1PCRE2GREP(1) General Commands Manual PCRE2GREP(1)
2
6 pcre2grep - a grep with Perl-compatible regular expressions.
7
9 pcre2grep [options] [long options] [pattern] [path1 path2 ...]
10
12 pcre2grep searches files for character patterns, in the same way as
14 other grep commands do, but it uses the PCRE2 regular expression li‐
15 brary to support patterns that are compatible with the regular expres‐
16 sions of Perl 5. See pcre2syntax(3) for a quick-reference summary of
17 pattern syntax, or pcre2pattern(3) for a full description of the syntax
18 and semantics of the regular expressions that PCRE2 supports.
19 Patterns, whether supplied on the command line or in a separate file,
21 are given without delimiters. For example:
22 pcre2grep Thursday /etc/motd
24 If you attempt to use delimiters (for example, by surrounding a pattern
26 with slashes, as is common in Perl scripts), they are interpreted as
27 part of the pattern. Quotes can of course be used to delimit patterns
28 on the command line because they are interpreted by the shell, and in‐
29 deed quotes are required if a pattern contains white space or shell
30 metacharacters.
31 The first argument that follows any option settings is treated as the
33 single pattern to be matched when neither -e nor -f is present. Con‐
34 versely, when one or both of these options are used to specify pat‐
35 terns, all arguments are treated as path names. At least one of -e, -f,
36 or an argument pattern must be provided.
37 If no files are specified, pcre2grep reads the standard input. The
39 standard input can also be referenced by a name consisting of a single
40 hyphen. For example:
41 pcre2grep some-pattern file1 - file3
43 Input files are searched line by line. By default, each line that
45 matches a pattern is copied to the standard output, and if there is
46 more than one file, the file name is output at the start of each line,
47 followed by a colon. However, there are options that can change how
48 pcre2grep behaves. In particular, the -M option makes it possible to
49 search for strings that span line boundaries. What defines a line
50 boundary is controlled by the -N (--newline) option.
51 The amount of memory used for buffering files that are being scanned is
53 controlled by parameters that can be set by the --buffer-size and
54 --max-buffer-size options. The first of these sets the size of buffer
55 that is obtained at the start of processing. If an input file contains
56 very long lines, a larger buffer may be needed; this is handled by au‐
57 tomatically extending the buffer, up to the limit specified by --max-
58 buffer-size. The default values for these parameters can be set when
59 pcre2grep is built; if nothing is specified, the defaults are set to
60 20KiB and 1MiB respectively. An error occurs if a line is too long and
61 the buffer can no longer be expanded.
62 The block of memory that is actually used is three times the "buffer
64 size", to allow for buffering "before" and "after" lines. If the buffer
65 size is too small, fewer than requested "before" and "after" lines may
66 be output.
67 Patterns can be no longer than 8KiB or BUFSIZ bytes, whichever is the
69 greater. BUFSIZ is defined in <stdio.h>. When there is more than one
70 pattern (specified by the use of -e and/or -f), each pattern is applied
71 to each line in the order in which they are defined, except that all
72 the -e patterns are tried before the -f patterns.
73 By default, as soon as one pattern matches a line, no further patterns
75 are considered. However, if --colour (or --color) is used to colour the
76 matching substrings, or if --only-matching, --file-offsets, or --line-
77 offsets is used to output only the part of the line that matched (ei‐
78 ther shown literally, or as an offset), scanning resumes immediately
79 following the match, so that further matches on the same line can be
80 found. If there are multiple patterns, they are all tried on the re‐
81 mainder of the line, but patterns that follow the one that matched are
82 not tried on the earlier matched part of the line.
83 This behaviour means that the order in which multiple patterns are
85 specified can affect the output when one of the above options is used.
86 This is no longer the same behaviour as GNU grep, which now manages to
87 display earlier matches for later patterns (as long as there is no
88 overlap).
89 Patterns that can match an empty string are accepted, but empty string
91 matches are never recognized. An example is the pattern "(su‐
92 per)?(man)?", in which all components are optional. This pattern finds
93 all occurrences of both "super" and "man"; the output differs from
94 matching with "super|man" when only the matching substrings are being
95 shown.
