1 /****************************************************************
2 Copyright (C) Lucent Technologies 1997
5 Permission to use, copy, modify, and distribute this software and
6 its documentation for any purpose and without fee is hereby
7 granted, provided that the above copyright notice appear in all
8 copies and that both that the copyright notice and this
9 permission notice and warranty disclaimer appear in supporting
10 documentation, and that the name Lucent Technologies or any of
11 its entities not be used in advertising or publicity pertaining
12 to distribution of the software without specific, written prior
15 LUCENT DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE,
16 INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS.
17 IN NO EVENT SHALL LUCENT OR ANY OF ITS ENTITIES BE LIABLE FOR ANY
18 SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
19 WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER
20 IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION,
21 ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF
23 ****************************************************************/
25 This file lists all bug fixes, changes, etc., made since the AWK book
26 was sent to the printers in August 1987.
29 adjbuf minlen error (cannot be 0) in cat, resulting in NULL pbuf.
30 discovered by todd miller. also use-after-free issue with
31 tempfree in cat, thanks to Miguel Pineiro Jr and valgrind.
34 Various leaks and use-after-free issues plugged/fixed.
35 Thanks to Miguel Pineiro Jr. <mpj@pineiro.cc>.
38 Memory leak when assigning a string to some of the built-in
39 variables. allocated string erroneously marked DONTFREE.
40 Thanks to Miguel Pineiro Jr. <mpj@pineiro.cc>.
43 Historic bug: command-line "name=value" assignment had been
44 truncating its entry in ARGV. (circa 1989) Thanks to
45 Miguel Pineiro Jr. <mpj@pineiro.cc>.
48 Fixed file management memory leak that appears to have been
49 there since the files array was first initialized with stdin,
50 stdout, and stderr (circa 1992). Thanks to Miguel Pineiro Jr.
54 The error handling in closefile and closeall was mangled. Long
55 standing warnings had been made fatal and some fatal errors went
56 undetected. Thanks to Miguel Pineiro Jr. <mpj@pineiro.cc>.
59 getline accesses uninitialized data after getrec()
60 returns 0 on EOF and leaves the contents of buf unchanged.
61 Thanks to Volodymyr Gubarkov, and Todd C Miller.
64 The fix for #83 changed the code to insert 2 chars, but the
65 call to adjbuf just above it only allows for 1 char. This can
66 cause a heap buffer overflow.
69 As per IEEE Std 1003.1-2008, -F "str" is now consistent with
70 -v FS="str" when str is null. Thanks to Warner Losh.
73 Fix readrec's definition of a record. This fixes an issue
74 with NetBSD's RS regular expression support that can cause
75 an infinite read loop. Thanks to Miguel Pineiro Jr.
77 Fix regular expression RS ^-anchoring. RS ^-anchoring needs to
78 know if it is reading the first record of a file. This change
79 restores a missing line that was overlooked when porting NetBSD's
80 RS regex functionality. Thanks to Miguel Pineiro Jr.
82 Fix size computation in replace_repeat() for special case
83 REPEAT_WITH_Q. Thanks to Todd C. Miller.
86 Small fix so that awk will compile again with g++. Thanks to
90 Fix a decision bug with trailing stuff in lib.c:is_valid_number
91 after recent changes. Thanks to Ozan Yigit.
94 Fix problems converting inf and NaN values in lib.c:is_valid_number.
95 Enhance number to string conversion to do the right thing for
96 NaN and inf values. Things are now pretty much the same as in
97 gawk. (Found a gawk bug while we're at it.) Added a torture
98 test for these values. Thanks to Arnold Robbins. Allows closing
102 Merge PR #99, which gets the right header for strcasecmp.
103 Thanks to GitHub user michaelforney.
106 Merge PR #98: Disallow hex data. Allow only +nan, -nan,
107 +inf, -inf (case independent) to give NaN and infinity values.
108 Improve things so that string to double conversion is only
109 done once, yielding something of a speedup. This obviate
110 PR #95. Thanks to Arnold Robbins.
113 Fix to argument parsing to avoid printing spurious newlines.
114 Thanks to Todd Miller. Merges PR #97.
117 Add casts before all the calls to malloc/calloc/realloc in order
118 to get it to compile with g++. Thanks to Arnold Robbins.
121 Additional fixes for DJGPP. Thanks to Eli Zaretskii for
125 Merge PR #93, which adds casts to (void*) for debug prints
126 using the %p format specifier. Thanks to GitHub user YongHaoWu
127 ("Chris") for the fixes.
130 In run.c, use non-restartable multibyte routines to attain
131 portability to DJGPP. Should fix Issue 92. Thanks to Albert Wik
132 for the report and to Todd Miller for the suggested fix.
135 Merge PRs 88-91 which fix small bugs. Thanks to Todd Miller and
136 Tim van der Molen for the fixes.
138 In order to make life easier, we move exclusively to bison
139 as the parser generator.
142 Merge PRs 85 and 86 which fix regressions. Thanks to
143 Tim van der Molen for the fixes.
146 Merge PRs 82 and 84. The latter fixes issue #83. Thanks to
147 Todd Miller and awkfan77.
150 Clear errno before calling errcheck to avoid any spurious errors
151 left over from previous calls that may have set it. Thanks to
152 Todd Miller for the fix, from PR #80.
154 Fix Issue #78 by allowing \r to follow floating point numbers in
155 lib.c:is_number. Thanks to GitHub user ajcarr for the report
156 and to Arnold Robbins for the fix.
