1 #!/usr/local/bin/perl -w
4 ##############################################################################
8 ## Jeffrey Friedl (jfriedl@omron.co.jp), Dec 1994.
9 ## Copyright 19.... ah hell, just take it.
12 ## A combo of find and grep -- more or less do a 'grep' on a whole
13 ## directory tree. Fast, with lots of options. Much more powerful than
14 ## the simple "find ... | xargs grep ....". Has a full man page.
15 ## Powerfully customizable.
17 ## This file is big, but mostly comments and man page.
19 ## See man page for usage info.
20 ## Return value: 2=error, 1=nothing found, 0=something found.
23 $version = "950918.5";
26 ## Changed all 'sysread' to 'read' because Linux perl's don't seem
33 ## Added -nice (due to Lionel Cons <Lionel.Cons@cern.ch>)
34 ## Removed any leading "./" from name.
35 ## Added default flags for ~/.search, including TTY, -nice, -list, etc.
36 ## Program name now has path removed when printed in diagnostics.
37 ## Added simple tilde-expansion to -dir arg.
38 ## Added -dskip, etc. Fixed -iregex bug.
39 ## Changed -dir to be additive, adding -ddir.
40 ## Now screen out devices, pipes, and sockets.
41 ## More tidying and lots of expanding of the man page
50 if (exists $ENV{'HOME'}) {
51 $rc_file = join('/', $ENV{'HOME'}, ".search");
59 ## Make sure we've got a regex.
60 ## Don't need one if -find or -showrc was specified.
61 $!=2, die "expecting regex arguments.\n"
62 if $FIND_ONLY == 0 && $showrc == 0 && @ARGV == 0;
64 &prepare_to_search($rc_file);
66 &import_program if !defined &dodir; ## BIG key to speed.
68 ## do search while there are directories to be done.
69 &dodir(shift(@todo)) while @todo;
71 &clear_message if $VERBOSE && $STDERR_IS_TTY;
73 ###############################################################################
77 ## initialize variables that might be reset by command-line args
78 $DOREP=0; ## set true by -dorep (redo multi-hardlink files)
79 $DOREP=1 if $^O eq 'MSWin32';
80 $DO_SORT=0; ## set by -sort (sort files in a dir before checking)
81 $FIND_ONLY=0; ## set by -find (don't search files)
82 $LIST_ONLY=0; ## set true by -l (list filenames only)
83 $NEWER=0; ## set by -newer, "-mtime -###"
84 $NICE=0; ## set by -nice (print human-readable output)
85 $NOLINKS=0; ## set true by -nolinks (don't follow symlinks)
86 $OLDER=0; ## set by -older, "-mtime ###"
87 $PREPEND_FILENAME=1; ## set false by -h (don't prefix lines with filename)
88 $REPORT_LINENUM=0; ## set true by -n (show line numbers)
89 $VERBOSE=0; ## set to a value by -v, -vv, etc. (verbose messages)
90 $WHY=0; ## set true by -why, -vvv+ (report why skipped)
91 $XDEV=0; ## set true by -xdev (stay on one filesystem)
92 $all=0; ## set true by -all (don't skip many kinds of files)
93 $iflag = ''; ## set to 'i' by -i (ignore case);
94 $norc=0; ## set by -norc (don't load rc file)
95 $showrc=0; ## set by -showrc (show what happens with rc file)
96 $underlineOK=0; ## set true by -u (watch for underline stuff)
97 $words=0; ## set true by -w (match whole-words only)
98 $DELAY=0; ## inter-file delay (seconds)
99 $retval=1; ## will set to 0 if we find anything.
101 ## various elements of stat() that we might access
106 $VV_PRINT_COUNT = 50; ## with -vv, print every VV_PRINT_COUNT files, or...
107 $VV_SIZE = 1024*1024; ## ...every VV_SIZE bytes searched
108 $vv_print = $vv_size = 0; ## running totals.
110 ## set default options, in case the rc file wants them
111 $opt{'TTY'}= 1 if -t STDOUT;
113 ## want to know this for debugging message stuff
114 $STDERR_IS_TTY = -t STDERR ? 1 : 0;
115 $STDERR_SCREWS_STDOUT = ($STDERR_IS_TTY && -t STDOUT) ? 1 : 0;
117 $0 =~ s,.*/,,; ## clean up $0 for any diagnostics we'll be printing.
125 while (@ARGV && $ARGV[0] =~ m/^-/)
129 if ($arg eq '-version' || ($VERBOSE && $arg eq '-help')) {
130 print qq/Jeffrey's file search, version "$version".\n/;
131 exit(0) unless $arg eq '-help';
133 if ($arg eq '-help') {
134 print <<INLINE_LITERAL_TEXT;
135 usage: $0 [options] [-e] [PerlRegex ....]
136 OPTIONS TELLING *WHERE* TO SEARCH:
137 -dir DIR start search at the named directory (default is current dir).
138 -xdev stay on starting file system.
139 -sort sort the files in each directory before processing.
140 -nolinks don't follow symbolic links.
141 OPTIONS TELLING WHICH FILES TO EVEN CONSIDER:
142 -mtime # consider files modified > # days ago (-# for < # days old)
143 -newer FILE consider files modified more recently than FILE (also -older)
144 -name GLOB consider files whose name matches pattern (also -regex).
145 -skip GLOB opposite of -name: identifies files to not consider.
146 -path GLOB like -name, but for files whose whole path is described.
147 -dpath/-dregex/-dskip versions for selecting or pruning directories.
148 -all don't skip any files marked to be skipped by the startup file.
149 -x<SPECIAL> (see manual, and/or try -showrc).
150 -why report why a file isn't checked (also implied by -vvvv).
151 OPTIONS TELLING WHAT TO DO WITH FILES THAT WILL BE CONSIDERED:
152 -f | -find just list files (PerlRegex ignored). Default is to grep them.
153 -ff | -ffind Does a faster -find (implies -find -all -dorep)
154 OPTIONS CONTROLLING HOW THE SEARCH IS DONE (AND WHAT IS PRINTED):
155 -l | -list only list files with matches, not the lines themselves.
156 -nice | -nnice print more "human readable" output.
157 -n prefix each output line with its line number in the file.
158 -h don't prefix output lines with file name.
159 -u also look "inside" manpage-style underlined text
160 -i do case-insensitive searching.
161 -w match words only (as defined by perl's \\b).
163 -v, -vv, -vvv various levels of message verbosity.
164 -e end of options (in case a regex looks like an option).
165 -showrc show what the rc file sets, then exit.
166 -norc don't load the rc file.
167 -dorep check files with multiple hard links multiple times.
