1 Fri Apr 2 17:31:59 1993 Jim Blandy (jimb@totoro.cs.oberlin.edu)
3 * Released version 0.12.
5 * regex.c (regerror): If errcode is zero, that's not a valid
6 error code, according to POSIX, but return "Success."
8 * regex.c (regerror): Remember to actually fetch the message
11 * regex.c (regex_compile): Don't use the trick for ".*\n" on
12 ".+\n". Since the latter involves laying an extra choice
13 point, the backward jump isn't adjusted properly.
15 Thu Mar 25 21:35:18 1993 Jim Blandy (jimb@totoro.cs.oberlin.edu)
17 * regex.c (regex_compile): In the handle_open and handle_close
18 sections, clear pending_exact to zero.
20 Tue Mar 9 12:03:07 1993 Jim Blandy (jimb@wookumz.gnu.ai.mit.edu)
22 * regex.c (re_search_2): In the loop which searches forward
23 using fastmap, don't forget to cast the character from the
24 string to an unsigned before using it as an index into the
27 Thu Jan 14 15:41:46 1993 David J. MacKenzie (djm@kropotkin.gnu.ai.mit.edu)
29 * regex.h: Never define const; let the callers do it.
30 configure.in: Don't define USING_AUTOCONF.
32 Wed Jan 6 20:49:29 1993 Jim Blandy (jimb@geech.gnu.ai.mit.edu)
34 * regex.c (regerror): Abort if ERRCODE is out of range.
36 Sun Dec 20 16:19:10 1992 Jim Blandy (jimb@totoro.cs.oberlin.edu)
38 * configure.in: Arrange to #define USING_AUTOCONF.
39 * regex.h: If USING_AUTOCONF is #defined, don't mess with
40 `const' at all; autoconf has taken care of it.
42 Mon Dec 14 21:40:39 1992 David J. MacKenzie (djm@kropotkin.gnu.ai.mit.edu)
44 * regex.h (RE_SYNTAX_AWK): Fix typo. From Arnold Robbins.
46 Sun Dec 13 20:35:39 1992 Jim Blandy (jimb@totoro.cs.oberlin.edu)
48 * regex.c (compile_range): Fetch the range start and end by
49 casting the pattern pointer to an `unsigned char *' before
52 Sat Dec 12 09:41:01 1992 Jim Blandy (jimb@totoro.cs.oberlin.edu)
54 * regex.c: Undo change of 12/7/92; it's better for Emacs to
55 #define HAVE_CONFIG_H.
57 Fri Dec 11 22:00:34 1992 Jim Meyering (meyering@hal.gnu.ai.mit.edu)
59 * regex.c: Define and use isascii-protected ctype.h macros.
61 Fri Dec 11 05:10:38 1992 Jim Blandy (jimb@totoro.cs.oberlin.edu)
63 * regex.c (re_match_2): Undo Karl's November 10th change; it
64 keeps the group in :\(.*\) from matching :/ properly.
66 Mon Dec 7 19:44:56 1992 Jim Blandy (jimb@wookumz.gnu.ai.mit.edu)
68 * regex.c: #include config.h if either HAVE_CONFIG_H or emacs
71 Tue Dec 1 13:33:17 1992 David J. MacKenzie (djm@goldman.gnu.ai.mit.edu)
73 * regex.c [HAVE_CONFIG_H]: Include config.h.
75 Wed Nov 25 23:46:02 1992 David J. MacKenzie (djm@goldman.gnu.ai.mit.edu)
77 * regex.c (regcomp): Add parens around bitwise & for clarity.
78 Initialize preg->allocated to prevent segv.
80 Tue Nov 24 09:22:29 1992 David J. MacKenzie (djm@goldman.gnu.ai.mit.edu)
82 * regex.c: Use HAVE_STRING_H, not USG.
83 * configure.in: Check for string.h, not USG.
85 Fri Nov 20 06:33:24 1992 Karl Berry (karl@cs.umb.edu)
87 * regex.c (SIGN_EXTEND_CHAR) [VMS]: Back out of this change,
88 since Roland Roberts now says it was a localism.
90 Mon Nov 16 07:01:36 1992 Karl Berry (karl@cs.umb.edu)
92 * regex.h (const) [!HAVE_CONST]: Test another cpp symbol (from
93 Autoconf) before zapping const.
95 Sun Nov 15 05:36:42 1992 Jim Blandy (jimb@wookumz.gnu.ai.mit.edu)
97 * regex.c, regex.h: Changes for VMS from Roland B Roberts
98 <roberts@nsrl31.nsrl.rochester.edu>.
100 Thu Nov 12 11:31:15 1992 Karl Berry (karl@cs.umb.edu)
102 * Makefile.in (distfiles): Include INSTALL.
104 Tue Nov 10 09:29:23 1992 Karl Berry (karl@cs.umb.edu)
106 * regex.c (re_match_2): At maybe_pop_jump, if at end of string
107 and pattern, just quit the matching loop.
109 * regex.c (LETTER_P): Rename to `WORDCHAR_P'.
111 * regex.c (AT_STRINGS_{BEG,END}): Take `d' as an arg; change
114 * regex.c (re_match_2) [!emacs]: In wordchar and notwordchar
117 Wed Nov 4 15:43:58 1992 Karl Berry (karl@hal.gnu.ai.mit.edu)
119 * regex.h (const) [!__STDC__]: Don't define if it's already defined.
121 Sat Oct 17 19:28:19 1992 Karl Berry (karl@cs.umb.edu)
123 * regex.c (bcmp, bcopy, bzero): Only #define if they are not
126 * configure.in: Use AC_CONST.
128 Thu Oct 15 08:39:06 1992 Karl Berry (karl@cs.umb.edu)
130 * regex.h (const) [!const]: Conditionalize.
132 Fri Oct 2 13:31:42 1992 Karl Berry (karl@cs.umb.edu)
134 * regex.h (RE_SYNTAX_ED): New definition.
136 Sun Sep 20 12:53:39 1992 Karl Berry (karl@cs.umb.edu)
138 * regex.[ch]: remove traces of `longest_p' -- dumb idea to put
139 this into the pattern buffer, as it means parallelism loses.
141 * Makefile.in (config.status): use sh to run configure --no-create.
143 * Makefile.in (realclean): OK, don't remove configure.
145 Sat Sep 19 09:05:08 1992 Karl Berry (karl@hayley)
147 * regex.c (PUSH_FAILURE_POINT, POP_FAILURE_POINT) [DEBUG]: keep
148 track of how many failure points we push and pop.
149 (re_match_2) [DEBUG]: declare variables for that, and print results.
150 (DEBUG_PRINT4): new macro.
152 * regex.h (re_pattern_buffer): new field `longest_p' (to
153 eliminate backtracking if the user doesn't need it).
154 * regex.c (re_compile_pattern): initialize it (to 1).
155 (re_search_2): set it to zero if register information is not needed.
156 (re_match_2): if it's set, don't backtrack.
158 * regex.c (re_search_2): update fastmap only after checking that
159 the pattern is anchored.
161 * regex.c (re_match_2): do more debugging at maybe_pop_jump.
163 * regex.c (re_search_2): cast result of TRANSLATE for use in
166 Thu Sep 17 19:47:16 1992 Karl Berry (karl@geech.gnu.ai.mit.edu)
170 Wed Sep 16 08:17:10 1992 Karl Berry (karl@hayley)
172 * regex.c (INIT_FAIL_STACK): rewrite as statements instead of a
173 complicated comma expr, to avoid compiler warnings (and also
175 (re_compile_fastmap, re_match_2): change callers.
177 * regex.c (POP_FAILURE_POINT): cast pop of regstart and regend
178 to avoid compiler warnings.
180 * regex.h (RE_NEWLINE_ORDINARY): remove this syntax bit, and
182 * regex.c (at_{beg,end}line_loc_p): go the last mile: remove
183 the RE_NEWLINE_ORDINARY case which made the ^ in \n^ be an anchor.
185 Tue Sep 15 09:55:29 1992 Karl Berry (karl@hayley)
187 * regex.c (at_begline_loc_p): new fn.
188 (at_endline_loc_p): simplify at_endline_op_p.
189 (regex_compile): in ^/$ cases, call the above.
191 * regex.c (POP_FAILURE_POINT): rewrite the fn as a macro again,
192 as lord's profiling indicates the function is 20% of the time.
193 (re_match_2): callers changed.
195 * configure.in (AC_MEMORY_H): remove, since we never use memcpy et al.
197 Mon Sep 14 17:49:27 1992 Karl Berry (karl@hayley)
199 * Makefile.in (makeargs): include MFLAGS.
201 Sun Sep 13 07:41:45 1992 Karl Berry (karl@hayley)
203 * regex.c (regex_compile): in \1..\9 case, make it always
204 invalid to use \<digit> if there is no preceding <digit>th subexpr.
205 * regex.h (RE_NO_MISSING_BK_REF): remove this syntax bit.
207 * regex.c (regex_compile): remove support for invalid empty groups.
208 * regex.h (RE_NO_EMPTY_GROUPS): remove this syntax bit.
210 * regex.c (FREE_VARIABLES) [!REGEX_MALLOC]: define as alloca (0),
213 * regex.h (RE_SYNTAX_POSIX_SED): don't bother with this.
215 Sat Sep 12 13:37:21 1992 Karl Berry (karl@hayley)
217 * README: incorporate emacs.diff.
219 * regex.h (_RE_ARGS) [!__STDC__]: define as empty parens.
221 * configure.in: add AC_ALLOCA.
223 * Put test files in subdir test, documentation in subdir doc.
224 Adjust Makefile.in and configure.in accordingly.
226 Thu Sep 10 10:29:11 1992 Karl Berry (karl@hayley)
228 * regex.h (RE_SYNTAX_{POSIX_,}SED): new definitions.
230 Wed Sep 9 06:27:09 1992 Karl Berry (karl@hayley)
234 Tue Sep 8 07:32:30 1992 Karl Berry (karl@hayley)
236 * xregex.texinfo: put the day of month into the date.
238 * Makefile.in (realclean): remove Texinfo-generated files.
239 (distclean): remove empty sorted index files.
240 (clean): remove dvi files, etc.
242 * configure.in: test for more Unix variants.
244 * fileregex.c: new file.
245 Makefile.in (fileregex): new target.
247 * iregex.c (main): move variable decls to smallest scope.
249 * regex.c (FREE_VARIABLES): free reg_{,info_}dummy.
250 (re_match_2): check that the allocation for those two succeeded.
252 * regex.c (FREE_VAR): replace FREE_NONNULL with this.
253 (FREE_VARIABLES): call it.
254 (re_match_2) [REGEX_MALLOC]: initialize all our vars to NULL.
256 * tregress.c (do_match): generalize simple_match.
257 (SIMPLE_NONMATCH): new macro.
258 (SIMPLE_MATCH): change from routine.
260 * Makefile.in (regex.texinfo): make file readonly, so we don't
263 * many files (re_default_syntax): rename to `re_syntax_options';
264 call re_set_syntax instead of assigning to the variable where
267 Mon Sep 7 10:12:16 1992 Karl Berry (karl@hayley)
269 * syntax.skel: don't use prototypes.
271 * {configure,Makefile}.in: new files.
273 * regex.c: include <string.h> `#if USG || STDC_HEADERS'; remove
274 obsolete test for `POSIX', and test for BSRTING.
275 Include <strings.h> if we are not USG or STDC_HEADERS.
276 Do not include <unistd.h>. What did we ever need that for?
278 * regex.h (RE_NO_EMPTY_ALTS): remove this.
279 (RE_SYNTAX_AWK): remove from here, too.
280 * regex.c (regex_compile): remove the check.
281 * xregex.texinfo (Alternation Operator): update.
282 * other.c (test_others): remove tests for this.
284 * regex.h (RE_DUP_MAX): undefine if already defined.
286 * regex.h: (RE_SYNTAX_POSIX*): redo to allow more operators, and
287 define new syntaxes with the minimal set.
289 * syntax.skel (main): used sscanf instead of scanf.
291 * regex.h (RE_SYNTAX_*GREP): new definitions from mike.
293 * regex.c (regex_compile): initialize the upper bound of
294 intervals at the beginning of the interval, not the end.
295 (From pclink@qld.tne.oz.au.)
297 * regex.c (handle_bar): rename to `handle_alt', for consistency.
299 * regex.c ({store,insert}_{op1,op2}): new routines (except the last).
300 ({STORE,INSERT}_JUMP{,2}): macros to replace the old routines,
301 which took arguments in different orders, and were generally weird.
303 * regex.c (PAT_PUSH*): rename to `BUF_PUSH*' -- we're not
304 appending info to the pattern!
306 Sun Sep 6 11:26:49 1992 Karl Berry (karl@hayley)
308 * regex.c (regex_compile): delete the variable
309 `following_left_brace', since we never use it.
311 * regex.c (print_compiled_pattern): don't print the fastmap if
314 * regex.c (re_compile_fastmap): handle
315 `on_failure_keep_string_jump' like `on_failure_jump'.
317 * regex.c (re_match_2): in `charset{,_not' case, cast the bit
318 count to unsigned, not unsigned char, in case we have a full
321 * tregress.c (simple_match): remove.
322 (simple_test): rename as `simple_match'.
323 (simple_compile): print the error string if the compile failed.
325 * regex.c (DO_RANGE): rewrite as a function, `compile_range', so
326 we can debug it. Change pattern characters to unsigned char
327 *'s, and change the range variable to an unsigned.
328 (regex_compile): change calls.
330 Sat Sep 5 17:40:49 1992 Karl Berry (karl@hayley)
332 * regex.h (_RE_ARGS): new macro to put in argument lists (if
333 ANSI) or omit them (if K&R); don't declare routines twice.
335 * many files (obscure_syntax): rename to `re_default_syntax'.
337 Fri Sep 4 09:06:53 1992 Karl Berry (karl@hayley)
339 * GNUmakefile (extraclean): new target.
340 (realclean): delete the info files.
342 Wed Sep 2 08:14:42 1992 Karl Berry (karl@hayley)
346 Sun Aug 23 06:53:15 1992 Karl Berry (karl@hayley)
348 * regex.[ch] (re_comp): no const in the return type (from djm).
350 Fri Aug 14 07:25:46 1992 Karl Berry (karl@hayley)
352 * regex.c (DO_RANGE): declare variables as unsigned chars, not
353 signed chars (from jimb).
355 Wed Jul 29 18:33:53 1992 Karl Berry (karl@claude.cs.umb.edu)
359 * GNUmakefile (distclean): do not remove regex.texinfo.
360 (realclean): remove it here.
362 * tregress.c (simple_test): initialize buf.buffer.
364 Sun Jul 26 08:59:38 1992 Karl Berry (karl@hayley)
366 * regex.c (push_dummy_failure): new opcode and corresponding
367 case in the various routines. Pushed at the end of
370 * regex.c (jump_past_next_alt): rename to `jump_past_alt', for
372 (no_pop_jump): rename to `jump'.
374 * regex.c (regex_compile) [DEBUG]: terminate printing of pattern
379 * tregress.c (simple_{compile,match,test}): routines to simplify all
382 * tregress.c: test for matching as much as possible.
384 Fri Jul 10 06:53:32 1992 Karl Berry (karl@hayley)
388 Wed Jul 8 06:39:31 1992 Karl Berry (karl@hayley)
390 * regex.c (SIGN_EXTEND_CHAR): #undef any previous definition, as
391 ours should always work properly.
393 Mon Jul 6 07:10:50 1992 Karl Berry (karl@hayley)
395 * iregex.c (main) [DEBUG]: conditionalize the call to
396 print_compiled_pattern.
398 * iregex.c (main): initialize buf.buffer to NULL.
399 * tregress (test_regress): likewise.
401 * regex.c (alloca) [sparc]: #if on HAVE_ALLOCA_H instead.
403 * tregress.c (test_regress): didn't have jla's test quite right.
405 Sat Jul 4 09:02:12 1992 Karl Berry (karl@hayley)
407 * regex.c (re_match_2): only REGEX_ALLOCATE all the register
408 vectors if the pattern actually has registers.
409 (match_end): new variable to avoid having to use best_regend[0].
411 * regex.c (IS_IN_FIRST_STRING): rename to FIRST_STRING_P.
413 * regex.c: doc fixes.
415 * tregess.c (test_regress): new fastmap test forwarded by rms.
417 * tregress.c (test_regress): initialize the fastmap field.
419 * tregress.c (test_regress): new test from jla that aborted
422 Fri Jul 3 09:10:05 1992 Karl Berry (karl@hayley)
424 * tregress.c (test_regress): add tests for translating charsets,
427 * GNUmakefile (common): add alloca.o.
428 * alloca.c: new file, copied from bison.
430 * other.c (test_others): remove var `buf', since it's no longer used.
432 * Below changes from ro@TechFak.Uni-Bielefeld.DE.
434 * tregress.c (test_regress): initialize buf.allocated.
436 * regex.c (re_compile_fastmap): initialize `succeed_n_p'.
438 * GNUmakefile (regex): depend on $(common).
440 Wed Jul 1 07:12:46 1992 Karl Berry (karl@hayley)
444 * regex.c: doc fixes.
446 Mon Jun 29 08:09:47 1992 Karl Berry (karl@fosse)
448 * regex.c (pop_failure_point): change string vars to
449 `const char *' from `unsigned char *'.
