":", or that we're in one
7786 ;; of QT's "slots" declarations. Allow '(' for the sake of macro
7787 ;; arguments. FIXME: Should build this regexp from the language
7788 ;; constants.
7790 ;; public: protected: private:
7800 ;; QT double keyword like "protected slots:" or goto target.
7819 'qt-1kwd-colon
7831 ;; Check that `c-nonlabel-token-key' doesn't match anywhere.
7841 t)
7844 ;; Record the identifiers in the label for fontification, unless
7845 ;; it begins with `c-label-kwds' in which case the following
7846 ;; identifiers are part of a (constant) expression that
7847 ;; shouldn't be fontified.
7859 ;; Not a label.
7861 label-type))
7864 ;; Assuming the point is at the beginning of a token, try to move
7865 ;; forward to the end of the Objective-C directive that starts
7866 ;; there. Return t if a directive was fully recognized, otherwise
7867 ;; the point is moved as far as one could be successfully parsed and
7868 ;; nil is returned.
7869 ;;
7870 ;; This function records identifier ranges on
7871 ;; `c-record-type-identifiers' and `c-record-ref-identifiers' if
7872 ;; `c-record-type-identifiers' is non-nil.
7873 ;;
7874 ;; This function might do hidden buffer changes.
7877 start-char
7879 lim
7880 ;; Turn off recognition of angle bracket arglists while parsing
7881 ;; types here since the protocol reference list might then be
7882 ;; considered part of the preceding name or superclass-name.
7893 t)
7902 ;; Handle the name of the class itself.
7904 ;; (c-forward-token-2) ; 2006/1/13 This doesn't move if the token's
7905 ;; at EOB.
7912 ;; Look for ": superclass-name" or "( category-name )".
7923 ;; Look for a protocol reference list.
7929 t))))
7935 t)
7938 nil)))
7941 ;; Go to the first non-whitespace after the colon that starts a
7942 ;; multiple inheritance introduction. Optional LIM is the farthest
7943 ;; back we should search.
7944 ;;
7945 ;; This function might do hidden buffer changes.
7953 ;; Return nil if we aren't in a method definition, otherwise the
7954 ;; position of the initial [+-].
7955 ;;
7956 ;; This function might do hidden buffer changes.
7962 ))
7964 ;; Contributed by Kevin Ryde <user42@zip.com.au>.
7966 ;; Return non-nil if point is within a gcc \"asm\" block.
7967 ;;
7968 ;; This should be called with point inside an argument list.
7969 ;;
7970 ;; Only one level of enclosing parentheses is considered, so for
7971 ;; instance `nil' is returned when in a function call within an asm
7972 ;; operand.
7973 ;;
7974 ;; This function might do hidden buffer changes.
7986 a function.
7988 More precisely, being at the top-level means that point is either
7989 outside any enclosing block (such as a function definition), or
7990 directly inside a class, namespace or other block that contains
7991 another declaration level.
7993 If point is not at the top-level (e.g. it is inside a method
7994 definition), then nil is returned. Otherwise, if point is at a
7995 top-level not enclosed within a class definition, t is returned.
7996 Otherwise, a 2-vector is returned where the zeroth element is the
7997 buffer position of the start of the class declaration, and the first
7998 element is the buffer position of the enclosing class's opening
7999 brace.
8001 Note that this function might do hidden buffer changes. See the
8002 comment at the start of cc-engine.el for more info."
8008 ;; Return non-nil if the point is in the region after the argument
8009 ;; list of a function and its opening brace (or semicolon in case it
8010 ;; got no body). If there are K&R style argument declarations in
8011 ;; that region, the point has to be inside the first one for this
8012 ;; function to recognize it.
8013 ;;
8014 ;; If successful, the point is moved to the first token after the
8015 ;; function header (see `c-forward-decl-or-cast-1' for details) and
8016 ;; the position of the opening paren of the function arglist is
8017 ;; returned.
8018 ;;
8019 ;; The point is clobbered if not successful.
8020 ;;
8021 ;; LIM is used as bound for backward buffer searches.
8022 ;;
8023 ;; This function might do hidden buffer changes.
8036 ;; There should not be a '=' or ',' between beg and the
8037 ;; start of the declaration since that means we were in the
8038 ;; "expression part" of the declaration.
8043 ;; Check that there's an arglist paren in the
8044 ;; declaration.
8047 ;; The declarator is a paren expression, so skip past it
8048 ;; so that we don't get stuck on that instead of the
8049 ;; function arglist.
8053 ;; Don't trip up on "operator ()".
8060 ;; Return the position of the first argument declaration if point is
8061 ;; inside a K&R style argument declaration list, nil otherwise.
8062 ;; `c-recognize-knr-p' is not checked. If LIM is non-nil, it's a
8063 ;; position that bounds the backward search for the argument list.
8064 ;;
8065 ;; Point must be within a possible K&R region, e.g. just before a top-level
8066 ;; "{". It must be outside of parens and brackets. The test can return
8067 ;; false positives otherwise.
8068 ;;
8069 ;; This function might do hidden buffer changes.
8073 ;; If we're in a macro, our search range is restricted to it. Narrow to
8074 ;; the searchable range.
8078 before-lparen after-rparen
8080 ; we give up
8083 ;; Search backwards for the defun's argument list. We give up if we
8084 ;; encounter a "}" (end of a previous defun) an "=" (which can't be in
8085 ;; a knr region) or BOB.
8086 ;;
8087 ;; The criterion for a paren structure being the arg list is:
8088 ;; o - there is non-WS stuff after it but before any "{"; AND
8089 ;; o - the token after it isn't a ";" AND
8090 ;; o - it is preceded by either an identifier (the function name) or
8091 ;; a macro expansion like "DEFUN (...)"; AND
8092 ;; o - its content is a non-empty comma-separated list of identifiers
8093 ;; (an empty arg list won't have a knr region).
8094 ;;
8095 ;; The following snippet illustrates these rules:
8096 ;; int foo (bar, baz, yuk)
8097 ;; int bar [] ;
8098 ;; int (*baz) (my_type) ;
8099 ;; int (*) (void) (*yuk) (void) ;
8100 ;; {
8111 ; parens/brackets.
8115 ;; We're inside a paren. Could it be our argument list....?
8121 ;; FIXME!!! What about macros between the parens? 2007/01/20
8124 ;; It can't be the arg list if next token is ; or {
8129 ;; Is the thing preceding the list an identifier (the
8130 ;; function name), or a macro expansion?
8140 ;; Have we got a non-empty list of comma-separated
8141 ;; identifiers?
8155 ;; ...Yes. We've identified the function's argument list.
8161 ;; ...No. The current parens aren't the function's arg list.
8168 ;; skip forward over conditional at point, including any predicate
8169 ;; statements in parentheses. No error checking is performed.
8170 ;;
8171 ;; This function might do hidden buffer changes.
8173 ;; else if()
8178 ;; do, else, try, finally
8183 ;; for, if, while, switch, catch, synchronized, foreach
8187 ;; If looking at the token after a conditional then return the
8188 ;; position of its start, otherwise return nil.
8189 ;;
8190 ;; This function might do hidden buffer changes.