169 print "Use -v -help for more verbose help.\n" unless $VERBOSE;
170 print "This script file is also a man page.\n" unless $stripped;
171 print <<INLINE_LITERAL_TEXT if $VERBOSE;
173 If -f (or -find) given, PerlRegex is optional and ignored.
174 Otherwise, will search for files with lines matching any of the given regexes.
176 Combining things like -name and -mtime implies boolean AND.
177 However, duplicating things (such as -name '*.c' -name '*.txt') implies OR.
179 -mtime may be given floating point (i.e. 1.5 is a day and a half).
180 -iskip/-idskip/-ipath/... etc are case-insensitive versions.
182 If any letter in -newer/-older is upper case, "or equal" is
183 inserted into the test.
185 You can always find the latest version on the World Wide Web in
186 http://www.wg.omron.co.jp/~jfriedl/perl/
190 $DOREP=1, next if $arg eq '-dorep'; ## do repeats
191 $DO_SORT=1, next if $arg eq '-sort'; ## sort files
192 $NOLINKS=1, next if $arg eq '-nolinks'; ## no sym. links
193 $PREPEND_FILENAME=0, next if $arg eq '-h'; ## no filename prefix
194 $REPORT_LINENUM=1, next if $arg eq '-n'; ## show line numbers
195 $WHY=1, next if $arg eq '-why'; ## tell why skipped
196 $XDEV=1, next if $arg eq '-xdev'; ## don't leave F.S.
197 $all=1,$opt{'-all'}=1,next if $arg eq '-all'; ## don't skip *.Z, etc
198 $iflag='i', next if $arg eq '-i'; ## ignore case
199 $norc=1, next if $arg eq '-norc'; ## don't load rc file
200 $showrc=1, next if $arg eq '-showrc'; ## show rc file
201 $underlineOK=1, next if $arg eq '-u'; ## look throuh underln.
202 $words=1, next if $arg eq '-w'; ## match "words" only
203 &strip if $arg eq '-strip'; ## dump this program
204 last if $arg eq '-e';
205 $DELAY=$1, next if $arg =~ m/-delay(\d+)/;
207 $FIND_ONLY=1, next if $arg =~/^-f(ind)?$/;## do "find" only
209 $FIND_ONLY=1, $DOREP=1, $all=1,
210 next if $arg =~/^-ff(ind)?$/;## fast -find
211 $LIST_ONLY=1,$opt{'-list'}=1,
212 next if $arg =~/^-l(ist)?$/;## only list files
214 if ($arg =~ m/^-(v+)$/) { ## verbosity
215 $VERBOSE =length($1);
216 foreach $len (1..$VERBOSE) { $opt{'-'.('v' x $len)}=1 }
219 if ($arg =~ m/^-(n+)ice$/) { ## "nice" output
221 foreach $len (1..$NICE) { $opt{'-'.('n' x $len).'ice'}=1 }
225 if ($arg =~ m/^-(i?)(d?)skip$/) {
226 local($i) = $1 eq 'i';
227 local($d) = $2 eq 'd';
228 $! = 2, die qq/$0: expecting glob arg to -$arg\n/ unless @ARGV;
229 foreach (split(/\s+/, shift @ARGV)) {
242 if ($arg =~ m/^-(i?)(d?)(regex|path|name)$/) {
243 local($i) = $1 eq 'i';
244 $! = 2, die qq/$0: expecting arg to -$arg\n/ unless @ARGV;
245 foreach (split(/\s+/, shift @ARGV)) {
246 $iname{join(',', $arg, $_)}=1 if $i;
247 $name{join(',', $arg, $_)}=1;
252 if ($arg =~ m/^-d?dir$/) {
254 $! = 2, die qq/$0: expecting filename arg to -$arg\n/ unless @ARGV;
255 $start = shift(@ARGV);
256 $start =~ s#^~(/+|$)#$ENV{'HOME'}$1# if defined $ENV{'HOME'};
257 $! = 2, die qq/$0: can't find ${arg}'s "$start"\n/ unless -e $start;
258 $! = 2, die qq/$0: ${arg}'s "$start" not a directory.\n/ unless -d _;
259 undef(@todo), $opt{'-ddir'}=1 if $arg eq '-ddir';
264 if ($arg =~ m/^-(new|old)er$/i) {
265 $! = 2, die "$0: expecting filename arg to -$arg\n" unless @ARGV;
266 local($file, $time) = shift(@ARGV);
267 $! = 2, die qq/$0: can't stat -${arg}'s "$file"./
268 unless $time = (stat($file))[$STAT_MTIME];
269 local($upper) = $arg =~ tr/A-Z//;
270 if ($arg =~ m/new/i) {
271 $time++ unless $upper;
272 $NEWER = $time if $NEWER < $time;
274 $time-- unless $upper;
275 $OLDER = $time if $OLDER == 0 || $OLDER > $time;
280 if ($arg =~ m/-mtime/) {
281 $! = 2, die "$0: expecting numerical arg to -$arg\n" unless @ARGV;
282 local($days) = shift(@ARGV);
283 $! = 2, die qq/$0: inappropriate arg ($days) to $arg\n/ if $days==0;
286 local($time) = $^T + $days;
287 $NEWER = $time if $NEWER < $time;
289 local($time) = $^T - $days;
290 $OLDER = $time if $OLDER == 0 || $OLDER > $time;
295 ## special user options
296 if ($arg =~ m/^-x(.+)/) {
297 foreach (split(/[\s,]+/, $1)) { $user_opt{$_} = $opt{$_}= 1; }
301 $! = 2, die "$0: unknown arg [$arg]\n";
306 ## Given a filename glob, return a regex.
307 ## If the glob has no globbing chars (no * ? or [..]), then
308 ## prepend an effective '*' to it.
313 local(@parts) = $glob =~ m/\\.|[*?]|\[]?[^]]*]|[^[\\*?]+/g;
316 if ($_ eq '*' || $_ eq '?') {
318 $trueglob=1; ## * and ? are a real glob
319 } elsif (substr($_, 0, 1) eq '[') {
320 $trueglob=1; ## [..] is a real glob
322 s/^\\//; ## remove any leading backslash;
323 s/\W/\\$&/g; ## now quote anything dangerous;
326 unshift(@parts, '.*') unless $trueglob;
327 join('', '^', @parts, '$');
330 sub prepare_to_search
332 local($rc_file) = @_;
334 $HEADER_BYTES=0; ## Might be set nonzero in &read_rc;
335 $last_message_length = 0; ## For &message and &clear_message.
337 &read_rc($rc_file, $showrc) unless $norc;
340 $NEXT_DIR_ENTRY = $DO_SORT ? 'shift @files' : 'readdir(DIR)';
341 $WHY = 1 if $VERBOSE > 3; ## Arg -vvvv or above implies -why.