96 If the LC_ALL or LC_CTYPE environment variable is set, pcre2grep uses
98 the value to set a locale when calling the PCRE2 library. The --locale
99 option can be used to override this.
100
102 It is possible to compile pcre2grep so that it uses libz or libbz2 to
104 read compressed files whose names end in .gz or .bz2, respectively. You
105 can find out whether your pcre2grep binary has support for one or both
106 of these file types by running it with the --help option. If the appro‐
107 priate support is not present, all files are treated as plain text. The
108 standard input is always so treated. When input is from a compressed
109 .gz or .bz2 file, the --line-buffered option is ignored.
110
112 By default, a file that contains a binary zero byte within the first
114 1024 bytes is identified as a binary file, and is processed specially.
115 However, if the newline type is specified as NUL, that is, the line
116 terminator is a binary zero, the test for a binary file is not applied.
117 See the --binary-files option for a means of changing the way binary
118 files are handled.
119
121 Patterns passed from the command line are strings that are terminated
123 by a binary zero, so cannot contain internal zeros. However, patterns
124 that are read from a file via the -f option may contain binary zeros.
125
127 The order in which some of the options appear can affect the output.
129 For example, both the -H and -l options affect the printing of file
130 names. Whichever comes later in the command line will be the one that
131 takes effect. Similarly, except where noted below, if an option is
132 given twice, the later setting is used. Numerical values for options
133 may be followed by K or M, to signify multiplication by 1024 or
134 1024*1024 respectively.
135 -- This terminates the list of options. It is useful if the next
137 item on the command line starts with a hyphen but is not an
138 option. This allows for the processing of patterns and file
139 names that start with hyphens.
140 -A number, --after-context=number
142 Output up to number lines of context after each matching
143 line. Fewer lines are output if the next match or the end of
144 the file is reached, or if the processing buffer size has
145 been set too small. If file names and/or line numbers are be‐
146 ing output, a hyphen separator is used instead of a colon for
147 the context lines. A line containing "--" is output between
148 each group of lines, unless they are in fact contiguous in
149 the input file. The value of number is expected to be rela‐
150 tively small. When -c is used, -A is ignored.
151 -a, --text
153 Treat binary files as text. This is equivalent to --binary-
154 files=text.
155 --allow-lookaround-bsk
157 PCRE2 now forbids the use of \K in lookarounds by default, in
158 line with Perl. This option causes pcre2grep to set the
159 PCRE2_EXTRA_ALLOW_LOOKAROUND_BSK option, which enables this
160 somewhat dangerous usage.
161 -B number, --before-context=number
163 Output up to number lines of context before each matching
164 line. Fewer lines are output if the previous match or the
165 start of the file is within number lines, or if the process‐
166 ing buffer size has been set too small. If file names and/or
167 line numbers are being output, a hyphen separator is used in‐
168 stead of a colon for the context lines. A line containing
169 "--" is output between each group of lines, unless they are
170 in fact contiguous in the input file. The value of number is
171 expected to be relatively small. When -c is used, -B is ig‐
172 nored.
173 --binary-files=word
175 Specify how binary files are to be processed. If the word is
176 "binary" (the default), pattern matching is performed on bi‐
177 nary files, but the only output is "Binary file <name>
178 matches" when a match succeeds. If the word is "text", which
179 is equivalent to the -a or --text option, binary files are
180 processed in the same way as any other file. In this case,
181 when a match succeeds, the output may be binary garbage,
182 which can have nasty effects if sent to a terminal. If the
183 word is "without-match", which is equivalent to the -I op‐
184 tion, binary files are not processed at all; they are assumed
185 not to be of interest and are skipped without causing any
186 output or affecting the return code.
187 --buffer-size=number
189 Set the parameter that controls how much memory is obtained
190 at the start of processing for buffering files that are being
191 scanned. See also --max-buffer-size below.
192 -C number, --context=number
194 Output number lines of context both before and after each
195 matching line. This is equivalent to setting both -A and -B
196 to the same value.