8201 ;; If the point is after an operator identifier that isn't handled
8202 ;; like an ordinary symbol (i.e. like "operator =" in C++) then the
8203 ;; position of the start of that identifier is returned. nil is
8204 ;; returned otherwise. The point may be anywhere in the syntactic
8205 ;; whitespace after the last token of the operator identifier.
8206 ;;
8207 ;; This function might do hidden buffer changes.
8219 ;; Assuming point is at a brace that opens a statement block of some
8220 ;; kind, move to the proper anchor point for that block. It might
8221 ;; need to be adjusted further by c-add-stmt-syntax, but the
8222 ;; position at return is suitable as start position for that
8223 ;; function.
8224 ;;
8225 ;; This function might do hidden buffer changes.
8232 ;; Assuming point is at a brace that opens the block of a top level
8233 ;; declaration of some kind, move to the proper anchor point for
8234 ;; that block.
8235 ;;
8236 ;; This function might do hidden buffer changes.
8241 ;; Search forward for the end of the "header" of the current
8242 ;; declaration. That's the position where the definition body
8243 ;; starts, or the first variable initializer, or the ending
8244 ;; semicolon. I.e. search forward for the closest following
8245 ;; (syntactically relevant) '{', '=' or ';' token. Point is left
8246 ;; _after_ the first found token, or at point-max if none is found.
8247 ;;
8248 ;; This function might do hidden buffer changes.
8253 ;; In C++ we need to take special care to handle operator
8254 ;; tokens and those pesky template brackets.
8259 ;; Handle operator identifiers, i.e. ignore any
8260 ;; operator token preceded by "operator".
8267 t
8269 nil)))))
8278 ;; Go to the beginning of the current declaration, or the beginning
8279 ;; of the previous one if already at the start of it. Point won't
8280 ;; be moved out of any surrounding paren. Return a cons cell of the
8281 ;; form (MOVE . KNR-POS). MOVE is like the return value from
8282 ;; `c-beginning-of-statement-1'. If point skipped over some K&R
8283 ;; style argument declarations (and they are to be recognized) then
8284 ;; KNR-POS is set to the start of the first such argument
8285 ;; declaration, otherwise KNR-POS is nil. If LIM is non-nil, it's a
8286 ;; position that bounds the backward search.
8287 ;;
8288 ;; NB: Cases where the declaration continues after the block, as in
8289 ;; "struct foo { ... } bar;", are currently recognized as two
8290 ;; declarations, e.g. "struct foo { ... }" and "bar;" in this case.
8291 ;;
8292 ;; This function might do hidden buffer changes.
8298 ;; `c-beginning-of-statement-1' stops at a block start, but we
8299 ;; want to continue if the block doesn't begin a top level
8300 ;; construct, i.e. if it isn't preceded by ';', '}', ':', bob,
8301 ;; or an open paren.
8303 ;; Go back one "statement" each time round the loop until we're just
8304 ;; after a ;, }, or :, or at BOB or the start of a macro or start of
8305 ;; an ObjC method. This will move over a multiple declaration whose
8306 ;; components are comma separated.
8308 ;; Must check with c-opt-method-key in ObjC mode.
8318 ;; Check that we don't move from the first thing in a
8319 ;; macro to its header.
8325 move tentative-move))
8330 ;; Handle K&R argdecls. Back up after the "statement" jumped
8331 ;; over by `c-beginning-of-statement-1', unless it was the
8332 ;; function body, in which case we're sitting on the opening
8333 ;; brace now. Then test if we're in a K&R argdecl region and
8334 ;; that we started at the other side of the first argdecl in
8335 ;; it.
8345 'previous
8347 knr-argdecl-start))
8350 ;; `c-beginning-of-statement-1' counts each brace block as a separate
8351 ;; statement, so the result will be 'previous if we've moved over any.
8352 ;; So change our result back to 'same if necessary.
8353 ;;
8354 ;; If they were brace list initializers we might not have moved over a
8355 ;; declaration boundary though, so change it to 'same if we've moved
8356 ;; past a '=' before '{', but not ';'. (This ought to be integrated
8357 ;; into `c-beginning-of-statement-1', so we avoid this extra pass which
8358 ;; potentially can search over a large amount of text.). Take special
8359 ;; pains not to get mislead by C++'s "operator=", and the like.
8362 c++-template-syntax-table
8368 ; no more, or get to one which isn't an "operator ="
8371 c-overloadable-operators-regexp
8372 c-opt-op-identifier-prefix
8387 ;; Assuming point is at the start of a declaration (as detected by
8388 ;; e.g. `c-beginning-of-decl-1'), go to the end of it. Unlike
8389 ;; `c-beginning-of-decl-1', this function handles the case when a
8390 ;; block is followed by identifiers in e.g. struct declarations in C
8391 ;; or C++. If a proper end was found then t is returned, otherwise
8392 ;; point is moved as far as possible within the current sexp and nil
8393 ;; is returned. This function doesn't handle macros; use
8394 ;; `c-end-of-macro' instead in those cases.
8395 ;;
8396 ;; This function might do hidden buffer changes.
8399 c++-template-syntax-table
8407 ;; Stopped at the ';' in a K&R argdecl section which is
8408 ;; detected using the same criteria as in
8409 ;; `c-beginning-of-decl-1'. Move to the following block
8410 ;; start.
8414 ;; Encountered a block in the declaration. Jump over it.
8425 ;; Check for `c-opt-block-decls-with-vars-key'
8426 ;; before the first paren.
8429 c-opt-block-decls-with-vars-key
8431 lim t t t)
8434 ;; Check that the first following paren is
8435 ;; the block.
8437 lim t t t)
8439 ;; The declaration doesn't have any of the
8440 ;; `c-opt-block-decls-with-vars' keywords in the
8441 ;; beginning, so it ends here at the end of the block.
8449 nil)))
8452 ;; Assuming the point is at an open brace, check if it starts a
8453 ;; block that contains another declaration level, i.e. that isn't a
8454 ;; statement block or a brace list, and if so return non-nil.
8455 ;;
8456 ;; If the check is successful, the return value is the start of the
8457 ;; keyword that tells what kind of construct it is, i.e. typically
8458 ;; what `c-decl-block-key' matched. Also, if GOTO-START is set then
8459 ;; the point will be at the start of the construct, before any
8460 ;; leading specifiers, otherwise it's at the returned position.
8461 ;;
8462 ;; The point is clobbered if the check is unsuccessful.
8463 ;;
8464 ;; CONTAINING-SEXP is the position of the open of the surrounding
8465 ;; paren, or nil if none.
8466 ;;
8467 ;; The optional LIMIT limits the backward search for the start of
8468 ;; the construct. It's assumed to be at a syntactically relevant
8469 ;; position.
8470 ;;
8471 ;; If any template arglists are found in the searched region before
8472 ;; the open brace, they get marked with paren syntax.
8473 ;;
8474 ;; This function might do hidden buffer changes.
8481 ;; Could be at the end of a template arglist.
8489 ;; Note: Can't get bogus hits inside template arglists below since they
8490 ;; have gotten paren syntax above.
8492 ;; If `goto-start' is set we begin by searching for the
8493 ;; first possible position of a leading specifier list.
8494 ;; The `c-decl-block-key' search continues from there since
8495 ;; we know it can't match earlier.