342 @todo = ('.') if @todo == 0; ## Where we'll start looking
344 ## see if any user options were specified that weren't accounted for
345 foreach $opt (keys %user_opt) {
346 next if defined $seen_opt{$opt};
347 warn "warning: -x$opt never considered.\n";
350 die "$0: multiple time constraints exclude all possible files.\n"
351 if ($NEWER && $OLDER) && ($NEWER > $OLDER);
354 ## Process any -skip/-iskip args that had been given
357 foreach $glob (keys %skip) {
358 $i = defined($iskip{$glob}) ? 'i': '';
359 push(@skip_test, '$name =~ m/'. &glob_to_regex($glob). "/$i");
362 $SKIP_TEST = join('||',@skip_test);
365 $DO_SKIP_TEST = $SKIP_TEST = 0;
369 ## Process any -dskip/-idskip args that had been given
372 foreach $glob (keys %dskip) {
373 $i = defined($idskip{$glob}) ? 'i': '';
374 push(@dskip_test, '$name =~ m/'. &glob_to_regex($glob). "/$i");
377 $DSKIP_TEST = join('||',@dskip_test);
380 $DO_DSKIP_TEST = $DSKIP_TEST = 0;
385 ## Process any -name, -path, -regex, etc. args that had been given.
389 foreach $key (keys %name) {
390 local($type, $pat) = split(/,/, $key, 2);
391 local($i) = defined($iname{$key}) ? 'i' : '';
392 if ($type =~ /regex/) {
394 $test = "\$name =~ m!^$pat\$!$i";
396 local($var) = $type eq 'name' ? '$name' : '$file';
397 $test = "$var =~ m/". &glob_to_regex($pat). "/$i";
399 if ($type =~ m/^-i?d/) {
400 push(@dname_test, $test);
402 push(@name_test, $test);
406 $GLOB_TESTS = join('||', @name_test);
410 $GLOB_TESTS = $DO_GLOB_TESTS = 0;
413 $DGLOB_TESTS = join('||', @dname_test);
416 $DGLOB_TESTS = $DO_DGLOB_TESTS = 0;
421 ## Process any 'magic' things from the startup file.
423 if (@magic_tests && $HEADER_BYTES) {
424 ## the $magic' one is for when &dodir is not inlined
425 $tests = join('||',@magic_tests);
426 $MAGIC_TESTS = " { package magic; \$val = ($tests) }";
434 ## Prepare regular expressions.
441 ## need to have $* set, but perl5 just won''t shut up about it.
450 ## Until I figure out a better way to deal with it,
451 ## We have to worry about a regex like [^xyz] when doing $LIST_ONLY.
452 ## Such a regex *will* match \n, and if I'm pulling in multiple
453 ## lines, it can allow lines to match that would otherwise not match.
455 ## Therefore, if there is a '[^' in a regex, we can NOT take a chance
456 ## an use the fast listonly.
458 $CAN_USE_FAST_LISTONLY = $LIST_ONLY;
461 local($underline_glue) = ($] >= 5) ? '(:?_\cH)?' : '(_\cH)?';
463 $regex = shift(@ARGV);
465 ## If watching for underlined things too, add another regex.
468 if ($regex =~ m/[?*+{}()\\.|^\$[]/) {
469 warn "$0: warning, can't underline-safe ``$regex''.\n";
471 $regex = join($underline_glue, split(//, $regex));
475 ## If nothing special in the regex, just use index...
476 ## is quite a bit faster.
477 if (($iflag eq '') && ($words == 0) &&
478 $regex !~ m/[?*+{}()\\.|^\$[]/)
480 push(@regex_tests, "(index(\$_, q+$regex+)>=0)");
483 $regex =~ s#[\$\@\/]\w#\\$&#;
485 if ($regex =~ m/\|/) {
486 ## could be dangerous -- see if we can wrap in parens.
487 if ($regex =~ m/\\\d/) {
488 warn "warning: -w and a | in a regex is dangerous.\n"
490 $regex = join($regex, '(', ')');
493 $regex = join($regex, '\b', '\b');
495 $CAN_USE_FAST_LISTONLY = 0 if substr($regex, "[^") >= 0;
496 push(@regex_tests, "m/$regex/$iflag$mflag");
499 ## If we're done, but still have @extra to do, get set for that.
500 if (@ARGV == 0 && @extra) {
501 @ARGV = @extra; ## now deal with the extra stuff.
502 $underlineOK = 0; ## but no more of this.
503 undef @extra; ## or this.
507 $REGEX_TEST = join('||', @regex_tests);
508 ## print STDERR $REGEX_TEST, "\n"; exit;
510 ## must be doing -find -- just give something syntactically correct.
516 ## Make sure we can read the first item(s).
518 foreach $start (@todo) {
519 $! = 2, die qq/$0: can't stat "$start"\n/
520 unless ($dev,$inode) = (stat($start))[$STAT_DEV,$STAT_INODE];
522 if (defined $dir_done{"$dev,$inode"}) {
523 ## ignore the repeat.
524 warn(qq/ignoring "$start" (same as "$dir_done{"$dev,$inode"}").\n/)
529 ## if -xdev was given, remember the device.
530 $xdev{$dev} = 1 if $XDEV;
532 ## Note that we won't want to do it again
533 $dir_done{"$dev,$inode"} = $start;
539 ## See the comment above the __END__ above the 'sub dodir' below.
544 print STDERR "$0: internal error (@_)\n";
548 ## Read from data, up to next __END__. This will be &dodir.
549 local($/) = "\n__END__";
553 $prog =~ s/\beval\b//g; ## remove any 'eval'
555 ## Inline uppercase $-variables by their current values.
557 $prog =~ s/\$([A-Z][A-Z0-9_]{2,}\b)/
558 &bad($1) if !defined ${$main::{$1}}; ${$main::{$1}};/eg;
560 $prog =~ s/\$([A-Z][A-Z0-9_]{2,}\b)/local(*VAR) = $_main{$1};
561 &bad($1) if !defined $VAR; $VAR;/eg;
564 eval $prog; ## now do it. This will define &dodir;
565 $!=2, die "$0 internal error: $@\n" if $@;
568 ###########################################################################
571 ## Read the .search file:
572 ## Blank lines and lines that are only #-comments ignored.
573 ## Newlines may be escaped to create long lines
574 ## Other lines are directives.
576 ## A directive may begin with an optional tag in the form <...>
577 ## Things inside the <...> are evaluated as with:
578 ## <(this || that) && must>
580 ## -xmust -xthis or -xmust -xthat
581 ## were specified on the command line (order doesn't matter, though)
582 ## A directive is not done if there is a tag and it's false.
583 ## Any characters but whitespace and &|()>,! may appear after an -x
584 ## (although "-xdev" is special). -xmust,this is the same as -xmust -xthis.