451 * regex.c: consolidate debugging stuff.
452 (print_partial_compiled_pattern): avoid enum clash.
454 Mon Jun 29 07:50:27 1992 Karl Berry (karl@hayley)
456 * xmalloc.c: new file.
457 * GNUmakefile (common): add it.
459 * iregex.c (print_regs): new routine (from jimb).
462 Sat Jun 27 10:50:59 1992 Jim Blandy (jimb@pogo.cs.oberlin.edu)
464 * xregex.c (re_match_2): When we have accepted a match and
465 restored d from best_regend[0], we need to set dend
466 appropriately as well.
468 Sun Jun 28 08:48:41 1992 Karl Berry (karl@hayley)
470 * tregress.c: rename from regress.c.
472 * regex.c (print_compiled_pattern): improve charset case to ease
474 Also, don't distinguish between Emacs and non-Emacs
475 {not,}wordchar opcodes.
477 * regex.c (print_fastmap): move here.
479 * regex.c (print_{{partial,}compiled_pattern,double_string}):
480 rename from ..._printer. Change calls here and in test.c.
482 * regex.c: create from xregex.c and regexinc.c for once and for
483 all, and change the debug fns to be extern, instead of static.
484 * GNUmakefile: remove traces of xregex.c.
485 * test.c: put in externs, instead of including regexinc.c.
487 * xregex.c: move interactive main program and scanstring to iregex.c.
488 * iregex.c: new file.
489 * upcase.c, printchar.c: new files.
491 * various doc fixes and other cosmetic changes throughout.
493 * regexinc.c (compiled_pattern_printer): change variable name,
495 (partial_compiled_pattern_printer): print other info about the
496 compiled pattern, besides just the opcodes.
497 * xregex.c (regex_compile) [DEBUG]: print the compiled pattern
500 * xregex.c (re_compile_fastmap): in the duplicate case, set
501 `can_be_null' and return.
502 Also, set `bufp->can_be_null' according to a new variable,
504 Also, rewrite main while loop to not test `p != NULL', since
505 we never set it that way.
506 Also, eliminate special `can_be_null' value for the endline case.
507 (re_search_2): don't test for the special value.
508 * regex.h (struct re_pattern_buffer): remove the definition.
510 Sat Jun 27 15:00:40 1992 Karl Berry (karl@hayley)
512 * xregex.c (re_compile_fastmap): remove the `RE_' from
513 `REG_RE_MATCH_NULL_AT_END'.
514 Also, assert the fastmap in the pattern buffer is non-null.
515 Also, reset `succeed_n_p' after we've
516 paid attention to it, instead of every time through the loop.
517 Also, in the `anychar' case, only clear fastmap['\n'] if the
518 syntax says to, and don't return prematurely.
519 Also, rearrange cases in some semblance of a rational order.
520 * regex.h (REG_RE_MATCH_NULL_AT_END): remove the `RE_' from the name.
522 * other.c: take bug reports from here.
523 * regress.c: new file for them.
524 * GNUmakefile (test): add it.
525 * main.c (main): new possible test.
526 * test.h (test_type): new value in enum.
528 Thu Jun 25 17:37:43 1992 Karl Berry (karl@hayley)
530 * xregex.c (scanstring) [test]: new function from jimb to allow some
532 (main) [test]: call it (on the string, not the pattern).
534 * xregex.c (main): make return type `int'.
536 Wed Jun 24 10:43:03 1992 Karl Berry (karl@hayley)
538 * xregex.c (pattern_offset_t): change to `int', for the benefit
539 of patterns which compile to more than 2^15 bytes.
541 * xregex.c (GET_BUFFER_SPACE): remove spurious braces.
543 * xregex.texinfo (Using Registers): put in a stub to ``document''
545 * regex.h (re_set_registers) [!__STDC__]: declare.
546 * xregex.c (re_set_registers): declare K&R style (also move to a
547 different place in the file).
549 Mon Jun 8 18:03:28 1992 Jim Blandy (jimb@pogo.cs.oberlin.edu)
551 * regex.h (RE_NREGS): Doc fix.
553 * xregex.c (re_set_registers): New function.
554 * regex.h (re_set_registers): Declaration for new function.
556 Fri Jun 5 06:55:18 1992 Karl Berry (karl@hayley)
558 * main.c (main): `return 0' instead of `exit (0)'. (From Paul Eggert)
560 * regexinc.c (SIGN_EXTEND_CHAR): cast to unsigned char.
561 (extract_number, EXTRACT_NUMBER): don't bother to cast here.
563 Tue Jun 2 07:37:53 1992 Karl Berry (karl@hayley)
567 * Change copyrights to `1985, 89, ...'.
569 * regex.h (REG_RE_MATCH_NULL_AT_END): new macro.
570 * xregex.c (re_compile_fastmap): initialize `can_be_null' to
571 `p==pend', instead of in the test at the top of the loop (as
572 it was, it was always being set).
573 Also, set `can_be_null'=1 if we would jump to the end of the
574 pattern in the `on_failure_jump' cases.
575 (re_search_2): check if `can_be_null' is 1, not nonzero. This
576 was the original test in rms' regex; why did we change this?
578 * xregex.c (re_compile_fastmap): rename `is_a_succeed_n' to
581 Sat May 30 08:09:08 1992 Karl Berry (karl@hayley)
583 * xregex.c (re_compile_pattern): declare `regnum' as `unsigned',
584 not `regnum_t', for the benefit of those patterns with more
587 * xregex.c: rename `failure_stack' to `fail_stack', for brevity;
588 likewise for `match_nothing' to `match_null'.
590 * regexinc.c (REGEX_REALLOCATE): take both the new and old
591 sizes, and copy only the old bytes.
592 * xregex.c (DOUBLE_FAILURE_STACK): pass both old and new.
593 * This change from Thorsten Ohl.
595 Fri May 29 11:45:22 1992 Karl Berry (karl@hayley)
597 * regexinc.c (SIGN_EXTEND_CHAR): define as `(signed char) c'
598 instead of relying on __CHAR_UNSIGNED__, to work with
599 compilers other than GCC. From Per Bothner.
601 * main.c (main): change return type to `int'.
603 Mon May 18 06:37:08 1992 Karl Berry (karl@hayley)
605 * regex.h (RE_SYNTAX_AWK): typo in RE_RE_UNMATCHED...
607 Fri May 15 10:44:46 1992 Karl Berry (karl@hayley)
611 Sun May 3 13:54:00 1992 Karl Berry (karl@hayley)
613 * regex.h (struct re_pattern_buffer): now it's just `regs_allocated'.
614 (REGS_UNALLOCATED, REGS_REALLOCATE, REGS_FIXED): new constants.
615 * xregex.c (regexec, re_compile_pattern): set the field appropriately.
616 (re_match_2): and use it. bufp can't be const any more.
618 Fri May 1 15:43:09 1992 Karl Berry (karl@hayley)
620 * regexinc.c: unconditionally include <sys/types.h>, first.
622 * regex.h (struct re_pattern_buffer): rename
623 `caller_allocated_regs' to `regs_allocated_p'.
624 * xregex.c (re_compile_pattern): same change here.
626 (re_match_2): reallocate registers if necessary.
628 Fri Apr 10 07:46:50 1992 Karl Berry (karl@hayley)
630 * regex.h (RE_SYNTAX{_POSIX,}_AWK): new definitions from Arnold.
632 Sun Mar 15 07:34:30 1992 Karl Berry (karl at hayley)
634 * GNUmakefile (dist): versionize regex.{c,h,texinfo}.
636 Tue Mar 10 07:05:38 1992 Karl Berry (karl at hayley)
640 * xregex.c (PUSH_FAILURE_POINT): always increment the failure id.
641 (DEBUG_STATEMENT) [DEBUG]: execute the statement even if `debug'==0.
643 * xregex.c (pop_failure_point): if the saved string location is
644 null, keep the current value.
645 (re_match_2): at fail, test for a dummy failure point by
646 checking the restored pattern value, not string value.
647 (re_match_2): new case, `on_failure_keep_string_jump'.
648 (regex_compile): output this opcode in the .*\n case.
649 * regexinc.c (re_opcode_t): define the opcode.
650 (partial_compiled_pattern_pattern): add the new case.
652 Mon Mar 9 09:09:27 1992 Karl Berry (karl at hayley)
654 * xregex.c (regex_compile): optimize .*\n to output an
655 unconditional jump to the ., instead of pushing failure points
656 each time through the loop.
658 * xregex.c (DOUBLE_FAILURE_STACK): compute the maximum size
659 ourselves (and correctly); change callers.
661 Sun Mar 8 17:07:46 1992 Karl Berry (karl at hayley)
663 * xregex.c (failure_stack_elt_t): change to `const char *', to
666 * regex.h (re_set_syntax): declare this.
668 * xregex.c (pop_failure_point) [DEBUG]: conditionally pass the
669 original strings and sizes; change callers.
671 Thu Mar 5 16:35:35 1992 Karl Berry (karl at claude.cs.umb.edu)
673 * xregex.c (regnum_t): new type for register/group numbers.
674 (compile_stack_elt_t, regex_compile): use it.
676 * xregex.c (regexec): declare len as `int' to match re_search.
678 * xregex.c (re_match_2): don't declare p1 twice.
680 * xregex.c: change `while (1)' to `for (;;)' to avoid silly
683 * regex.h [__STDC__]: use #if, not #ifdef.
685 * regexinc.c (REGEX_REALLOCATE): cast the result of alloca to
686 (char *), to avoid warnings.
688 * xregex.c (regerror): declare variable as const.
690 * xregex.c (re_compile_pattern, re_comp): define as returning a const
692 * regex.h (re_compile_pattern, re_comp): likewise.
694 Thu Mar 5 15:57:56 1992 Karl Berry (karl@hal)
696 * xregex.c (regcomp): declare `syntax' as unsigned.
698 * xregex.c (re_match_2): try to avoid compiler warnings about
699 unsigned comparisons.
701 * GNUmakefile (test-xlc): new target.
703 * regex.h (reg_errcode_t): remove trailing comma from definition.
704 * regexinc.c (re_opcode_t): likewise.
706 Thu Mar 5 06:56:07 1992 Karl Berry (karl at hayley)
708 * GNUmakefile (dist): add version numbers automatically.
709 (versionfiles): new variable.
710 (regex.{c,texinfo}): don't add version numbers here.
711 * regex.h: put in placeholder instead of the version number.
713 Fri Feb 28 07:11:33 1992 Karl Berry (karl at hayley)
715 * xregex.c (re_error_msg): declare const, since it is.
717 Sun Feb 23 05:41:57 1992 Karl Berry (karl at fosse)
719 * xregex.c (PAT_PUSH{,_2,_3}, ...): cast args to avoid warnings.
720 (regex_compile, regexec): return REG_NOERROR, instead
722 (boolean): define as char, and #define false and true.
723 * regexinc.c (STREQ): cast the result.
725 Sun Feb 23 07:45:38 1992 Karl Berry (karl at hayley)
727 * GNUmakefile (test-cc, test-hc, test-pcc): new targets.
729 * regex.inc (extract_number, extract_number_and_incr) [DEBUG]:
730 only define if we are debugging.
732 * xregex.c [_AIX]: do #pragma alloca first if necessary.
733 * regexinc.c [_AIX]: remove the #pragma from here.
735 * regex.h (reg_syntax_t): declare as unsigned, and redo the enum
736 as #define's again. Some compilers do stupid things with enums.
738 Thu Feb 20 07:19:47 1992 Karl Berry (karl at hayley)
742 * xregex.c, regex.h (newline_anchor_match_p): rename to
743 `newline_anchor'; dumb idea to change the name.
745 Tue Feb 18 07:09:02 1992 Karl Berry (karl at hayley)
747 * regexinc.c: go back to original, i.e., don't include
748 <string.h> or define strchr.
749 * xregex.c (regexec): don't bother with adding characters after
750 newlines to the fastmap; instead, just don't use a fastmap.
751 * xregex.c (regcomp): set the buffer and fastmap fields to zero.
753 * xregex.texinfo (GNU r.e. compiling): have to initialize more
756 * regex.h (struct re_pattern_buffer): rename `newline_anchor' to
757 `newline_anchor_match_p', as we're back to two cases.
758 * xregex.c (regcomp, re_compile_pattern, re_comp): change
760 (re_match_2): at begline and endline, POSIX is not a special
761 case anymore; just check newline_anchor_match_p.
763 Thu Feb 13 16:29:33 1992 Karl Berry (karl at hayley)
765 * xregex.c (*empty_string*): rename to *null_string*, for brevity.
767 Wed Feb 12 06:36:22 1992 Karl Berry (karl at hayley)
769 * xregex.c (re_compile_fastmap): at endline, don't set fastmap['\n'].
770 (re_match_2): rewrite the begline/endline cases to take account
771 of the new field newline_anchor.
773 Tue Feb 11 14:34:55 1992 Karl Berry (karl at hayley)
775 * regexinc.c [!USG etc.]: include <strings.h> and define strchr
778 * xregex.c (re_search_2): when searching backwards, declare `c'
779 as a char and use casts when using it as an array subscript.
781 * xregex.c (regcomp): if REG_NEWLINE, set
782 RE_HAT_LISTS_NOT_NEWLINE. Set the `newline_anchor' field
784 (regex_compile): compile [^...] as matching a \n according to
786 (regexec): if doing REG_NEWLINE stuff, compile a fastmap and add
787 characters after any \n's to the newline.
788 * regex.h (RE_HAT_LISTS_NOT_NEWLINE): new syntax bit.
789 (struct re_pattern_buffer): rename `posix_newline' to
790 `newline_anchor', define constants for its values.
792 Mon Feb 10 07:22:50 1992 Karl Berry (karl at hayley)
794 * xregex.c (re_compile_fastmap): combine the code at the top and
795 bottom of the loop, as it's essentially identical.
797 Sun Feb 9 10:02:19 1992 Karl Berry (karl at hayley)
799 * xregex.texinfo (POSIX Translate Tables): remove this, as it
800 doesn't match the spec.
802 * xregex.c (re_compile_fastmap): if we finish off a path, go
803 back to the top (to set can_be_null) instead of returning
806 * xregex.texinfo: changes from bob.
808 Sat Feb 1 07:03:25 1992 Karl Berry (karl at hayley)
810 * xregex.c (re_search_2): doc fix (from rms).
812 Fri Jan 31 09:52:04 1992 Karl Berry (karl at hayley)
814 * xregex.texinfo (GNU Searching): clarify the range arg.
816 * xregex.c (re_match_2, at_endline_op_p): add extra parens to
817 get rid of GCC 2's (silly, IMHO) warning about && within ||.
819 * xregex.c (common_op_match_empty_string_p): use
820 MATCH_NOTHING_UNSET_VALUE, not -1.
822 Thu Jan 16 08:43:02 1992 Karl Berry (karl at hayley)
824 * xregex.c (SET_REGS_MATCHED): only set the registers from
827 * regexinc.c (MIN): new macro.
828 * xregex.c (re_match_2): only check min (num_regs,
829 regs->num_regs) when we set the returned regs.
831 * xregex.c (re_match_2): set registers after the first
832 num_regs to -1 before we return.
834 Tue Jan 14 16:01:42 1992 Karl Berry (karl at hayley)
836 * xregex.c (re_match_2): initialize max (RE_NREGS, re_nsub + 1)
837 registers (from rms).
839 * xregex.c, regex.h: don't abbreviate `19xx' to `xx'.
841 * regexinc.c [!emacs]: include <sys/types.h> before <unistd.h>.
842 (from ro@thp.Uni-Koeln.DE).
844 Thu Jan 9 07:23:00 1992 Karl Berry (karl at hayley)
846 * xregex.c (*unmatchable): rename to `match_empty_string_p'.
847 (CAN_MATCH_NOTHING): rename to `REG_MATCH_EMPTY_STRING_P'.
849 * regexinc.c (malloc, realloc): remove prototypes, as they can
850 cause clashes (from rms).
852 Mon Jan 6 12:43:24 1992 Karl Berry (karl at claude.cs.umb.edu)
856 Sun Jan 5 10:50:38 1992 Karl Berry (karl at hayley)
858 * xregex.texinfo: bring more or less up-to-date.
859 * GNUmakefile (regex.texinfo): generate from regex.h and
861 * include.awk: new file.
863 * xregex.c: change all calls to the fn extract_number_and_incr
866 * xregex.c (re_match_2) [emacs]: in at_dot, use PTR_CHAR_POS + 1,
867 instead of bf_* and sl_*. Cast d to unsigned char *, to match
868 the declaration in Emacs' buffer.h.
869 [emacs19]: in before_dot, at_dot, and after_dot, likewise.
871 * regexinc.c: unconditionally include <sys/types.h>.
873 * regexinc.c (alloca) [!alloca]: Emacs config files sometimes
874 define this, so don't define it if it's already defined.
876 Sun Jan 5 06:06:53 1992 Karl Berry (karl at fosse)
878 * xregex.c (re_comp): fix type conflicts with regex_compile (we
879 haven't been compiling this).
881 * regexinc.c (SIGN_EXTEND_CHAR): use `__CHAR_UNSIGNED__', not
884 * regexinc.c (NULL) [!NULL]: define it (as zero).