159 In fldbld(), make sure that inputFS is set before trying to
160 use it. Thanks to Steffen Nurpmeso <steffen@sdaoden.eu>
164 Fix checks for compilers that can handle noreturn. Thanks to
165 GitHub user enh-google for pointing it out. Closes Issue #79.
168 Handle old compilers that don't support C11 (for noreturn).
169 Thanks to Arnold Robbins.
172 Use <stdnoreturn.h> and noreturn instead of GCC attributes.
173 Thanks to GitHub user awkfan77. Closes PR #77.
176 More cleanups from Christos Zoulas: notably backslash continuation
177 inside strings removes the newline and a fix for RS = "^a".
178 Fix for address sanitizer-found problem. Thanks to GitHub user
182 More small cleanups from Christos Zoulas.
185 Additional cleanups from Christos Zoulas. It's no longer necessary
186 to use the -y flag to bison.
189 Additional small cleanups from Christos Zoulas. awk is now
190 a little more robust about reporting I/O errors upon exit.
193 Merge PR #70, which avoids use of variable length arrays. Thanks
194 to GitHub user michaelforney. Fix issue #60 ({0} in interval
195 expressions doesn't work). Also get all tests working again.
196 Thanks to Arnold Robbins.
199 A number of small cleanups from Christos Zoulas. Add the close
200 on exec flag to files/pipes opened for redirection; courtesy of
204 If POSIXLY_CORRECT is set in the environment, then sub and gsub
205 use POSIX rules for multiple backslashes. This fixes Issue #66,
206 while maintaining backwards compatibility.
209 Input/output errors on closing files are now fatal instead of
210 mere warnings. Thanks to Martijn Dekker <martijn@inlv.org>.
213 Fix a bug in the concatentation of two string constants into
214 one done in the grammar. Fixes GitHub issue #61. Thanks
215 to GitHub user awkfan77 for pointing out the direction for
216 the fix. New test T.concat added to the test suite.
217 Fix a few memory leaks reported by valgrind, as well.
220 Fix a bug whereby a{0,3} could match four a's. Thanks to
221 "Anonymous AWK fan" for the report.
224 Further printf-related fixes for 32 bit systems.
225 Thanks again to Christos Zoulas.
228 Fix the return value of sprintf("%d") on 32 bit systems.
229 Thanks to Jim Lowe for the report and to Christos Zoulas
233 Convert a number of Boolean integer variables into
234 actual bools. Convert compile_time variable into an
235 enum and simplify some of the related code. Thanks
239 Fix from Ori Bernstein to get UTF-8 characters instead of
240 bytes when FS = "". This is currently the only bit of
241 the One True Awk that understands multibyte characters.
242 From Arnold Robbins, apply some cleanups in the test suite.
245 More fixes and cleanups from NetBSD, courtesy of Christos
246 Zoulas. Merges PRs 54 and 55.
249 Import second round of code cleanups from NetBSD. Much thanks
250 to Christos Zoulas (GitHub user zoulasc). Merges PR 53.
251 Add an optimization for string concatenation, also from
255 Import code cleanups from NetBSD. Much thanks to Christos
256 Zoulas (GitHub user zoulasc). Merges PR 51.
259 Import code from NetBSD awk that implements RS as a regular
263 Fixes for various array / memory overruns found via gcc's
264 -fsanitize=unknown. Thanks to Alexander Richardson (GitHub
265 user arichardson). Merges PRs 47 and 48.
268 Import grammar optimization from NetBSD: Two string constants
269 concatenated together get turned into a single string.
272 Support POSIX-specified C-style escape sequences "\a" (alarm)
273 and "\v" (vertical tab) in command line arguments and regular
274 expressions, further to the support for them in strings added on
275 Apr 9, 1989. These now no longer match as literal "a" and "v"
276 characters (as they don't on other awk implementations).
277 Thanks to Martijn Dekker.
280 Pull in a number of code cleanups and minor fixes from
281 Warner Losh's bsd-ota branch. The only user visible change
282 is the use of random(3) as the random number generator.
283 Thanks to Warner Losh for collecting all these fixes in
284 one easy place to get them from.
287 Fix field splitting to use FS value as of the time a record
288 was read or assigned to. Thanks to GitHub user Cody Mello (melloc)
289 for the fix. (Merged from his branch, via PR #42.) Updated
290 testdir/T.split per said PR as well.
293 Extract awktest.tar into testdir directory. Add some very
294 simple mechanics to the makefile for running the tests and
295 for cleaning up. No changes to awk itself.
298 Disallow deleting SYMTAB and its elements, which creates
299 use-after-free bugs. Thanks to GitHub user Cody Mello (melloc)
300 for the fix. (Merged from PR #43.)
303 Allow unmatched right parenthesis in a regular expression to
304 be treated literally. Fixes Issue #40. Thanks to GitHub user
305 Warner Losh (bsdimp) for the report. Thanks to Arnold Robbins
309 Fix check for command line arguments to no longer require that
310 first character after '=' not be another '='. Reverts change of
311 August 11, 1989. Thanks to GitHub user Jamie Landeg Jones for
312 pointing out the issue; from Issue #38.
315 Update awktest.tar(p.50) to use modern options to sort. Needed
316 for Android development. Thanks to GitHub user mohd-akram (Mohamed
317 Akram). From Issue #33.
320 Added very simplistic support for cross-compiling in the
321 makefile. We are NOT going to go in the direction of the
322 autotools, though. Thanks to GitHub user nee-san for
323 the basic change. (Merged from PR #34.)
326 Added support for POSIX-standard interval expressions (a.k.a.
327 bounds, a.k.a. repetition expressions) in regular expressions,
328 backported (via NetBSD) from Apple awk-24 (20070501).
329 Thanks to Martijn Dekker <martijn@inlv.org> for the port.
330 (Merged from PR #30.)