585 ## Something like -x~ would make <~> true, and <!~> false.
587 ## Directives are in the form:
589 ## magic : NUMBYTES : EXPR
592 ## The STRING is parsed like a Bourne shell command line, and the
593 ## options are used as if given on the command line.
594 ## No comments are allowed on 'option' lines.
596 ## # skip objects and libraries
597 ## option: -skip '.o .a'
598 ## # skip emacs *~ and *# files, unless -x~ given:
599 ## <!~> option: -skip '~ #'
602 ## EXPR can be pretty much any perl (comments allowed!).
603 ## If it evaluates to true for any particular file, it is skipped.
604 ## The only info you'll have about a file is the variable $H, which
605 ## will have at least the first NUMBYTES of the file (less if the file
606 ## is shorter than that, of course, and maybe more). You'll also have
607 ## any variables you set in previous 'magic' lines.
609 ## magic: 6 : ($x6 = substr($H, 0, 6)) eq 'GIF87a'
610 ## magic: 6 : $x6 eq 'GIF89a'
612 ## magic: 6 : (($x6 = substr($H, 0, 6)) eq 'GIF87a' ## old gif \
613 ## || $x6 eq 'GIF89a' ## new gif
614 ## (the above two sets are the same)
615 ## ## Check the first 32 bytes for "binarish" looking bytes.
616 ## ## Don't blindly dump on any high-bit set, as non-ASCII text
617 ## ## often has them set. \x80 and \xff seem to be special, though.
618 ## ## Require two in a row to not get things like perl's $^T.
619 ## ## This is known to get *.Z, *.gz, pkzip, *.elc and about any
620 ## ## executable you'll find.
621 ## magic: 32 : $H =~ m/[\x00-\x06\x10-\x1a\x1c-\x1f\x80\xff]{2}/
625 local($file, $show) = @_;
626 local($line_num, $ln, $tag) = 0;
627 local($use_default, @default) = 0;
629 { package magic; $^W= 0; } ## turn off warnings for when we run EXPR's
631 unless (open(RC, "$file")) {
633 $file = "<internal default startup file>";
634 ## no RC file -- use this default.
635 @default = split(/\n/,<<'--------INLINE_LITERAL_TEXT');
636 magic: 32 : $H =~ m/[\x00-\x06\x10-\x1a\x1c-\x1f\x80\xff]{2}/
637 option: -skip '.a .elc .gz .o .pbm .xbm .dvi'
638 option: -iskip '.com .exe .lib .pdb .tarz .zip .z .lzh .jpg .jpeg .gif .uu'
639 <!~> option: -skip '~ #'
640 --------INLINE_LITERAL_TEXT
644 ## Make an eval error pretty.
646 sub clean_eval_error {
648 s/ in file \(eval\) at line \d+,//g; ## perl4-style error
649 s/ at \(eval \d+\) line \d+,//g; ## perl5-style error
650 $_ = $` if m/\n/; ## remove all but first line
654 print "reading RC file: $file\n" if $show;
656 while (defined($_ = ($use_default ? shift(@default) : <RC>))) {
657 $ln = ++$line_num; ## note starting line num.
658 $_ .= <RC>, $line_num++ while s/\\\n?$/\n/; ## allow continuations
659 next if /^\s*(#.*)?$/; ## skip blank or comment-only lines.
662 ## look for an initial <...> tag.
663 if (s/^\s*<([^>]*)>//) {
664 ## This simple s// will make the tag ready to eval.
665 ($tag = $msg = $1) =~
667 $seen_opt{$&}=1; ## note seen option
668 "defined(\$opt{q>$&>})" ## (q>> is safe quoting here)
671 ## see if the tag is true or not, abort this line if not.
672 $dothis = (eval $tag);
673 $!=2, die "$file $ln <$msg>: $_".&clean_eval_error($@) if $@;
676 $msg =~ s/[^\s&|(!)]+/-x$&/;
677 $msg =~ s/\s*!\s*/ no /g;
678 $msg =~ s/\s*&&\s*/ and /g;
679 $msg =~ s/\s*\|\|\s*/ or /g;
680 $msg =~ s/^\s+//; $msg =~ s/\s+$//;
681 $do = $dothis ? "(doing because $msg)" :
688 if (m/^\s*option\s*:\s*/) {
689 next if $all && !$show; ## -all turns off these checks;
693 print " $do option: $_\n" if $show;
694 local($0) = "$0 ($file)"; ## for any error message.
698 ## Parse $_ as a Bourne shell line -- fill @ARGV
702 push(@ARGV, $this) if defined $this;
706 $this = '' if !defined $this;
707 $this .= $1 while s/^'([^']*)'// ||
711 die "$file $ln: error parsing $orig at $_\n" if m/^\S/;
713 push(@ARGV, $this) if defined $this;
715 die qq/$file $ln: unused arg "@ARGV".\n/ if @ARGV;
719 if (m/^\s*magic\s*:\s*(\d+)\s*:\s*/) {
720 next if $all && !$show; ## -all turns off these checks;
721 local($bytes, $check) = ($1, $');
724 $check =~ s/\n?$/\n/;
725 print " $do contents: $check";
727 ## Check to make sure the thing at least compiles.
728 eval "package magic; (\$H = '1'x \$main'bytes) && (\n$check\n)\n";
729 $! = 2, die "$file $ln: ".&clean_eval_error($@) if $@;
731 $HEADER_BYTES = $bytes if $bytes > $HEADER_BYTES;
732 push(@magic_tests, "(\n$check\n)");
735 $! = 2, die "$file $ln: unknown command\n";
742 if (!$STDERR_IS_TTY) {
743 print STDERR $_[0], "\n";
746 $thislength = length($text);
747 if ($thislength >= $last_message_length) {
748 print STDERR $text, "\r";
750 print STDERR $text, ' 'x ($last_message_length-$thislength),"\r";
752 $last_message_length = $thislength;
758 print STDERR ' ' x $last_message_length, "\r" if $last_message_length;
759 $vv_print = $vv_size = $last_message_length = 0;
763 ## Output a copy of this program with comments, extra whitespace, and
764 ## the trailing man page removed. On an ultra slow machine, such a copy
765 ## might load faster (but I can't tell any difference on my machine).
768 seek(DATA, 0, 0) || die "$0: can't reset internal pointer.\n";
770 print, next if /INLINE_LITERAL_TEXT/.../INLINE_LITERAL_TEXT/;
771 ## must mention INLINE_LITERAL_TEXT on this line!
772 s/\#\#.*|^\s+|\s+$//; ## remove cruft
773 last if $_ eq '.00;';
774 next if ($_ eq '') || ($_ eq "'di'") || ($_ eq "'ig00'");
775 s/\$stripped=0;/\$stripped=1;/;
776 s/\s\s+/ /; ## squish multiple whitespaces down to one.