886 * regexinc.c (extract_number): remove the temporaries.
888 Sun Jan 5 07:50:14 1992 Karl Berry (karl at hayley)
890 * regex.h (regerror) [!__STDC__]: return a size_t, not a size_t *.
892 * xregex.c (PUSH_FAILURE_POINT, ...): declare `destination' as
893 `char *' instead of `void *', to match alloca declaration.
895 * xregex.c (regerror): use `size_t' for the intermediate values
896 as well as the return type.
898 * xregex.c (regexec): cast the result of malloc.
900 * xregex.c (regexec): don't initialize `private_preg' in the
901 declaration, as old C compilers can't do that.
903 * xregex.c (main) [test]: declare printchar void.
905 * xregex.c (assert) [!DEBUG]: define this to do nothing, and
906 remove #ifdef DEBUG's from around asserts.
908 * xregex.c (re_match_2): remove error message when not debugging.
910 Sat Jan 4 09:45:29 1992 Karl Berry (karl at hayley)
912 * other.c: test the bizarre duplicate case in re_compile_fastmap
915 * test.c (general_test): don't test registers beyond the end of
916 correct_regs, as well as regs.
918 * xregex.c (regex_compile): at handle_close, don't assign to
919 *inner_group_loc if we didn't push a start_memory (because the
920 group number was too big). In fact, don't push or pop the
921 inner_group_offset in that case.
923 * regex.c: rename to xregex.c, since it's not the whole thing.
924 * regex.texinfo: likewise.
925 * GNUmakefile: change to match.
927 * regex.c [DEBUG]: only include <stdio.h> if debugging.
929 * regexinc.c (SIGN_EXTEND_CHAR) [CHAR_UNSIGNED]: if it's already
930 defined, don't redefine it.
932 * regex.c: define _GNU_SOURCE at the beginning.
933 * regexinc.c (isblank) [!isblank]: define it.
934 (isgraph) [!isgraph]: change conditional to this, and remove the
937 * regex.c (regex_compile): add `blank' character class.
939 * regex.c (regex_compile): don't use a uchar variable to loop
940 through all characters.
942 * regex.c (regex_compile): at '[', improve logic for checking
943 that we have enough space for the charset.
945 * regex.h (struct re_pattern_buffer): declare translate as char
946 * again. We only use it as an array subscript once, I think.
948 * regex.c (TRANSLATE): new macro to cast the data character
950 (num_internal_regs): rename to `num_regs'.
952 Fri Jan 3 07:58:01 1992 Karl Berry (karl at hayley)
954 * regex.h (struct re_pattern_buffer): declare `allocated' and
955 `used' as unsigned long, since these are never negative.
957 * regex.c (compile_stack_element): rename to compile_stack_elt_t.
958 (failure_stack_element): similarly.
960 * regexinc.c (TALLOC, RETALLOC): new macros to simplify
961 allocation of arrays.
963 * regex.h (re_*) [__STDC__]: don't declare string args unsigned
964 char *; that makes them incompatible with string constants.
965 (struct re_pattern_buffer): declare the pattern and translate
966 table as unsigned char *.
967 * regex.c (most routines): use unsigned char vs. char consistently.
969 * regex.h (re_compile_pattern): do not declare the length arg as
971 * regex.c (re_compile_pattern): likewise.
973 * regex.c (POINTER_TO_REG): rename to `POINTER_TO_OFFSET'.
975 * regex.h (re_registers): declare `start' and `end' as
976 `regoff_t', instead of `int'.
978 * regex.c (regexec): if either of the malloc's for the register
979 information fail, return failure.
981 * regex.h (RE_NREGS): define this again, as 30 (from jla).
982 (RE_ALLOCATE_REGISTERS): remove this.
983 (RE_SYNTAX_*): remove it from definitions.
984 (re_pattern_buffer): remove `return_default_num_regs', add
985 `caller_allocated_regs'.
986 * regex.c (re_compile_pattern): clear no_sub and
987 caller_allocated_regs in the pattern.
988 (regcomp): set caller_allocated_regs.
989 (re_match_2): do all register allocation at the end of the
990 match; implement new semantics.
992 * regex.c (MAX_REGNUM): new macro.
993 (regex_compile): at handle_open and handle_close, if the group
994 number is too large, don't push the start/stop memory.
996 Thu Jan 2 07:56:10 1992 Karl Berry (karl at hayley)
998 * regex.c (re_match_2): if the back reference is to a group that
999 never matched, then goto fail, not really_fail. Also, don't
1000 test if the pattern can match the empty string. Why did we
1002 (really_fail): this label no longer needed.
1004 * regexinc.c [STDC_HEADERS]: use only this to test if we should
1007 * regex.c (DO_RANGE, regex_compile): translate in all cases
1008 except the single character after a \.
1010 * regex.h (RE_AWK_CLASS_HACK): rename to
1011 RE_BACKSLASH_ESCAPE_IN_LISTS.
1012 * regex.c (regex_compile): change use.
1014 * regex.c (re_compile_fastmap): do not translate the characters
1015 again; we already translated them at compilation. (From ylo@ngs.fi.)
1017 * regex.c (re_match_2): in case for at_dot, invert sense of
1018 comparison and find the character number properly. (From
1019 worley@compass.com.)
1020 (re_match_2) [emacs]: remove the cases for before_dot and
1021 after_dot, since there's no way to specify them, and the code
1022 is wrong (judging from this change).
1024 Wed Jan 1 09:13:38 1992 Karl Berry (karl at hayley)
1026 * psx-{interf,basic,extend}.c, other.c: set `t' as the first
1027 thing, so that if we run them in sucession, general_test's
1028 kludge to see if we're doing POSIX tests works.
1030 * test.h (test_type): add `all_test'.
1031 * main.c: add case for `all_test'.
1033 * regexinc.c (partial_compiled_pattern_printer,
1034 double_string_printer): don't print anything if we're passed null.
1036 * regex.c (PUSH_FAILURE_POINT): do not scan for the highest and
1037 lowest active registers.
1038 (re_match_2): compute lowest/highest active regs at start_memory and
1040 (NO_{LOW,HIGH}EST_ACTIVE_REG): new sentinel values.
1041 (pop_failure_point): return the lowest/highest active reg values
1042 popped; change calls.
1044 * regex.c [DEBUG]: include <assert.h>.
1045 (various routines) [DEBUG]: change conditionals to assertions.
1047 * regex.c (DEBUG_STATEMENT): new macro.
1048 (PUSH_FAILURE_POINT): use it to increment num_regs_pushed.
1049 (re_match_2) [DEBUG]: only declare num_regs_pushed if DEBUG.
1051 * regex.c (*can_match_nothing): rename to *unmatchable.
1053 * regex.c (re_match_2): at stop_memory, adjust argument reading.
1055 * regex.h (re_pattern_buffer): declare `can_be_null' as a 2-bit
1058 * regex.h (re_pattern_buffer): declare `buffer' unsigned char *;
1059 no, dumb idea. The pattern can have signed number.
1061 * regex.c (re_match_2): in maybe_pop_jump case, skip over the
1062 right number of args to the group operators, and don't do
1063 anything with endline if posix_newline is not set.
1065 * regex.c, regexinc.c (all the things we just changed): go back
1066 to putting the inner group count after the start_memory,
1067 because we need it in the on_failure_jump case in re_match_2.
1068 But leave it after the stop_memory also, since we need it
1069 there in re_match_2, and we don't have any way of getting back
1070 to the start_memory.
1072 * regexinc.c (partial_compiled_pattern_printer): adjust argument
1073 reading for start/stop_memory.
1074 * regex.c (re_compile_fastmap, group_can_match_nothing): likewise.
1076 Tue Dec 31 10:15:08 1991 Karl Berry (karl at hayley)
1078 * regex.c (bits list routines): remove these.
1079 (re_match_2): get the number of inner groups from the pattern,
1080 instead of keeping track of it at start and stop_memory.
1081 Put the count after the stop_memory, not after the
1083 (compile_stack_element): remove `fixup_inner_group' member,
1084 since we now put it in when we can compute it.
1085 (regex_compile): at handle_open, don't push the inner group
1086 offset, and at handle_close, don't pop it.
1088 * regex.c (level routines): remove these, and their uses in
1089 regex_compile. This was another manifestation of having to find
1090 $'s that were endlines.
1092 * regex.c (regexec): this does searching, not matching (a
1093 well-disguised part of the standard). So rewrite to use
1094 `re_search' instead of `re_match'.
1095 * psx-interf.c (test_regexec): add tests to, uh, match.
1097 * regex.h (RE_TIGHT_ALT): remove this; nobody uses it.
1098 * regex.c: remove the code that was supposed to implement it.
1100 * other.c (test_others): ^ and $ never match newline characters;
1101 RE_CONTEXT_INVALID_OPS doesn't affect anchors.
1103 * psx-interf.c (test_regerror): update for new error messages.
1105 * psx-extend.c: it's now ok to have an alternative be just a $,
1106 so remove all the tests which supposed that was invalid.
1108 Wed Dec 25 09:00:05 1991 Karl Berry (karl at hayley)
1110 * regex.c (regex_compile): in handle_open, don't skip over ^ and
1111 $ when checking for an empty group. POSIX has changed the
1113 * psx-extend.c (test_posix_extended): thus, move (^$) tests to
1116 * regexinc.c (boolean): move from here to test.h and regex.c.
1117 * test files: declare verbose, omit_register_tests, and
1118 test_should_match as boolean.
1120 * psx-interf.c (test_posix_c_interface): remove the `c_'.
1123 * psx-basic.c (test_posix_basic): ^ ($) is an anchor after
1124 (before) an open (close) group.
1126 * regex.c (re_match_2): in endline, correct precedence of
1127 posix_newline condition.
1129 Tue Dec 24 06:45:11 1991 Karl Berry (karl at hayley)
1131 * test.h: incorporate private-tst.h.
1132 * test files: include test.h, not private-tst.h.
1134 * test.c (general_test): set posix_newline to zero if we are
1135 doing POSIX tests (unfortunately, it's difficult to call
1136 regcomp in this case, which is what we should really be doing).
1138 * regex.h (reg_syntax_t): make this an enumeration type which
1139 defines the syntax bits; renames re_syntax_t.
1141 * regex.c (at_endline_op_p): don't preincrement p; then if it's
1142 not an empty string op, we lose.
1144 * regex.h (reg_errcode_t): new enumeration type of the error
1146 * regex.c (regex_compile): return that type.
1148 * regex.c (regex_compile): in [, initialize
1149 just_had_a_char_class to false; somehow I had changed this to
1152 * regex.h (RE_NO_CONSECUTIVE_REPEATS): remove this, since we
1153 don't use it, and POSIX doesn't require this behavior anymore.
1154 * regex.c (regex_compile): remove it from here.
1156 * regex.c (regex_compile): remove the no_op insertions for
1157 verify_and_adjust_endlines, since that doesn't exist anymore.
1159 * regex.c (regex_compile) [DEBUG]: use printchar to print the
1160 pattern, so unprintable bytes will print properly.
1162 * regex.c: move re_error_msg back.
1163 * test.c (general_test): print the compile error if the pattern
1166 Mon Dec 23 08:54:53 1991 Karl Berry (karl at hayley)
1168 * regexinc.c: move re_error_msg here.
1170 * regex.c (re_error_msg): the ``message'' for success must be
1171 NULL, to keep the interface to re_compile_pattern the same.
1172 (regerror): if the msg is null, use "Success".
1174 * rename most test files for consistency. Change Makefile
1177 * test.c (most routines): add casts to (unsigned char *) when we
1178 call re_{match,search}{,_2}.
1180 Sun Dec 22 09:26:06 1991 Karl Berry (karl at hayley)
1182 * regex.c (re_match_2): declare string args as unsigned char *
1183 again; don't declare non-pointer args const; declare the
1184 pattern buffer const.
1185 (re_match): likewise.
1186 (re_search_2, re_search): likewise, except don't declare the
1187 pattern const, since we make a fastmap.
1188 * regex.h [__STDC__]: change prototypes.
1190 * regex.c (regex_compile): return an error code, not a string.
1191 (re_err_list): new table to map from error codes to string.
1192 (re_compile_pattern): return an element of re_err_list.
1193 (regcomp): don't test all the strings.
1194 (regerror): just use the list.
1195 (put_in_buffer): remove this.
1197 * regex.c (equivalent_failure_points): remove this.
1199 * regex.c (re_match_2): don't copy the string arguments into
1200 non-const pointers. We never alter the data.
1202 * regex.c (re_match_2): move assignment to `is_a_jump_n' out of
1203 the main loop. Just initialize it right before we do
1206 * regex.[ch] (re_match_2): don't declare the int parameters const.
1208 Sat Dec 21 08:52:20 1991 Karl Berry (karl at hayley)
1210 * regex.h (re_syntax_t): new type; declare to be unsigned
1211 (previously we used int, but since we do bit operations on
1212 this, unsigned is better, according to H&S).
1213 (obscure_syntax, re_pattern_buffer): use that type.
1214 * regex.c (re_set_syntax, regex_compile): likewise.
1216 * regex.h (re_pattern_buffer): new field `posix_newline'.
1217 * regex.c (re_comp, re_compile_pattern): set to zero.
1218 (regcomp): set to REG_NEWLINE.
1219 * regex.h (RE_HAT_LISTS_NOT_NEWLINE): remove this (we can just
1220 check `posix_newline' instead.)
1222 * regex.c (op_list_type, op_list, add_op): remove these.
1223 (verify_and_adjust_endlines): remove this.
1224 (pattern_offset_list_type, *pattern_offset* routines): and these.
1225 These things all implemented the nonleading/nontrailing position
1226 code, which was very long, had a few remaining problems, and
1227 is no longer needed. So...
1229 * regexinc.c (STREQ): new macro to abbreviate strcmp(,)==0, for
1230 brevity. Change various places in regex.c to use it.
1232 * regex{,inc}.c (enum regexpcode): change to a typedef
1233 re_opcode_t, for brevity.
1235 * regex.h (re_syntax_table) [SYNTAX_TABLE]: remove this; it
1236 should only be in regex.c, I think, since we don't define it
1237 in this case. Maybe it should be conditional on !SYNTAX_TABLE?
1239 * regexinc.c (partial_compiled_pattern_printer): simplify and
1240 distinguish the emacs/not-emacs (not)wordchar cases.
1242 Fri Dec 20 08:11:38 1991 Karl Berry (karl at hayley)
1244 * regexinc.c (regexpcode) [emacs]: only define the Emacs opcodes
1245 if we are ifdef emacs.
1247 * regex.c (BUF_PUSH*): rename to PAT_PUSH*.
1249 * regex.c (regex_compile): in $ case, go back to essentially the
1250 original code for deciding endline op vs. normal char.
1251 (at_endline_op_p): new routine.
1252 * regex.h (RE_ANCHORS_ONLY_AT_ENDS, RE_CONTEXT_INVALID_ANCHORS,
1253 RE_REPEATED_ANCHORS_AWAY, RE_NO_ANCHOR_AT_NEWLINE): remove
1254 these. POSIX has simplified the rules for anchors in draft
1256 (RE_NEWLINE_ORDINARY): new syntax bit.
1257 (RE_CONTEXT_INDEP_ANCHORS): change description to be compatible
1259 * regex.texinfo (Syntax Bits): remove the descriptions.
1261 Mon Dec 16 08:12:40 1991 Karl Berry (karl at hayley)
1263 * regex.c (re_match_2): in jump_past_next_alt, unconditionally
1264 goto no_pop. The only register we were finding was one which
1265 enclosed the whole alternative expression, not one around an
1266 individual alternative. So we were never doing what we
1267 thought we were doing, and this way makes (|a) against the
1270 * regex.c (regex_compile): remove `highest_ever_regnum', and
1271 don't restore regnum from the stack; just put it into a
1272 temporary to put into the stop_memory. Otherwise, groups
1273 aren't numbered consecutively.
1275 * regex.c (is_in_compile_stack): rename to
1276 `group_in_compile_stack'; remove unnecessary test for the
1279 * regex.c (re_match_2): in on_failure_jump, skip no_op's before
1280 checking for the start_memory, in case we were called from
1283 Sun Dec 15 16:20:48 1991 Karl Berry (karl at hayley)
1285 * regex.c (regex_compile): in duplicate case, use
1286 highest_ever_regnum instead of regnum, since the latter is
1287 reverted at stop_memory.
1289 * regex.c (re_match_2): in on_failure_jump, if the * applied to
1290 a group, save the information for that group and all inner
1291 groups (by making it active), even though we're not inside it
1294 Sat Dec 14 09:50:59 1991 Karl Berry (karl at hayley)
1296 * regex.c (PUSH_FAILURE_ITEM, POP_FAILURE_ITEM): new macros.
1297 Use them instead of copying the stack manipulating a zillion
1300 * regex.c (PUSH_FAILURE_POINT, pop_failure_point) [DEBUG]: save
1301 and restore a unique identification value for each failure point.
1303 * regexinc.c (partial_compiled_pattern_printer): don't print an
1304 extra / after duplicate commands.
1306 * regex.c (regex_compile): in back-reference case, allow a back
1307 reference to register `regnum'. Otherwise, even `\(\)\1'
1308 fails, since regnum is 1 at the back-reference.