333 Merge PRs as follows:
334 #12: Avoid undefined behaviour when using ctype(3) functions in
335 relex(). Thanks to GitHub user iamleot.
336 #31: Make getline handle numeric strings, and update FIXES. Thanks
337 to GitHub user Arnold Robbins (arnoldrobbins)
338 #32: maketab: support build systems with read-only source. Thanks
342 Make getline handle numeric strings properly in all cases.
346 Merged a number of small fixes from GitHub pull requests.
347 Thanks to GitHub users Arnold Robbins (arnoldrobbins),
348 Cody Mello (melloc) and Christoph Junghans (junghans).
349 PR numbers: 13-21, 23, 24, 27.
352 Added test in maketab.c to prevent generating a proctab entry
353 for YYSTYPE_IS_DEFINED. It was harmless but some gcc settings
354 generated a warning message. Thanks to Nan Xiao for report.
357 Disallow '$' in printf formats; arguments evaluated in order
358 and printed in order.
360 Added some casts to silence warnings on debugging printfs.
364 A long list of fixes courtesy of Arnold Robbins,
365 to whom profound thanks.
367 1. ofs-rebuild: OFS value used to rebuild the record was incorrect.
368 Fixed August 19, 2014. Revised fix August 2018.
370 2. system-status: Instead of a floating-point division by 256, use
371 the wait(2) macros to create a reasonable exit status.
372 Fixed March 12, 2016.
374 3. space: Use provided xisblank() function instead of ispace() for
375 matching [[:blank:]].
377 4. a-format: Add POSIX standard %a and %A to supported formats. Check
378 at runtime that this format is available.
380 5. decr-NF: Decrementing NF did not change $0. This is a decades-old
381 bug. There are interactions with the old and new value of OFS as well.
382 Most of the fix came from the NetBSD awk.
384 6. string-conv: String conversions of scalars were sticky. Once a
385 conversion to string happened, even with OFMT, that value was used until
386 a new numeric value was assigned, even if OFMT differed from CONVFMT,
387 and also if CONVFMT changed.
389 7. unary-plus: Unary plus on a string constant returned the string.
390 Instead, it should convert the value to numeric and give that value.
392 Also added Arnold's tests for these to awktest.tar as T.arnold.
395 fixed mangled awktest.tar (thanks, Arnold), posted all
396 current (very minor) fixes to github / onetrueawk
400 Updated some broken tests (beebe.tar, T.lilly)
401 [thanks to Arnold Robbins]
404 buffer overflow in error reporting; thanks to tobias ulmer
405 and john-mark gurney for spotting it and the fix.
408 cleaned up a handful of tests that didn't seem to actually
409 test for correct behavior: T.latin1, T.gawk.
412 added ,NULL initializer to static Cells in run.c; not really
413 needed but cleaner. Thanks to Michael Bombardieri.
416 fiddled makefile to get correct yacc and bison flags. pick yacc
417 (linux) or bison (mac) as necessary.
419 added __attribute__((__noreturn__)) to a couple of lines in
420 proto.h, to silence someone's enthusiastic checker.
422 fixed obscure call by value bug in split(a[1],a) reported on
423 9fans. the management of temporary values is just a mess; i
424 took a shortcut by making an extra string copy. thanks
425 to paul patience and arnold robbins for passing it on and for
428 tiny fiddle in setfval to eliminate -0 results in T.expr, which
429 has irritated me for 20+ years.
432 another fix to avoid core dump with delete(ARGV); again, many thanks
436 split(s, a, //) now behaves the same as split(s, a, "")
439 /pat/, \n /pat/ {...} is now legal, though bad style to use.
441 added checks to new -v code that permits -vnospace; thanks to
442 ruslan ermilov for spotting this and providing the patch.
444 removed fixed limit on number of open files; thanks to aleksey
445 cheusov and christos zoulos.
447 fixed day 1 bug that resurrected deleted elements of ARGV when
448 used as filenames (in lib.c).
450 minor type fiddles to make gcc -Wall -pedantic happier (but not
451 totally so); turned on -fno-strict-aliasing in makefile.
454 added #ifdef for isblank.
455 now allows -ffoo as well as -f foo arguments.
459 after advice from todd miller, kevin lo, ruslan ermilov,
460 and arnold robbins, changed srand() to return the previous
461 seed (which is 1 on the first call of srand). the seed is
462 an Awkfloat internally though converted to unsigned int to
463 pass to the library srand(). thanks, everyone.
465 fixed a subtle (and i hope low-probability) overflow error
466 in fldbld, by adding space for one extra \0. thanks to
467 robert bassett for spotting this one and providing a fix.
469 removed the files related to compilation on windows. i no
470 longer have anything like a current windows environment, so
471 i can't test any of it.
474 fixed long-standing overflow bug in run.c; many thanks to
475 nelson beebe for spotting it and providing the fix.
477 fixed bug that didn't parse -vd=1 properly; thanks to santiago
478 vila for spotting it.
481 i give up. replaced isblank with isspace in b.c; there are
482 no consistent header files.
485 fixed a long-standing issue with when FS takes effect. a
486 change to FS is now noticed immediately for subsequent splits.
488 changed the name getline() to awkgetline() to avoid yet another
489 name conflict somewhere.
492 temporarily for now defined HAS_ISBLANK, since that seems to
493 be the best way through the thicket. isblank arrived in C99,
494 but seems to be arriving at different systems at different
498 fixed typo in b.c that set tmpvec wrongly. no one had ever
499 run into the problem, apparently. thanks to alistair crooks.
502 minor fix in lib.c: increase inputFS to 100, change malloc
505 fixed memory fault caused by out of order test in setsval.
507 thanks to david o'brien, freebsd, for both fixes.
510 fiddle in makefile to fix for BSD make; thanks to igor sobrado.
