rsalz@uunet.uu.net (Rich Salz) (02/10/89)
Submitted-by: Mike Haertel <mike@wheaties.ai.mit.edu>
Posting-number: Volume 17, Issue 98
Archive-name: gnugrep/part01
[ This replaces BM, etc., and all the other Grep's that are floating
around the net, and gathering dust in archives. Spread this around!
The "AUTHORS" section in the manual page is quite nice. --r$ ]
This is GNU e?grep version 1.2, hopefully the "fastest grep in the west."
This new version now includes a manual page and a fix for NFS environments.
Due to some control characters being present, line 121 of the file
tests/spencer.tests may not extract properly. It should look as
follows, where ^A, ^B, and ^C represent characters \001, \002, and \003
respectively: "0:a[^A-^C]?c:a^Bc"
#! /bin/sh
# This is a shell archive. Remove anything before this line, then unpack
# it by saving it into a file and typing "sh file". To overwrite existing
# files, type "sh file -c". You can also feed this as standard input via
# unshar, or by typing "sh <file", e.g.. If this archive is complete, you
# will see the following message at the end:
# "End of shell archive."
# Contents: README Makefile tests tests/khadafy.lines
# tests/khadafy.regexp tests/regress.sh tests/scriptgen.awk
# tests/spencer.tests README.cray alloca.c grep.man getopt.c
#
PATH=/bin:/usr/bin:/usr/ucb ; export PATH
if test -f 'README' -a "${1}" != "-c" ; then
echo shar: Will not clobber existing file \"'README'\"
else
echo shar: Extracting \"'README'\" \(7427 characters\)
sed "s/^X//" >'README' <<'END_OF_FILE'
XThis README documents GNU e?grep version 1.2.
X
XChanges needed to the makefile under various perversions of Unix are
Xdescribed therein.
X
XIf the type "char" is unsigned on your machine, you will have to fix
Xthe definition of the macro SIGN_EXTEND_CHAR() in regex.c. A reasonable
Xdefinition might be:
X #define SIGN_EXTEND_CHAR(c) ((c)>(char)127?(c)-256:(c))
X
XGNU e?grep is provided "as is" with no warranty. The exact terms
Xunder which you may use and (re)distribute this program are detailed
Xin a comment at the top of grep.c.
X
XGNU e?grep is based on a fast lazy-state deterministic matcher (about
Xtwice as fast as stock Unix egrep) hybridized with a Boyer-Moore-Gosper
Xsearch for a fixed string that eliminates impossible text from being
Xconsidered by the full regexp matcher without necessarily having to
Xlook at every character. The result is typically many times faster
Xthan Unix grep or egrep. (Regular expressions containing backreferencing
Xmay run more slowly, however.)
X
XGNU e?grep attempts, as closely as possible, to understand compatibly
Xthe regexp syntaxes of the Unix programs it replaces. The following table
Xdetails the various special characters understood in both the grep and
Xegrep incarnations:
X
X(grep) (egrep) (explanation)
X . . matches any single character except newline
X \? ? postfix operator; preceeding item is optional
X * * postfix operator; preceeding item 0 or more times
X \+ + postfix operator; preceeding item 1 or more times
X \| | infix operator; matches either argument
X ^ ^ matches the empty string at the beginning of a line
X $ $ matches the empty string at the end of a line
X \< \< matches the empty string at the beginning of a word
X \> \> matches the empty string at the end of a word
X [chars] [chars] match any character in the given class; if the
X first character after [ is ^, match any character
X not in the given class; a range of characters may
X be specified by <first>-<last>; for example, \W
X (below) is equivalent to the class [^A-Za-z0-9]
X \( \) ( ) parentheses are used to override operator precedence
X \<1-9> \<1-9> \<n> matches a repeat of the text matched earlier
X in the regexp by the subexpression inside the
X nth opening parenthesis
X \ \ any special character may be preceded by a backslash
X to match it literally
X
X(the following are for compatibility with GNU Emacs)
X \b \b matches the empty string at the edge of a word
X \B \B matches the empty string if not at the edge of a word
X \w \w matches word-constituent characters (letters & digits)
X \W \W matches characters that are not word-constituent
X
XOperator precedence is (highest to lowest) ?, *, and +, concatenation,
Xand finally |. All other constructs are syntactically identical to
Xnormal characters. For the truly interested, a comment in dfa.c describes
Xthe exact grammar understood by the parser.
X
XGNU e?grep understands the following command line options:
X -A <num> print <num> lines of context after every matching line
X -B <num> print <num> lines of context before every matching line
X -C print 2 lines of context on each side of every match
X -<num> print <num> lines of context on each side
X -V print the version number on stderr
X -b print every match preceded by its byte offset
X -c print a total count of matching lines only
X -e <expr> search for <expr>; useful if <expr> begins with -
X -f <file> take <expr> from the given <file>
X -h don't display filenames on matches
X -i ignore case difference when comparing strings
X -l list files containing matches only
X -n print each match preceded by its line number
X -s run silently producing no output except error messages
X -v print only lines that contain no matches for the <expr>
X -w print only lines where the match is a complete word
X -x print only lines where the match is a whole line
X
XThe options understood by GNU e?grep are meant to be (nearly) compatible
Xwith both the BSD and System V versions of grep and egrep.
X
XThe following incompatibilities with other versions of grep exist:
X the context-dependent meaning of * is not quite the same (grep only)
X -b prints a byte offset instead of a block offset
X the \{m,n\} construct of System V grep is not implemented
X
XGNU e?grep has been thoroughly debugged and tested by several people
Xover a period of several months; we think it's a reliable beast or we
Xwouldn't distribute it. If by some fluke of the universe you discover
Xa bug, send a detailed description (including options, regular
Xexpressions, and a copy of an input file that can reproduce it) to me,
Xmike@wheaties.ai.mit.edu.
X
XGNU e?grep is brought to you by the efforts of several people:
X
X Mike Haertel wrote the deterministic regexp code and the bulk
X of the program.
X
X James A. Woods is responsible for the hybridized search strategy
X of using Boyer-Moore-Gosper fixed-string search as a filter
X before calling the general regexp matcher.
X
X Arthur David Olson contributed code that finds fixed strings for
X the aforementioned BMG search for a large class of regexps.
X
X Richard Stallman wrote the backtracking regexp matcher that is
X used for \<digit> backreferences, as well as the getopt that
X is provided for 4.2BSD sites. The backtracking matcher was
X originally written for GNU Emacs.
X
X D. A. Gwyn wrote the C alloca emulation that is provided so
X System V machines can run this program. (Alloca is used only
X by RMS' backtracking matcher, and then only rarely, so there
X is no loss if your machine doesn't have a "real" alloca.)
X
X Scott Anderson and Henry Spencer designed the regression tests
X used in the "regress" script.
X
X Paul Placeway wrote the manual page, based on this README.