1310 * regex.c (re_match_2): in fail, don't examine the pattern if we
1313 * test_private.h: rename to private_tst.h. Change includes.
1315 * regex.c (extend_bits_list): compute existing size for realloc
1316 in bytes, not blocks.
1318 * regex.c (re_match_2): in jump_past_next_alt, the for loop was
1319 missing its (empty) statement. Even so, some register tests
1320 still fail, although in a different way than in the previous change.
1322 Fri Dec 13 15:55:08 1991 Karl Berry (karl at hayley)
1324 * regex.c (re_match_2): in jump_past_next_alt, unconditionally
1325 goto no_pop, since we weren't properly detecting if the
1326 alternative matched something anyway. No, we need to not jump
1327 to keep the register values correct; just change to not look at
1328 register zero and not test RE_NO_EMPTY_ALTS (which is a
1329 compile-time thing).
1331 * regex.c (SET_REGS_MATCHED): start the loop at 1, since we never
1332 care about register zero until the very end. (I think.)
1334 * regex.c (PUSH_FAILURE_POINT, pop_failure_point): go back to
1335 pushing and popping the active registers, instead of only doing
1336 the registers before a group: (fooq|fo|o)*qbar against fooqbar
1337 fails, since we restore back into the middle of group 1, yet it
1338 isn't active, because the previous restore clobbered the active flag.
1340 Thu Dec 12 17:25:36 1991 Karl Berry (karl at hayley)
1342 * regex.c (PUSH_FAILURE_POINT): do not call
1343 `equivalent_failure_points' after all; it causes the registers
1344 to be ``wrong'' (according to POSIX), and an infinite loop on
1345 `((a*)*)*' against `ab'.
1347 * regex.c (re_compile_fastmap): don't push `pend' on the failure
1350 Tue Dec 10 10:30:03 1991 Karl Berry (karl at hayley)
1352 * regex.c (PUSH_FAILURE_POINT): if pushing same failure point that
1353 is on the top of the stack, fail.
1354 (equivalent_failure_points): new routine.
1356 * regex.c (re_match_2): add debug statements for every opcode we
1359 * regex.c (regex_compile/handle_close): restore
1360 `fixup_inner_group_count' and `regnum' from the stack.
1362 Mon Dec 9 13:51:15 1991 Karl Berry (karl at hayley)
1364 * regex.c (PUSH_FAILURE_POINT): declare `this_reg' as int, so
1365 unsigned arithmetic doesn't happen when we don't want to save
1368 Tue Dec 3 08:11:10 1991 Karl Berry (karl at hayley)
1370 * regex.c (extend_bits_list): divide size by bits/block.
1372 * regex.c (init_bits_list): remove redundant assignmen to
1375 * regexinc.c (partial_compiled_pattern_printer): don't do *p++
1376 twice in the same expr.
1378 * regex.c (re_match_2): at on_failure_jump, use the correct
1379 pattern positions for getting the stuff following the start_memory.
1381 * regex.c (struct register_info): remove the bits_list for the
1382 inner groups; make that a separate variable.
1384 Mon Dec 2 10:42:07 1991 Karl Berry (karl at hayley)
1386 * regex.c (PUSH_FAILURE_POINT): don't pass `failure_stack' as an
1387 arg; change callers.
1389 * regex.c (PUSH_FAILURE_POINT): print items in order they are
1391 (pop_failure_point): likewise.
1393 * regex.c (main): prompt for the pattern and string.
1395 * regex.c (FREE_VARIABLES) [!REGEX_MALLOC]: declare as nothing;
1396 remove #ifdefs from around calls.
1398 * regex.c (extract_number, extract_number_and_incr): declare static.
1400 * regex.c: remove the canned main program.
1402 * Makefile (COMMON): add main.o.
1404 Tue Sep 24 06:26:51 1991 Kathy Hargreaves (kathy at fosse)
1406 * regex.c (re_match_2): Made `pend' and `dend' not register variables.
1407 Only set string2 to string1 if string1 isn't null.
1408 Send address of p, d, regstart, regend, and reg_info to
1410 Put in more debug statements.
1412 * regex.c [debug]: Added global variable.
1413 (DEBUG_*PRINT*): Only print if `debug' is true.
1414 (DEBUG_DOUBLE_STRING_PRINTER): Changed DEBUG_STRING_PRINTER's
1416 Changed some comments.
1417 (PUSH_FAILURE_POINT): Moved and added some debugging statements.
1418 Was saving regstart on the stack twice instead of saving both
1419 regstart and regend; remedied this.
1420 [NUM_REGS_ITEMS]: Changed from 3 to 4, as now save lowest and
1421 highest active registers instead of highest used one.
1422 [NUM_NON_REG_ITEMS]: Changed name of NUM_OTHER_ITEMS to this.
1423 (NUM_FAILURE_ITEMS): Use active registers instead of number 0
1424 through highest used one.
1425 (re_match_2): Have pop_failure_point put things in the variables.
1426 (pop_failure_point): Have it do what the fail case in re_match_2
1427 did with the failure stack, instead of throwing away the stuff
1428 popped off. re_match_2 can ignore results when it doesn't
1432 Thu Sep 5 13:23:28 1991 Kathy Hargreaves (kathy at fosse)
1434 * regex.c (banner): Changed copyright years to be separate.
1436 * regex.c [CHAR_UNSIGNED]: Put __ at both ends of this name.
1437 [DEBUG, debug_count, *debug_p, DEBUG_PRINT_1, DEBUG_PRINT_2,
1438 DEBUG_COMPILED_PATTERN_PRINTER ,DEBUG_STRING_PRINTER]:
1439 defined these for debugging.
1440 (extract_number): Added this (debuggable) routine version of
1441 the macro EXTRACT_NUMBER. Ditto for EXTRACT_NUMBER_AND_INCR.
1442 (re_compile_pattern): Set return_default_num_regs if the
1443 syntax bit RE_ALLOCATE_REGISTERS is set.
1444 [REGEX_MALLOC]: Renamed USE_ALLOCA to this.
1445 (BUF_POP): Got rid of this, as don't ever use it.
1446 (regex_compile): Made the type of `pattern' not be register.
1447 If DEBUG, print the pattern to compile.
1448 (re_match_2): If had a `$' in the pattern before a `^' then
1449 don't record the `^' as an anchor.
1450 Put (enum regexpcode) before references to b, as suggested
1451 [RE_NO_BK_BRACES]: Changed RE_NO_BK_CURLY_BRACES to this.
1452 (remove_pattern_offset): Removed this unused routine.
1453 (PUSH_FAILURE_POINT): Changed to only save active registers.
1454 Put in debugging statements.
1455 (re_compile_fastmap): Made `pattern' not a register variable.
1456 Use routine for extracting numbers instead of macro.
1457 (re_match_2): Made `p', `mcnt' and `mcnt2' not register variables.
1458 Added `num_regs_pushed' for debugging.
1459 Only malloc registers if the syntax bit RE_ALLOCATE_REGISTERS is set.
1460 Put in debug statements.
1461 Put the macro NOTE_INNER_GROUP's code inline, as it was the
1462 only called in one place.
1463 For debugging, extract numbers using routines instead of macros.
1464 In case fail: only restore pushed active registers, and added
1465 debugging statements.
1466 (pop_failure_point): Test for underfull stack.
1467 (group_can_match_nothing, common_op_can_match_nothing): For
1468 debugging, extract numbers using routines instead of macros.
1469 (regexec): Changed formal parameters to not be prototypes.
1470 Don't initialize `regs' or `private_preg' in their declarations.
1472 Tue Jul 23 18:38:36 1991 Kathy Hargreaves (kathy at hayley)
1474 * regex.h [RE_CONTEX_INDEP_OPS]: Moved the anchor stuff out of
1476 [RE_UNMATCHED_RIGHT_PAREN_ORD]: Defined this bit.
1477 [RE_CONTEXT_INVALID_ANCHORS]: Defined this bit.
1478 [RE_CONTEXT_INDEP_ANCHORS]: Defined this bit.
1479 Added RE_CONTEXT_INDEP_ANCHORS to all syntaxes which had
1480 RE_CONTEXT_INDEP_OPS.
1481 Took RE_ANCHORS_ONLY_AT_ENDS out of the POSIX basic syntax.
1482 Added RE_UNMATCHED_RIGHT_PAREN_ORD to the POSIX extended
1484 Took RE_REPEATED_ANCHORS_AWAY out of the POSIX extended syntax.
1485 Defined REG_NOERROR (which will probably have to go away again).
1486 Changed the type `off_t' to `regoff_t'.
1488 * regex.c: Changed some commments.
1489 (regex_compile): Added variable `had_an_endline' to keep track
1490 of if hit a `$' since the beginning of the pattern or the last
1491 alternative (if any).
1492 Changed RE_CONTEXT_INVALID_OPS and RE_CONTEXT_INDEP_OPS to
1493 RE_CONTEXT_INVALID_ANCHORS and RE_CONTEXT_INDEP_ANCHORS where
1495 Put a `no_op' in the pattern if a repeat is only zero or one
1496 times; in this case and if it is many times (whereupon a jump
1497 backwards is pushed instead), keep track of the operator for
1498 verify_and_adjust_endlines.
1499 If RE_UNMATCHED_RIGHT_PAREN is set, make an unmatched
1500 close-group operator match `)'.
1501 Changed all error exits to exit (1).
1502 (remove_pattern_offset): Added this routine, but don't use it.
1503 (verify_and_adjust_endlines): At top of routine, if initialize
1504 routines run out of memory, return true after setting
1505 enough_memory false.
1506 At end of endline, et al. case, don't set *p to no_op.
1507 Repetition operators also set the level and active groups'
1508 match statuses, unless RE_REPEATED_ANCHORS_AWAY is set.
1509 (get_group_match_status): Put a return in front of call to get_bit.
1510 (re_compile_fastmap): Changed is_a_succeed_n to a boolean.
1511 If at end of pattern, then if the failure stack isn't empty,
1512 go back to the failure point.
1513 In *jump* case, only pop the stack if what's on top of it is
1514 where we've just jumped to.
1515 (re_search_2): Return -2 instead of val if val is -2.
1516 (group_can_match_nothing, alternative_can_match_nothing,
1517 common_op_can-match_nothing): Now pass in reg_info for the
1519 (re_match_2): Don't skip over the next alternative also if
1520 empty alternatives aren't allowed.
1521 In fail case, if failed to a backwards jump that's part of a
1522 repetition loop, pop the current failure point and use the
1524 (pop_failure_point): Check that there's as many register items
1525 on the failure stack as the stack says there are.
1526 (common_op_can_match_nothing): Added variables `ret' and
1527 `reg_no' so can set reg_info for the group encountered.
1528 Also break without doing anything if hit a no_op or the other
1529 kinds of `endline's.
1530 If not done already, set reg_info in start_memory case.
1531 Put in no_pop_jump for an optimized succeed_n of zero repetitions.
1532 In succeed_n case, if the number isn't zero, then return false.
1533 Added `duplicate' case.
1535 Sat Jul 13 11:27:38 1991 Kathy Hargreaves (kathy at hayley)
1537 * regex.h (REG_NOERROR): Added this error code definition.
1539 * regex.c: Took some redundant parens out of macros.
1540 (enum regexpcode): Added jump_past_next_alt.
1541 Wrapped some macros in `do..while (0)'.
1542 Changed some comments.
1543 (regex_compile): Use `fixup_alt_jump' instead of `fixup_jump'.
1544 Use `maybe_pop_jump' instead of `maybe_pop_failure_jump'.
1545 Use `jump_past_next_alt' instead of `no_pop_jump' when at the
1546 end of an alternative.
1547 (re_match_2): Used REGEX_ALLOCATE for the registers stuff.
1548 In stop_memory case: Add more boolean tests to see if the
1550 Added jump_past_next_alt case, which doesn't jump over the
1551 next alternative if the last one didn't match anything.
1552 Unfortunately, to make this work with, e.g., `(a+?*|b)*'
1553 against `bb', I also had to pop the alternative's failure
1554 point, which in turn broke backtracking!
1555 In fail case: Detect a dummy failure point by looking at
1556 failure_stack.avail - 2, not stack[-2].
1557 (pop_failure_point): Only pop if the stack isn't empty; don't
1558 give an error if it is. (Not sure yet this is correct.)
1559 (group_can_match_nothing): Make it return a boolean instead of int.
1560 Make it take an argument indicating the end of where it should look.
1561 If find a group that can match nothing, set the pointer
1562 argument to past the group in the pattern.
1563 Took out cases which can share with alternative_can_match_nothing
1564 and call common_op_can_match_nothing.
1565 Took ++ out of switch, so could call common_op_can_match_nothing.
1566 Wrote lots more for on_failure_jump case to handle alternatives.
1567 Main loop now doesn't look for matching stop_memory, but
1568 rather the argument END; return true if hit the matching
1569 stop_memory; this way can call itself for inner groups.
1570 (alternative_can_match_nothing): Added for alternatives.
1571 (common_op_can_match_nothing): Added for previous two routines'
1573 (regerror): Returns a message saying there's no error if gets
1576 Wed Jul 3 10:43:15 1991 Kathy Hargreaves (kathy at hayley)
1578 * regex.c: Removed unnecessary enclosing parens from several macros.
1579 Put `do..while (0)' around a few.
1580 Corrected some comments.
1581 (INIT_FAILURE_STACK_SIZE): Deleted in favor of using
1583 (INIT_FAILURE_STACK, DOUBLE_FAILURE_STACK, PUSH_PATTERN_OP,
1584 PUSH_FAILURE_POINT): Made routines of the same name (but with all
1585 lowercase letters) into these macros, so could use `alloca'
1586 when USE_ALLOCA is defined. The reason is stated below for
1587 bits lists. Deleted analogous routines.
1588 (re_compile_fastmap): Added variable void *destination for
1590 (re_match_2): Added variable void *destination for REGEX_REALLOCATE.
1591 Used the failure stack macros in place of the routines.
1592 Detected a dummy failure point by inspecting the failure stack's
1593 (avail - 2)th element, not failure_stack.stack[-2]. This bug
1594 arose when used the failure stack macros instead of the routines.
1596 * regex.c [USE_ALLOCA]: Put this conditional around previous
1597 alloca stuff and defined these to work differently depending
1598 on whether or not USE_ALLOCA is defined:
1599 (REGEX_ALLOCATE): Uses either `alloca' or `malloc'.
1600 (REGEX_REALLOCATE): Uses either `alloca' or `realloc'.
1601 (INIT_BITS_LIST, EXTEND_BITS_LIST, SET_BIT_TO_VALUE): Defined
1602 macro versions of routines with the same name (only with all
1603 lowercase letters) so could use `alloc' in re_match_2. This
1604 is to prevent core leaks when C-g is used in Emacs and to make
1605 things faster and avoid storage fragmentation. These things
1606 have to be macros because the results of `alloca' go away with
1607 the routine by which it's called.
1608 (BITS_BLOCK_SIZE, BITS_BLOCK, BITS_MASK): Moved to above the
1609 above-mentioned macros instead of before the routines defined
1610 below regex_compile.
1611 (set_bit_to_value): Compacted some code.
1612 (reg_info_type): Changed inner_groups field to be bits_list_type
1613 so could be arbitrarily long and thus handle arbitrary nesting.
1614 (NOTE_INNER_GROUP): Put `do...while (0)' around it so could
1616 Changed code to use bits lists.
1617 Added variable void *destination for REGEX_REALLOCATE (whose call
1618 is several levels in).
1619 Changed variable name of `this_bit' to `this_reg'.
1620 (FREE_VARIABLES): Only define and use if USE_ALLOCA is defined.
1621 (re_match_2): Use REGEX_ALLOCATE instead of malloc.
1622 Instead of setting INNER_GROUPS of reg_info to zero, have to
1623 use INIT_BITS_LIST and return -2 (and free variables if
1624 USE_ALLOCA isn't defined) if it fails.
1626 Fri Jun 28 13:45:07 1991 Karl Berry (karl at hayley)
1628 * regex.c (re_match_2): set value of `dend' when we restore `d'.
1630 * regex.c: remove declaration of alloca.
1632 * regex.c (MISSING_ISGRAPH): rename to `ISGRAPH_MISSING'.
1634 * regex.h [_POSIX_SOURCE]: remove these conditionals; always
1636 * regex.c (_POSIX_SOURCE): change conditionals to use `POSIX'
1639 Sat Jun 1 16:56:50 1991 Kathy Hargreaves (kathy at hayley)
1641 * regex.*: Changed RE_CONTEXTUAL_* to RE_CONTEXT_*,
1642 RE_TIGHT_VBAR to RE_TIGHT_ALT, RE_NEWLINE_OR to
1643 RE_NEWLINE_ALT, and RE_DOT_MATCHES_NEWLINE to RE_DOT_NEWLINE.
1645 Wed May 29 09:24:11 1991 Karl Berry (karl at hayley)
1647 * regex.texinfo (POSIX Pattern Buffers): cross-reference the
1648 correct node name (Match-beginning-of-line, not ..._line).
1649 (Syntax Bits): put @code around all syntax bits.
1651 Sat May 18 16:29:58 1991 Karl Berry (karl at hayley)
1653 * regex.c (global): add casts to keep broken compilers from
1654 complaining about malloc and realloc calls.