8498 open-brace t t)
8500 t)
8501 t)
8507 ;; Exclude cases where we matched what would ordinarily
8508 ;; be a block declaration keyword, except where it's not
8509 ;; legal because it's part of a "compound keyword" like
8510 ;; "enum class". Of course, if c-after-brace-list-key
8511 ;; is nil, we can skip the test.
8521 ;; Found a keyword that can't be a type?
8524 ;; Can be a type too, in which case it's the return type of a
8525 ;; function (under the assumption that no declaration level
8526 ;; block construct starts with a type).
8529 ;; Jumped over a type, but it could be a declaration keyword
8530 ;; followed by the declared identifier that we've jumped over
8531 ;; instead (e.g. in "class Foo {"). If it indeed is a type
8532 ;; then we should be at the declarator now, so check for a
8533 ;; valid declarator start.
8534 ;;
8535 ;; Note: This doesn't cope with the case when a declared
8536 ;; identifier is followed by e.g. '(' in a language where '('
8537 ;; also might be part of a declarator expression. Currently
8538 ;; there's no such language.
8542 ;; In Pike a list of modifiers may be followed by a brace
8543 ;; to make them apply to many identifiers. Note that the
8544 ;; match data will be empty on return in this case.
8550 ;; Use this variant to avoid yet another special regexp.
8554 t)))
8556 ;; Got a match.
8559 ;; Back up over any preceding specifiers and their clauses
8560 ;; by going forward from `first-specifier-pos', which is the
8561 ;; earliest possible position where the specifier list can
8562 ;; start.
8568 ;; Accept any plain symbol token on the ground that
8569 ;; it's a specifier masked through a macro (just
8570 ;; like `c-forward-decl-or-cast-1' skip forward over
8571 ;; such tokens).
8572 ;;
8573 ;; Could be more restrictive wrt invalid keywords,
8574 ;; but that'd only occur in invalid code so there's
8575 ;; no use spending effort on it.
8581 ;; Can't parse a declaration preamble and is still
8582 ;; before `kwd-start'. That means `first-specifier-pos'
8583 ;; was in some earlier construct. Search again.
8587 ;; Got no preamble before the block declaration keyword.
8593 kwd-start)))
8596 ;; Check if the closest containing paren sexp is a declaration
8597 ;; block, returning a 2 element vector in that case. Aref 0
8598 ;; contains the bufpos at boi of the class key line, and aref 1
8599 ;; contains the bufpos of the open brace. This function is an
8600 ;; obsolete wrapper for `c-looking-at-decl-block'.
8601 ;;
8602 ;; This function might do hidden buffer changes.
8610 nil))
8615 ;; Return the buffer position of the most enclosing decl-block brace (in the
8616 ;; sense of c-looking-at-decl-block) in the PAREN-STATE structure, or nil if
8617 ;; none was found.
8626 open-brace))
8629 ;; Return the position of the L-brace if point is inside a brace list
8630 ;; initialization of an array, etc. This is an approximate function,
8631 ;; designed for speed over accuracy. It will not find every bracelist, but
8632 ;; a non-nil result is reliable. We simply search for "= {" (naturally with
8633 ;; syntactic whitespace allowed). PAREN-STATE is the normal thing that it
8634 ;; is everywhere else.
8643 b-pos)))
8646 ;; We're at the token after what might be a type prefixed with a colon. Try
8647 ;; moving backward over this type and the colon. On success, return t and
8648 ;; leave point before colon; on failure, leave point unchanged. Will clobber
8649 ;; match data.
8666 nil)
8668 t)
8670 t)
8674 t)))
8677 ;; We're at a "{". Move back to the enum-like keyword that starts this
8678 ;; declaration and return t, otherwise don't move and return nil.
8680 up-sexp-pos before-identifier)
8696 t)
8702 t)
8708 ;; return the buffer position of the beginning of the brace list
8709 ;; statement if we're inside a brace list, otherwise return nil.
8710 ;; CONTAINING-SEXP is the buffer pos of the innermost containing
8711 ;; paren. PAREN-STATE is the remainder of the state of enclosing
8712 ;; braces
8713 ;;
8714 ;; N.B.: This algorithm can potentially get confused by cpp macros
8715 ;; placed in inconvenient locations. It's a trade-off we make for
8716 ;; speed.
8717 ;;
8718 ;; This function might do hidden buffer changes.
8720 ;; This will pick up brace list declarations.
8724 ;; this will pick up array/aggregate init lists, even if they are nested.
8727 ;; Pike can have class definitions anywhere, so we must
8728 ;; check for the class key here.
8730 c-decl-block-key))
8731 bufpos braceassignp lim next-containing macro-start)
8733 containing-sexp)
8744 ;; We're in an in-expression block of some kind. Do not
8745 ;; check nesting. We deliberately set the limit to the
8746 ;; containing sexp, so that c-looking-at-inexpr-block
8747 ;; doesn't check for an identifier before it.
8749 ;; see if the open brace is preceded by = or [...] in
8750 ;; this statement, but watch out for operator=
8753 ;; Checks to do only on the first sexp before the brace.
8756 ;; In Java, an initialization brace list may follow
8757 ;; directly after "new Foo[]", so check for a "new"
8758 ;; earlier.
8764 ;; Carry on looking if this is an
8765 ;; identifier (may contain "." in Java)
8766 ;; or another "[]" sexp.
8769 ;; Checks to do on all sexps before the brace, up to the
8770 ;; beginning of the statement.
8778 ;; We've seen a =, but must check earlier tokens so
8779 ;; that it isn't something that should be ignored.
8785 ;; Check for operator =
8788 nil)
8789 ;; Check for `<opchar>= in Pike.
8792 ;; Special case for Pikes
8793 ;; `[]=, since '[' is not in
8794 ;; the punctuation class.
8797 nil)
8799 ;; make sure we're not in a C++ template
8800 ;; argument assignment
8812 ))))
8813 nil)
8820 ;; We've hit the beginning of the aggregate list.
8825 ;; Brace lists can't contain a semicolon, so we're done.
8830 ;; We've a macro whose expansion starts with the '{'.
8831 ;; Heuristically, if we have a ';' in it we've not got a
8832 ;; brace list, otherwise we have.
8839 containing-sexp nil)
8842 ;; Go up one level
8844 lim nil
8845 next-containing nil)))))
8847 bufpos))
8848 ))
8851 ;; If we're looking at the start of a pike-style list, i.e., `({Â })',
8852 ;; `([Â ])', `(<Â >)', etc., a cons of a cons of its starting and ending
8853 ;; positions and its entry in c-special-brace-lists is returned, nil
8854 ;; otherwise. The ending position is nil if the list is still open.
8855 ;; LIM is the limit for forward search. The point may either be at
8856 ;; the `(' or at the following paren character. Tries to check the
8857 ;; matching closer, but assumes it's correct if no balanced paren is
8858 ;; found (i.e. the case `({ ... } ... )' is detected as _not_ being
8859 ;; a special brace list).
8860 ;;
8861 ;; This function might do hidden buffer changes.
8866 inner-beg end type)
8881 nil)))))
8891 ;; Check balancing of the inner paren
8892 ;; below.
8895 t)
8896 ;; If the inner char isn't a paren then
8897 ;; we can't check balancing, so just
8898 ;; check the char before the outer
8899 ;; closing paren.