783 ## Just to shut up -w. Never executed.
787 1 || &dummy || &dir_done || &bad || &message || $NEXT_DIR_ENTRY ||
788 $DELAY || $VV_SIZE || $VV_PRINT_COUNT || $STDERR_SCREWS_STDOUT ||
789 @files || @files || $magic'H || $magic'H || $xdev{''} || &clear_message;
794 ## If the following __END__ is in place, what follows will be
795 ## inlined when the program first starts up. Any $ variable name
796 ## all in upper case, specifically, any string matching
797 ## \$([A-Z][A-Z0-9_]{2,}\b
798 ## will have the true value for that variable inlined. Also, any 'eval' is
801 ## The idea is that when the whole thing is then eval'ed to define &dodir,
802 ## the perl optimizer will make all the decisions that are based upon
803 ## command-line options (such as $VERBOSE), since they'll be inlined as
806 ## Also, and here's the big win, the tests for matching the regex, and a
807 ## few others, are all inlined. Should be blinding speed here.
809 ## See the read from <DATA> above for where all this takes place.
810 ## But all-in-all, you *want* the __END__ here. Comment it out only for
817 ## Given a directory, check all "appropriate" files in it.
818 ## Shove any subdirectories into the global @todo, so they'll be done
821 ## Be careful about adding any upper-case variables, as they are subject
822 ## to being inlined. See comments above the __END__ above.
827 $dir =~ s,/+$,,; ## remove any trailing slash.
828 unless (opendir(DIR, "$dir/.")) {
829 &clear_message if $VERBOSE && $STDERR_SCREWS_STDOUT;
830 warn qq($0: can't opendir "$dir/".\n);
836 $vv_print = $vv_size = 0;
839 @files = sort readdir(DIR) if $DO_SORT;
841 while (defined($name = eval $NEXT_DIR_ENTRY))
843 next if $name eq '.' || $name eq '..'; ## never follow these.
845 ## create full relative pathname.
846 $file = $dir eq '.' ? $name : "$dir/$name";
848 ## if link and skipping them, do so.
849 if ($NOLINKS && -l $file) {
850 warn qq/skip (symlink): $file\n/ if $WHY;
854 ## skip things unless files or directories
855 unless (-f $file || -d _) {
857 $why = (-S _ && "socket") ||
859 (-b _ && "block special")||
860 (-c _ && "char special") || "somekinda special";
861 warn qq/skip ($why): $file\n/;
866 ## skip things we can't read
869 $why = (-l $file) ? "follow" : "read";
870 warn qq/skip (can't $why): $file\n/;
875 ## skip things that are empty
876 unless (-s _ || -d _) {
877 warn qq/skip (empty): $file\n/ if $WHY;
881 ## Note file device & inode. If -xdev, skip if appropriate.
882 ($dev, $inode) = (stat(_))[$STAT_DEV, $STAT_INODE];
883 if ($XDEV && defined $xdev{$dev}) {
884 warn qq/skip (other device): $file\n/ if $WHY;
889 ## special work for a directory
891 ## Do checks for directory file endings.
892 if ($DO_DSKIP_TEST && (eval $DSKIP_TEST)) {
893 warn qq/skip (-dskip): $file\n/ if $WHY;
896 ## do checks for -name/-regex/-path tests
897 if ($DO_DGLOB_TESTS && !(eval $DGLOB_TESTS)) {
898 warn qq/skip (dirname): $file\n/ if $WHY;
902 ## _never_ redo a directory
903 if (defined $dir_done{$id} and $^O ne 'MSWin32') {
904 warn qq/skip (did as "$dir_done{$id}"): $file\n/ if $WHY;
907 $dir_done{$id} = $file; ## mark it done.
908 unshift(@todo, $file); ## add to the list to do.
911 if ($WHY == 0 && $VERBOSE > 1) {
912 if ($VERBOSE>2||$vv_print++>$VV_PRINT_COUNT||($vv_size+=-s _)>$VV_SIZE){
914 $vv_print = $vv_size = 0;
918 ## do time-related tests
919 if ($NEWER || $OLDER) {
920 $_ = (stat(_))[$STAT_MTIME];
921 if ($NEWER && $_ < $NEWER) {
922 warn qq/skip (too old): $file\n/ if $WHY;
925 if ($OLDER && $_ > $OLDER) {
926 warn qq/skip (too new): $file\n/ if $WHY;
931 ## do checks for file endings
932 if ($DO_SKIP_TEST && (eval $SKIP_TEST)) {
933 warn qq/skip (-skip): $file\n/ if $WHY;
937 ## do checks for -name/-regex/-path tests
938 if ($DO_GLOB_TESTS && !(eval $GLOB_TESTS)) {
939 warn qq/skip (filename): $file\n/ if $WHY;
944 ## If we're not repeating files,
945 ## skip this one if we've done it, or note we're doing it.
947 if (defined $file_done{$id}) {
948 warn qq/skip (did as "$file_done{$id}"): $file\n/ if $WHY;
951 $file_done{$id} = $file;
954 if ($DO_MAGIC_TESTS) {
955 if (!open(FILE_IN, $file)) {
956 &clear_message if $VERBOSE && $STDERR_SCREWS_STDOUT;
957 warn qq/$0: can't open: $file\n/;
960 unless (read(FILE_IN, $magic'H, $HEADER_BYTES)) {
961 &clear_message if $VERBOSE && $STDERR_SCREWS_STDOUT;
962 warn qq/$0: can't read from "$file"\n"/;
970 warn qq/skip (magic): $file\n/ if $WHY;
973 seek(FILE_IN, 0, 0); ## reset for later <FILE_IN>
976 if ($WHY != 0 && $VERBOSE > 1) {
977 if ($VERBOSE>2||$vv_print++>$VV_PRINT_COUNT||($vv_size+=-s _)>$VV_SIZE){
979 $vv_print = $vv_size = 0;
988 &clear_message if $VERBOSE && $STDERR_SCREWS_STDOUT;
990 $retval=0; ## we've found something
991 close(FILE_IN) if $DO_MAGIC_TESTS;
994 ## if we weren't doing magic tests, file won't be open yet...
995 if (!$DO_MAGIC_TESTS && !open(FILE_IN, $file)) {
996 &clear_message if $VERBOSE && $STDERR_SCREWS_STDOUT;
997 warn qq/$0: can't open: $file\n/;
1000 if ($LIST_ONLY && $CAN_USE_FAST_LISTONLY) {
1002 ## This is rather complex, but buys us a LOT when we're just
1003 ## listing files and not the individual internal lines.
1005 local($size) = 4096; ## block-size in which to do reads
1006 local($nl); ## will point to $_'s ending newline.