1656 * regex.c (isgraph) [MISSING_ISGRAPH]: change test to this,
1657 instead of `#ifndef isgraph', since broken compilers can't
1658 have both a macro and a symbol by the same name.
1660 * regex.c (re_comp, re_exec) [_POSIX_SOURCE]: do not define.
1661 (regcomp, regfree, regexec, regerror) [_POSIX_SOURCE && !emacs]:
1662 only define in this case.
1664 Mon May 6 17:37:04 1991 Kathy Hargreaves (kathy at hayley)
1666 * regex.h (re_search, re_search_2): Changed BUFFER to not be const.
1668 * regex.c (re_compile_pattern): `^' is in a leading position if
1669 it precedes a newline.
1670 (various routines): Added or changed header comments.
1671 (double_pattern_offsets_list): Changed name from
1672 `extend_pattern_offsets_list'.
1673 (adjust_pattern_offsets_list): Changed return value from
1675 (verify_and_adjust_endlines): Now returns `true' and `false'
1677 `$' is in a leading position if it follows a newline.
1678 (set_bit_to_value, get_bit_value): Exit with error if POSITION < 0
1679 so now calling routines don't have to.
1680 (init_failure_stack, inspect_failure_stack_top,
1681 pop_failure_stack_top, push_pattern_op, double_failure_stack):
1682 Now return value unsigned instead of boolean.
1683 (re_search, re_search_2): Changed BUFP to not be const.
1684 (re_search_2): Added variable const `private_bufp' to send to
1686 (push_failure_point): Made return value unsigned instead of boolean.
1688 Sat May 4 15:32:22 1991 Kathy Hargreaves (kathy at hayley)
1690 * regex.h (re_compile_fastmap): Added extern for this.
1691 Changed some comments.
1693 * regex.c (re_compile_pattern): In case handle_bar: put invalid
1694 pattern test before levels matching stuff.
1695 Changed some commments.
1696 Added optimizing test for detecting an empty alternative that
1697 ends with a trailing '$' at the end of the pattern.
1698 (re_compile_fastmap): Moved failure_stack stuff to before this
1699 so could use it. Made its stack dynamic.
1700 Made it return an int so that it could return -2 if its stack
1701 couldn't be allocated.
1702 Added to header comment (about the return values).
1703 (init_failure_stack): Wrote so both re_match_2 and
1704 re_compile_fastmap could use it similar stacks.
1705 (double_failure_stack): Added for above reasons.
1706 (push_pattern_op): Wrote for re_compile_fastmap.
1707 (re_search_2): Now return -2 if re_compile_fastmap does.
1708 (re_match_2): Made regstart and regend type failure_stack_element*.
1709 (push_failure_point): Made pattern_place and string_place type
1710 failure_stack_element*.
1711 Call double_failure_stack now.
1712 Return true instead of 1.
1714 Wed May 1 12:57:21 1991 Kathy Hargreaves (kathy at hayley)
1716 * regex.c (remove_intervening_anchors): Avoid erroneously making
1717 ops into no_op's by making them no_op only when they're beglines.
1718 (verify_and_adjust_endlines): Don't make '$' a normal character
1719 if it's before a newline.
1720 Look for the endline op in *p, not p[1].
1721 (failure_stack_element): Added this declaration.
1722 (failure_stack_type): Added this declaration.
1723 (INIT_FAILURE_STACK_SIZE, FAILURE_STACK_EMPTY,
1724 FAILURE_STACK_PTR_EMPTY, REMAINING_AVAIL_SLOTS): Added for
1726 (FAILURE_ITEM_SIZE, PUSH_FAILURE_POINT): Deleted.
1727 (FREE_VARIABLES): Now free failure_stack.stack instead of stackb.
1728 (re_match_2): deleted variables `initial_stack', `stackb',
1729 `stackp', and `stacke' and added `failure_stack' to replace them.
1730 Replaced calls to PUSH_FAILURE_POINT with those to
1732 (push_failure_point): Added for re_match_2.
1733 (pop_failure_point): Rewrote to use a failure_stack_type of stack.
1734 (can_match_nothing): Moved definition to below re_match_2.
1735 (bcmp_translate): Moved definition to below re_match_2.
1737 Mon Apr 29 14:20:54 1991 Kathy Hargreaves (kathy at hayley)
1739 * regex.c (enum regexpcode): Added codes endline_before_newline
1740 and repeated_endline_before_newline so could detect these
1741 types of endlines in the intermediate stages of a compiled
1743 (INIT_FAILURE_ALLOC): Renamed NFAILURES to this and set it to 5.
1744 (BUF_PUSH): Put `do {...} while 0' around this.
1745 (BUF_PUSH_2): Defined this to cut down on expansion of EXTEND_BUFFER.
1746 (regex_compile): Changed some comments.
1747 Now push endline_before_newline if find a `$' before a newline
1749 If a `$' might turn into an ordinary character, set laststart
1751 In '^' case, if syntax bit RE_TIGHT_VBAR is set, then for `^'
1752 to be in a leading position, it must be first in the pattern.
1753 Don't have to check in one of the else clauses that it's not set.
1754 If RE_CONTEXTUAL_INDEP_OPS isn't set but RE_ANCHORS_ONLY_AT_ENDS
1755 is, make '^' a normal character if it isn't first in the pattern.
1756 Can only detect at the end if a '$' after an alternation op is a
1757 trailing one, so can't immediately detect empty alternatives
1758 if a '$' follows a vbar.
1759 Added a picture of the ``success jumps'' in alternatives.
1760 Have to set bufp->used before calling verify_and_adjust_endlines.
1761 Also do it before returning all error strings.
1762 (remove_intervening_anchors): Now replaces the anchor with
1763 repeated_endline_before_newline if it's an endline_before_newline.
1764 (verify_and_adjust_endlines): Deleted SYNTAX parameter (could
1765 use bufp's) and added GROUP_FORWARD_MATCH_STATUS so could
1766 detect back references referring to empty groups.
1767 Added variable `bend' to point past the end of the pattern buffer.
1768 Added variable `previous_p' so wouldn't have to reinspect the
1769 pattern buffer to see what op we just looked at.
1770 Added endline_before_newline and repeated_endline_before_newline
1772 When checking if in a trailing position, added case where '$'
1773 has to be at the pattern's end if either of the syntax bits
1774 RE_ANCHORS_ONLY_AT_ENDS or RE_TIGHT_VBAR are set.
1775 Since `endline' can have the intermediate form `endline_in_repeat',
1776 have to change it to `endline' if RE_REPEATED_ANCHORS_AWAY
1778 Now disallow empty alternatives with trailing endlines in them
1779 if RE_NO_EMPTY_ALTS is set.
1780 Now don't make '$' an ordinary character if it precedes a newline.
1781 Don't make it an ordinary character if it's before a newline.
1782 Back references now affect the level matching something only if
1783 they refer to nonempty groups.
1784 (can_match_nothing): Now increment p1 in the switch, which
1785 changes many of the cases, but makes the code more like what
1786 it was derived from.
1787 Adjust the return statement to reflect above.
1788 (struct register_info): Made `can_match_nothing' field an int
1789 instead of a bit so could have -1 in it if never set.
1790 (MAX_FAILURE_ITEMS): Changed name from MAX_NUM_FAILURE_ITEMS.
1791 (FAILURE_ITEM_SIZE): Defined how much space a failure items uses.
1792 (PUSH_FAILURE_POINT): Changed variable `last_used_reg's name
1793 to `highest_used_reg'.
1794 Added variable `num_stack_items' and changed `len's name to
1796 Test failure stack limit in terms of number of items in it, not
1797 in terms of its length. rms' fix tested length against number
1798 of items, which was a misunderstanding.
1799 Use `realloc' instead of `alloca' to extend the failure stack.
1800 Use shifts instead of multiplying by 2.
1801 (FREE_VARIABLES): Free `stackb' instead of `initial_stack', as
1802 might may have been reallocated.
1803 (re_match_2): When mallocing `initial_stack', now multiply
1804 the number of items wanted (what was there before) by
1806 (pop_failure_point): Need this procedure form of the macro of
1807 the same name for debugging, so left it in and deleted the
1809 (recomp): Don't free the pattern buffer's translate field.
1811 Mon Apr 15 09:47:47 1991 Kathy Hargreaves (kathy at hayley)
1813 * regex.h (RE_DUP_MAX): Moved to outside of #ifdef _POSIX_SOURCE.
1814 * regex.c (#include <sys/types.h>): Removed #ifdef _POSIX_SOURCE
1816 (malloc, realloc): Made return type void* #ifdef __STDC__.
1817 (enum regexpcode): Added endline_in_repeat for the compiler's
1818 use; this never ends up on the final compiled pattern.
1819 (INIT_PATTERN_OFFSETS_LIST_SIZE): Initial size for
1820 pattern_offsets_list_type.
1821 (pattern_offset_type): Type for pattern offsets.
1822 (pattern_offsets_list_type): Type for keeping a list of
1824 (anchor_list_type): Changed to above type.
1825 (PATTERN_OFFSETS_LIST_PTR_FULL): Tests if a pattern offsets
1827 (ANCHOR_LIST_PTR_FULL): Changed to above.
1828 (BIT_BLOCK_SIZE): Changed to BITS_BLOCK_SIZE and moved to
1829 above bits list routines below regex_compile.
1830 (op_list_type): Defined to be pattern_offsets_list_type.
1831 (compile_stack_type): Changed offsets to be
1832 pattern_offset_type instead of unsigned.
1833 (pointer): Changed the name of all structure fields from this
1835 (COMPILE_STACK_FULL): Changed so the stack is full if `avail'
1836 is equal to `size' instead of `size' - 1.
1837 (GET_BUFFER_SPACE): Changed `>=' to `>' in the while statement.
1838 (regex_compile): Added variable `enough_memory' so could check
1839 that routine that verifies '$' positions could return an
1841 (group_count): Deleted this variable, as `regnum' already does
1843 (op_list): Added this variable to keep track of operations
1844 needed for verifying '$' positions.
1845 (anchor_list): Now initialize using routine
1846 `init_pattern_offsets_list'.
1847 Consolidated the three bits_list initializations.
1848 In case '$': Instead of trying to go past constructs which can
1849 follow '$', merely detect the special case where it has to be
1850 at the pattern's end, fix up any fixup jumps if necessary,
1851 record the anchor if necessary and add an `endline' (and
1852 possibly two `no-op's) to the pattern; will call a routine at
1853 the end to verify if it's in a valid position or not.
1854 (init_pattern_offsets_list): Added to initialize pattern
1856 (extend_anchor_list): Renamed this extend_pattern_offsets_list
1857 and renamed parameters and internal variables appropriately.
1858 (add_pattern_offset): Added this routine which both
1859 record_anchor_position and add_op call.
1860 (adjust_pattern_offsets_list): Add this routine to adjust by
1861 some increment all the pattern offsets a list of such after a
1863 (record_anchor_position): Now send in offset instead of
1864 calculating it and just call add_pattern_offset.
1865 (adjust_anchor_list): Replaced by above routine.
1866 (remove_intervening_anchors): If the anchor is an `endline'
1867 then replace it with `endline_in_repeat' instead of `no_op'.
1868 (add_op): Added this routine to call in regex_compile
1869 wherever push something relevant to verifying '$' positions.
1870 (verify_and_adjust_endlines): Added routine to (1) verify that
1871 '$'s in a pattern buffer (represented by `endline') were in
1872 valid positions and (2) whether or not they were anchors.
1873 (BITS_BLOCK_SIZE): Renamed BIT_BLOCK_SIZE and moved to right
1874 above bits list routines.
1875 (BITS_BLOCK): Defines which array element of a bits list the
1876 bit corresponding to a given position is in.
1877 (BITS_MASK): Has a 1 where the bit (in a bit list array element)
1878 for a given position is.
1880 Mon Apr 1 12:09:06 1991 Kathy Hargreaves (kathy at hayley)
1882 * regex.c (BIT_BLOCK_SIZE): Defined this for using with
1883 bits_list_type, abstracted from level_list_type so could use
1884 for more things than just the level match status.
1885 (regex_compile): Renamed `level_list' variable to
1886 `level_match_status'.
1887 Added variable `group_match_status' of type bits_list_type.
1888 Kept track of whether or not for all groups any of them
1889 matched other than the empty string, so detect if a back
1890 reference in front of a '^' made it nonleading or not.
1891 Do this by setting a match status bit for all active groups
1892 whenever leave a group that matches other than the empty string.
1893 Could detect which groups are active by going through the
1894 stack each time, but or-ing a bits list of active groups with
1895 a bits list of group match status is faster, so make a bits
1896 list of active groups instead.
1897 Have to check that '^' isn't in a leading position before
1898 going to normal_char.
1899 Whenever set level match status of the current level, also set
1900 the match status of all active groups.
1901 Increase the group count and make that group active whenever
1903 When close a group, only set the next level down if the
1904 current level matches other than the empty string, and make
1905 the current group inactive.
1906 At a back reference, only set a level's match status if the
1907 group to which the back reference refers matches other than
1909 (init_bits_list): Added to initialize a bits list.
1910 (get_level_value): Deleted this. (Made into
1911 get_level_match_status.)
1912 (extend_bits_list): Added to extend a bits list. (Made this
1913 from deleted routine `extend_level_list'.)
1914 (get_bit): Added to get a bit value from a bits list. (Made
1915 this from deleted routine `get_level_value'.)
1916 (set_bit_to_value): Added to set a bit in a bits list. (Made
1917 this from deleted routine `set_level_value'.)
1918 (get_level_match_status): Added this to get the match status
1919 of a given level. (Made from get_level_value.)
1920 (set_this_level, set_next_lower_level): Made all routines
1921 which set bits extend the bits list if necessary, thus they
1922 now return an unsigned value to indicate whether or not the
1923 reallocation failed.
1924 (increase_level): No longer extends the level list.
1925 (make_group_active): Added to mark as active a given group in
1926 an active groups list.
1927 (make_group_inactive): Added to mark as inactive a given group
1928 in an active groups list.
1929 (set_match_status_of_active_groups): Added to set the match
1930 status of all currently active groups.
1931 (get_group_match_status): Added to get a given group's match status.
1932 (no_levels_match_anything): Removed the paramenter LEVEL.
1933 (PUSH_FAILURE_POINT): Added rms' bug fix and changed RE_NREGS
1934 to num_internal_regs.
1936 Sun Mar 31 09:04:30 1991 Kathy Hargreaves (kathy at hayley)
1938 * regex.h (RE_ANCHORS_ONLY_AT_ENDS): Added syntax so could
1939 constrain '^' and '$' to only be anchors if at the beginning
1940 and end of the pattern.
1941 (RE_SYNTAX_POSIX_BASIC): Added the above bit.
1943 * regex.c (enum regexcode): Changed `unused' to `no_op'.
1944 (this_and_lower_levels_match_nothing): Deleted forward reference.
1945 (regex_compile): case '^': if the syntax bit RE_ANCHORS_ONLY_AT_ENDS
1946 is set, then '^' is only an anchor if at the beginning of the
1947 pattern; only record anchor position if the syntax bit
1948 RE_REPEATED_ANCHORS_AWAY is set; the '^' is a normal char if
1949 the syntax bit RE_ANCHORS_ONLY_AT_END is set and we're not at
1950 the beginning of the pattern (and neither RE_CONTEXTUAL_INDEP_OPS
1951 nor RE_CONTEXTUAL_INDEP_OPS syntax bits are set).
1952 Only adjust the anchor list if the syntax bit
1953 RE_REPEATED_ANCHORS_AWAY is set.
1955 * regex.c (level_list_type): Use to detect when '^' is
1956 in a leading position.
1957 (regex_compile): Added level_list_type level_list variable in
1958 which we keep track of whether or not a grouping level (in its
1959 current or most recent incarnation) matches anything besides the
1960 empty string. Set the bit for the i-th level when detect it
1961 should match something other than the empty string and the bit
1962 for the (i-1)-th level when leave the i-th group. Clear all
1963 bits for the i-th and higher levels if none of 0--(i - 1)-th's
1964 bits are set when encounter an alternation operator on that
1965 level. If no levels are set when hit a '^', then it is in a
1966 leading position. We keep track of which level we're at by
1967 increasing a variable current_level whenever we encounter an
1968 open-group operator and decreasing it whenever we encounter a
1969 close-group operator.
1970 Have to adjust the anchor list contents whenever insert
1971 something ahead of them (such as on_failure_jump's) in the
1973 (adjust_anchor_list): Adjusts the offsets in an anchor list by
1974 a given increment starting at a given start position.
1975 (get_level_value): Returns the bit setting of a given level.
1976 (set_level_value): Sets the bit of a given level to a given value.
1977 (set_this_level): Sets (to 1) the bit of a given level.
1978 (set_next_lower_level): Sets (to 1) the bit of (LEVEL - 1) for a
1980 (clear_this_and_higher_levels): Clears the bits for a given
1981 level and any higher levels.
1982 (extend_level_list): Adds sizeof(unsigned) more bits to a level list.
1983 (increase_level): Increases by 1 the value of a given level variable.
1984 (decrease_level): Decreases by 1 the value of a given level variable.