X
XIf you are interested in improving this program, you may wish to try
Xany of the following:
X
X1. Make backreferencing \<digit> faster. Right now, backreferencing is
X handled by calling the Emacs backtracking matcher to verify the partial
X match. This is slow; if the DFA routines could handle backreferencing
X themselves a speedup on the order of three to four times might occur
X in those cases where the backtracking matcher is called to verify nearly
X every line. Also, some portability problems due to the inclusion of the
X emacs matcher would be solved because it could then be eliminated.
X Note that expressions with backreferencing are not true regular
X expressions, and thus are not equivalent to any DFA. So this is hard.
X
X2. There is a bug in the backtracking matcher, regex.c, such that the |
X operator is not properly commutative. Let x and y be arbitrary
X regular expressions, and suppose both x and y have matches at
X some point in the target text. Then the regexp x|y should select
X the longest of the two matches. With the backtracking matcher, if the
X first match succeeds it does not even try the second, even though
X the second may be a longer match. This is obviously of no concern
X for grep, which does not care exactly where or how long a match is,
X so long as it knows it is there. On the other hand, the backtracking
X matcher is used in GNU AWK, wherein its behavior can only be considered
X a bug.
X
X3. Handle POSIX style regexps. I'm not sure if this could be called an
X improvement; some of the things on regexps in the POSIX draft I have
X seen are pretty sickening. But it would be useful in the interests of
X conforming to the standard.
END_OF_FILE
if test 7427 -ne `wc -c <'README'`; then
echo shar: \"'README'\" unpacked with wrong size!
fi
# end of 'README'
fi
if test -f 'Makefile' -a "${1}" != "-c" ; then
echo shar: Will not clobber existing file \"'Makefile'\"
else
echo shar: Extracting \"'Makefile'\" \(510 characters\)
sed "s/^X//" >'Makefile' <<'END_OF_FILE'
X#
X# Makefile for GNU e?grep
X#
X
X# Add -DUSG for System V.
XCFLAGS = -O
X
X# You may add getopt.o if your C library lacks getopt(); note that
X# 4.3BSD getopt() is said to be somewhat broken.
X# Add alloca.o if your machine does not support alloca().
XOBJS = grep.o dfa.o regex.o
X
Xall: regress
X
Xregress: grep
X cd tests; sh regress.sh
X
Xgrep: $(OBJS)
X $(CC) $(CFLAGS) -o grep $(OBJS)
X rm -f egrep
X ln grep egrep
X
Xclean:
X rm -f grep egrep *.o core tests/core tests/tmp.script
X
Xdfa.o grep.o: dfa.h
Xgrep.o regex.o: regex.h
END_OF_FILE
if test 510 -ne `wc -c <'Makefile'`; then
echo shar: \"'Makefile'\" unpacked with wrong size!
fi
# end of 'Makefile'
fi
if test ! -d 'tests' ; then
echo shar: Creating directory \"'tests'\"
mkdir 'tests'
fi
if test -f 'tests/khadafy.lines' -a "${1}" != "-c" ; then
echo shar: Will not clobber existing file \"'tests/khadafy.lines'\"
else
echo shar: Extracting \"'tests/khadafy.lines'\" \(728 characters\)
sed "s/^X//" >'tests/khadafy.lines' <<'END_OF_FILE'
X1) Muammar Qaddafi
X2) Mo'ammar Gadhafi
X3) Muammar Kaddafi
X4) Muammar Qadhafi
X5) Moammar El Kadhafi
X6) Muammar Gadafi
X7) Mu'ammar al-Qadafi
X8) Moamer El Kazzafi
X9) Moamar al-Gaddafi
X10) Mu'ammar Al Qathafi
X11) Muammar Al Qathafi
X12) Mo'ammar el-Gadhafi
X13) Moamar El Kadhafi
X14) Muammar al-Qadhafi
X15) Mu'ammar al-Qadhdhafi
X16) Mu'ammar Qadafi
X17) Moamar Gaddafi
X18) Mu'ammar Qadhdhafi
X19) Muammar Khaddafi
X20) Muammar al-Khaddafi
X21) Mu'amar al-Kadafi
X22) Muammar Ghaddafy
X23) Muammar Ghadafi
X24) Muammar Ghaddafi
X25) Muamar Kaddafi
X26) Muammar Quathafi
X27) Muammar Gheddafi
X28) Muamar Al-Kaddafi
X29) Moammar Khadafy
X30) Moammar Qudhafi
X31) Mu'ammar al-Qaddafi
X32) Mulazim Awwal Mu'ammar Muhammad Abu Minyar al-Qadhafi
END_OF_FILE
if test 728 -ne `wc -c <'tests/khadafy.lines'`; then
echo shar: \"'tests/khadafy.lines'\" unpacked with wrong size!
fi
# end of 'tests/khadafy.lines'
fi
if test -f 'tests/khadafy.regexp' -a "${1}" != "-c" ; then
echo shar: Will not clobber existing file \"'tests/khadafy.regexp'\"
else
echo shar: Extracting \"'tests/khadafy.regexp'\" \(66 characters\)
sed "s/^X//" >'tests/khadafy.regexp' <<'END_OF_FILE'
XM[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]
END_OF_FILE
if test 66 -ne `wc -c <'tests/khadafy.regexp'`; then
echo shar: \"'tests/khadafy.regexp'\" unpacked with wrong size!
fi
# end of 'tests/khadafy.regexp'
fi
if test -f 'tests/regress.sh' -a "${1}" != "-c" ; then
echo shar: Will not clobber existing file \"'tests/regress.sh'\"
else
echo shar: Extracting \"'tests/regress.sh'\" \(455 characters\)
sed "s/^X//" >'tests/regress.sh' <<'END_OF_FILE'
X#! /bin/sh -
X
Xfailures=0
X
X# The Khadafy test is brought to you by Scott Anderson . . .
X../egrep -f khadafy.regexp khadafy.lines > khadafy.out
Xif cmp khadafy.lines khadafy.out
Xthen
X rm khadafy.out
Xelse
X echo Khadafy test failed -- output left on khadafy.out
X failures=1
Xfi
X
X# . . . and the following by Henry Spencer.