8914 ;; Return non-nil if between two statements or declarations, assuming
8915 ;; point is not inside a literal or comment.
8916 ;;
8917 ;; Obsolete - `c-at-statement-start-p' or `c-at-expression-start-p'
8918 ;; are recommended instead.
8919 ;;
8920 ;; This function might do hidden buffer changes.
8925 ;; Return non-nil if we're looking at the beginning of a block
8926 ;; inside an expression. The value returned is actually a cons of
8927 ;; either 'inlambda, 'inexpr-statement or 'inexpr-class and the
8928 ;; position of the beginning of the construct.
8929 ;;
8930 ;; LIM limits the backward search. CONTAINING-SEXP is the start
8931 ;; position of the closest containing list. If it's nil, the
8932 ;; containing paren isn't used to decide whether we're inside an
8933 ;; expression or not. If both LIM and CONTAINING-SEXP are used, LIM
8934 ;; needs to be farther back.
8935 ;;
8936 ;; If CHECK-AT-END is non-nil then extra checks at the end of the
8937 ;; brace block might be done. It should only be used when the
8938 ;; construct can be assumed to be complete, i.e. when the original
8939 ;; starting position was further down than that.
8940 ;;
8941 ;; This function might do hidden buffer changes.
8946 ;; Look at the character after point only as a last resort
8947 ;; when we can't disambiguate.
8967 ;; If the class definition is at the start of
8968 ;; a statement, we don't consider it an
8969 ;; in-expression class.
8977 ;; Also, in Pike we treat it as an
8978 ;; in-expression class if it's used in an
8979 ;; object clone expression.
8995 nil)
9003 ;; Accept several square bracket sexps for
9004 ;; Java array initializations.
9012 block-follows
9013 containing-sexp
9022 nil
9025 res))))
9028 ;; Returns non-nil if we're looking at the end of an in-expression
9029 ;; block, otherwise the same as `c-looking-at-inexpr-block'.
9030 ;; PAREN-STATE is the paren state relevant at the current position.
9031 ;;
9032 ;; This function might do hidden buffer changes.
9034 ;; We currently only recognize a block.
9037 containing-sexp)
9046 paren-state)
9047 containing-sexp)))))
9050 ;; Is there a "virtual semicolon" at POS or point?
9051 ;; (See cc-defs.el for full details of "virtual semicolons".)
9052 ;;
9053 ;; This is true when point is at the last non syntactic WS position on the
9054 ;; line, there is a macro call last on the line, and this particular macro's
9055 ;; name is defined by the regexp `c-vs-macro-regexp' as not needing a
9056 ;; semicolon.
9064 c-macro-with-semi-re
9067 ;; Check we've got nothing after this except comments and empty lines
9068 ;; joined by escaped EOLs.
9081 nil)
9089 ;; Check for one of the listed macros being before point.
9093 t))
9101 \f
9102 ;; `c-guess-basic-syntax' and the functions that precedes it below
9103 ;; implements the main decision tree for determining the syntactic
9104 ;; analysis of the current line of code.
9106 ;; Dynamically bound to t when `c-guess-basic-syntax' is called during
9107 ;; auto newline analysis.
9111 ;; BRACEPOS is the position of a brace in a construct like "namespace
9112 ;; Bar {". Return the anchor point in this construct; this is the
9113 ;; earliest symbol on the brace's line which isn't earlier than
9114 ;; "namespace".
9115 ;;
9116 ;; Currently (2007-08-17), "like namespace" means "matches
9117 ;; c-other-block-decl-kwds". It doesn't work with "class" or "struct"
9118 ;; or anything like that.
9128 ;; A simple function to prepend a new syntax element to
9129 ;; `c-syntactic-context'. Using `setq' on it is unsafe since it
9130 ;; should always be dynamically bound but since we read it first
9131 ;; we'll fail properly anyway if this function is misused.
9133 c-syntactic-context)))
9136 ;; Like `c-add-syntax' but appends to the end of the syntax list.
9137 ;; (Normally not necessary.)
9142 syntax-extra-args
9143 stop-at-boi-only
9144 containing-sexp
9145 paren-state)
9146 ;; Add the indicated SYNTAX-SYMBOL to `c-syntactic-context', extending it as
9147 ;; needed with further syntax elements of the types `substatement',
9148 ;; `inexpr-statement', `arglist-cont-nonempty', `statement-block-intro', and
9149 ;; `defun-block-intro'.
9150 ;;
9151 ;; Do the generic processing to anchor the given syntax symbol on
9152 ;; the preceding statement: Skip over any labels and containing
9153 ;; statements on the same line, and then search backward until we
9154 ;; find a statement or block start that begins at boi without a
9155 ;; label or comment.
9156 ;;
9157 ;; Point is assumed to be at the prospective anchor point for the
9158 ;; given SYNTAX-SYMBOL. More syntax entries are added if we need to
9159 ;; skip past open parens and containing statements. Most of the added
9160 ;; syntax elements will get the same anchor point - the exception is
9161 ;; for an anchor in a construct like "namespace"[*] - this is as early
9162 ;; as possible in the construct but on the same line as the {.
9163 ;;
9164 ;; [*] i.e. with a keyword matching c-other-block-decl-kwds.
9165 ;;
9166 ;; SYNTAX-EXTRA-ARGS are a list of the extra arguments for the
9167 ;; syntax symbol. They are appended after the anchor point.
9168 ;;
9169 ;; If STOP-AT-BOI-ONLY is nil, we can stop in the middle of the line
9170 ;; if the current statement starts there.
9171 ;;
9172 ;; Note: It's not a problem if PAREN-STATE "overshoots"
9173 ;; CONTAINING-SEXP, i.e. contains info about parens further down.
9174 ;;
9175 ;; This function might do hidden buffer changes.
9178 ;; This is by far the most common case, so let's give it special
9179 ;; treatment.
9184 ;; Set when we're on a label, so that we don't stop there.
9185 ;; FIXME: To be complete we should check if we're on a label
9186 ;; now at the start.
9187 on-label)
9189 ;; Use point as the anchor point for "namespace", "extern", etc.
9193 syntax-extra-args)
9195 ;; Loop while we have to back out of containing blocks.
9200 ;; Loop while we have to back up statements.
9202 on-label
9205 ;; Skip past any comments that stands between the
9206 ;; statement start and boi.
9214 ;; Skip to the beginning of this statement or backward
9215 ;; another one.
9223 ;; If we didn't move we're at the start of a block and
9224 ;; have to continue outside it.
9233 ;; Special case to avoid deeper and deeper indentation
9234 ;; of "else if" clauses.
9235 )
9240 ;; If stop-at-boi-only is nil, we shouldn't back up
9241 ;; over previous or containing statements to try to
9242 ;; reach boi, so go back to the last position and
9243 ;; exit.
9250 ;; If we've moved into another statement then we
9251 ;; should no longer try to stop in the middle of a
9252 ;; line.
9255 ;; Record this as a substatement if we skipped up one
9256 ;; level.
9259 )))
9261 containing-sexp)
9263 ;; Now we have to go out of this block.
9266 ;; Don't stop in the middle of a special brace list opener
9267 ;; like "({".
9279 ;; Analyze the construct in front of the block we've stepped out
9280 ;; from and add the right syntactic element for it.