513 fixed some null pointer refs calling adjbuf.
516 fixed a bug in matching the null RE in sub and gsub. thanks to al aho
517 who actually did the fix (in b.c), and to wolfgang seeberg for finding
518 it and providing a very compact test case.
520 fixed quotation in b.c; thanks to Hal Pratt and the Princeton Dante
523 removed some no-effect asserts in run.c.
525 fiddled maketab.c to not complain about bison-generated values.
527 removed the obsolete -V argument; fixed --version to print the
530 fixed wording and an outright error in the usage message; thanks to igor
531 sobrado and jason mcintyre.
533 fixed a bug in -d that caused core dump if no program followed.
536 dropped mac.code from makefile; there are few non-MacOSX
540 system() not flagged as unsafe in the unadvertised -safe option.
541 found it while enhancing tests before shipping the ;login: article.
542 practice what you preach.
544 removed the 9-years-obsolete -mr and -mf flags.
546 added -version and --version options.
548 core dump on linux with BEGIN {nextfile}, now fixed.
550 removed some #ifdef's in run.c and lex.c that appear to no
554 modified lib.c so that values of $0 et al are preserved in the END
555 block, apparently as required by posix. thanks to havard eidnes
556 for the report and code.
559 fixed infinite loop in parsing, originally found by brian tsang.
560 thanks to arnold robbins for a suggestion that started me
564 prevent overflow of -f array in main, head off potential error in
565 call of SYNTAX(), test malloc return in lib.c, all with thanks to
569 cranked up size of NCHARS; coverity thinks it can be overrun with
570 smaller size, and i think that's right. added some assertions to b.c
571 to catch places where it might overrun. the RE code is still fragile.
574 fixed a couple of overflow problems with ridiculous field numbers:
575 e.g., print $(2^32-1). thanks to ruslan ermilov, giorgos keramidas
576 and david o'brien at freebsd.org for patches. this really should
577 be re-done from scratch.
580 fixed another 25-year-old RE bug, in split. it's another failure
581 to (re-)initialize. thanks to steve fisher for spotting this and
582 providing a good test case.
585 fixed a bug in regular expressions that dates (so help me) from 1977;
586 it's been there from the beginning. an anchored longest match that
587 was longer than the number of states triggered a failure to initialize
588 the machine properly. many thanks to moinak ghosh for not only finding
589 this one but for providing a fix, in some of the most mysterious
592 fixed a storage leak in call() that appears to have been there since
593 1983 or so -- a function without an explicit return that assigns a
594 string to a parameter leaked a Cell. thanks to moinak ghosh for
595 spotting this very subtle one.
598 fixed, thanks to andrey chernov and ruslan ermilov, a bug in lex.c
599 that mis-handled the character 255 in input. (it was being compared
600 to EOF with a signed comparison.)
603 fixed (i think) the long-standing botch that included the beginning of
604 line state ^ for RE's in the set of valid characters; this led to a
605 variety of odd problems, including failure to properly match certain
606 regular expressions in non-US locales. thanks to ruslan for keeping
610 n-th try at getting internationalization right, with thanks to volker
611 kiefel, arnold robbins and ruslan ermilov for advice, though they
612 should not be blamed for the outcome. according to posix, "." is the
613 radix character in programs and command line arguments regardless of
614 the locale; otherwise, the locale should prevail for input and output
615 of numbers. so it's intended to work that way.
617 i have rescinded the attempt to use strcoll in expanding shorthands in
618 regular expressions (cclenter). its properties are much too
619 surprising; for example [a-c] matches aAbBc in locale en_US but abBcC
620 in locale fr_CA. i can see how this might arise by implementation
621 but i cannot explain it to a human user. (this behavior can be seen
622 in gawk as well; we're leaning on the same library.)
624 the issue appears to be that strcoll is meant for sorting, where
625 merging upper and lower case may make sense (though note that unix
626 sort does not do this by default either). it is not appropriate
627 for regular expressions, where the goal is to match specific
628 patterns of characters. in any case, the notations [:lower:], etc.,
629 are available in awk, and they are more likely to work correctly in
632 a moratorium is hereby declared on internationalization changes.
633 i apologize to friends and colleagues in other parts of the world.
634 i would truly like to get this "right", but i don't know what
635 that is, and i do not want to keep making changes until it's clear.
638 fixed bug that permitted non-terminated RE, as in "awk /x".
641 subtle change to split: if source is empty, number of elems
642 is always 0 and the array is not set.
645 added some parens to isblank, in another attempt to make things
646 internationally portable.
649 the internationalization changes, somewhat modified, are now
650 reinstated. in theory awk will now do character comparisons
651 and case conversions in national language, but "." will always
652 be the decimal point separator on input and output regardless
653 of national language. isblank(){} has an #ifndef.
655 this no longer compiles on windows: LC_MESSAGES isn't defined
658 fixed subtle behavior in field and record splitting: if FS is
659 a single character and RS is not empty, \n is NOT a separator.
660 this tortuous reading is found in the awk book; behavior now
661 matches gawk and mawk.
664 for the moment, the internationalization changes of nov 29 are
665 rolled back -- programs like x = 1.2 don't work in some locales,
666 because the parser is expecting x = 1,2. until i understand this
667 better, this will have to wait.
670 modified b.c (with tiny changes in main and run) to support
671 locales, using strcoll and iswhatever tests for posix character
672 classes. thanks to ruslan ermilov (ru@freebsd.org) for code.
673 the function isblank doesn't seem to have propagated to any
674 header file near me, so it's there explicitly. not properly
675 tested on non-ascii character sets by me.