X
Xawk -F: -f scriptgen.awk spencer.tests > tmp.script
X
Xif sh tmp.script
Xthen
X rm tmp.script
X exit $failures
Xelse
X rm tmp.script
X exit 1
Xfi
END_OF_FILE
if test 455 -ne `wc -c <'tests/regress.sh'`; then
echo shar: \"'tests/regress.sh'\" unpacked with wrong size!
fi
# end of 'tests/regress.sh'
fi
if test -f 'tests/scriptgen.awk' -a "${1}" != "-c" ; then
echo shar: Will not clobber existing file \"'tests/scriptgen.awk'\"
else
echo shar: Extracting \"'tests/scriptgen.awk'\" \(274 characters\)
sed "s/^X//" >'tests/scriptgen.awk' <<'END_OF_FILE'
XBEGIN { print "failures=0"; }
X!/^#/ && NF == 3 {
X print "echo '" $3 "' | ../egrep -e '" $2 "' > /dev/null 2>&1";
X print "if [ $? != " $1 " ]"
X print "then"
X printf "\techo Spencer test \\#%d failed\n", ++n
X print "\tfailures=1"
X print "fi"
X}
XEND { print "exit $failures"; }
END_OF_FILE
if test 274 -ne `wc -c <'tests/scriptgen.awk'`; then
echo shar: \"'tests/scriptgen.awk'\" unpacked with wrong size!
fi
# end of 'tests/scriptgen.awk'
fi
if test -f 'tests/spencer.tests' -a "${1}" != "-c" ; then
echo shar: Will not clobber existing file \"'tests/spencer.tests'\"
else
echo shar: Extracting \"'tests/spencer.tests'\" \(1748 characters\)
sed "s/^X//" >'tests/spencer.tests' <<'END_OF_FILE'
X0:abc:abc
X1:abc:xbc
X1:abc:axc
X1:abc:abx
X0:abc:xabcy
X0:abc:ababc
X0:ab*c:abc
X0:ab*bc:abc
X0:ab*bc:abbc
X0:ab*bc:abbbbc
X0:ab+bc:abbc
X1:ab+bc:abc
X1:ab+bc:abq
X0:ab+bc:abbbbc
X0:ab?bc:abbc
X0:ab?bc:abc
X1:ab?bc:abbbbc
X0:ab?c:abc
X0:^abc$:abc
X1:^abc$:abcc
X0:^abc:abcc
X1:^abc$:aabc
X0:abc$:aabc
X0:^:abc
X0:$:abc
X0:a.c:abc
X0:a.c:axc
X0:a.*c:axyzc
X1:a.*c:axyzd
X1:a[bc]d:abc
X0:a[bc]d:abd
X1:a[b-d]e:abd
X0:a[b-d]e:ace
X0:a[b-d]:aac
X0:a[-b]:a-
X2:a[b-]:a-
X1:a[b-a]:-
X2:a[]b:-
X2:a[:-
X0:a]:a]
X0:a[]]b:a]b
X0:a[^bc]d:aed
X1:a[^bc]d:abd
X0:a[^-b]c:adc
X1:a[^-b]c:a-c
X1:a[^]b]c:a]c
X0:a[^]b]c:adc
X0:ab|cd:abc
X0:ab|cd:abcd
X0:()ef:def
X0:()*:-
X1:*a:-
X0:^*:-
X0:$*:-
X1:(*)b:-
X1:$b:b
X2:a\:-
X0:a\(b:a(b
X0:a\(*b:ab
X0:a\(*b:a((b
X1:a\x:a\x
X2:abc):-
X2:(abc:-
X0:((a)):abc
X0:(a)b(c):abc
X0:a+b+c:aabbabc
X0:a**:-
X0:a*?:-
X0:(a*)*:-
X0:(a*)+:-
X0:(a|)*:-
X0:(a*|b)*:-
X0:(a+|b)*:ab
X0:(a+|b)+:ab
X0:(a+|b)?:ab
X0:[^ab]*:cde
X0:(^)*:-
X0:(ab|)*:-
X2:)(:-
X1:abc:
X1:abc:
X0:a*:
X0:([abc])*d:abbbcd
X0:([abc])*bcd:abcd
X0:a|b|c|d|e:e
X0:(a|b|c|d|e)f:ef
X0:((a*|b))*:-
X0:abcd*efg:abcdefg
X0:ab*:xabyabbbz
X0:ab*:xayabbbz
X0:(ab|cd)e:abcde
X0:[abhgefdc]ij:hij
X1:^(ab|cd)e:abcde
X0:(abc|)ef:abcdef
X0:(a|b)c*d:abcd
X0:(ab|ab*)bc:abc
X0:a([bc]*)c*:abc
X0:a([bc]*)(c*d):abcd
X0:a([bc]+)(c*d):abcd
X0:a([bc]*)(c+d):abcd
X0:a[bcd]*dcdcde:adcdcde
X1:a[bcd]+dcdcde:adcdcde
X0:(ab|a)b*c:abc
X0:((a)(b)c)(d):abcd
X0:[A-Za-z_][A-Za-z0-9_]*:alpha
X0:^a(bc+|b[eh])g|.h$:abh
X0:(bc+d$|ef*g.|h?i(j|k)):effgz
X0:(bc+d$|ef*g.|h?i(j|k)):ij
X1:(bc+d$|ef*g.|h?i(j|k)):effg
X1:(bc+d$|ef*g.|h?i(j|k)):bcdd
X0:(bc+d$|ef*g.|h?i(j|k)):reffgz
X1:((((((((((a)))))))))):-
X0:(((((((((a))))))))):a
X1:multiple words of text:uh-uh
X0:multiple words:multiple words, yeah
X0:(.*)c(.*):abcde
X1:\((.*),:(.*)\)
X1:[k]:ab
X0:abcd:abcd
X0:a(bc)d:abcd
X0:a[-]?c:ac
X0:(....).*\1:beriberi
END_OF_FILE
echo shar: 3 control characters may be missing from \"'tests/spencer.tests'\"
if test 1748 -ne `wc -c <'tests/spencer.tests'`; then
echo shar: \"'tests/spencer.tests'\" unpacked with wrong size!
fi
# end of 'tests/spencer.tests'
fi
if test -f 'README.cray' -a "${1}" != "-c" ; then
echo shar: Will not clobber existing file \"'README.cray'\"
else
echo shar: Extracting \"'README.cray'\" \(2730 characters\)
sed "s/^X//" >'README.cray' <<'END_OF_FILE'
X(Message inbox:135)
XDate: Mon, 17 Oct 88 16:53:33 PDT
XTo: mike@wheaties.ai.mit.edu
Xcc: darin%pioneer@eos.arc.nasa.gov, luzmoor@violet.berkeley.edu
XFrom: James A. Woods <jaw@eos.arc.nasa.gov>
XSubject: README.cray for GNU e?grep
X
XI just sent this out to comp.unix.cray:
X
X-------------------------------------------------------------------
XFrom: jaw@eos.UUCP (James A. Woods)
XNewsgroups: comp.unix.cray
XSubject: GNU e?grep on Cray machines
XMessage-ID: <1750@eos.UUCP>
XDate: 17 Oct 88 23:47:29 GMT
XOrganization: NASA Ames Research Center, California
XLines: 66
X
X# "What comes after silicon? Oh, gallium arsenide, I'd guess. And after
X that, there's a thing called indium phosphide."