1007 local($read); ## will be how many bytes read.
1008 local($_) = ''; ## Starts out empty
1009 local($hold); ## (see below)
1011 while (($read = read(FILE_IN,$_,$size,length($_)))||length($_))
1014 ## if read a full block, but no newline, need to read more.
1015 while ($read == $size && ($nl = rindex($_, "\n")) < 0) {
1016 push(@parts, $_); ## save that part
1017 $read = read(FILE_IN, $_, $size); ## keep trying
1021 ## If we had to save parts, must now combine them together.
1022 ## adjusting $nl to reflect the now-larger $_. This should
1023 ## be a lot more efficient than using any kind of .= in the
1027 local($lastlen) = length($_); #only need if $nl >= 0
1028 $_ = join('', @parts, $_);
1029 $nl = length($_) - ($lastlen - $nl) if $nl >= 0;
1033 ## If we're at the end of the file, then we can use $_ as
1034 ## is. Otherwise, we need to remove the final partial-line
1035 ## and save it so that it'll be at the beginning of the
1036 ## next read (where the rest of the line will be layed in
1037 ## right after it). $hold will be what we should save
1040 if ($read != $size || $nl < 0) {
1043 $hold = substr($_, $nl + 1);
1044 substr($_, $nl + 1) = '';
1048 ## Now have a bunch of full lines in $_. Use it.
1050 if (eval $REGEX_TEST) {
1051 &clear_message if $VERBOSE && $STDERR_SCREWS_STDOUT;
1053 $retval=0; ## we've found something
1058 ## Prepare for next read....
1062 } else { ## else not using faster block scanning.....
1064 $lines_printed = 0 if $NICE;
1067 next unless (eval $REGEX_TEST);
1070 ## We found a matching line.
1073 &clear_message if $VERBOSE && $STDERR_SCREWS_STDOUT;
1078 ## prepare to print line.
1079 if ($NICE && $lines_printed++ == 0) {
1080 print '-' x 70, "\n" if $NICE > 1;
1085 ## Print all the prelim stuff. This looks less efficient
1086 ## than it needs to be, but that's so that when the eval
1087 ## is compiled (and the tests are optimized away), the
1088 ## result will be less actual PRINTs than the more natural
1089 ## way of doing these tests....
1092 if ($REPORT_LINENUM) {
1097 } elsif ($REPORT_LINENUM && $PREPEND_FILENAME) {
1098 print "$file,:$.: ";
1099 } elsif ($PREPEND_FILENAME) {
1101 } elsif ($REPORT_LINENUM) {
1105 print "\n" unless m/\n$/;
1108 print "\n" if ($NICE > 1) && $lines_printed;
1119 'di \" finish diversion--previous line must be blank
1120 .nr nl 0-1 \" fake up transition to first page again
1121 .nr % 0 \" start at page 1
1122 .\"__________________NORMAL_MAN_PAGE_BELOW_________________
1124 .TH search 1 "Dec 17, 1994"
1126 search \- search files (a'la grep) in a whole directory tree.
1128 search [ grep-like and find-like options] [regex ....]
1131 is more or less a combo of 'find' and 'grep' (although the regular
1132 expression flavor is that of the perl being used, which is closer to
1133 egrep's than grep's).
1136 does generally the same kind of thing that
1138 find <blah blah> | xargs egrep <blah blah>
1142 more powerful and efficient (and intuitive, I think).
1144 This manual describes
1146 as of version "941227.4". You can always find the latest version at
1148 http://www.wg.omron.co.jp/~jfriedl/perl/index.html
1152 Basic use is simple:
1156 will search files in the current directory, and all sub directories, for
1157 files that have "jeff" in them. The lines will be listed with the
1158 containing file's name prepended.
1160 If you list more than one regex, such as with
1162 % search jeff Larry Randal+ 'Stoc?k' 'C.*son'
1164 then a line containing any of the regexes will be listed.
1165 This makes it effectively the same as
1167 % search 'jeff|Larry|Randal+|Stoc?k|C.*son'
1169 However, listing them separately is much more efficient (and is easier
1172 Note that in the case of these examples, the
1174 (list whole-words only) option would be useful.
1176 Normally, various kinds of files are automatically removed from consideration.
1177 If it has has a certain ending (such as ".tar", ".Z", ".o", .etc), or if
1178 the beginning of the file looks like a binary, it'll be excluded.
1179 You can control exactly how this works -- see below. One quick way to
1180 override this is to use the
1182 option, which means to consider all the files that would normally be
1183 automatically excluded.
1184 Or, if you're curious, you can use
1186 to have notes about what files are skipped (and why) printed to stderr.
1188 .SH "BASIC OVERVIEW"
1189 Normally, the search starts in the current directory, considering files in
1194 file to control ways to automatically exclude files.
1195 If you don't have this file, a default one will kick in, which automatically
1200 (among others) to exclude those kinds of files (which you probably want to
1201 skip when searching for text, as is normal).
1202 Files that look to be be binary will also be excluded.
1204 Files ending with "#" and "~" will also be excluded unless the
1210 to show what kinds of files will normally be skipped.
1211 See the section on the startup file
1216 option to indicate you want to consider all files that would otherwise be
1217 skipped by the startup file.
1219 Based upon various other flags (see "WHICH FILES TO CONSIDER" below),
1220 more files might be removed from consideration. For example
1224 will exclude files that aren't at least three days old (change the 3 to -3
1225 to exclude files that are more than three days old), while
1229 would exclude any file beginning with a dot (of course, '.' and '..' are
1230 special and always excluded).
1232 If you'd like to see what files are being excluded, and why, you can get the
1237 If a file makes it past all the checks, it is then "considered".
1238 This usually means it is greped for the regular expressions you gave
1239 on the command line.
1241 If any of the regexes match a line, the line is printed.
1244 is given, just the filename is printed. Or, if
1246 is given, a somewhat more (human-)readable output is generated.
1248 If you're searching a huge tree and want to keep informed about how
1249 the search is progressing,
1251 will print (to stderr) the current directory being searched.
1254 will also print the current file "every so often", which could be useful
1255 if a directory is huge. Using
1257 will print the update with every file.
1259 Below is the full listing of options.
1261 .SH "OPTIONS TELLING *WHERE* TO SEARCH"
1264 Start searching at the named directory instead of the current directory.
1267 arguments are given, multiple trees will be searched.
1272 except it flushes any previous
1274 directories (i.e. "-dir A -dir B -dir C" will search A, B, and C, while
1275 "-dir A -ddir B -dir C" will search only B and C. This might be of use
1276 in the startup file (see that section below).