9283 step-type)
9286 ;; Stepped out of a parenthesis block, so we're in an
9287 ;; expression now.
9299 ;; Stepped out of an in-expression statement. This
9300 ;; syntactic element won't get an anchor pos.
9303 ;; A parenthesis normally belongs to an arglist.
9311 on-label nil))
9313 ;; Stepped out of a brace block.
9325 containing-sexp)))
9327 'defun-block-intro
9329 nil))
9333 c-other-decl-block-key-in-symbols-alist))
9340 ;; Always done if the open brace was at boi. The
9341 ;; c-beginning-of-statement-1 call above is necessary
9342 ;; anyway, to decide the type of block-intro to add.
9345 ))
9347 ;; Fill in the current point as the anchor for all the symbols
9348 ;; added above.
9357 )))
9360 containing-decl-open
9361 containing-decl-start
9362 containing-decl-kwd
9363 paren-state)
9364 ;; The inclass and class-close syntactic symbols are added in
9365 ;; several places and some work is needed to fix everything.
9366 ;; Therefore it's collected here.
9367 ;;
9368 ;; This function might do hidden buffer changes.
9373 containing-decl-open)
9375 ;; Ought to use `c-add-stmt-syntax' instead of backing up to boi
9376 ;; here, but we have to do like this for compatibility.
9386 char-after-ip
9387 beg-of-same-or-containing-stmt
9388 containing-sexp
9389 paren-state)
9390 ;; This function contains the decision tree reached through both
9391 ;; cases 18 and 10. It's a continued statement or top level
9392 ;; construct of some kind.
9393 ;;
9394 ;; This function might do hidden buffer changes.
9401 ;; (CASE A removed.)
9402 ;; CASE B: open braces for class or brace-lists
9409 ;; CASE B.1: class-open
9416 ;; CASE B.2: brace-list-open
9421 indent-point t t t)))
9422 ;; The most semantically accurate symbol here is
9423 ;; brace-list-open, but we normally report it simply as a
9424 ;; statement-cont. The reason is that one normally adjusts
9425 ;; brace-list-open for brace lists as top-level constructs,
9426 ;; and brace lists inside statements is a completely different
9427 ;; context. C.f. case 5A.3.
9430 ;; Turn off the dwim above when we're
9431 ;; analyzing the nature of the brace
9432 ;; for the auto newline feature.
9433 'brace-list-open
9435 nil nil
9436 containing-sexp paren-state))
9438 ;; CASE B.3: The body of a function declared inside a normal
9439 ;; block. Can occur e.g. in Pike and when using gcc
9440 ;; extensions, but watch out for macros followed by blocks.
9441 ;; C.f. cases E, 16F and 17G.
9447 ;; Turn off recognition of constructs that lacks a
9448 ;; type in this case, since that's more likely to be
9449 ;; a macro followed by a block.
9452 containing-sexp paren-state))
9454 ;; CASE B.4: Continued statement with block open. The most
9455 ;; accurate analysis is perhaps `statement-cont' together with
9456 ;; `block-open' but we play DWIM and use `substatement-open'
9457 ;; instead. The rationale is that this typically is a macro
9458 ;; followed by a block which makes it very similar to a
9459 ;; statement with a substatement block.
9462 containing-sexp paren-state))
9463 ))
9465 ;; CASE C: iostream insertion or extraction operator
9469 ;; If there is no preceding streamop in the statement
9470 ;; then indent this line as a normal statement-cont.
9474 t))))
9476 ;; CASE E: In the "K&R region" of a function declared inside a
9477 ;; normal block. C.f. case B.3.
9479 ;; Check that the next token is a '{'. This works as
9480 ;; long as no language that allows nested function
9481 ;; definitions allows stuff like member init lists, K&R
9482 ;; declarations or throws clauses there.
9483 ;;
9484 ;; Note that we do a forward search for something ahead
9485 ;; of the indentation line here. That's not good since
9486 ;; the user might not have typed it yet. Unfortunately
9487 ;; it's exceedingly tricky to recognize a function
9488 ;; prototype in a code block without resorting to this.
9496 ;; Turn off recognition of constructs that lacks a
9497 ;; type in this case, since that's more likely to be
9498 ;; a macro followed by a block.
9501 containing-sexp paren-state))
9503 ;;CASE F: continued statement and the only preceding items are
9504 ;;annotations.
9512 t)
9519 ;; CASE G: a template list continuation?
9520 ;; Mostly a duplication of case 5D.3 to fix templates-19:
9538 ;; In a nested template arglist.
9544 ;; FIXME: Should use c-add-stmt-syntax, but it's not yet
9545 ;; template aware.
9548 ;; CASE D: continued statement.
9552 containing-sexp paren-state))
9553 )))
9555 ;; The next autoload was added by RMS on 2005/8/9 - don't know why (ACM,
9556 ;; 2005/11/29).
9557 ;;;###autoload
9559 "Return the syntactic context of the current line."
9565 open-paren-in-column-0-is-defun-start
9566 ;; A whole ugly bunch of various temporary variables. Have
9567 ;; to declare them here since it's not possible to declare
9568 ;; a variable with only the scope of a cond test and the
9569 ;; following result clauses, and most of this function is a
9570 ;; single gigantic cond. :P
9571 literal char-before-ip before-ws-ip char-after-ip macro-start
9572 in-macro-expr c-syntactic-context placeholder c-in-literal-cache
9573 step-type tmpsymbol keyword injava-inher special-brace-list tmp-pos
9574 containing-<
9575 ;; The following record some positions for the containing
9576 ;; declaration block if we're directly within one:
9577 ;; `containing-decl-open' is the position of the open
9578 ;; brace. `containing-decl-start' is the start of the
9579 ;; declaration. `containing-decl-kwd' is the keyword
9580 ;; symbol of the keyword that tells what kind of block it
9581 ;; is.
9582 containing-decl-open
9583 containing-decl-start
9584 containing-decl-kwd
9585 ;; The open paren of the closest surrounding sexp or nil if
9586 ;; there is none.
9587 containing-sexp
9588 ;; The position after the closest preceding brace sexp
9589 ;; (nested sexps are ignored), or the position after
9590 ;; `containing-sexp' if there is none, or (point-min) if
9591 ;; `containing-sexp' is nil.
9592 lim
9593 ;; The paren state outside `containing-sexp', or at
9594 ;; `indent-point' if `containing-sexp' is nil.
9596 ;; There's always at most one syntactic element which got
9597 ;; an anchor pos. It's stored in syntactic-relpos.
9598 syntactic-relpos
9601 ;; Check if we're directly inside an enclosing declaration
9602 ;; level block.
9611 containing-sexp)
9612 t)))
9615 containing-sexp nil)
9620 ;; Init some position variables.
9633 ;; Ignore balanced paren. The next entry
9634 ;; can't be another one.
9637 ;; If there is no surrounding open paren then
9638 ;; put the last balanced pair back on paren-state.
9640 containing-sexp nil)))
9644 ;; If we're in a parenthesis list then ',' delimits the
9645 ;; "statements" rather than being an operator (with the
9646 ;; exception of the "for" clause). This difference is
9647 ;; typically only noticeable when statements are used in macro
9648 ;; arglists.