X -- Seymour Cray, Datamation interview, circa 1980
X
X Now that most Cray software development is done on Crays themselves,
Xthanks to Unix, GNU e?grep should come in handy. Of course, if you're
Xscanning GENBANK for the Human Genome Project at 10 MB/second (the raw
XX/MP Unix I/O rate), you really do need the speed.
X
X Sample, from one of the Ames Cray 2 machines:
X
X stokes> time ./egrep astrian web2 # GNU egrep
X alabastrian
X Lancastrian
X Zoroastrian
X Zoroastrianism
X 0.5980u 0.0772s 0:01 35%
X stokes> time /usr/bin/egrep astrian web2 # ATT egrep
X alabastrian
X Lancastrian
X Zoroastrian
X Zoroastrianism
X 7.6765u 0.1373s 0:15 49%
X
X(web2 is a 2.4 MB wordlist, standard on BSD Unix.)
X
X To bring up GNU E?GREP, ftp Mike Haertel's version 1.1 package from
X'prep.ai.mit.edu' or 'ames.arc.nasa.gov'. Mention -DUSG in the Makefile,
Xand specify
X
X #define SIGN_EXTEND_CHAR(c) ((c)>(char)127?(c)-256:(c))
X
Xin regex.c. [Cray characters, like MIPS chars, are unsigned, but the
Xcompiler won't allow ... #define SIGN_EXTEND_CHAR(c) ((signed char) (c))]
X
X However, at least on the Cray 2, there's a compiler bug involving the
Xincrement operator in complex expressions, which requires the following
Xmodification (also in regex.c):
X
Xchange
X m->elems[m->nelem++].constraint |= s2->elems[j++].constraint;
Xto
X m->elems[m->nelem].constraint |= s2->elems[j].constraint;
X m->nelem++;
X j++;
X
XThanks go to Darin Okuyama of NASA ARC for providing this workaround.
X
X-- James A. Woods (ames!jaw)
X NASA Ames Research Center
X
XP.S.
XThough Crays are not at their best pushing bytes, the timing difference
Xis even more exaggerated with heavier regexpr processing, to wit:
X
X time ./egrep -i 'as.*Trian' web2
X ...
X 0.7677u 0.0769s 0:01 44%
Xvs.
X time /usr/bin/egrep -i 'as.*Trian' web2
X ...
X 16.1327u 0.1379s 0:32 49%
X
Xwhich is a mite unfair given a known System 5 egrep -i gaffe. You get
Xextra credit for vectorizing the inner loop of the Boyer/Moore/Gosper
Xcode, though changing all chars to ints might help also.
END_OF_FILE
if test 2730 -ne `wc -c <'README.cray'`; then
echo shar: \"'README.cray'\" unpacked with wrong size!
fi
# end of 'README.cray'
fi
if test -f 'alloca.c' -a "${1}" != "-c" ; then
echo shar: Will not clobber existing file \"'alloca.c'\"
else
echo shar: Extracting \"'alloca.c'\" \(5090 characters\)
sed "s/^X//" >'alloca.c' <<'END_OF_FILE'
X/*
X alloca -- (mostly) portable public-domain implementation -- D A Gwyn
X
X This implementation of the PWB library alloca() function,
X which is used to allocate space off the run-time stack so
X that it is automatically reclaimed upon procedure exit,
X was inspired by discussions with J. Q. Johnson of Cornell.
X
X It should work under any C implementation that uses an
X actual procedure stack (as opposed to a linked list of
X frames). There are some preprocessor constants that can
X be defined when compiling for your specific system, for
X improved efficiency; however, the defaults should be okay.
X
X The general concept of this implementation is to keep
X track of all alloca()-allocated blocks, and reclaim any
X that are found to be deeper in the stack than the current
X invocation. This heuristic does not reclaim storage as
X soon as it becomes invalid, but it will do so eventually.
X
X As a special case, alloca(0) reclaims storage without
X allocating any. It is a good idea to use alloca(0) in
X your main control loop, etc. to force garbage collection.
X*/
X#ifndef lint
Xstatic char SCCSid[] = "@(#)alloca.c 1.1"; /* for the "what" utility */
X#endif
X
X#ifdef emacs
X#include "config.h"
X#ifdef static
X/* actually, only want this if static is defined as ""
X -- this is for usg, in which emacs must undefine static
X in order to make unexec workable
X */
X#ifndef STACK_DIRECTION
Xyou
Xlose
X-- must know STACK_DIRECTION at compile-time
X#endif /* STACK_DIRECTION undefined */
X#endif static
X#endif emacs
X
X#ifdef X3J11
Xtypedef void *pointer; /* generic pointer type */
X#else
Xtypedef char *pointer; /* generic pointer type */
X#endif
X
X#define NULL 0 /* null pointer constant */
X
Xextern void free();
Xextern pointer malloc();
X
X/*
X Define STACK_DIRECTION if you know the direction of stack
X growth for your system; otherwise it will be automatically
X deduced at run-time.
X
X STACK_DIRECTION > 0 => grows toward higher addresses
X STACK_DIRECTION < 0 => grows toward lower addresses
X STACK_DIRECTION = 0 => direction of growth unknown
X*/
X
X#ifndef STACK_DIRECTION
X#define STACK_DIRECTION 0 /* direction unknown */
X#endif
X
X#if STACK_DIRECTION != 0
X
X#define STACK_DIR STACK_DIRECTION /* known at compile-time */
X
X#else /* STACK_DIRECTION == 0; need run-time code */
X
Xstatic int stack_dir; /* 1 or -1 once known */
X#define STACK_DIR stack_dir
X
Xstatic void
Xfind_stack_direction (/* void */)
X{
X static char *addr = NULL; /* address of first
X `dummy', once known */
X auto char dummy; /* to get stack address */
X
X if (addr == NULL)
X { /* initial entry */
X addr = &dummy;
X
X find_stack_direction (); /* recurse once */
X }
X else /* second entry */
X if (&dummy > addr)
X stack_dir = 1; /* stack grew upward */
X else
X stack_dir = -1; /* stack grew downward */
X}
X
X#endif /* STACK_DIRECTION == 0 */
X
X/*
X An "alloca header" is used to:
X (a) chain together all alloca()ed blocks;
X (b) keep track of stack depth.
X
X It is very important that sizeof(header) agree with malloc()
X alignment chunk size. The following default should work okay.
X*/
X
X#ifndef ALIGN_SIZE
X#define ALIGN_SIZE sizeof(double)
X#endif
X
Xtypedef union hdr
X{
X char align[ALIGN_SIZE]; /* to force sizeof(header) */
X struct
X {
X union hdr *next; /* for chaining headers */
X char *deep; /* for stack depth measure */
X } h;
X} header;
X
X/*
X alloca( size ) returns a pointer to at least `size' bytes of
X storage which will be automatically reclaimed upon exit from
X the procedure that called alloca(). Originally, this space
X was supposed to be taken from the current stack frame of the
X caller, but that method cannot be made to work for some
X implementations of C, for example under Gould's UTX/32.