1279 Stay on the same filesystem as the starting directory/directories.
1282 Sort the items in a directory before processing them.
1283 Normally they are processed in whatever order they happen to be read from
1287 Don't follow symbolic links. Normally they're followed.
1289 .SH "OPTIONS CONTROLLING WHICH FILES TO CONSIDER AND EXCLUDE"
1292 Only consider files that were last changed more than
1299 has '-' prepended, i.e. "-mtime -2.5" means to consider files that
1300 have been changed in the last two and a half days).
1303 Only consider files that have not changed since
1306 If there is any upper case in the "-older", "or equal" is added to the sense
1307 of the test. Therefore, "search -older ./file regex" will never consider
1308 "./file", while "search -Older ./file regex" will.
1310 If a file is a symbolic link, the time used is that of the file and not the
1318 Only consider files that match the shell filename pattern
1320 The check is only done on a file's name (use
1322 to check the whole path, and use
1324 to check directory names).
1326 Multiple specifications can be given by separating them with spaces, a'la
1330 to consider C source and header files.
1333 doesn't contain any special pattern characters, a '*' is prepended.
1334 This last example could have been given as
1338 It could also be given as
1344 -name '*.c' -name '*.h'
1351 but in this last case, you have to be sure to supply the leading '*'.
1356 except the entire path is checked against the pattern.
1359 Considers files whose names (not paths) match the given perl regex
1363 Case-insensitive version of
1367 Case-insensitive version of
1370 .BI -iregex " REGEX"
1371 Case-insensitive version of
1376 Only search down directories whose path matches the given pattern (this
1377 doesn't apply to the initial directory given by
1382 -dir /usr/man -dpath /usr/man/man*
1384 would completely skip
1385 "/usr/man/cat1", "/usr/man/cat2", etc.
1388 Skips directories whose name (not path) matches the given pattern.
1391 -dir /usr/man -dskip cat*
1393 would completely skip any directory in the tree whose name begins with "cat"
1394 (including "/usr/man/cat1", "/usr/man/cat2", etc.).
1396 .BI -dregex " REGEX"
1399 but the pattern is a full perl regex. Note that this quite different
1402 which considers only file names (not paths). This option considers
1403 full directory paths (not just names). It's much more useful this way.
1404 Sorry if it's confusing.
1407 This option exists, but is probably not very useful. It probably wants to
1408 be like the '-below' or something I mention in the "TODO" section.
1411 Case-insensitive version of
1415 Case-insensitive version of
1418 .BI -idregex " REGEX"
1419 Case-insensitive version of
1423 Ignore any 'magic' or 'option' lines in the startup file.
1424 The effect is that all files that would otherwise be automatically
1425 excluded are considered.
1428 Arguments starting with
1432 explained elsewhere) do special interaction with the
1434 startup file. Something like
1438 will turn on "flag1" and "flag2" in the startup file (and is
1439 the same as "-xflag1,flag2"). You can use this to write your own
1440 rules for what kinds of files are to be considered.
1442 For example, the internal-default startup file contains the line
1444 <!~> option: -skip '~ #'
1446 This means that if the
1455 The effect is that emacs temp and backup files are not normally
1456 considered, but you can included them with the -x~ flag.
1458 You can write your own rules to customize
1460 in powerful ways. See the STARTUP FILE section below.
1463 Print a message (to stderr) when and why a file is not considered.
1465 .SH "OPTIONS TELLING WHAT TO DO WITH FILES THAT WILL BE CONSIDERED"
1471 This option changes the basic action of
1474 Normally, if a file is considered, it is searched
1475 for the regular expressions as described earlier. However, if this option
1476 is given, the filename is printed and no searching takes place. This turns
1478 into a 'find' of some sorts.
1480 In this case, no regular expressions are needed on the command line
1481 (any that are there are silently ignored).
1483 This is not intended to be a replacement for the 'find' program,
1485 you in understanding just what files are getting past the exclusion checks.
1486 If you really want to use it as a sort of replacement for the 'find' program,
1487 you might want to use
1489 so that it doesn't waste time checking to see if the file is binary, etc
1490 (unless you really want that, of course).
1494 none of the "GREP-LIKE OPTIONS" (below) matter.
1496 As a replacement for 'find',
1498 is probably a bit slower (or in the case of GNU find, a lot slower --
1502 However, "search -ffind"
1503 might be more useful than 'find' when options such as
1505 are used (at least until 'find' gets such functionality).
1510 A faster more 'find'-like find. Does
1514 .SH "GREP-LIKE OPTIONS"
1515 These options control how a searched file is accessed,
1516 and how things are printed.
1519 Ignore letter case when matching.
1522 Consider only whole-word matches ("whole word" as defined by perl's "\\b"
1526 If the regex(es) is/are simple, try to modify them so that they'll work
1527 in manpage-like underlined text (i.e. like _^Ht_^Hh_^Hi_^Hs).
1528 This is very rudimentary at the moment.
1534 Don't print matching lines, but the names of files that contain matching
1535 lines. This will likely be *much* faster, as special optimizations are
1536 made -- particularly with large files.
1539 Pepfix each line by its line number.
1542 Not a grep-like option, but similar to
1546 will have the output be a bit more human-readable, with matching lines printed
1547 slightly indented after the filename, a'la
1551 somedir/somefile: line with foo in it
1552 somedir/somefile: some food for thought
1553 anotherdir/x: don't be a buffoon!
1563 some food for thought
1569 This option due to Lionel Cons.
1574 Prefix each file's output by a rule line, and follow with an extra blank line.
1577 Don't prepend each output line with the name of the file
1587 Print the usage information.
1590 Print the version information and quit.
1593 Set the level of message verbosity.
1595 will print a note whenever a new directory is entered.
1597 will also print a note "every so often". This can be useful to see
1598 what's happening when searching huge directories.
1600 will print a new with every file.
1608 This ends the options, and can be useful if the regex begins with '-'.
1611 Shows what is being considered in the startup file, then exits.
1614 Normally, an identical file won't be checked twice (even with multiple
1615 hard or symbolic links). If you're just trying to do a fast
1617 the bookkeeping to remember which files have been seen is not desirable,
1618 so you can eliminate the bookkeeping with this flag.
1623 starts up, it processes the directives in
1625 If no such file exists, a default
1626 internal version is used.
1628 The internal version looks like:
1631 magic: 32 : $H =~ m/[\ex00-\ex06\ex10-\ex1a\ex1c-\ex1f\ex80\exff]{2}/
1632 option: -skip '.a .COM .elc .EXE .gz .o .pbm .xbm .dvi'
1633 option: -iskip '.tarz .zip .z .lzh .jpg .jpeg .gif .uu'
1634 <!~> option: -skip '~ #'
1637 If you wish to create your own "~/.search",
1638 you might consider copying the above, and then working from there.