678 modified run/format() and tran/getsval() to do a slightly better
679 job on using OFMT for output from print and CONVFMT for other
680 number->string conversions, as promised by posix and done by
681 gawk and mawk. there are still places where it doesn't work
682 right if CONVFMT is changed; by then the STR attribute of the
683 variable has been irrevocably set. thanks to arnold robbins for
686 fixed subtle bug in format that could get core dump. thanks to
687 Jaromir Dolecek <jdolecek@NetBSD.org> for finding and fixing.
688 minor cleanup in run.c / format() at the same time.
690 added some tests for null pointers to debugging printf's, which
691 were never intended for external consumption. thanks to dave
692 kerns (dkerns@lucent.com) for pointing this out.
694 GNU compatibility: an empty regexp matches anything (thanks to
695 dag-erling smorgrav, des@ofug.org). subject to reversion if
696 this does more harm than good.
698 pervasive small changes to make things more const-correct, as
699 reported by gcc's -Wwrite-strings. as it says in the gcc manual,
700 this may be more nuisance than useful. provoked by a suggestion
701 and code from arnaud desitter, arnaud@nimbus.geog.ox.ac.uk
703 minor documentation changes to note that this now compiles out
704 of the box on Mac OS X.
707 changed types in posix chars structure to quiet solaris cc.
710 fflush() or fflush("") flushes all files and pipes.
712 length(arrayname) returns number of elements; thanks to
713 arnold robbins for suggestion.
715 added a makefile.win to make it easier to build on windows.
716 based on dan allen's buildwin.bat.
719 added support for posix character class names like [:digit:],
720 which are not exactly shorter than [0-9] and perhaps no more
721 portable. thanks to dag-erling smorgrav for code.
724 removed -m option; no longer needed, and it was actually
725 broken (noted thanks to volker kiefel).
728 fixed an appalling bug in gettok: any sequence of digits, +,-, E, e,
729 and period was accepted as a valid number if it started with a period.
730 this would never have happened with the lex version.
732 other 1-character botches, now fixed, include a bare $ and a
733 bare " at the end of the input.
736 more (const char *) casts in b.c and tran.c to silence warnings.
739 fixed a bug introduced in august 1997 that caused expressions
740 like $f[1] to be syntax errors. thanks to arnold robbins for
741 noticing this and providing a fix.
744 fixed some nextfile bugs: not handling all cases. thanks to
745 arnold robbins for pointing this out. new regressions added.
747 close() is now a function. it returns whatever the library
748 fclose returns, and -1 for closing a file or pipe that wasn't
752 permit \n explicitly in character classes; won't work right
753 if comes in as "[\n]" but ok as /[\n]/, because of multiple
754 processing of \'s. thanks to arnold robbins.
757 minor fiddles in tran.c to keep compilers happy about uschar.
758 thanks to norman wilson.
761 yet another attempt at making 8-bit input work, with another
762 band-aid in b.c (member()), and some (uschar) casts to head
763 off potential errors in subscripts (like isdigit). also
764 changed HAT to NCHARS-2. thanks again to santiago vila.
766 changed maketab.c to ignore apparently out of range definitions
767 instead of halting; new freeBSD generates one. thanks to
768 jon snader <jsnader@ix.netcom.com> for pointing out the problem.
771 fixed an 8-bit problem in b.c by making several char*'s into
772 unsigned char*'s. not clear i have them all yet. thanks to
773 Santiago Vila <sanvila@unex.es> for the bug report.
776 finally found and fixed a memory leak in function call; it's
777 been there since functions were added ~1983. thanks to
778 jon bentley for the test case that found it.
780 added test in envinit to catch environment "variables" with
781 names beginning with '='; thanks to Berend Hasselman.
784 added test in defn() to catch function foo(foo), which
785 otherwise recurses until core dump. thanks to arnold
786 robbins for noticing this.
789 added *bp in gettok in lex.c; appears possible to exit function
790 without terminating the string. thanks to russ cox.
793 added function stdinit() to run to initialize files[] array,
794 in case stdin, etc., are not constants; some compilers care.
797 replaced the ERROR ... FATAL, etc., macros with functions
798 based on vprintf, to avoid problems caused by overrunning
799 fixed-size errbuf array. thanks to ralph corderoy for the
800 impetus, and for pointing out a string termination bug in
804 fixed bug that caused occasional core dumps with commandline
805 variable with value ending in \. (thanks to nelson beebe for
809 with code kindly provided by Bruce Lilly, awk now parses
810 /=/ and similar constructs more sensibly in more places.
811 Bruce also provided some helpful test cases.
814 changed true/false to True/False in run.c to make it
815 easier to compile with C++. Added some casts on malloc
816 and realloc to be honest about casts; ditto. changed
817 ltype int to long in struct rrow to reduce some 64-bit
818 complaints; other changes scattered throughout for the
819 same purpose. thanks to Nelson Beebe for these portability
822 removed some horrible pointer-int casting in b.c and elsewhere
823 by adding ptoi and itonp to localize the casts, which are
824 all benign. fixed one incipient bug that showed up on sgi
827 reset lineno for new source file; include filename in error
828 message. also fixed line number error in continuation lines.
829 (thanks to Nelson Beebe for both of these.)
832 Nelson Beebe notes that irix 5.3 yacc dies with a bogus
833 error; use a newer version or switch to bison, since sgi
834 is unlikely to fix it.
837 changed isnumber to is_number to avoid the problem caused by
838 versions of ctype.h that include the name isnumber.
840 distribution now includes a script for building on a Mac,
844 fixed memory leaks in run.c (call) and tran.c (setfval).
845 thanks to Stephen Nutt for finding these and providing the fixes.
848 replaced srand argument by (unsigned int) in run.c;
849 avoids problem on Mac and potentially on Unix & Windows.