1985 (lower_levels_match_nothing): Checks if any levels lower than
1986 the given one match anything.
1987 (no_levels_match_anything): Checks if any levels match anything.
1988 (re_match_2): At case wordbeg: before looking at d-1, check that
1989 we're not at the string's beginning.
1990 At case wordend: Added some illuminating parentheses.
1992 Mon Mar 25 13:58:51 1991 Kathy Hargreaves (kathy at hayley)
1994 * regex.h (RE_NO_ANCHOR_AT_NEWLINE): Changed syntax bit name
1995 from RE_ANCHOR_NOT_NEWLINE because an anchor never matches the
1996 newline itself, just the empty string either before or after it.
1997 (RE_REPEATED_ANCHORS_AWAY): Added this syntax bit for ignoring
1998 anchors inside groups which are operated on by repetition
2000 (RE_DOT_MATCHES_NEWLINE): Added this bit so the match-any-character
2001 operator could match a newline when it's set.
2002 (RE_SYNTAX_POSIX_BASIC): Set RE_DOT_MATCHES_NEWLINE in this.
2003 (RE_SYNTAX_POSIX_EXTENDED): Set RE_DOT_MATCHES_NEWLINE and
2004 RE_REPEATED_ANCHORS_AWAY in this.
2005 (regerror): Changed prototypes to new POSIX spec.
2007 * regex.c (anchor_list_type): Added so could null out anchors inside
2009 (ANCHOR_LIST_PTR_FULL): Added for above type.
2010 (compile_stack_element): Changed name from stack_element.
2011 (compile_stack_type): Changed name from compile_stack.
2012 (INIT_COMPILE_STACK_SIZE): Changed name from INIT_STACK_SIZE.
2013 (COMPILE_STACK_EMPTY): Changed name from STACK_EMPTY.
2014 (COMPILE_STACK_FULL): Changed name from STACK_FULL.
2015 (regex_compile): Changed SYNTAX parameter to non-const.
2016 Changed variable name `stack' to `compile_stack'.
2017 If syntax bit RE_REPEATED_ANCHORS_AWAY is set, then naively put
2018 anchors in a list when encounter them and then set them to
2019 `unused' when detect they are within a group operated on by a
2020 repetition operator. Need something more sophisticated than
2021 this, as they should only get set to `unused' if they are in
2022 positions where they would be anchors. Also need a better way to
2023 detect contextually invalid anchors.
2024 Changed some commments.
2025 (is_in_compile_stack): Changed name from `is_in_stack'.
2026 (extend_anchor_list): Added to do anchor stuff.
2027 (record_anchor_position): Added to do anchor stuff.
2028 (remove_intervening_anchors): Added to do anchor stuff.
2029 (re_match_2): Now match a newline with the match-any-character
2030 operator if RE_DOT_MATCHES_NEWLINE is set.
2031 Compacted some code.
2032 (regcomp): Added new POSIX newline information to the header
2034 If REG_NEWLINE cflag is set, then now unset RE_DOT_MATCHES_NEWLINE
2036 (put_in_buffer): Added to do new POSIX regerror spec. Called
2038 (regerror): Changed to take a pattern buffer, error buffer and
2039 its size, and return type `size_t', the size of the full error
2040 message, and the first ERRBUF_SIZE - 1 characters of the full
2041 error message in the error buffer.
2043 Wed Feb 27 16:38:33 1991 Kathy Hargreaves (kathy at hayley)
2045 * regex.h (#include <sys/types.h>): Removed this as new POSIX
2046 standard has the user include it.
2047 (RE_SYNTAX_POSIX_BASIC and RE_SYNTAX_POSIX_EXTENDED): Removed
2048 RE_HAT_LISTS_NOT_NEWLINE as new POSIX standard has the cflag
2049 REG_NEWLINE now set this. Similarly, added syntax bit
2050 RE_ANCHOR_NOT_NEWLINE as this is now unset by REG_NEWLINE.
2051 (RE_SYNTAX_POSIX_BASIC): Removed syntax bit
2052 RE_NO_CONSECUTIVE_REPEATS as POSIX now allows them.
2054 * regex.c (#include <sys/types.h>): Added this as new POSIX
2055 standard has the user include it instead of us putting it in
2057 (extern char *re_syntax_table): Made into an extern so the
2058 user could allocate it.
2059 (DO_RANGE): If don't find a range end, now goto invalid_range_end
2060 instead of unmatched_left_bracket.
2061 (regex_compile): Made variable SYNTAX non-const.????
2062 Reformatted some code.
2063 (re_compile_fastmap): Moved is_a_succeed_n's declaration to
2065 Compacted some code.
2066 (SET_NEWLINE_FLAG): Removed and put inline.
2067 (regcomp): Made variable `syntax' non-const so can unset
2068 RE_ANCHOR_NOT_NEWLINE syntax bit if cflag RE_NEWLINE is set.
2069 If cflag RE_NEWLINE is set, set the RE_HAT_LISTS_NOT_NEWLINE
2070 syntax bit and unset RE_ANCHOR_NOT_NEWLINE one of `syntax'.
2072 Wed Feb 20 16:33:38 1991 Kathy Hargreaves (kathy at hayley)
2074 * regex.h (RE_NO_CONSECUTIVE_REPEATS): Changed name from
2075 RE_NO_CONSEC_REPEATS.
2076 (REG_ENESTING): Deleted this POSIX return value, as the stack
2078 (struct re_pattern_buffer): Changed some comments.
2079 (re_compile_pattern): Changed a comment.
2080 Deleted check on stack upper bound and corresponding error.
2081 Now when there's no interval contents and it's the end of the
2082 pattern, go to unmatched_left_curly_brace instead of end_of_pattern.
2083 Removed nesting_too_deep error, as the stack is now unbounded.
2084 (regcomp): Removed REG_ENESTING case, as the stack is now unbounded.
2085 (regerror): Removed REG_ENESTING case, as the stack is now unbounded.
2087 * regex.c (MAX_STACK_SIZE): Deleted because don't need upper
2088 bound on array indexed with an unsigned number.
2090 Sun Feb 17 15:50:24 1991 Kathy Hargreaves (kathy at hayley)
2092 * regex.h: Changed and added some comments.
2094 * regex.c (init_syntax_once): Made `_' a word character.
2095 (re_compile_pattern): Added a comment.
2096 (re_match_2): Redid header comment.
2097 (regexec): With header comment about PMATCH, corrected and
2098 removed details found regex.h, adding a reference.
2100 Fri Feb 15 09:21:31 1991 Kathy Hargreaves (kathy at hayley)
2102 * regex.c (DO_RANGE): Removed argument parentheses.
2103 Now get untranslated range start and end characters and set
2104 list bits for the translated (if at all) versions of them and
2105 all characters between them.
2106 (re_match_2): Now use regs->num_regs instead of num_regs_wanted
2108 (regcomp): Now build case-fold translate table using isupper
2109 and tolower facilities so will work on foreign language characters.
2111 Sat Feb 9 16:40:03 1991 Kathy Hargreaves (kathy at hayley)
2113 * regex.h (RE_HAT_LISTS_NOT_NEWLINE): Changed syntax bit name
2114 from RE_LISTS_NOT_NEWLINE as it only affects nonmatching lists.
2115 Changed all references to the match-beginning-of-string
2116 operator to match-beginning-of-line operator, as this is what
2118 (RE_NO_CONSEC_REPEATS): Added this syntax bit.
2119 (RE_SYNTAX_POSIX_BASIC): Added above bit to this.
2120 (REG_PREMATURE_END): Changed name to REG_EEND.
2121 (REG_EXCESS_NESTING): Changed name to REG_ENESTING.
2122 (REG_TOO_BIG): Changed name to REG_ESIZE.
2123 (REG_INVALID_PREV_RE): Deleted this return POSIX value.
2124 Added and changed some comments.
2126 * regex.c (re_compile_pattern): Now sets the pattern buffer's
2127 `return_default_num_regs' field.
2128 (typedef struct stack_element, stack_type, INIT_STACK_SIZE,
2129 MAX_STACK_SIZE, STACK_EMPTY, STACK_FULL): Added for regex_compile.
2130 (INIT_BUF_SIZE): Changed value from 28 to 32.
2131 (BUF_PUSH): Changed name from BUFPUSH.
2132 (MAX_BUF_SIZE): Added so could use in many places.
2133 (IS_CHAR_CLASS_STRING): Replaced is_char_class with this.
2134 (regex_compile): Added a stack which could grow dynamically
2135 and which has struct elements.
2136 Go back to initializing `zero_times_ok' and `many_time_ok' to
2137 0 and |=ing them inside the loop.
2138 Now disallow consecutive repetition operators if the syntax
2139 bit RE_NO_CONSEC_REPEATS is set.
2140 Now detect trailing backslash when the compiler is expecting a
2142 Changed calls to GET_BUFFER_SPACE which asked for 6 to ask for
2143 3, as that's all they needed.
2144 Now check for trailing backslash inside lists.
2145 Now disallow an empty alternative right before an end-of-line
2147 Now get buffer space before leaving space for a fixup jump.
2148 Now check if at pattern end when at open-interval operator.
2149 Added some comments.
2150 Now check if non-interval repetition operators follow an
2151 interval one if the syntax bit RE_NO_CONSEC_REPEATS is set.
2152 Now only check if what precedes an interval repetition
2153 operator isn't a regular expression which matches one
2154 character if the syntax bit RE_NO_CONSEC_REPEATS is set.
2155 Now return "Unmatched [ or [^" instead of "Unmatched [".
2156 (is_in_stack): Added to check if a given register number is in
2158 (re_match_2): If initial variable allocations fail, return -2,
2160 Now set reg's `num_regs' field when allocating regs.
2161 Now before allocating them, free regs->start and end if they
2162 aren't NULL and return -2 if either allocation fails.
2163 Now use regs->num_regs instead of num_regs_wanted to control
2165 Now increment past the newline when matching it with an
2166 end-of-line operator.
2167 (recomp): Added to the header comment.
2168 Now return REG_ESUBREG if regex_compile returns "Unmatched [
2169 or [^" instead of doing so if it returns "Unmatched [".
2170 Now return REG_BADRPT if in addition to returning "Missing
2171 preceding regular expression", regex_compile returns "Invalid
2172 preceding regular expression".
2173 Now return new return value names (see regex.h changes).
2174 (regexec): Added to header comment.
2175 Initialize regs structure.
2176 Now match whole string.
2177 Now always free regs.start and regs.end instead of just when
2179 (regerror): Now return "Regex error: Unmatched [ or [^.\n"
2180 instead of "Regex error: Unmatched [.\n".
2181 Now return "Regex error: Preceding regular expression either
2182 missing or not simple.\n" instead of "Regex error: Missing
2183 preceding regular expression.\n".
2184 Removed REG_INVALID_PREV_RE case (it got subsumed into the
2187 Thu Jan 17 09:52:35 1991 Kathy Hargreaves (kathy at hayley)
2189 * regex.h: Changed a comment.
2191 * regex.c: Changed and added large header comments.
2192 (re_compile_pattern): Now if detect that `laststart' for an
2193 interval points to a byte code for a regular expression which
2194 matches more than one character, make it an internal error.
2195 (regerror): Return error message, don't print it.
2197 Tue Jan 15 15:32:49 1991 Kathy Hargreaves (kathy at hayley)
2199 * regex.h (regcomp return codes): Added GNU ones.
2200 Updated some comments.
2202 * regex.c (DO_RANGE): Changed `obscure_syntax' to `syntax'.
2203 (regex_compile): Added `following_left_brace' to keep track of
2204 where pseudo interval following a valid interval starts.
2205 Changed some instances that returned "Invalid regular
2206 expression" to instead return error strings coinciding with
2208 Changed some comments.
2209 Now consider only things between `[:' and `:]' to be possible
2210 character class names.
2211 Now a character class expression can't end a pattern; at
2212 least a `]' must close the list.
2213 Now if the syntax bit RE_NO_BK_CURLY_BRACES is set, then a
2214 valid interval must be followed by yet another to get an error
2215 for preceding an interval (in this case, the second one) with
2216 a regular expression that matches more than one character.
2217 Now if what follows a valid interval begins with a open
2218 interval operator but doesn't begin a valid interval, then set
2219 following_left_bracket to it, put it in C and go to
2221 Added some comments.
2222 Return "Invalid character class name" instead of "Invalid
2224 (regerror): Return messages for all POSIX error codes except
2225 REG_ECOLLATE and REG_NEWLINE, along with all GNU error codes.
2226 Added `break's after all cases.
2227 (main): Call re_set_syntax instead of setting `obscure_syntax'
2230 Sat Jan 12 13:37:59 1991 Kathy Hargreaves (kathy at hayley)
2232 * regex.h (Copyright): Updated date.
2233 (#include <sys/types.h>): Include unconditionally.
2234 (RE_CANNOT_MATCH_NEWLINE): Deleted this syntax bit.
2235 (RE_SYNTAX_POSIX_BASIC, RE_SYNTAX_POSIX_EXTENDED): Removed
2236 setting the RE_ANCHOR_NOT_NEWLINE syntax bit from these.
2237 Changed and added some comments.
2238 (struct re_pattern_buffer): Changed some flags from chars to bits.
2239 Added field `syntax'; holds which syntax pattern was compiled with.
2240 Added bit flag `return_default_num_regs'.
2241 (externs for GNU and Berkeley UNIX routines): Added `const's to
2242 parameter types to be compatible with POSIX.
2243 (#define const): Added to support old C compilers.
2245 * regex.c (Copyright): Updated date.
2246 (enum regexpcode): Deleted `newline'.
2247 (regex_compile): Renamed re_compile_pattern to this, added a
2248 syntax parameter so it can set the pattern buffer's `syntax'
2250 Made `pattern', and `size' `const's so could pass to POSIX
2251 interface routines; also made `const' whatever interval
2252 variables had to be to make this work.
2253 Changed references to `obscure_syntax' to new parameter `syntax'.
2254 Deleted putting `newline' in buffer when see `\n'.
2255 Consider invalid character classes which have nothing wrong
2256 except the character class name; if so, return character-class error.
2257 (is_char_class): Added routine for regex_compile.
2258 (re_compile_pattern): added a new one which calls
2259 regex_compile with `obscure_syntax' as the actual parameter
2260 for the formal `syntax'.
2261 Gave this the old routine's header comments.
2262 Made `pattern', and `size' `const's so could use POSIX interface
2264 (re_search, re_search_2, re_match, re_match_2): Changed
2266 (re_search_2, re_match_2): Changed `mstop' to `stop'.
2267 (re_search, re_search_2): Made all parameters except `regs'
2268 `const's so could use POSIX interface routines parameters.
2269 (re_search_2): Added private copies of `const' parameters so
2270 could change their values.
2271 (re_match_2): Made all parameters except `regs' `const's so
2272 could use POSIX interface routines parameters.
2273 Changed `size1' and `size2' parameters to `size1_arg' and
2274 `size2_arg' and so could change; added local `size1' and
2275 `size2' and set to these.
2276 Added some comments.
2277 Deleted `newline' case.
2278 `begline' can also possibly match if `d' contains a newline;
2279 if it does, we have to increment d to point past the newline.
2280 Replaced references to `obscure_syntax' with `bufp->syntax'.
2281 (re_comp, re_exec): Made parameter `s' a `const' so could use POSIX
2282 interface routines parameters.
2283 Now call regex_compile, passing `obscure_syntax' via the
2285 (re_exec): Made local `len' a `const' so could pass to re_search.
2286 (regcomp): Added header comment.
2287 Added local `syntax' to set and pass to regex_compile rather
2288 than setting global `obscure_syntax' and passing it.
2289 Call regex_compile with its `syntax' parameter rather than
2291 Return REG_ECTYPE if character-class error.
2292 (regexec): Don't initialize `regs' to anything.
2293 Made `private_preg' a nonpointer so could set to what the
2294 constant `preg' points.
2295 Initialize `private_preg's `return_default_num_regs' field to
2296 zero because want to return `nmatch' registers, not however
2297 many there are subexpressions in the pattern.
2298 Also test if `nmatch' > 0 to see if should pass re_match `regs'.
2300 Tue Jan 8 15:57:17 1991 Kathy Hargreaves (kathy at hayley)
2302 * regex.h (struct re_pattern_buffer): Reworded comment.
2304 * regex.c (EXTEND_BUFFER): Also reset beg_interval.
2305 (re_search_2): Return val if val = -2.
2306 (NUM_REG_ITEMS): Listed items in comment.
2307 (NUM_OTHER_ITEMS): Defined this for using in > 1 definition.
2308 (MAX_NUM_FAILURE_ITEMS): Replaced `+ 2' with NUM_OTHER_ITEMS.
2309 (NUM_FAILURE_ITEMS): As with definition above and added to
2311 (PUSH_FAILURE_POINT): Replaced `* 2's with `<< 1's.
2312 (re_match_2): Test with equality with 1 to see pbufp->bol and
2315 Fri Jan 4 15:07:22 1991 Kathy Hargreaves (kathy at hayley)
2317 * regex.h (struct re_pattern_buffer): Reordered some fields.
2318 Updated some comments.
2319 Added not_bol and not_eol fields.
2320 (extern regcomp, regexec, regerror): Added return types.
2321 (extern regfree): Added `extern'.