9652 ;; cache char before and after indent point, and move point to
9653 ;; the most likely position to perform the majority of tests
9662 ;; are we in a literal?
9665 ;; now figure out syntactic qualities of the current line
9668 ;; CASE 1: in a string.
9672 ;; CASE 2: in a C or C++ style comment.
9674 ;; This is a kludge for XEmacs where we use
9675 ;; `buffer-syntactic-context', which doesn't correctly
9676 ;; recognize "\*/" to end a block comment.
9677 ;; `parse-partial-sexp' which is used by
9678 ;; `c-literal-limits' will however do that in most
9679 ;; versions, which results in that we get nil from
9680 ;; `c-literal-limits' even when `c-in-literal' claims
9681 ;; we're inside a comment.
9685 ;; CASE 3: in a cpp preprocessor macro continuation.
9695 ;; If at the beginning of the body of a #define
9696 ;; directive then analyze as cpp-define-intro
9697 ;; only. Go on with the syntactic analysis
9698 ;; otherwise. in-macro-expr is set if we're in a
9699 ;; cpp expression, i.e. before the #define body
9700 ;; or anywhere in a non-#define directive.
9707 nil)))))
9711 ;; CASE 11: an else clause?
9715 containing-sexp paren-state))
9717 ;; CASE 12: while closure of a do/while construct?
9725 containing-sexp paren-state))
9727 ;; CASE 13: A catch or finally clause? This case is simpler
9728 ;; than if-else and do-while, because a block is required
9729 ;; after every try, catch and finally.
9737 t)
9741 t)
9744 t))
9749 containing-sexp paren-state))
9751 ;; CASE 18: A substatement we can recognize by keyword.
9761 ;; Ought to cache the result from the
9762 ;; c-beginning-of-statement-1 calls here.
9772 ;; Step up to the containing statement if we
9773 ;; stayed in the same one.
9781 ;; There was no containing statement after all.
9783 placeholder))
9785 ;; Require a parenthesis after these keywords.
9786 ;; Necessary to catch e.g. synchronized in Java,
9787 ;; which can be used both as statement and
9788 ;; modifier.
9794 ;; CASE 18A: Simple substatement.
9800 containing-sexp paren-state))
9806 containing-sexp paren-state))
9809 containing-sexp paren-state))))
9811 ;; CASE 18B: Some other substatement. This is shared
9812 ;; with case 10.
9814 char-after-ip
9815 placeholder
9816 lim
9817 paren-state)))
9819 ;; CASE 14: A case or default label
9825 containing-sexp))
9828 ;; Got a bogus label at the top level. In lack of better
9829 ;; alternatives, anchor it on (point-min).
9832 ;; CASE 15: any other label
9839 containing-decl-open
9840 containing-decl-start
9841 containing-decl-kwd
9842 paren-state))
9843 ;; Append access-label with the same anchor point as
9844 ;; inclass gets.
9850 containing-sexp))
9855 ;; If the surrounding statement is a switch then
9856 ;; let's analyze all labels as switch labels, so
9857 ;; that they get lined up consistently.
9858 'case-label
9864 ;; A label on the top level. Treat it as a class
9865 ;; context. (point-min) is the closest we get to the
9866 ;; class open brace.
9869 ;; CASE 4: In-expression statement. C.f. cases 7B, 16A and
9870 ;; 17E.
9873 containing-sexp
9874 ;; Have to turn on the heuristics after
9875 ;; the point even though it doesn't work
9876 ;; very well. C.f. test case class-16.pike.
9877 t))
9882 ;; It's a statement block or an anonymous class.
9884 ;; It's a Pike lambda. Check whether we are between the
9885 ;; lambda keyword and the argument list or at the defun
9886 ;; opener.
9888 'inline-open
9894 paren-state)
9898 ;; CASE 5: Line is inside a declaration level block or at top level.
9902 ;; CASE 5A: we are looking at a defun, brace list, class,
9903 ;; or inline-inclass method opening brace
9910 ;; CASE 5A.1: Non-class declaration block open.
9925 ;; Special case for extern-lang-open.
9926 'extern-lang-open
9928 nil t containing-sexp paren-state))
9930 ;; CASE 5A.2: we are looking at a class opening brace
9939 ;; CASE 5A.3: brace list open
9964 ))))
9968 ;; We're in Java and have found that the open brace
9969 ;; belongs to a "new Foo[]" initialization list,
9970 ;; which means the brace list is part of an
9971 ;; expression and not a top level definition. We
9972 ;; therefore treat it as any topmost continuation
9973 ;; even though the semantically correct symbol still
9974 ;; is brace-list-open, on the same grounds as in
9975 ;; case B.2.
9981 ;; CASE 5A.4: inline defun open
9987 containing-decl-open
9988 containing-decl-start
9989 containing-decl-kwd
9990 paren-state))
9992 ;; CASE 5A.5: ordinary defun open
10000 ;; Bogus to use bol here, but it's the legacy. (Resolved,
10001 ;; 2007-11-09)
10002 ))))
10004 ;; CASE 5R: Member init list. (Used to be part of CASE 5B.1)
10005 ;; Note there is no limit on the backward search here, since member
10006 ;; init lists can, in practice, be very large.
10012 ;; There is no preceding member init clause.
10013 ;; Indent relative to the beginning of indentation
10014 ;; for the topmost-intro line that contains the
10015 ;; prototype's open paren.
10018 ;; Indent relative to the first member init clause.
10023 ;; CASE 5B: After a function header but before the body (or
10024 ;; the ending semicolon if there's no body).
10031 ;; CASE 5B.1: Member init list.
10033 ;; There is no preceding member init clause.
10034 ;; Indent relative to the beginning of indentation
10035 ;; for the topmost-intro line that contains the
10036 ;; prototype's open paren.
10040 ;; CASE 5B.2: K&R arg decl intro
10047 containing-decl-open
10048 containing-decl-start
10049 containing-decl-kwd
10050 paren-state)))
10052 ;; CASE 5B.4: Nether region after a C++ or Java func
10053 ;; decl, which could include a `throws' declaration.
10057 )))
10059 ;; CASE 5C: inheritance line. could be first inheritance
10060 ;; line, or continuation of a multiple inheritance
10068 ;; watch out for scope operator
10073 )))
10082 ;; for Java
10087 cont done)
10092 done t))
10096 )
10098 injava-inher)
10101 ))
10104 ;; CASE 5C.1: non-hanging colon on an inher intro
10108 ;; don't add inclass symbol since relative point already
10109 ;; contains any class offset
10110 )
10112 ;; CASE 5C.2: hanging colon on an inher intro
10118 containing-decl-open
10119 containing-decl-start
10120 containing-decl-kwd
10121 paren-state)))
10123 ;; CASE 5C.3: in a Java implements/extends
10137 )))
10139 ;; CASE 5C.4: a continued inheritance line
10143 ;; don't add inclass symbol since relative point already
10144 ;; contains any class offset
10145 )))
10147 ;; CASE 5P: AWK pattern or function or continuation
10148 ;; thereof.
10154 'topmost-intro-cont
10156 nil nil
10157 containing-sexp paren-state))
10159 ;; CASE 5D: this could be a top-level initialization, a
10160 ;; member init list continuation, or a template argument
10161 ;; list continuation.