852 added a few (int) casts to silence useless compiler warnings.
853 e.g., errorflag= in run.c jump().
855 added proctab.c to the bundle outout; one less thing
856 to have to compile out of the box.
858 added calls to _popen and _pclose to the win95 stub for
859 pipes (thanks to Steve Adams for this helpful suggestion).
860 seems to work, though properties are not well understood
861 by me, and it appears that under some circumstances the
862 pipe output is truncated. Be careful.
865 fixed a couple of bugs in getrec: could fail to update $0
866 after a getline var; because inputFS wasn't initialized,
867 could split $0 on every character, a misleading diversion.
869 fixed caching bug in makedfa: LRU was actually removing
872 thanks to ross ridge for finding these, and for providing
876 fixed potential bug in readrec: might fail to update record
877 pointer after growing. thanks to dan levy for spotting this
878 and suggesting the fix.
881 added -V to print version number and die.
883 [notify dave kerns, dkerns@dacsoup.ih.lucent.com]
886 subtle silent bug in lex.c: if the program ended with a number
887 longer than 1 digit, part of the input would be pushed back and
888 parsed again because token buffer wasn't terminated right.
889 example: awk 'length($0) > 10'. blush. at least i found it
893 s/adelete/awkdelete/: SGI uses this in malloc.h.
894 thanks to nelson beebe for pointing this one out.
897 fixed some bugs in sub and gsub when replacement includes \\.
898 this is a dark, horrible corner, but at least now i believe that
899 the behavior is the same as gawk and the intended posix standard.
900 thanks to arnold robbins for advice here.
903 somewhat regretfully, replaced the ancient lex-based lexical
904 analyzer with one written in C. it's longer, generates less code,
905 and more portable; the old one depended too much on mysterious
906 properties of lex that were not preserved in other environments.
907 in theory these recognize the same language.
909 now using strtod to test whether a string is a number, instead of
910 the convoluted original function. should be more portable and
911 reliable if strtod is implemented right.
913 removed now-pointless optimization in makefile that tries to avoid
914 recompilation when awkgram.y is changed but symbols are not.
916 removed most fixed-size arrays, though a handful remain, some
917 of which are unchecked. you have been warned.
920 with some trepidation, replaced the ancient code that managed
921 fields and $0 in fixed-size arrays with arrays that grow on
922 demand. there is still some tension between trying to make this
923 run fast and making it clean; not sure it's right yet.
925 the ill-conceived -mr and -mf arguments are now useful only
926 for debugging. previous dynamic string code removed.
928 numerous other minor cleanups along the way.
931 using code provided by dan levy (to whom profuse thanks), replaced
932 fixed-size arrays and awkward kludges by a fairly uniform mechanism
933 to grow arrays as needed for printf, sub, gsub, etc.
936 falling off the end of a function returns "" and 0, not 0.
937 thanks to arnold robbins.
940 replaced several fixed-size arrays by dynamically-created ones
941 in run.c; added overflow tests to some previously unchecked cases.
942 getline, toupper, tolower.
944 getline code is still broken in that recursive calls may wind
945 up using the same space. [fixed later]
947 increased RECSIZE to 8192 to push problems further over the horizon.
949 added \r to \n as input line separator for programs, not data.
952 modified format() to permit explicit printf("%c", 0) to include
953 a null byte in output. thanks to ken stailey for the fix.
955 added a "-safe" argument that disables file output (print >,
956 print >>), process creation (cmd|getline, print |, system), and
957 access to the environment (ENVIRON). this is a first approximation
958 to a "safe" version of awk, but don't rely on it too much. thanks
959 to joan feigenbaum and matt blaze for the inspiration long ago.
962 fixed long-standing bug in sub, gsub(/a/, "\\\\&"); thanks to
966 fixed awful bug in new field splitting; didn't get all the places
967 where input was done.
970 changed field-splitting to conform to posix definition: fields are
971 split using the value of FS at the time of input; it used to be
972 the value when the field or NF was first referred to, a much less
973 predictable definition. thanks to arnold robbins for encouragement
974 to do the right thing.
977 fixed appalling but apparently unimportant bug in parsing octal
978 numbers in reg exprs.
980 explicit hex in reg exprs now limited to 2 chars: \xa, \xaa.
983 cleaned up some declarations so gcc -Wall is now almost silent.
985 makefile now includes backup copies of ytab.c and lexyy.c in case
986 one makes before looking; it also avoids recreating lexyy.c unless
989 s/aprintf/awkprint, s/asprintf/awksprintf/ to avoid some name clashes
990 with unwisely-written header files.
992 thanks to jeffrey friedl for several of these.
995 an attempt to rationalize the (unsigned) char issue. almost all
996 instances of unsigned char have been removed; the handful of places
997 in b.c where chars are used as table indices have been hand-crafted.
998 added some latin-1 tests to the regression, but i'm not confident;
999 none of my compilers seem to care much. thanks to nelson beebe for
1000 pointing out some others that do care.
1003 removed all register declarations.
1005 enhanced split(), as in gawk, etc: split(s, a, "") splits s into
1006 a[1]...a[length(s)] with each character a single element.
1008 made the same changes for field-splitting if FS is "".
1010 added nextfile, as in gawk: causes immediate advance to next
1011 input file. (thanks to arnold robbins for inspiration and code).
1013 small fixes to regexpr code: can now handle []], [[], and
1014 variants; [] is now a syntax error, rather than matching
1015 everything; [z-a] is now empty, not z. far from complete
1016 or correct, however. (thanks to jeffrey friedl for pointing out
1017 some awful behaviors.)
1020 replaced uchar by uschar everywhere; apparently some compilers
1021 usurp this name and this causes conflicts.