X*/
X
Xstatic header *last_alloca_header = NULL; /* -> last alloca header */
X
Xpointer
Xalloca (size) /* returns pointer to storage */
X unsigned size; /* # bytes to allocate */
X{
X auto char probe; /* probes stack depth: */
X register char *depth = &probe;
X
X#if STACK_DIRECTION == 0
X if (STACK_DIR == 0) /* unknown growth direction */
X find_stack_direction ();
X#endif
X
X /* Reclaim garbage, defined as all alloca()ed storage that
X was allocated from deeper in the stack than currently. */
X
X {
X register header *hp; /* traverses linked list */
X
X for (hp = last_alloca_header; hp != NULL;)
X if (STACK_DIR > 0 && hp->h.deep > depth
X || STACK_DIR < 0 && hp->h.deep < depth)
X {
X register header *np = hp->h.next;
X
X free ((pointer) hp); /* collect garbage */
X
X hp = np; /* -> next header */
X }
X else
X break; /* rest are not deeper */
X
X last_alloca_header = hp; /* -> last valid storage */
X }
X
X if (size == 0)
X return NULL; /* no allocation required */
X
X /* Allocate combined header + user data storage. */
X
X {
X register pointer new = malloc (sizeof (header) + size);
X /* address of header */
X
X ((header *)new)->h.next = last_alloca_header;
X ((header *)new)->h.deep = depth;
X
X last_alloca_header = (header *)new;
X
X /* User storage begins just after header. */
X
X return (pointer)((char *)new + sizeof(header));
X }
X}
END_OF_FILE
if test 5090 -ne `wc -c <'alloca.c'`; then
echo shar: \"'alloca.c'\" unpacked with wrong size!
fi
# end of 'alloca.c'
fi
if test -f 'grep.man' -a "${1}" != "-c" ; then
echo shar: Will not clobber existing file \"'grep.man'\"
else
echo shar: Extracting \"'grep.man'\" \(6853 characters\)
sed "s/^X//" >'grep.man' <<'END_OF_FILE'
X.TH GREP 1 "1988 December 13" "GNU Project"
X.UC 4
X.SH NAME
Xgrep, egrep \- print lines matching a regular expression
X.SH SYNOPSIS
X.B grep
X[
X.B \-CVbchilnsvwx
X]
X[
X.B \-\c
X.I num
X]
X[
X.B \-AB
X.I num
X]
X[ [
X.B \-e
X]
X.I expr
X|
X.B \-f
X.I file
X] [
X.I "files ..."
X]
X.SH DESCRIPTION
X.I Grep
Xsearches the files listed in the arguments (or standard
Xinput if no files are given) for all lines that contain a match for
Xthe given
X.IR expr .
XIf any lines match, they are printed.
X.PP
XAlso, if any matches were found,
X.I grep
Xwill exit with a status of 0, but if no matches were found it will exit
Xwith a status of 1. This is useful for building shell scripts that
Xuse
X.I grep
Xas a condition for, for example, the
X.I if
Xstatement.
X.PP
XWhen invoked as
X.I egrep
Xthe syntax of the
X.I expr
Xis slightly different; See below.
X.br
X.SH "REGULAR EXPRESSIONS"
X.RS 2.5i
X.ta 1i; 2i
X.sp
X.ti -2.0i
X(grep) (egrep) (explanation)
X.sp
X.ti -2.0i
X\fIc\fP \fIc\fP a single (non-meta) character matches itself.
X.sp
X.ti -2.0i
X\&. . matches any single character except newline.
X.sp
X.ti -2.0i
X\\? ? postfix operator; preceeding item is optional.
X.sp
X.ti -2.0i
X\(** \(** postfix operator; preceeding item 0 or
Xmore times.
X.sp
X.ti -2.0i
X\\+ + postfix operator; preceeding item 1 or
Xmore times.
X.sp
X.ti -2.0i
X\\| | infix operator; matches either
Xargument.
X.sp
X.ti -2.0i
X^ ^ matches the empty string at the beginning of a line.
X.sp
X.ti -2.0i
X$ $ matches the empty string at the end of a line.
X.sp
X.ti -2.0i
X\\< \\< matches the empty string at the beginning of a word.
X.sp
X.ti -2.0i
X\\> \\> matches the empty string at the end of a word.
X.sp
X.ti -2.0i
X[\fIchars\fP] [\fIchars\fP] match any character in the given class; if the
Xfirst character after [ is ^, match any character
Xnot in the given class; a range of characters may
Xbe specified by \fIfirst\-last\fP; for example, \\W
X(below) is equivalent to the class [^A\-Za\-z0\-9]
X.sp
X.ti -2.0i
X\\( \\) ( ) parentheses are used to override operator precedence.
X.sp
X.ti -2.0i
X\\\fIdigit\fP \\\fIdigit\fP \\\fIn\fP matches a repeat of the text
Xmatched earlier in the regexp by the subexpression inside the nth
Xopening parenthesis.
X.sp
X.ti -2.0i
X\\ \\ any special character may be preceded
Xby a backslash to match it literally.
X.sp
X.ti -2.0i
X(the following are for compatibility with GNU Emacs)
X.sp
X.ti -2.0i
X\\b \\b matches the empty string at the edge of a word.
X.sp
X.ti -2.0i
X\\B \\B matches the empty string if not at the edge of a word.
X.sp
X.ti -2.0i
X\\w \\w matches word-constituent characters (letters & digits).
X.sp
X.ti -2.0i
X\\W \\W matches characters that are not word-constituent.
X.RE
X.PP
XOperator precedence is (highest to lowest) ?, \(**, and +, concatenation,
Xand finally |. All other constructs are syntactically identical to
Xnormal characters. For the truly interested, the file dfa.c describes
X(and implements) the exact grammar understood by the parser.
X.SH OPTIONS
X.TP
X.BI \-A " num"
Xprint <num> lines of context after every matching line
X.TP
X.BI \-B " num"
Xprint
X.I num
Xlines of context before every matching line
X.TP
X.B \-C
Xprint 2 lines of context on each side of every match
X.TP
X.BI \- num
Xprint
X.I num
Xlines of context on each side of every match
X.TP
X.B \-V
Xprint the version number on the diagnostic output
X.TP
X.B \-b
Xprint every match preceded by its byte offset
X.TP
X.B \-c
Xprint a total count of matching lines only
X.TP
X.BI \-e " expr"
Xsearch for
X.IR expr ;
Xuseful if
X.I expr
Xbegins with \-
X.TP
X.BI \-f " file"
Xsearch for the expression contained in
X.I file
X.TP
X.B \-h
Xdon't display filenames on matches
X.TP
X.B \-i
Xignore case difference when comparing strings
X.TP
X.B \-l
Xlist files containing matches only
X.TP
X.B \-n
Xprint each match preceded by its line number
X.TP
X.B \-s
Xrun silently producing no output except error messages
X.TP
X.B \-v
Xprint only lines that contain no matches for the <expr>
X.TP
X.B \-w
Xprint only lines where the match is a complete word
X.TP
X.B \-x
Xprint only lines where the match is a whole line
X.SH "SEE ALSO"
Xemacs(1), ed(1), sh(1),
X.I "GNU Emacs Manual"
X.SH INCOMPATIBILITIES
XThe following incompatibilities with UNIX
X.I grep
Xexist:
X.PP
X.RS 0.5i
XThe context-dependent meaning of \(** is not quite the same (grep only).