10163 ;; Note: We use the fact that lim is always after any
10164 ;; preceding brace sexp.
10174 t))
10179 t)
10185 nil)
10186 t))
10188 ;; NB: No c-after-special-operator-id stuff in this
10189 ;; clause - we assume only C++ needs it.
10194 ;; CASE 5D.3: perhaps a template list continuation?
10209 ;; In a nested template arglist.
10215 ;; FIXME: Should use c-add-stmt-syntax, but it's not yet
10216 ;; template aware.
10219 ;; CASE 5D.4: perhaps a multiple inheritance line?
10231 t)
10236 ;; CASE 5D.5: Continuation of the "expression part" of a
10237 ;; top level construct. Or, perhaps, an unrecognized construct.
10249 ;; A preceding comma at the top level means that a
10250 ;; new variable declaration starts here. Use
10251 ;; topmost-intro-cont for it, for consistency with
10252 ;; the first variable declaration. C.f. case 5N.
10255 nil nil containing-sexp paren-state))
10256 ))
10258 ;; CASE 5F: Close of a non-class declaration level block.
10262 ;; This is inconsistent: Should use `containing-decl-open'
10263 ;; here if it's at boi, like in case 5J.
10267 ;; Special case for compatibility with the
10268 ;; extern-lang syntactic symbols.
10269 'extern-lang-close
10272 nil t
10274 paren-state))
10276 ;; CASE 5G: we are looking at the brace which closes the
10277 ;; enclosing nested class decl
10282 containing-decl-open
10283 containing-decl-start
10284 containing-decl-kwd
10285 paren-state))
10287 ;; CASE 5H: we could be looking at subsequent knr-argdecls
10294 ;; Do an extra check to avoid tripping up on
10295 ;; statements that occur in invalid contexts
10296 ;; (e.g. in macro bodies where we don't really
10297 ;; know the context of what we're looking at).
10304 ;; CASE 5I: ObjC method definition.
10309 ;; Handle the case when it's the first (non-comment)
10310 ;; thing in the buffer. Can't look for a 'same return
10311 ;; value from cbos1 since ObjC directives currently
10312 ;; aren't recognized fully, so that we get 'same
10313 ;; instead of 'previous if it moved over a preceding
10314 ;; directive.
10318 ;; CASE 5N: At a variable declaration that follows a class
10319 ;; definition or some other block declaration that doesn't
10320 ;; end at the closing '}'. C.f. case 5D.5.
10329 ;; Speed up the backward search a bit.
10334 ;; The '}' is unbalanced.
10335 nil
10340 containing-sexp paren-state))
10342 ;; NOTE: The point is at the end of the previous token here.
10344 ;; CASE 5J: we are at the topmost level, make
10345 ;; sure we skip back past any access specifiers
10347 ;; A macro continuation line is never at top level.
10369 nil)))))
10370 ;; For historic reasons we anchor at bol of the last
10371 ;; line of the previous declaration. That's clearly
10372 ;; highly bogus and useless, and it makes our lives hard
10373 ;; to remain compatible. :P
10384 ;; Special case for compatibility with the
10385 ;; extern-lang syntactic symbols.
10386 'inextern-lang
10389 nil t
10391 paren-state))
10393 containing-decl-open
10394 containing-decl-start
10395 containing-decl-kwd
10396 paren-state)))
10398 macro-start
10403 ;; CASE 5K: we are at an ObjC method definition
10404 ;; continuation line.
10413 ;; CASE 5L: we are at the first argument of a template
10414 ;; arglist that begins on the previous line.
10422 ;; CASE 5Q: we are at a statement within a macro.
10427 ;;CASE 5N: We are at a topmost continuation line and the only
10428 ;;preceding items are annotations.
10435 t)
10441 ;; CASE 5M: we are at a topmost continuation line
10452 ))
10454 ;; (CASE 6 has been removed.)
10456 ;; CASE 7: line is an expression, not a statement. Most
10457 ;; likely we are either in a function prototype or a function
10458 ;; call argument list
10466 ;; CASE 7A: we are looking at the arglist closing paren.
10467 ;; C.f. case 7F.
10479 paren-state))
10481 ;; CASE 7B: Looking at the opening brace of an
10482 ;; in-expression block or brace list. C.f. cases 4, 16A
10483 ;; and 17E.
10487 paren-state))
10491 placeholder
10494 paren-state)
10495 containing-sexp)))
10496 ;; placeholder is nil if it's a block directly in
10497 ;; a function arglist. That makes us skip out of
10498 ;; this case.
10499 )))
10504 paren-state)
10508 ;; CASE 7C: we are looking at the first argument in an empty
10509 ;; argument list. Use arglist-close if we're actually
10510 ;; looking at a close paren or bracket.
10522 paren-state))
10524 ;; CASE 7D: we are inside a conditional test clause. treat
10525 ;; these things as statements
10535 ))
10537 ;; CASE 7E: maybe a continued ObjC method call. This is the
10538 ;; case when we are inside a [] bracketed exp, and what
10539 ;; precede the opening bracket is not an identifier.
10547 )))
10549 ;; CASE 7F: we are looking at an arglist continuation line,
10550 ;; but the preceding argument is on the same line as the
10551 ;; opening paren. This case includes multi-line
10552 ;; mathematical paren groupings, but we could be on a
10553 ;; for-list continuation line. C.f. case 7A.
10570 paren-state))
10572 ;; CASE 7G: we are looking at just a normal arglist
10573 ;; continuation line
10576 ))
10578 ;; CASE 8: func-local multi-inheritance line
10588 ;; CASE 8A: non-hanging colon on an inher intro
10593 ;; CASE 8B: hanging colon on an inher intro
10597 ;; CASE 8C: a continued inheritance line
10601 )))
10603 ;; CASE 9: we are inside a brace-list
10613 ;; CASE 9A: In the middle of a special brace list opener.
10633 ;; CASE 9B: brace-list-close brace
10635 ;; Check special brace list closer.
10642 ;; We were between the special close char and the `)'.
10645 ;; We were before the special close char.
10649 ;; Normal brace list check.
10660 ;; Prepare for the rest of the cases below by going to the
10661 ;; token following the opening brace
10674 ;; CASE 9C: we're looking at the first line in a brace-list
10685 ;; CASE 9D: this is just a later brace-list-entry or
10686 ;; brace-entry-open
10695 ))
10696 ))))
10698 ;; CASE 10: A continued statement or top level construct.
10708 ;; This is shared with case 18.
10710 char-after-ip
10711 placeholder
10712 containing-sexp
10713 paren-state))
10715 ;; CASE 16: block close brace, possibly closing the defun or
10716 ;; the class
10718 ;; From here on we have the next containing sexp in lim.
10723 ;; CASE 16E: Closing a statement block? This catches
10724 ;; cases where it's preceded by a statement keyword,
10725 ;; which works even when used in an "invalid" context,
10726 ;; e.g. a macro argument.
10731 ;; CASE 16A: closing a lambda defun or an in-expression
10732 ;; block? C.f. cases 4, 7B and 17E.
10735 nil))
10737 'inline-close
10747 paren-state)
10751 ;; CASE 16B: does this close an inline or a function in
10752 ;; a non-class declaration level block?