1023 fixed call to time in run.c (bltin); arg is time_t *.
1025 replaced horrible pointer/long punning in b.c by a legitimate
1026 union. should be safer on 64-bit machines and cleaner everywhere.
1027 (thanks to nelson beebe for pointing out some of these problems.)
1029 replaced nested comments by #if 0...#endif in run.c, lib.c.
1031 removed getsval, setsval, execute macros from run.c and lib.c.
1032 machines are 100x faster than they were when these macros were
1035 revised filenames: awk.g.y => awkgram.y, awk.lx.l => awklex.l,
1036 y.tab.[ch] => ytab.[ch], lex.yy.c => lexyy.c, all in the aid of
1037 portability to nameless systems.
1039 "make bundle" now includes yacc and lex output files for recipients
1040 who don't have yacc or lex.
1043 initialized Cells in setsymtab more carefully; some fields
1044 were not set. (thanks to purify, all of whose complaints i
1045 think i now understand.)
1047 fixed at least one error in gsub that looked at -1-th element
1048 of an array when substituting for a null match (e.g., $).
1050 delete arrayname is now legal; it clears the elements but leaves
1051 the array, which may not be the right behavior.
1053 modified makefile: my current make can't cope with the test used
1054 to avoid unnecessary yacc invocations.
1057 added dynamically growing strings to awk.lx.l and b.c
1058 to permit regular expressions to be much bigger.
1059 the state arrays can still overflow.
1062 detect duplicate arguments in function definitions (mdm).
1065 trivial fix to printf to limit string size in sub().
1068 fixed yet another subtle self-assignment problem:
1069 $1 = $2; $1 = $1 clobbered $1.
1071 Regression tests now use private echo, to avoid quoting problems.
1074 changed error() to print line number as %d, not %g.
1077 cosmetic changes: increased sizes of some arrays,
1078 reworded some error messages.
1080 added CONVFMT as in posix (just replaced OFMT in getsval)
1082 FILENAME is now "" until the first thing that causes a file
1086 deleted yyunput and yyoutput from proto.h;
1087 different versions of lex give these different declarations.
1090 added -mr N and -mf N options: more record and fields.
1091 these really ought to adjust automatically.
1093 cleaned up some error messages; "out of space" now means
1094 malloc returned NULL in all cases.
1096 changed rehash so that if it runs out, it just returns;
1097 things will continue to run slow, but maybe a bit longer.
1100 remove redundant close of stdin when using -f -.
1102 got rid of core dump with -d; awk -d just prints date.
1105 added explicit check for /dev/std(in,out,err) in redirection.
1106 unlike gawk, no /dev/fd/n yet.
1108 added (file/pipe) builtin. hard to test satisfactorily.
1112 recompile after abortive changes; should be unchanged.
1115 die-casting time: converted to ansi C, installed that.
1118 fixed storage leak in freefa, failing to recover [N]CCL.
1119 thanks to Bill Jones (jones@cs.usask.ca)
1122 use RAND_MAX instead of literal in builtin().
1125 cranked up some fixed-size arrays in b.c, and added a test for
1126 overflow in penter. thanks to mark larsen.
1129 increased buffer in gsub. a very crude fix to a general problem.
1130 and again on Sep 26.
1133 enforce variable name syntax for commandline variables: has to
1134 start with letter or _.
1137 allow newline after ; in for statements.
1140 fixed so that in self-assignment like $1=$1, side effects
1141 like recomputing $0 take place. (this is getting subtle.)
1144 better test for detecting too-long output record.
1147 better defense against very long printf strings.
1148 made break and continue illegal outside of loops.
1151 removed extra arg on gettemp, tempfree. minor error message rewording.
1154 fixed silly bug in hex parsing in hexstr().
1155 removed an apparently unnecessary test in isnumber().
1156 warn about weird printf conversions.
1157 fixed unchecked array overwrite in relex().
1159 changed for (i in array) to access elements in sorted order.
1160 then unchanged it -- it really does run slower in too many cases.
1161 left the code in place, commented out.
1164 check error status on all writes, to avoid banging on full disks.
1167 awk -f - reads the program from stdin.
1170 failed to set numeric state on $0 in cmd|getline context in run.c.
1173 fixed sleazy test for integrality in getsval; use modf.
1176 fixed sleazy buggy code in lib.c that looked (incorrectly) for
1177 too long input lines.
1180 fixed the bug on p. 198 in which it couldn't deduce that an
1181 argument was an array in some contexts. replaced the error
1182 message in intest() by code that damn well makes it an array.
1185 fixed horrible bug: types and values were not preserved in
1186 some kinds of self-assignment. (in assign().)
1189 changed NCHARS to 256 to handle 8-bit characters in strings
1190 presented to match(), etc.
1193 changed struct rrow (awk.h) to use long instead of int for lval,
1194 since cfoll() stores a pointer in it. now works better when int's
1195 are smaller than pointers!
1198 AVA fixed the grammar so that ! is uniformly of the same precedence as
1199 unary + and -. This renders illegal some constructs like !x=y, which
1200 now has to be parenthesized as !(x=y), and makes others work properly:
1201 !x+y is (!x)+y, and x!y is x !y, not two pattern-action statements.
1202 (These problems were pointed out by Bob Lenk of Posix.)
1204 Added \x to regular expressions (already in strings).
1205 Limited octal to octal digits; \8 and \9 are not octal.
1206 Centralized the code for parsing escapes in regular expressions.
1207 Added a bunch of tests to T.re and T.sub to verify some of this.
1210 fixed null pointer dereference bug in main.c: -F[nothing]. sigh.
1212 restored srand behavior: it returns the current seed.