X.PP
X.B \-b
Xprints a byte offset instead of a block offset.
X.PP
XThe {\fIm,n\fP} construct of System V grep is not implemented.
X.PP
X.SH BUGS
XGNU \fIe?grep\fP has been thoroughly debugged and tested by several people
Xover a period of several months; we think it's a reliable beast or we
Xwouldn't distribute it. If by some fluke of the universe you discover
Xa bug, send a detailed description (including options, regular
Xexpressions, and a copy of an input file that can reproduce it) to me,
Xmike@wheaties.ai.mit.edu.
X.PP
XThere is also a newsgroup, gnu.utils.bug, for reporting FSF utility
Xprograms' bugs and fixes; but before reporting something as a bug,
Xplease try to be sure that it really is a bug, not a misunderstanding
Xor a deliberate feature. Also, include the version number of the
Xutility program you are running in \fIevery\fR bug report that you
Xsend in. Please do not send anything but bug reports to this
Xnewsgroup.
X.PP
X.SH AVAILABILITY
X.PP
XGNU
X.I grep
Xis free; anyone may redistribute copies of
X.I grep
Xto
Xanyone under the terms stated in the
XGNU General Public License,
Xa copy of which may be found in each copy of
X.IR "GNU Emacs" .
XSee also the comment at the beginning of the source code file grep.c.
X.PP
XCopies of GNU
X.I grep
Xmay sometimes be received packaged with distributions of Unix systems,
Xbut it is never included in the scope of any license covering those
Xsystems. Such inclusion violates the terms on which distribution
Xis permitted. In fact, the primary purpose of the General Public
XLicense is to prohibit anyone from attaching any other restrictions
Xto redistribution of any of the Free Software Foundation programs.
X.SH AUTHORS
XMike Haertel wrote the deterministic regexp code and the bulk
Xof the program.
X.PP
XJames A. Woods is responsible for the hybridized search strategy
Xof using Boyer-Moore-Gosper fixed-string search as a filter
Xbefore calling the general regexp matcher.
X.PP
XArthur David Olson contributed code that finds fixed strings for
Xthe aforementioned BMG search for a large class of regexps.
X.PP
XRichard Stallman wrote the backtracking regexp matcher that is
Xused for \\fIdigit\fP backreferences, as well as the getopt that
Xis provided for 4.2BSD sites. The backtracking matcher was
Xoriginally written for GNU Emacs.
X.PP
XD. A. Gwyn wrote the C alloca emulation that is provided so
XSystem V machines can run this program. (Alloca is used only
Xby RMS' backtracking matcher, and then only rarely, so there
Xis no loss if your machine doesn't have a "real" alloca.)
X.PP
XScott Anderson and Henry Spencer designed the regression tests
Xused in the "regress" script.
X.PP
XPaul Placeway wrote the original version of this manual page.
END_OF_FILE
if test 6853 -ne `wc -c <'grep.man'`; then
echo shar: \"'grep.man'\" unpacked with wrong size!
fi
# end of 'grep.man'
fi
if test -f 'getopt.c' -a "${1}" != "-c" ; then
echo shar: Will not clobber existing file \"'getopt.c'\"
else
echo shar: Extracting \"'getopt.c'\" \(16596 characters\)
sed "s/^X//" >'getopt.c' <<'END_OF_FILE'
X/* Getopt for GNU.
X Copyright (C) 1987 Free Software Foundation, Inc.
X
X NO WARRANTY
X
X BECAUSE THIS PROGRAM IS LICENSED FREE OF CHARGE, WE PROVIDE ABSOLUTELY
XNO WARRANTY, TO THE EXTENT PERMITTED BY APPLICABLE STATE LAW. EXCEPT
XWHEN OTHERWISE STATED IN WRITING, FREE SOFTWARE FOUNDATION, INC,
XRICHARD M. STALLMAN AND/OR OTHER PARTIES PROVIDE THIS PROGRAM "AS IS"
XWITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING,
XBUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
XFITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY
XAND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE PROGRAM PROVE
XDEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR OR
XCORRECTION.
X
X IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW WILL RICHARD M.
XSTALLMAN, THE FREE SOFTWARE FOUNDATION, INC., AND/OR ANY OTHER PARTY
XWHO MAY MODIFY AND REDISTRIBUTE THIS PROGRAM AS PERMITTED BELOW, BE
XLIABLE TO YOU FOR DAMAGES, INCLUDING ANY LOST PROFITS, LOST MONIES, OR
XOTHER SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE
XUSE OR INABILITY TO USE (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR
XDATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY THIRD PARTIES OR
XA FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS) THIS
XPROGRAM, EVEN IF YOU HAVE BEEN ADVISED OF THE POSSIBILITY OF SUCH
XDAMAGES, OR FOR ANY CLAIM BY ANY OTHER PARTY.
X
X GENERAL PUBLIC LICENSE TO COPY
X
X 1. You may copy and distribute verbatim copies of this source file
Xas you receive it, in any medium, provided that you conspicuously and
Xappropriately publish on each copy a valid copyright notice "Copyright
X (C) 1987 Free Software Foundation, Inc."; and include following the
Xcopyright notice a verbatim copy of the above disclaimer of warranty
Xand of this License. You may charge a distribution fee for the
Xphysical act of transferring a copy.
X
X 2. You may modify your copy or copies of this source file or
Xany portion of it, and copy and distribute such modifications under
Xthe terms of Paragraph 1 above, provided that you also do the following:
X
X a) cause the modified files to carry prominent notices stating
X that you changed the files and the date of any change; and
X
X b) cause the whole of any work that you distribute or publish,
X that in whole or in part contains or is a derivative of this
X program or any part thereof, to be licensed at no charge to all
X third parties on terms identical to those contained in this
X License Agreement (except that you may choose to grant more
X extensive warranty protection to third parties, at your option).
X
X c) You may charge a distribution fee for the physical act of
X transferring a copy, and you may at your option offer warranty
X protection in exchange for a fee.