10759 nil))
10769 ;; CASE 16F: Can be a defun-close of a function declared
10770 ;; in a statement block, e.g. in Pike or when using gcc
10771 ;; extensions, but watch out for macros followed by
10772 ;; blocks. Let it through to be handled below.
10773 ;; C.f. cases B.3 and 17G.
10779 ;; Turn off recognition of constructs that
10780 ;; lacks a type in this case, since that's more
10781 ;; likely to be a macro followed by a block.
10788 ;; CASE 16C: If there is an enclosing brace then this is
10789 ;; a block close since defun closes inside declaration
10790 ;; level blocks have been handled above.
10792 ;; If the block is preceded by a case/switch label on
10793 ;; the same line, we anchor at the first preceding label
10794 ;; at boi. The default handling in c-add-stmt-syntax
10795 ;; really fixes it better, but we do like this to keep
10796 ;; the indentation compatible with version 5.28 and
10797 ;; earlier. C.f. case 17H.
10804 ;; c-backward-to-block-anchor not necessary here; those
10805 ;; situations are handled in case 16E above.
10808 ;; CASE 16D: Only top level defun close left.
10814 paren-state))
10815 ))
10817 ;; CASE 19: line is an expression, not a statement, and is directly
10818 ;; contained by a template delimiter. Most likely, we are in a
10819 ;; template arglist within a statement. This case is based on CASE
10820 ;; 7. At some point in the future, we may wish to create more
10821 ;; syntactic symbols such as `template-intro',
10822 ;; `template-cont-nonempty', etc., and distinguish between them as we
10823 ;; do for `arglist-intro' etc. (2009-12-07).
10829 ; '<') before indent-point.
10837 paren-state))
10839 ;; CASE 17: Statement or defun catchall.
10842 ;; Back up statements until we find one that starts at boi.
10845 containing-sexp)))
10849 nil)
10854 ;; CASE 17B: continued statement
10858 containing-sexp paren-state))
10860 ;; CASE 17A: After a case/default label?
10868 'statement-case-open
10870 nil t containing-sexp paren-state))
10872 ;; CASE 17D: any old statement
10879 containing-sexp paren-state)
10883 ;; CASE 17I: Inside a substatement block.
10885 ;; The following tests are all based on containing-sexp.
10887 ;; From here on we have the next containing sexp in lim.
10892 lim paren-state)
10896 ;; CASE 17E: first statement in an in-expression block.
10897 ;; C.f. cases 4, 7B and 16A.
10900 nil))
10902 'defun-block-intro
10911 paren-state)
10917 ;; CASE 17F: first statement in an inline, or first
10918 ;; statement in a top-level defun. we can tell this is it
10919 ;; if there are no enclosing braces that haven't been
10920 ;; narrowed out by a class (i.e. don't use bod here).
10923 paren-state)))
10929 nil))))
10934 ;; CASE 17G: First statement in a function declared inside
10935 ;; a normal block. This can occur in Pike and with
10936 ;; e.g. the gcc extensions, but watch out for macros
10937 ;; followed by blocks. C.f. cases B.3 and 16F.
10943 ;; Turn off recognition of constructs that lacks
10944 ;; a type in this case, since that's more likely
10945 ;; to be a macro followed by a block.
10951 lim paren-state))
10953 ;; CASE 17H: First statement in a block.
10955 ;; If the block is preceded by a case/switch label on the
10956 ;; same line, we anchor at the first preceding label at
10957 ;; boi. The default handling in c-add-stmt-syntax is
10958 ;; really fixes it better, but we do like this to keep the
10959 ;; indentation compatible with version 5.28 and earlier.
10960 ;; C.f. case 16C.
10967 ;; c-backward-to-block-anchor not necessary here; those
10968 ;; situations are handled in case 17I above.
10970 lim paren-state))
10973 ))
10974 )
10976 ;; now we need to look at any modifiers
10980 ;; are we looking at a comment only line?
10985 ;; we might want to give additional offset to friends (in C++).
10990 ;; Set syntactic-relpos.
10996 nil)
10997 t))
11000 ;; Start of or a continuation of a preprocessor directive?
11015 ;; If inside a cpp expression, i.e. anywhere in a
11016 ;; cpp directive except a #define body, we only let
11017 ;; through the syntactic analysis that is internal
11018 ;; in the expression. That means the arglist
11019 ;; elements, if they are anchored inside the cpp
11020 ;; expression.
11029 ;; Inside a #define body and the syntactic analysis is
11030 ;; anchored on the start of the #define. In this case
11031 ;; we add cpp-define-intro to get the extra
11032 ;; indentation of the #define body.
11035 ;; return the syntax
11036 c-syntactic-context)))
11038 \f
11039 ;; Indentation calculation.
11042 ;; offset can be a number, a function, a variable, a list, or one of
11043 ;; the symbols + or -
11044 ;;
11045 ;; This function might do hidden buffer changes.
11064 langelem symbol))
11070 offset symbol)
11071 nil)
11085 Error evaluating offset %S for %s: \
11086 Cannot combine absolute offset %S with relative %S in `%s' method"
11092 Error evaluating offset %S for %s: \
11093 Cannot combine relative offset %S with absolute %S in `%s' method"
11098 res))
11116 Error evaluating offset %S for %s: \
11117 Cannot combine absolute offsets %S and %S in `add' method"
11121 res))
11130 res))))
11137 nil))))
11141 res
11143 offset symbol res)
11144 nil)))
11147 ;; Get offset from LANGELEM which is a list beginning with the
11148 ;; syntactic symbol and followed by any analysis data it provides.
11149 ;; That data may be zero or more elements, but if at least one is
11150 ;; given then the first is the anchor position (or nil). The symbol
11151 ;; is matched against `c-offsets-alist' and the offset calculated
11152 ;; from that is returned.
11153 ;;
11154 ;; This function might do hidden buffer changes.
11164 offset
11168 ))
11171 ;; This is a compatibility wrapper for `c-calc-offset' in case
11172 ;; someone is calling it directly. It takes an old style syntactic
11173 ;; element on the form (SYMBOL . ANCHOR-POS) and converts it to the
11174 ;; new list form.
11175 ;;
11176 ;; This function might do hidden buffer changes.
11183 ;; Calculate the syntactic indentation from a syntactic description
11184 ;; as returned by `c-guess-syntax'.
11185 ;;
11186 ;; Note that topmost-intro always has an anchor position at bol, for
11187 ;; historical reasons. It's often used together with other symbols
11188 ;; that has more sane positions. Since we always use the first
11189 ;; found anchor position, we rely on that these other symbols always
11190 ;; precede topmost-intro in the LANGELEMS list.
11191 ;;
11192 ;; This function might do hidden buffer changes.
11200 ;; Got an absolute column that overrides any indentation
11201 ;; we've collected so far, but not the relative
11202 ;; indentation we might get for the nested structures
11203 ;; further down the langelems list.
11207 ;; Got a relative change of the current calculated
11208 ;; indentation.
11211 ;; Use the anchor position from the first syntactic
11212 ;; element with one.
11222 indent)))
11224 \f
11227 ;;; Local Variables:
11228 ;;; indent-tabs-mode: t
11229 ;;; tab-width: 8
11230 ;;; End:
11231 ;;; cc-engine.el ends here