2323 * regex.c (min): Deleted unused macro.
2324 (re_match_2): Compacted some code.
2325 Removed call to macro `min' from `for' loop.
2326 Fixed so unused registers get filled with -1's.
2327 Fail if the pattern buffer's `not_bol' field is set and
2328 encounter a `begline'.
2329 Fail if the pattern buffer's `not_eol' field is set and
2330 encounter a `endline'.
2331 Deleted redundant check for empty stack in fail case.
2332 Don't free pattern buffer's components in re_comp.
2333 (regexec): Initialize variable regs.
2334 Added `private_preg' pattern buffer so could set `not_bol' and
2335 `not_eol' fields and hand to re_match.
2336 Deleted naive attempt to detect anchors.
2337 Set private pattern buffer's `not_bol' and `not_eol' fields
2338 according to eflags value.
2339 `nmatch' must also be > 0 for us to bother allocating
2340 registers to send to re_match and filling pmatch
2341 with their results after the call to re_match.
2342 Send private pattern buffer instead of argument to re_match.
2343 If use the registers, always free them and then set them to NULL.
2344 (regerror): Added this Posix routine.
2345 (regfree): Added this Posix routine.
2347 Tue Jan 1 15:02:45 1991 Kathy Hargreaves (kathy at hayley)
2349 * regex.h (RE_NREGS): Deleted this definition, as now the user
2350 can choose how many registers to have.
2351 (REG_NOTBOL, REG_NOTEOL): Defined these Posix eflag bits.
2352 (REG_NOMATCH, REG_BADPAT, REG_ECOLLATE, REG_ECTYPE,
2353 REG_EESCAPE, REG_ESUBREG, REG_EBRACK, REG_EPAREN, REG_EBRACE,
2354 REG_BADBR, REG_ERANGE, REG_ESPACE, REG_BADRPT, REG_ENEWLINE):
2355 Defined these return values for Posix's regcomp and regexec.
2356 Updated some comments.
2357 (struct re_pattern_buffer): Now typedef this as regex_t
2358 instead of the other way around.
2359 (struct re_registers): Added num_regs field. Made start and
2360 end fields pointers to char instead of fixed size arrays.
2361 (regmatch_t): Added this Posix register type.
2362 (regcomp, regexec, regerror, regfree): Added externs for these
2365 * regex.c (enum boolean): Typedefed this.
2366 (re_pattern_buffer): Reformatted some comments.
2367 (re_compile_pattern): Updated some comments.
2368 Always push start_memory and its attendant number whenever
2369 encounter a group, not just when its number is less than the
2370 previous maximum number of registers; same for stop_memory.
2371 Get 4 bytes of buffer space instead of 2 when pushing a
2373 (can_match_nothing): Added this to elaborate on and replace
2375 (reg_info_type): Made can_match_nothing field a bit instead of int.
2376 (MIN): Added for re_match_2.
2377 (re_match_2 macros): Changed all `for' loops which used
2378 RE_NREGS to now use num_internal_regs as upper bounds.
2379 (MAX_NUM_FAILURE_ITEMS): Use num_internal_regs instead of RE_NREGS.
2380 (POP_FAILURE_POINT): Added check for empty stack.
2381 (FREE_VARIABLES): Added this to free (and set to NULL)
2382 variables allocated in re_match_2.
2383 (re_match_2): Rearranged parameters to be in order.
2384 Added variables num_regs_wanted (how many registers the user wants)
2385 and num_internal_regs (how many groups there are).
2386 Allocated initial_stack, regstart, regend, old_regstart,
2387 old_regend, reginfo, best_regstart, and best_regend---all
2388 which used to be fixed size arrays. Free them all and return
2390 Free above variables if starting position pos isn't valid.
2391 Changed all `for' loops which used RE_NREGS to now use
2392 num_internal_regs as upper bounds---except for the loops which
2393 fill regs; then use num_regs_wanted.
2394 Allocate regs if the user has passed it and wants more than 0
2396 Set regs->start[i] and regs->end[i] to -1 if either
2397 regstart[i] or regend[i] equals -1, not just the first.
2398 Free allocated variables before returning.
2399 Updated some comments.
2400 (regcomp): Return REG_ESPACE, REG_BADPAT, REG_EPAREN when
2402 Free translate array.
2403 (regexec): Added this Posix interface routine.
2405 Mon Dec 24 14:21:13 1990 Kathy Hargreaves (kathy at hayley)
2407 * regex.h: If _POSIX_SOURCE is defined then #include <sys/types.h>.
2408 Added syntax bit RE_CANNOT_MATCH_NEWLINE.
2409 Defined Posix cflags: REG_EXTENDED, REG_NEWLINE, REG_ICASE, and
2411 Added fields re_nsub and no_sub to struct re_pattern_buffer.
2412 Typedefed regex_t to be `struct re_pattern_buffer'.
2414 * regex.c (CHAR_SET_SIZE): Defined this to be 256 and replaced
2415 incidences of this value with this constant.
2416 (re_compile_pattern): Added switch case for `\n' and put
2417 `newline' into the pattern buffer when encounter this.
2418 Increment the pattern_buffer's `re_nsub' field whenever open a
2420 (re_match_2): Match a newline with `newline'---provided the
2421 syntax bit RE_CANNOT_MATCH_NEWLINE isn't set.
2422 (regcomp): Added this Posix interface routine.
2423 (enum test_type): Added interface_test tag.
2424 (main): Added Posix interface test.
2426 Tue Dec 18 12:58:12 1990 Kathy Hargreaves (kathy at hayley)
2428 * regex.h (struct re_pattern_buffer): reformatted so would fit
2429 in texinfo documentation.
2431 Thu Nov 29 15:49:16 1990 Kathy Hargreaves (kathy at hayley)
2433 * regex.h (RE_NO_EMPTY_ALTS): Added this bit.
2434 (RE_SYNTAX_POSIX_EXTENDED): Added above bit.
2436 * regex.c (re_compile_pattern): Disallow empty alternatives only
2437 when RE_NO_EMPTY_ALTS is set, not when RE_CONTEXTUAL_INVALID_OPS is.
2438 Changed RE_NO_BK_CURLY_BRACES to RE_NO_BK_PARENS when testing
2439 for empty groups at label handle_open.
2440 At label handle_bar: disallow empty alternatives if RE_NO_EMPTY_ALTS
2442 Rewrote some comments.
2444 (re_compile_fastmap): cleaned up code.
2446 (re_search_2): Rewrote comment.
2448 (struct register_info): Added field `inner_groups'; it records
2449 which groups are inside of the current one.
2450 Added field can_match_nothing; it's set if the current group
2452 Added field ever_match_something; it's set if current group
2453 ever matched something.
2455 (INNER_GROUPS): Added macro to access inner_groups field of
2456 struct register_info.
2458 (CAN_MATCH_NOTHING): Added macro to access can_match_nothing
2459 field of struct register_info.
2461 (EVER_MATCHED_SOMETHING): Added macro to access
2462 ever_matched_something field of struct register_info.
2464 (NOTE_INNER_GROUP): Defined macro to record that a given group
2465 is inside of all currently active groups.
2467 (re_match_2): Added variables *p1 and mcnt2 (multipurpose).
2468 Added old_regstart and old_regend arrays to hold previous
2469 register values if they need be restored.
2470 Initialize added fields and variables.
2471 case start_memory: Find out if the group can match nothing.
2472 Save previous register values in old_restart and old_regend.
2473 Record that current group is inside of all currently active
2475 If the group is inside a loop and it ever matched anything,
2476 restore its registers to values before the last failed match.
2477 Restore the registers for the inner groups, too.
2478 case duplicate: Can back reference to a group that never
2479 matched if it can match nothing.
2481 Thu Nov 29 11:12:54 1990 Karl Berry (karl at hayley)
2483 * regex.c (bcopy, ...): define these if either _POSIX_SOURCE or
2484 STDC_HEADERS is defined; same for including <stdlib.h>.
2486 Sat Oct 6 16:04:55 1990 Kathy Hargreaves (kathy at hayley)
2488 * regex.h (struct re_pattern_buffer): Changed field comments.
2490 * regex.c (re_compile_pattern): Allow a `$' to precede an
2491 alternation operator (`|' or `\|').
2492 Disallow `^' and/or `$' in empty groups if the syntax bit
2493 RE_NO_EMPTY_GROUPS is set.
2494 Wait until have parsed a valid `\{...\}' interval expression
2495 before testing RE_CONTEXTUAL_INVALID_OPS to see if it's
2496 invalidated by that.
2497 Don't use RE_NO_BK_CURLY_BRACES to test whether or not a validly
2498 parsed interval expression is invalid if it has no preceding re;
2499 rather, use RE_CONTEXTUAL_INVALID_OPS.
2500 If an interval parses, but there is no preceding regular
2501 expression, yet the syntax bit RE_CONTEXTUAL_INDEP_OPS is set,
2502 then that interval can match the empty regular expression; if
2503 the bit isn't set, then the characters in the interval
2504 expression are parsed as themselves (sans the backslashes).
2505 In unfetch_interval case: Moved PATFETCH to above the test for
2506 RE_NO_BK_CURLY_BRACES being set, which would force a goto
2507 normal_backslash; the code at both normal_backsl and normal_char
2508 expect a character in `c.'
2510 Sun Sep 30 11:13:48 1990 Kathy Hargreaves (kathy at hayley)
2512 * regex.h: Changed some comments to use the terms used in the
2514 (RE_CONTEXTUAL_INDEP_OPS): Changed name from `RE_CONTEXT_INDEP_OPS'.
2515 (RE_LISTS_NOT_NEWLINE): Changed name from `RE_HAT_NOT_NEWLINE.'
2516 (RE_ANCHOR_NOT_NEWLINE): Added this syntax bit.
2517 (RE_NO_EMPTY_GROUPS): Added this syntax bit.
2518 (RE_NO_HYPHEN_RANGE_END): Deleted this syntax bit.
2519 (RE_SYNTAX_...): Reformatted.
2520 (RE_SYNTAX_POSIX_BASIC, RE_SYNTAX_EXTENDED): Added syntax bits
2521 RE_ANCHOR_NOT_NEWLINE and RE_NO_EMPTY_GROUPS, and deleted
2522 RE_NO_HYPHEN_RANGE_END.
2523 (RE_SYNTAX_POSIX_EXTENDED): Added syntax bit RE_DOT_NOT_NULL.
2525 * regex.c (bcopy, bcmp, bzero): Define if _POSIX_SOURCE is defined.
2526 (_POSIX_SOURCE): ifdef this, #include <stdlib.h>
2527 (#ifdef emacs): Changed comment of the #endif for the its #else
2528 clause to be `not emacs', not `emacs.'
2529 (no_pop_jump): Changed name from `jump'.
2530 (pop_failure_jump): Changed name from `finalize_jump.'
2531 (maybe_pop_failure_jump): Changed name from `maybe_finalize_jump'.
2532 (no_pop_jump_n): Changed name from `jump_n.'
2533 (EXTEND_BUFFER): Use shift instead of multiplication to double
2535 (DO_RANGE, recompile_pattern): Added macro to set the list bits
2537 (re_compile_pattern): Fixed grammar problems in some comments.
2538 Checked that RE_NO_BK_VBAR is set to make `$' valid before a `|'
2539 and not set to make it valid before a `\|'.
2540 Checked that RE_NO_BK_PARENS is set to make `$' valid before a ')'
2541 and not set to make it valid before a `\)'.
2542 Disallow ranges starting with `-', unless the range is the
2543 first item in a list, rather than disallowing ranges which end
2545 Disallow empty groups if the syntax bit RE_NO_EMPTY_GROUPS is set.
2546 Disallow nothing preceding `{' and `\{' if they represent the
2547 open-interval operator and RE_CONTEXTUAL_INVALID_OPS is set.
2548 (register_info_type): typedef-ed this using `struct register_info.'
2549 (SET_REGS_MATCHED): Compacted the code.
2550 (re_match_2): Made it fail if back reference a group which we've
2552 Made `^' not match a newline if the syntax bit
2553 RE_ANCHOR_NOT_NEWLINE is set.
2554 (really_fail): Added this label so could force a final fail that
2555 would not try to use the failure stack to recover.
2557 Sat Aug 25 14:23:01 1990 Kathy Hargreaves (kathy at hayley)
2559 * regex.h (RE_CONTEXTUAL_OPS): Changed name from RE_CONTEXT_OPS.
2560 (global): Rewrote comments and rebroke some syntax #define lines.
2562 * regex.c (isgraph): Added definition for sequents.
2563 (global): Now refer to character set lists as ``lists.''
2564 Rewrote comments containing ``\('' or ``\)'' to now refer to
2566 (RE_CONTEXTUAL_OPS): Changed name from RE_CONTEXT_OPS.
2568 (re_compile_pattern): Expanded header comment.
2570 Sun Jul 15 14:50:25 1990 Kathy Hargreaves (kathy at hayley)
2572 * regex.h (RE_CONTEX_INDEP_OPS): the comment's sense got turned
2573 around when we changed how it read; changed it to be correct.
2575 Sat Jul 14 16:38:06 1990 Kathy Hargreaves (kathy at hayley)
2577 * regex.h (RE_NO_EMPTY_BK_REF): changed name to
2578 RE_NO_MISSING_BK_REF, as this describes it better.
2580 * regex.c (re_compile_pattern): changed RE_NO_EMPTY_BK_REF
2581 to RE_NO_MISSING_BK_REF, as above.
2583 Thu Jul 12 11:45:05 1990 Kathy Hargreaves (kathy at hayley)
2585 * regex.h (RE_NO_EMPTY_BRACKETS): removed this syntax bit, as
2586 bracket expressions should *never* be empty regardless of the
2587 syntax. Removes this bit from RE_SYNTAX_POSIX_BASIC and
2588 RE_SYNTAX_POSIX_EXTENDED.
2590 * regex.c (SET_LIST_BIT): in the comment, now refer to character
2591 sets as (non)matching sets, as bracket expressions can now match
2592 other things in addition to characters.
2593 (re_compile_pattern): refer to groups as such instead of `\(...\)'
2594 or somesuch, because groups can now be enclosed in either plain
2595 parens or backslashed ones, depending on the syntax.
2596 In the '[' case, added a boolean just_had_a_char_class to detect
2597 whether or not a character class begins a range (which is invalid).
2598 Restore way of breaking out of a bracket expression to original way.
2599 Add way to detect a range if the last thing in a bracket
2600 expression was a character class.
2601 Took out check for c != ']' at the end of a character class in
2602 the else clause, as it had already been checked in the if part
2603 that also checked the validity of the string.
2604 Set or clear just_had_a_char_class as appropriate.
2605 Added some comments. Changed references to character sets to
2606 ``(non)matching lists.''
2608 Sun Jul 1 12:11:29 1990 Karl Berry (karl at hayley)
2610 * regex.h (BYTEWIDTH): moved back to regex.c.
2612 * regex.h (re_compile_fastmap): removed declaration; this
2613 shouldn't be advertised.
2615 Mon May 28 15:27:53 1990 Kathy Hargreaves (kathy at hayley)
2617 * regex.c (ifndef Sword): Made comments more specific.
2618 (global): include <stdio.h> so can write fatal messages on
2619 standard error. Replaced calls to assert with fprintfs to
2620 stderr and exit (1)'s.
2621 (PREFETCH): Reformatted to make more readable.
2622 (AT_STRINGS_BEG): Defined to test if we're at the beginning of
2623 the virtual concatenation of string1 and string2.
2624 (AT_STRINGS_END): Defined to test if at the end of the virtual
2625 concatenation of string1 and string2.
2626 (AT_WORD_BOUNDARY): Defined to test if are at a word boundary.
2627 (IS_A_LETTER(d)): Defined to test if the contents of the pointer D
2629 (re_match_2): Rewrote the wordbound, notwordbound, wordbeg, wordend,
2630 begbuf, and endbuf cases in terms of the above four new macros.
2631 Called SET_REGS_MATCHED in the matchsyntax, matchnotsyntax,
2632 wordchar, and notwordchar cases.
2634 Mon May 14 14:49:13 1990 Kathy Hargreaves (kathy at hayley)
2636 * regex.c (re_search_2): Fixed RANGE to not ever take STARTPOS
2637 outside of virtual concatenation of STRING1 and STRING2.
2638 Updated header comment as to this.
2639 (re_match_2): Clarified comment about MSTOP in header.
2641 Sat May 12 15:39:00 1990 Kathy Hargreaves (kathy at hayley)
2643 * regex.c (re_search_2): Checked for out-of-range STARTPOS.
2645 When searching backwards, not only get the character with which
2646 to compare to the fastmap from string2 if the starting position
2647 >= size1, but also if size1 is zero; this is so won't get a
2648 segmentation fault if string1 is null.
2649 Reformatted code at label advance.
2651 Thu Apr 12 20:26:21 1990 Kathy Hargreaves (kathy at hayley)
2653 * regex.h: Added #pragma once and #ifdef...endif __REGEXP_LIBRARY.
2654 (RE_EXACTN_VALUE): Added for search.c to use.
2655 Reworded some comments.
2657 regex.c: Punctuated some comments correctly.
2658 (NULL): Removed this.
2659 (RE_EXACTN_VALUE): Added for search.c to use.