X
X 3. You may copy and distribute this program or any portion of it in
Xcompiled, executable or object code form under the terms of Paragraphs
X1 and 2 above provided that you do the following:
X
X a) cause each such copy to be accompanied by the
X corresponding machine-readable source code, which must
X be distributed under the terms of Paragraphs 1 and 2 above; or,
X
X b) cause each such copy to be accompanied by a
X written offer, with no time limit, to give any third party
X free (except for a nominal shipping charge) a machine readable
X copy of the corresponding source code, to be distributed
X under the terms of Paragraphs 1 and 2 above; or,
X
X c) in the case of a recipient of this program in compiled, executable
X or object code form (without the corresponding source code) you
X shall cause copies you distribute to be accompanied by a copy
X of the written offer of source code which you received along
X with the copy you received.
X
X 4. You may not copy, sublicense, distribute or transfer this program
Xexcept as expressly provided under this License Agreement. Any attempt
Xotherwise to copy, sublicense, distribute or transfer this program is void and
Xyour rights to use the program under this License agreement shall be
Xautomatically terminated. However, parties who have received computer
Xsoftware programs from you with this License Agreement will not have
Xtheir licenses terminated so long as such parties remain in full compliance.
X
X 5. If you wish to incorporate parts of this program into other free
Xprograms whose distribution conditions are different, write to the Free
XSoftware Foundation at 675 Mass Ave, Cambridge, MA 02139. We have not yet
Xworked out a simple rule that can be stated here, but we will often permit
Xthis. We will be guided by the two goals of preserving the free status of
Xall derivatives of our free software and of promoting the sharing and reuse of
Xsoftware.
X
X
XIn other words, you are welcome to use, share and improve this program.
XYou are forbidden to forbid anyone else to use, share and improve
Xwhat you give them. Help stamp out software-hoarding! */
X
X/* This version of `getopt' appears to the caller like standard Unix `getopt'
X but it behaves differently for the user, since it allows the user
X to intersperse the options with the other arguments.
X
X As `getopt' works, it permutes the elements of `argv' so that,
X when it is done, all the options precede everything else. Thus
X all application programs are extended to handle flexible argument order.
X
X Setting the environment variable _POSIX_OPTION_ORDER disables permutation.
X Then the behavior is completely standard.
X
X GNU application programs can use a third alternative mode in which
X they can distinguish the relative order of options and other arguments. */
X
X#include <stdio.h>
X
X#ifdef sparc
X#include <alloca.h>
X#endif
X#ifdef USG
X#define bcopy(s, d, l) memcpy((d), (s), (l))
X#endif
X
X/* For communication from `getopt' to the caller.
X When `getopt' finds an option that takes an argument,
X the argument value is returned here.
X Also, when `ordering' is RETURN_IN_ORDER,
X each non-option ARGV-element is returned here. */
X
Xchar *optarg = 0;
X
X/* Index in ARGV of the next element to be scanned.
X This is used for communication to and from the caller
X and for communication between successive calls to `getopt'.
X
X On entry to `getopt', zero means this is the first call; initialize.
X
X When `getopt' returns EOF, this is the index of the first of the
X non-option elements that the caller should itself scan.
X
X Otherwise, `optind' communicates from one call to the next
X how much of ARGV has been scanned so far. */
X
Xint optind = 0;
X
X/* The next char to be scanned in the option-element
X in which the last option character we returned was found.
X This allows us to pick up the scan where we left off.
X
X If this is zero, or a null string, it means resume the scan
X by advancing to the next ARGV-element. */
X
Xstatic char *nextchar;
X
X/* Callers store zero here to inhibit the error message
X for unrecognized options. */
X
Xint opterr = 1;
X
X/* Describe how to deal with options that follow non-option ARGV-elements.
X
X UNSPECIFIED means the caller did not specify anything;
X the default is then REQUIRE_ORDER if the environment variable
X _OPTIONS_FIRST is defined, PERMUTE otherwise.
X
X REQUIRE_ORDER means don't recognize them as options.
X Stop option processing when the first non-option is seen.
X This is what Unix does.
X
X PERMUTE is the default. We permute the contents of `argv' as we scan,
X so that eventually all the options are at the end. This allows options
X to be given in any order, even with programs that were not written to
X expect this.
X
X RETURN_IN_ORDER is an option available to programs that were written
X to expect options and other ARGV-elements in any order and that care about
X the ordering of the two. We describe each non-option ARGV-element
X as if it were the argument of an option with character code zero.
X Using `-' as the first character of the list of option characters
X requests this mode of operation.
X
X The special argument `--' forces an end of option-scanning regardless
X of the value of `ordering'. In the case of RETURN_IN_ORDER, only
X `--' can cause `getopt' to return EOF with `optind' != ARGC. */
X
Xstatic enum { REQUIRE_ORDER, PERMUTE, RETURN_IN_ORDER } ordering;
X
X/* Handle permutation of arguments. */
X
X/* Describe the part of ARGV that contains non-options that have
X been skipped. `first_nonopt' is the index in ARGV of the first of them;
X `last_nonopt' is the index after the last of them. */
X
Xstatic int first_nonopt;
Xstatic int last_nonopt;
X
X/* Exchange two adjacent subsequences of ARGV.
X One subsequence is elements [first_nonopt,last_nonopt)
X which contains all the non-options that have been skipped so far.
X The other is elements [last_nonopt,optind), which contains all
X the options processed since those non-options were skipped.
X
X `first_nonopt' and `last_nonopt' are relocated so that they describe
X the new indices of the non-options in ARGV after they are moved. */
X
Xstatic void
Xexchange (argv)
X char **argv;
X{
X int nonopts_size
X = (last_nonopt - first_nonopt) * sizeof (char *);
X char **temp = (char **) alloca (nonopts_size);
X
X /* Interchange the two blocks of data in argv. */
X
X bcopy (&argv[first_nonopt], temp, nonopts_size);
X bcopy (&argv[last_nonopt], &argv[first_nonopt],
X (optind - last_nonopt) * sizeof (char *));
X bcopy (temp, &argv[first_nonopt + optind - last_nonopt],
X nonopts_size);
X
X /* Update records for the slots the non-options now occupy. */
X
X first_nonopt += (optind - last_nonopt);
X last_nonopt = optind;
X}
X
X/* Scan elements of ARGV (whose length is ARGC) for option characters
X given in OPTSTRING.
X
X If an element of ARGV starts with '-', and is not exactly "-" or "--",
X then it is an option element. The characters of this element
X (aside from the initial '-') are option characters. If `getopt'
X is called repeatedly, it returns successively each of theoption characters
X from each of the option elements.
X
X If `getopt' finds another option character, it returns that character,
X updating `optind' and `nextchar' so that the next call to `getopt' can
X resume the scan with the following option character or ARGV-element.
X
X If there are no more option characters, `getopt' returns `EOF'.