1640 There are two kinds of directives in a startup file: "magic" and "option".
1644 Option lines will automatically do the command-line options given.
1645 For example, the line
1649 in you startup file will turn on -v every time, without needing to type it
1650 on the command line.
1652 The text on the line after the "option:" directive is processed
1653 like the Bourne shell, so make sure to pay attention to quoting.
1655 option: -skip .exe .com
1657 will give an error (".com" by itself isn't a valid option), while
1659 option: -skip ".exe .com"
1661 will properly include it as part of -skip's argument.
1665 Magic lines are used to determine if a file should be considered a binary
1666 or not (the term "magic" refers to checking a file's magic number). These
1667 are described in more detail below.
1670 Blank lines and comments (lines beginning with '#') are allowed.
1672 If a line begins with <...>, then it's a check to see if the
1673 directive on the line should be done or not. The stuff inside the <...>
1674 can contain perl's && (and), || (or), ! (not), and parens for grouping,
1675 along with "flags" that might be indicated by the user with
1679 For example, using "-xfoo" will cause "foo" to be true inside the <...>
1680 blocks. Therefore, a line beginning with "<foo>" would be done only when
1681 "-xfoo" had been specified, while a line beginning with "<!foo>" would be
1682 done only when "-xfoo" is not specified (of course, a line without any <...>
1683 is done in either case).
1685 A realistic example might be
1689 This will cause -vv messages to be the default, but allow "-xv" to override.
1691 There are a few flags that are set automatically:
1695 true if the output is to the screen (as opposed to being redirected to a file).
1696 You can force this (as with all the other automatic flags) with -xTTY.
1699 True if -v was specified. If -vv was specified, both
1703 flags are true (and so on).
1706 True if -nice was specified. Same thing about -nnice as for -vv.
1710 true if -list (or -l) was given.
1713 true if -dir was given.
1716 Using this info, you might change the last example to
1719 <!v && !-v> option: -vv
1722 The added "&& !-v" means "and if the '-v' option not given".
1723 This will allow you to use "-v" alone on the command line, and not
1724 have this directive add the more verbose "-vv" automatically.
1727 Some other examples:
1729 <!-dir && !here> option: -dir ~/
1730 Effectively make the default directory your home directory (instead of the
1731 current directory). Using -dir or -xhere will undo this.
1733 <tex> option: -name .tex -dir ~/pub
1734 Create '-xtex' to search only "*.tex" files in your ~/pub directory tree.
1735 Actually, this could be made a bit better. If you combine '-xtex' and '-dir'
1736 on the command line, this directive will add ~/pub to the list, when you
1737 probably want to use the -dir directory only. You could do
1740 <tex> option: -name .tex
1741 <tex && !-dir> option: -dir ~/pub
1744 to will allow '-xtex' to work as before, but allow a command-line "-dir"
1745 to take precedence with respect to ~/pub.
1747 <fluff> option: -nnice -sort -i -vvv
1748 Combine a few user-friendly options into one '-xfluff' option.
1750 <man> option: -ddir /usr/man -v -w
1751 When the '-xman' option is given, search "/usr/man" for whole-words
1752 (of whatever regex or regexes are given on the command line), with -v.
1755 The lines in the startup file are executed from top to bottom, so something
1759 <both> option: -xflag1 -xflag2
1760 <flag1> option: ...whatever...
1761 <flag2> option: ...whatever...
1764 will allow '-xboth' to be the same as '-xflag1 -xflag2' (or '-xflag1,flag2'
1765 for that matter). However, if you put the "<both>" line below the others,
1766 they will not be true when encountered, so the result would be different
1767 (and probably undesired).
1769 The "magic" directives are used to determine if a file looks to be binary
1770 or not. The form of a magic line is
1772 magic: \fISIZE\fP : \fIPERLCODE\fP
1776 is the number of bytes of the file you need to check, and
1778 is the code to do the check. Within
1780 the variable $H will hold at least the first
1782 bytes of the file (unless the file is shorter than that, of course).
1783 It might hold more bytes. The perl should evaluate to true if the file
1784 should be considered a binary.
1788 magic: 6 : substr($H, 0, 6) eq 'GIF87a'
1790 to test for a GIF ("-iskip .gif" is better, but this might be useful
1791 if you have images in files without the ".gif" extension).
1793 Since the startup file is checked from top to bottom, you can be a bit
1796 magic: 6 : ($x6 = substr($H, 0, 6)) eq 'GIF87a'
1797 magic: 6 : $x6 eq 'GIF89a'
1799 You could also write the same thing as
1801 magic: 6 : (($x6 = substr($H, 0, 6)) eq 'GIF87a') || ## an old gif, or.. \e
1802 $x6 eq 'GIF89a' ## .. a new one.
1804 since newlines may be escaped.
1806 The default internal startup file includes
1808 magic: 32 : $H =~ m/[\ex00-\ex06\ex10-\ex1a\ex1c-\ex1f\ex80\exff]{2}/
1810 which checks for certain non-printable characters, and catches a large
1811 number of binary files, including most system's executables, linkable
1812 objects, compressed, tarred, and otherwise folded, spindled, and mutilated
1815 Another example might be
1817 ## an archive library
1818 magic: 17 : substr($H, 0, 17) eq "!<arch>\en__.SYMDEF"
1823 returns zero if lines (or files, if appropriate) were found,
1824 or if no work was requested (such as with
1826 Returns 1 if no lines (or files) were found.
1830 Things I'd like to add some day:
1832 + show surrounding lines (context).
1833 + highlight matched portions of lines.
1834 + add '-and', which can go between regexes to override
1835 the default logical or of the regexes.
1836 + add something like
1838 which will examine a tree and only consider files that
1839 lie in a directory deeper than one named by the pattern.
1840 + add 'warning' and 'error' directives.
1841 + add 'help' directive.
1844 If -xdev and multiple -dir arguments are given, any file in any of the
1845 target filesystems are allowed. It would be better to allow each filesystem
1846 for each separate tree.
1848 Multiple -dir args might also cause some confusing effects. Doing
1850 -dir some/dir -dir other
1852 will search "some/dir" completely, then search "other" completely. This
1853 is good. However, something like
1855 -dir some/dir -dir some/dir/more/specific
1857 will search "some/dir" completely *except for* "some/dir/more/specific",
1858 after which it will return and be searched. Not really a bug, but just sort
1861 File times (for -newer, etc.) of symbolic links are for the file, not the
1862 link. This could cause some misunderstandings.
1864 Probably more. Please let me know.
1866 Jeffrey Friedl, Omron Corp (jfriedl@omron.co.jp)
1868 http://www.wg.omron.co.jp/cgi-bin/j-e/jfriedl.html
1871 See http://www.wg.omron.co.jp/~jfriedl/perl/index.html