2660 (<ctype.h>): Moved this include to top of file.
2661 (<assert.h>): Added this include.
2662 (struct regexpcode): Assigned 0 to unused and 1 to exactn
2663 because of RE_EXACTN_VALUE.
2665 (various macros): Lined up backslashes near end of line.
2666 (insert_jump): Cleaned up the header comment.
2667 (re_search): Corrected the header comment.
2668 (re_search_2): Cleaned up and completed the header comment.
2669 (re_max_failures): Updated comment.
2670 (struct register_info): Constructed as bits so as to save space
2671 on the stack when pushing register information.
2672 (IS_ACTIVE): Macro for struct register_info.
2673 (MATCHED_SOMETHING): Macro for struct register_info.
2674 (NUM_REG_ITEMS): How many register information items for each
2675 register we have to push on the stack at each failure.
2676 (MAX_NUM_FAILURE_ITEMS): If push all the registers on failure,
2677 this is how many items we push on the stack.
2678 (PUSH_FAILURE_POINT): Now pushes whether or not the register is
2679 currently active, and whether or not it matched something.
2680 Checks that there's enough space allocated to accomodate all the
2681 items we currently want to push. (Before, a test for an empty
2682 stack sufficed because we always pushed and popped the same
2684 Replaced ``2'' with MAX_NUM_FAILURE_POINTS when ``2'' refers
2685 to how many things get pushed on the stack each time.
2686 When copy the stack into the newly allocated storage, now only copy
2689 (POP_FAILURE_POINT): Defined to use in places where put number
2690 of registers on the stack into a variable before using it to
2691 decrement the stack, so as to not confuse the compiler.
2692 (IS_IN_FIRST_STRING): Defined to check if a pointer points into
2694 (SET_REGS_MATCHED): Changed to use the struct register_info
2695 bits; also set the matched-something bit to false if the
2696 register isn't currently active. (This is a redundant setting.)
2697 (re_match_2): Cleaned up and completed the header comment.
2698 Updated the failure stack comment.
2699 Replaced the ``2'' with MAX_NUM_FAILURE_ITEMS in the static
2700 allocation of initial_stack, because now more than two (now up
2701 to MAX_FAILURE_ITEMS) items get pushed on the failure stack each
2704 Trashed restart_seg1, regend_seg1, best_regstart_seg1, and
2705 best_regend_seg1 because they could have erroneous information
2706 in them, such as when matching ``a'' (in string1) and ``ab'' (in
2707 string2) with ``(a)*ab''; before using IS_IN_FIRST_STRING to see
2708 whether or not the register starts or ends in string1,
2709 regstart[1] pointed past the end of string1, yet regstart_seg1
2711 Added variable reg_info of type struct register_info to keep
2712 track of currently active registers and whether or not they
2713 currently match anything.
2714 Commented best_regs_set.
2715 Trashed reg_active and reg_matched_something and put the
2716 information they held into reg_info; saves space on the stack.
2717 Replaced NULL with '\000'.
2718 In begline case, compacted the code.
2719 Used assert to exit if had an internal error.
2720 In begbuf case, because now force the string we're working on
2721 into string2 if there aren't two strings, now allow d == string2
2722 if there is no string1 (and the check for that is size1 == 0!);
2723 also now succeeds if there aren't any strings at all.
2724 (main, ifdef canned): Put test type into a variable so could
2725 change it while debugging.
2727 Sat Mar 24 12:24:13 1990 Kathy Hargreaves (kathy at hayley)
2729 * regex.c (GET_UNSIGNED_NUMBER): Deleted references to num_fetches.
2730 (re_compile_pattern): Deleted num_fetches because could keep
2731 track of the number of fetches done by saving a pointer into the
2733 Added variable beg_interval to be used as a pointer, as above.
2734 Assert that beg_interval points to something when it's used as above.
2735 Initialize succeed_n's to lower_bound because re_compile_fastmap
2737 (re_compile_fastmap): Deleted unnecessary variable is_a_jump_n.
2739 (re_match_2): Put number of registers on the stack into a
2740 variable before using it to decrement the stack, so as to not
2741 confuse the compiler.
2743 Used error routine instead of printf and exit.
2744 In exactn case, restored longer code from ``original'' regex.c
2745 which doesn't test translate inside a loop.
2747 * regex.h: Moved #define NULL and the enum regexpcode definition
2748 and to regex.c. Changed some comments.
2750 regex.c (global): Updated comments about compiling and for the
2751 re_compile_pattern jump routines.
2752 Added #define NULL and the enum regexpcode definition (from
2754 (enum regexpcode): Added set_number_at to reset the n's of
2755 succeed_n's and jump_n's.
2756 (re_set_syntax): Updated its comment.
2757 (re_compile_pattern): Moved its heading comment to after its macros.
2758 Moved its include statement to the top of the file.
2759 Commented or added to comments of its macros.
2760 In start_memory case: Push laststart value before adding
2761 start_memory and its register number to the buffer, as they
2762 might not get added.
2763 Added code to put a set_number_at before each succeed_n and one
2764 after each jump_n; rewrote code in what seemed a more
2765 straightforward manner to put all these things in the pattern so
2766 the succeed_n's would correctly jump to the set_number_at's of
2767 the matching jump_n's, and so the jump_n's would correctly jump
2768 to after the set_number_at's of the matching succeed_n's.
2769 Initialize succeed_n n's to -1.
2770 (insert_op_2): Added this to insert an operation followed by
2772 (re_compile_fastmap): Added set_number_at case.
2773 (re_match_2): Moved heading comment to after macros.
2774 Added mention of REGS to heading comment.
2775 No longer turn a succeed_n with n = 0 into an on_failure_jump,
2776 because n needs to be reset each time through a loop.
2777 Check to see if a succeed_n's n is set by its set_number_at.
2778 Added set_number_at case.
2779 Updated some comments.
2780 (main): Added another main to run posix tests, which is compiled
2781 ifdef both test and canned. (Old main is still compiled ifdef
2784 Tue Mar 19 09:22:55 1990 Kathy Hargreaves (kathy at hayley)
2786 * regex.[hc]: Change all instances of the word ``legal'' to
2787 ``valid'' and all instances of ``illegal'' to ``invalid.''
2789 Sun Mar 4 12:11:31 1990 Kathy Hargreaves (kathy at hayley)
2791 * regex.h: Added syntax bit RE_NO_EMPTY_RANGES which is set if
2792 an ending range point has to collate higher or equal to the
2793 starting range point.
2794 Added syntax bit RE_NO_HYPHEN_RANGE_END which is set if a hyphen
2795 can't be an ending range point.
2796 Set to two above bits in RE_SYNTAX_POSIX_BASIC and
2797 RE_SYNTAX_POSIX_EXTENDED.
2799 regex.c: (re_compile_pattern): Don't allow empty ranges if the
2800 RE_NO_EMPTY_RANGES syntax bit is set.
2801 Don't let a hyphen be a range end if the RE_NO_HYPHEN_RANGE_END
2803 (ESTACK_PUSH_2): renamed this PUSH_FAILURE_POINT and made it
2804 push all the used registers on the stack, as well as the number
2805 of the highest numbered register used, and (as before) the two
2807 (re_match_2): Fixed up comments.
2808 Added arrays best_regstart[], best_regstart_seg1[], best_regend[],
2809 and best_regend_seg1[] to keep track of the best match so far
2810 whenever reach the end of the pattern but not the end of the
2811 string, and there are still failure points on the stack with
2812 which to backtrack; if so, do the saving and force a fail.
2813 If reach the end of the pattern but not the end of the string,
2814 but there are no more failure points to try, restore the best
2815 match so far, set the registers and return.
2816 Compacted some code.
2817 In stop_memory case, if the subexpression we've just left is in
2818 a loop, push onto the stack the loop's on_failure_jump failure
2819 point along with the current pointer into the string (d).
2820 In finalize_jump case, in addition to popping the failure
2821 points, pop the saved registers.
2822 In the fail case, restore the registers, as well as the failure
2825 Sun Feb 18 15:08:10 1990 Kathy Hargreaves (kathy at hayley)
2827 * regex.c: (global): Defined a macro GET_BUFFER_SPACE which
2828 makes sure you have a specified number of buffer bytes
2830 Redefined the macro BUFPUSH to use this.
2833 (re_compile_pattern): Call GET_BUFFER_SPACE before storing or
2834 inserting any jumps.
2836 (re_match_2): Set d to string1 + pos and dend to end_match_1
2837 only if string1 isn't null.
2838 Force exit from a loop if it's around empty parentheses.
2839 In stop_memory case, if found some jumps, increment p2 before
2840 extracting address to which to jump. Also, don't need to know
2841 how many more times can jump_n.
2842 In begline case, d must equal string1 or string2, in that order,
2843 only if they are not null.
2844 In maybe_finalize_jump case, skip over start_memorys' and
2845 stop_memorys' register numbers, too.
2847 Thu Feb 15 15:53:55 1990 Kathy Hargreaves (kathy at hayley)
2849 * regex.c (BUFPUSH): off by one goof in deciding whether to
2852 Wed Jan 24 17:07:46 1990 Kathy Hargreaves (kathy at hayley)
2854 * regex.h: Moved definition of NULL to here.
2855 Got rid of ``In other words...'' comment.
2856 Added to some comments.
2858 regex.c: (re_compile_pattern): Tried to bulletproof some code,
2859 i.e., checked if backward references (e.g., p[-1]) were within
2860 the range of pattern.
2862 (re_compile_fastmap): Fixed a bug in succeed_n part where was
2863 getting the amount to jump instead of how many times to jump.
2865 (re_search_2): Changed the name of the variable ``total'' to
2867 Condensed some code.
2869 (re_match_2): Moved the comment about duplicate from above the
2870 start_memory case to above duplicate case.
2872 (global): Rewrote some comments.
2873 Added commandline arguments to testing.
2875 Wed Jan 17 11:47:27 1990 Kathy Hargreaves (kathy at hayley)
2877 * regex.c: (global): Defined a macro STORE_NUMBER which stores a
2878 number into two contiguous bytes. Also defined STORE_NUMBER_AND_INCR
2879 which does the same thing and then increments the pointer to the
2880 storage place to point after the number.
2881 Defined a macro EXTRACT_NUMBER which extracts a number from two
2882 continguous bytes. Also defined EXTRACT_NUMBER_AND_INCR which
2883 does the same thing and then increments the pointer to the
2884 source to point to after where the number was.
2886 Tue Jan 16 12:09:19 1990 Kathy Hargreaves (kathy at hayley)
2888 * regex.h: Incorporated rms' changes.
2889 Defined RE_NO_BK_REFS syntax bit which is set when want to
2890 interpret back reference patterns as literals.
2891 Defined RE_NO_EMPTY_BRACKETS syntax bit which is set when want
2892 empty bracket expressions to be illegal.
2893 Defined RE_CONTEXTUAL_ILLEGAL_OPS syntax bit which is set when want
2894 it to be illegal for *, +, ? and { to be first in an re or come
2895 immediately after a | or a (, and for ^ not to appear in a
2896 nonleading position and $ in a nontrailing position (outside of
2897 bracket expressions, that is).
2898 Defined RE_LIMITED_OPS syntax bit which is set when want +, ?
2899 and | to always be literals instead of ops.
2900 Fixed up the Posix syntax.
2901 Changed the syntax bit comments from saying, e.g., ``0 means...''
2902 to ``If this bit is set, it means...''.
2903 Changed the syntax bit defines to use shifts instead of integers.
2905 * regex.c: (global): Incorporated rms' changes.
2907 (re_compile_pattern): Incorporated rms' changes
2908 Made it illegal for a $ to appear anywhere but inside a bracket
2909 expression or at the end of an re when RE_CONTEXTUAL_ILLEGAL_OPS
2910 is set. Made the same hold for $ except it has to be at the
2911 beginning of an re instead of the end.
2912 Made the re "[]" illegal if RE_NO_EMPTY_BRACKETS is set.
2913 Made it illegal for | to be first or last in an re, or immediately
2914 follow another | or a (.
2915 Added and embellished some comments.
2916 Allowed \{ to be interpreted as a literal if RE_NO_BK_CURLY_BRACES
2918 Made it illegal for *, +, ?, and { to appear first in an re, or
2919 immediately follow a | or a ( when RE_CONTEXTUAL_ILLEGAL_OPS is set.
2920 Made back references interpreted as literals if RE_NO_BK_REFS is set.
2921 Made recursive intervals either illegal (if RE_NO_BK_CURLY_BRACES
2922 isn't set) or interpreted as literals (if is set), if RE_INTERVALS
2924 Made it treat +, ? and | as literals if RE_LIMITED_OPS is set.
2925 Cleaned up some code.
2927 Thu Dec 21 15:31:32 1989 Kathy Hargreaves (kathy at hayley)
2929 * regex.c: (global): Moved RE_DUP_MAX to regex.h and made it
2930 equal 2^15 - 1 instead of 1000.
2931 Defined NULL to be zero.
2932 Moved the definition of BYTEWIDTH to regex.h.
2933 Made the global variable obscure_syntax nonstatic so the tests in
2934 another file could use it.
2936 (re_compile_pattern): Defined a maximum length (CHAR_CLASS_MAX_LENGTH)
2937 for character class strings (i.e., what's between the [: and the
2939 Defined a macro SET_LIST_BIT(c) which sets the bit for C in a
2941 Took out comments that EXTEND_BUFFER clobbers C.
2942 Made the string "^" match itself, if not RE_CONTEXT_IND_OPS.
2943 Added character classes to bracket expressions.
2944 Change the laststart pointer saved with the start of each
2945 subexpression to point to start_memory instead of after the
2946 following register number. This is because the subexpression
2948 Added comments and compacted some code.
2949 Made intervals only work if preceded by an re matching a single
2950 character or a subexpression.
2951 Made back references to nonexistent subexpressions illegal if
2953 Made intervals work on the last preceding character of a
2954 concatenation of characters, e.g., ab{0,} matches abbb, not abab.
2955 Moved macro PREFETCH to outside the routine.
2957 (re_compile_fastmap): Added succeed_n to work analogously to
2958 on_failure_jump if n is zero and jump_n to work analogously to
2959 the other backward jumps.
2961 (re_match_2): Defined macro SET_REGS_MATCHED to set which
2962 current subexpressions had matches within them.
2963 Changed some comments.
2964 Added reg_active and reg_matched_something arrays to keep track
2965 of in which subexpressions currently have matched something.
2966 Defined MATCHING_IN_FIRST_STRING and replaced ``dend == end_match_1''
2967 with it to make code easier to understand.
2968 Fixed so can apply * and intervals to arbitrarily nested
2969 subexpressions. (Lots of previous bugs here.)
2970 Changed so won't match a newline if syntax bit RE_DOT_NOT_NULL is set.
2971 Made the upcase array nonstatic so the testing file could use it also.
2973 (main.c): Moved the tests out to another file.
2975 (tests.c): Moved all the testing stuff here.
2977 Sat Nov 18 19:30:30 1989 Kathy Hargreaves (kathy at hayley)
2979 * regex.c: (re_compile_pattern): Defined RE_DUP_MAX, the maximum
2980 number of times an interval can match a pattern.
2981 Added macro GET_UNSIGNED_NUMBER (used to get below):
2982 Added variables lower_bound and upper_bound for upper and lower
2983 bounds of intervals.
2984 Added variable num_fetches so intervals could do backtracking.
2985 Added code to handle '{' and "\{" and intervals.
2988 (store_jump_n): (Added) Stores a jump with a number following the
2989 relative address (for intervals).
2991 (insert_jump_n): (Added) Inserts a jump_n.
2993 (re_match_2): Defined a macro ESTACK_PUSH_2 for the error stack;
2994 it checks for overflow and reallocates if necessary.
2996 * regex.h: Added bits (RE_INTERVALS and RE_NO_BK_CURLY_BRACES)
2997 to obscure syntax to indicate whether or not
2998 a syntax handles intervals and recognizes either \{ and
2999 \} or { and } as operators. Also added two syntaxes
3000 RE_SYNTAX_POSIX_BASIC and RE_POSIX_EXTENDED and two command codes
3001 to the enumeration regexpcode; they are succeed_n and jump_n.
3003 Sat Nov 18 19:30:30 1989 Kathy Hargreaves (kathy at hayley)
3005 * regex.c: (re_compile_pattern): Defined INIT_BUFF_SIZE to get rid
3006 of repeated constants in code. Tested with value 1.
3007 Renamed PATPUSH as BUFPUSH, since it pushes things onto the
3008 buffer, not the pattern. Also made this macro extend the buffer
3009 if it's full (so could do the following):
3010 Took out code at top of loop that checks to see if buffer is going
3011 to be full after 10 additions (and reallocates if necessary).
3013 (insert_jump): Rearranged declaration lines so comments would read
3016 (re_match_2): Compacted exactn code and added more comments.
3018 (main): Defined macros TEST_MATCH and MATCH_SELF to do
3019 testing; took out loop so could use these instead.
3021 Tue Oct 24 20:57:18 1989 Kathy Hargreaves (kathy at hayley)
3023 * regex.c (re_set_syntax): Gave argument `syntax' a type.
3024 (store_jump, insert_jump): made them void functions.
3029 version-control: never
projects / dragonfly.git / blob