X Then `optind' is the index in ARGV of the first ARGV-element
X that is not an option. (The ARGV-elements have been permuted
X so that those that are not options now come last.)
X
X OPTSTRING is a string containing the legitimate option characters.
X A colon in OPTSTRING means that the previous character is an option
X that wants an argument. The argument is taken from the rest of the
X current ARGV-element, or from the following ARGV-element,
X and returned in `optarg'.
X
X If an option character is seen that is not listed in OPTSTRING,
X return '?' after printing an error message. If you set `opterr' to
X zero, the error message is suppressed but we still return '?'.
X
X If a char in OPTSTRING is followed by a colon, that means it wants an arg,
X so the following text in the same ARGV-element, or the text of the following
X ARGV-element, is returned in `optarg. Two colons mean an option that
X wants an optional arg; if there is text in the current ARGV-element,
X it is returned in `optarg'.
X
X If OPTSTRING starts with `-', it requests a different method of handling the
X non-option ARGV-elements. See the comments about RETURN_IN_ORDER, above. */
X
Xint
Xgetopt (argc, argv, optstring)
X int argc;
X char **argv;
X char *optstring;
X{
X /* Initialize the internal data when the first call is made.
X Start processing options with ARGV-element 1 (since ARGV-element 0
X is the program name); the sequence of previously skipped
X non-option ARGV-elements is empty. */
X
X if (optind == 0)
X {
X first_nonopt = last_nonopt = optind = 1;
X
X nextchar = 0;
X
X /* Determine how to handle the ordering of options and nonoptions. */
X
X if (optstring[0] == '-')
X ordering = RETURN_IN_ORDER;
X else if (getenv ("_POSIX_OPTION_ORDER") != 0)
X ordering = REQUIRE_ORDER;
X else
X ordering = PERMUTE;
X }
X
X if (nextchar == 0 || *nextchar == 0)
X {
X if (ordering == PERMUTE)
X {
X /* If we have just processed some options following some non-options,
X exchange them so that the options come first. */
X
X if (first_nonopt != last_nonopt && last_nonopt != optind)
X exchange (argv);
X else if (last_nonopt != optind)
X first_nonopt = optind;
X
X /* Now skip any additional non-options
X and extend the range of non-options previously skipped. */
X
X while (optind < argc
X && (argv[optind][0] != '-'
X || argv[optind][1] == 0))
X optind++;
X last_nonopt = optind;
X }
X
X /* Special ARGV-element `--' means premature end of options.
X Skip it like a null option,
X then exchange with previous non-options as if it were an option,
X then skip everything else like a non-option. */
X
X if (optind != argc && !strcmp (argv[optind], "--"))
X {
X optind++;
X
X if (first_nonopt != last_nonopt && last_nonopt != optind)
X exchange (argv);
X else if (first_nonopt == last_nonopt)
X first_nonopt = optind;
X last_nonopt = argc;
X
X optind = argc;
X }
X
X /* If we have done all the ARGV-elements, stop the scan
X and back over any non-options that we skipped and permuted. */
X
X if (optind == argc)
X {
X /* Set the next-arg-index to point at the non-options
X that we previously skipped, so the caller will digest them. */
X if (first_nonopt != last_nonopt)
X optind = first_nonopt;
X return EOF;
X }
X
X /* If we have come to a non-option and did not permute it,
X either stop the scan or describe it to the caller and pass it by. */
X
X if (argv[optind][0] != '-' || argv[optind][1] == 0)
X {
X if (ordering == REQUIRE_ORDER)
X return EOF;
X optarg = argv[optind++];
X return 0;
X }
X
X /* We have found another option-ARGV-element.
X Start decoding its characters. */
X
X nextchar = argv[optind] + 1;
X }
X
X /* Look at and handle the next option-character. */
X
X {
X char c = *nextchar++;
X char *temp = (char *) index (optstring, c);
X
X /* Increment `optind' when we start to process its last character. */
X if (*nextchar == 0)
X optind++;
X
X if (temp == 0 || c == ':')
X {
X if (opterr != 0)
X {
X if (c < 040 || c >= 0177)
X fprintf (stderr, "%s: unrecognized option, character code 0%o\n",
X argv[0], c);
X else
X fprintf (stderr, "%s: unrecognized option `-%c'\n",
X argv[0], c);
X }
X return '?';
X }
X if (temp[1] == ':')
X {
X if (temp[2] == ':')
X {
X /* This is an option that accepts an argument optionally. */
X if (*nextchar != 0)
X {
X optarg = nextchar;
X optind++;
X }
X else
X optarg = 0;
X nextchar = 0;
X }
X else
X {
X /* This is an option that requires an argument. */
X if (*nextchar != 0)
X {
X optarg = nextchar;
X /* If we end this ARGV-element by taking the rest as an arg,
X we must advance to the next element now. */
X optind++;
X }
X else if (optind == argc)
X {
X if (opterr != 0)
X fprintf (stderr, "%s: no argument for `-%c' option\n",
X argv[0], c);
X optarg = 0;
X }
X else
X /* We already incremented `optind' once;
X increment it again when taking next ARGV-elt as argument. */
X optarg = argv[optind++];
X nextchar = 0;
X }
X }
X return c;
X }
X}
X
X#ifdef TEST
X
X/* Compile with -DTEST to make an executable for use in testing
X the above definition of `getopt'. */
X
Xint
Xmain (argc, argv)
X int argc;
X char **argv;
X{
X char c;
X int digit_optind = 0;
X
X while (1)
X {
X int this_option_optind = optind;
X if ((c = getopt (argc, argv, "abc:d:0123456789")) == EOF)
X break;
X
X switch (c)
X {
X case '0':
X case '1':
X case '2':
X case '3':
X case '4':
X case '5':
X case '6':
X case '7':
X case '8':
X case '9':
X if (digit_optind != 0 && digit_optind != this_option_optind)
X printf ("digits occur in two different argv-elements.\n");
X digit_optind = this_option_optind;
X printf ("option %c\n", c);
X break;
X
X case 'a':
X printf ("option a\n");
X break;
X
X case 'b':
X printf ("option b\n");
X break;
X
X case 'c':
X printf ("option c with value `%s'\n", optarg);
X break;
X
X case '?':
X break;
X
X default:
X printf ("?? getopt returned character code 0%o ??\n", c);
X }
X }
X
X if (optind < argc)
X {
X printf ("non-option ARGV-elements: ");
X while (optind < argc)
X printf ("%s ", argv[optind++]);
X printf ("\n");
X }
X
X return 0;
X}
X
X#endif /* TEST */
END_OF_FILE
if test 16596 -ne `wc -c <'getopt.c'`; then
echo shar: \"'getopt.c'\" unpacked with wrong size!
fi
# end of 'getopt.c'
fi
echo shar: End of shell archive.
exit 0
--
Please send comp.sources.unix-related mail to rsalz@uunet.uu.net.