1215 srand now returns previous seed value (0 to start).
1218 fix potential problem in tran.c -- something was freed,
1219 then used in freesymtab.
1222 another try to get the max number of open files set with
1223 relatively machine-independent code.
1225 small fix to input() in case of multiple reads after EOF.
1228 FILENAME is now defined in the BEGIN block -- too many old
1231 "-" means stdin in getline as well as on the commandline.
1233 added a bunch of casts to the code to tell the truth about
1234 char * vs. unsigned char *, a right royal pain. added a
1235 setlocale call to the front of main, though probably no one
1236 has it usefully implemented yet.
1239 removed redundant relational tests against nullnode if parse
1240 tree already had a relational at that point.
1243 fixed bug: commandline variable assignment has to look like
1244 var=something. (consider the man page for =, in file =.1)
1246 changed number of arguments to functions to static arrays
1247 to avoid repeated malloc calls.
1250 restored -F (space) separator
1253 added -v x=1 y=2 ... for immediate commandline variable assignment;
1254 done before the BEGIN block for sure. they have to precede the
1255 program if the program is on the commandline.
1256 Modified Aug 2 to require a separate -v for each assignment.
1259 fixed ref-thru-zero bug in environment code in tran.c
1262 add newline to usage message.
1265 added some missing ansi printf conversion letters: %i %X %E %G.
1266 no sensible meaning for h or L, so they may not do what one expects.
1268 made %* conversions work.
1270 changed x^y so that if n is a positive integer, it's done
1271 by explicit multiplication, thus achieving maximum accuracy.
1272 (this should be done by pow() but it seems not to be locally.)
1273 done to x ^= y as well.
1276 ENVIRON array contains environment: if shell variable V=thing,
1277 ENVIRON["V"] is "thing"
1279 multiple -f arguments permitted. error reporting is naive.
1280 (they were permitted before, but only the last was used.)
1282 fixed a really stupid botch in the debugging macro dprintf
1284 fixed order of evaluation of commandline assignments to match
1285 what the book claims: an argument of the form x=e is evaluated
1286 at the time it would have been opened if it were a filename (p 63).
1287 this invalidates the suggested answer to ex 4-1 (p 195).
1289 removed some code that permitted -F (space) fieldseparator,
1290 since it didn't quite work right anyway. (restored aug 2)
1293 Line number now accumulated correctly for comment lines.
1296 Debugging output now includes a version date,
1297 if one compiles it into the source each time.
1300 Changed grammar to prohibit constants as 3rd arg of sub and gsub;
1301 prevents class of overwriting-a-constant errors. (Last one?)
1302 This invalidates the "banana" example on page 43 of the book.
1304 Added \a ("alert"), \v (vertical tab), \xhhh (hexadecimal),
1305 as in ANSI, for strings. Rescinded the sloppiness that permitted
1306 non-octal digits in \ooo. Warning: not all compilers and libraries
1307 will be able to deal with \x correctly.
1310 Fixed bug that caused tempcell list to contain a duplicate.
1314 Catches some more commandline errors in main.
1315 Removed redundant decl of modf in run.c (confuses some compilers).
1316 Warning: there's no single declaration of malloc, etc., in awk.h
1317 that seems to satisfy all compilers.
1320 Added a bit of code to error printing to avoid printing nulls.
1321 (Not clear that it actually would.)
1324 With fear and trembling, modified the grammar to permit
1325 multiple pattern-action statements on one line without
1326 an explicit separator. By definition, this capitulation
1327 to the ghost of ancient implementations remains undefined
1328 and thus subject to change without notice or apology.
1332 Fixed bug in call() that failed to recover storage.
1334 A warning is now generated if there are more arguments
1335 in the call than in the definition (in lieu of fixing
1336 another storage leak).
1339 Fixed %c: if expr is numeric, use numeric value;
1340 otherwise print 1st char of string value. still
1341 doesn't work if the value is 0 -- won't print \0.
1343 Added a few more checks for running out of malloc.
1346 Fixed bug in call() that freed local arrays twice.
1348 Fixed to handle deletion of non-existent array right;
1349 complains about attempt to delete non-array element.
1352 Now guarantees to evaluate all arguments of built-in
1353 functions, as in C; the appearance is that arguments
1354 are evaluated before the function is called. Places
1355 affected are sub (gsub was ok), substr, printf, and
1356 all the built-in arithmetic functions in bltin().
1357 A warning is generated if a bltin() is called with
1358 the wrong number of arguments.
1360 This requires changing makeprof on p167 of the book.
1363 setting FILENAME in BEGIN caused core dump, apparently
1364 because it was freeing space not allocated by malloc.
1367 fixed egregious error in toupper/tolower functions.
1368 still subject to rescinding, however.
1371 flush stdout before opening file or pipe
1374 performance bug in b.c/cgoto(): not freeing some sets of states.
1375 partial fix only right now, and the number of states increased
1376 to make it less obvious.
1379 check error status on close
1382 srand returns seed value it's using.
1386 Removed limit on depth of function calls.
1389 Fixed lib.c to permit _ in commandline variable names.
1392 main.c fixed to recognize -- as terminator of command-
1393 line options. Illegal options flagged.
1394 Error reporting slightly cleaned up.
1397 Newer C compilers apply a strict scope rule to extern
1398 declarations within functions. Two extern declarations in
1399 lib.c and tran.c have been moved to obviate this problem.
1402 Reluctantly added toupper and tolower functions.
1403 Subject to rescinding without notice.
1406 Error-message printer had printf(s) instead of
1407 printf("%s",s); got core dumps when the message
1411 Very long printf strings caused core dump;
1412 fixed aprintf, asprintf, format to catch them.
1413 Can still get a core dump in printf itself.