197 -c, --count
199 Do not output lines from the files that are being scanned;
200 instead output the number of lines that would have been
201 shown, either because they matched, or, if -v is set, because
202 they failed to match. By default, this count is exactly the
203 same as the number of lines that would have been output, but
204 if the -M (multiline) option is used (without -v), there may
205 be more suppressed lines than the count (that is, the number
206 of matches).
207 If no lines are selected, the number zero is output. If sev‐
209 eral files are are being scanned, a count is output for each
210 of them and the -t option can be used to cause a total to be
211 output at the end. However, if the --files-with-matches op‐
212 tion is also used, only those files whose counts are greater
213 than zero are listed. When -c is used, the -A, -B, and -C op‐
214 tions are ignored.
215 --colour, --color
217 If this option is given without any data, it is equivalent to
218 "--colour=auto". If data is required, it must be given in
219 the same shell item, separated by an equals sign.
220 --colour=value, --color=value
222 This option specifies under what circumstances the parts of a
223 line that matched a pattern should be coloured in the output.
224 By default, the output is not coloured. The value (which is
225 optional, see above) may be "never", "always", or "auto". In
226 the latter case, colouring happens only if the standard out‐
227 put is connected to a terminal. More resources are used when
228 colouring is enabled, because pcre2grep has to search for all
229 possible matches in a line, not just one, in order to colour
230 them all.
231 The colour that is used can be specified by setting one of
233 the environment variables PCRE2GREP_COLOUR, PCRE2GREP_COLOR,
234 PCREGREP_COLOUR, or PCREGREP_COLOR, which are checked in that
235 order. If none of these are set, pcre2grep looks for
236 GREP_COLORS or GREP_COLOR (in that order). The value of the
237 variable should be a string of two numbers, separated by a
238 semicolon, except in the case of GREP_COLORS, which must
239 start with "ms=" or "mt=" followed by two semicolon-separated
240 colours, terminated by the end of the string or by a colon.
241 If GREP_COLORS does not start with "ms=" or "mt=" it is ig‐
242 nored, and GREP_COLOR is checked.
243 If the string obtained from one of the above variables con‐
245 tains any characters other than semicolon or digits, the set‐
246 ting is ignored and the default colour is used. The string is
247 copied directly into the control string for setting colour on
248 a terminal, so it is your responsibility to ensure that the
249 values make sense. If no relevant environment variable is
250 set, the default is "1;31", which gives red.
251 -D action, --devices=action
253 If an input path is not a regular file or a directory, "ac‐
254 tion" specifies how it is to be processed. Valid values are
255 "read" (the default) or "skip" (silently skip the path).
256 -d action, --directories=action
258 If an input path is a directory, "action" specifies how it is
259 to be processed. Valid values are "read" (the default in
260 non-Windows environments, for compatibility with GNU grep),
261 "recurse" (equivalent to the -r option), or "skip" (silently
262 skip the path, the default in Windows environments). In the
263 "read" case, directories are read as if they were ordinary
264 files. In some operating systems the effect of reading a di‐
265 rectory like this is an immediate end-of-file; in others it
266 may provoke an error.
267 --depth-limit=number
269 See --match-limit below.
270 -e pattern, --regex=pattern, --regexp=pattern
272 Specify a pattern to be matched. This option can be used mul‐
273 tiple times in order to specify several patterns. It can also
274 be used as a way of specifying a single pattern that starts
275 with a hyphen. When -e is used, no argument pattern is taken
276 from the command line; all arguments are treated as file
277 names. There is no limit to the number of patterns. They are
278 applied to each line in the order in which they are defined
279 until one matches.
280 If -f is used with -e, the command line patterns are matched
282 first, followed by the patterns from the file(s), independent
283 of the order in which these options are specified. Note that
284 multiple use of -e is not the same as a single pattern with
285 alternatives. For example, X|Y finds the first character in a
286 line that is X or Y, whereas if the two patterns are given
287 separately, with X first, pcre2grep finds X if it is present,
288 even if it follows Y in the line. It finds Y only if there is
289 no X in the line. This matters only if you are using -o or
290 --colo(u)r to show the part(s) of the line that matched.
291 --exclude=pattern
293 Files (but not directories) whose names match the pattern are
294 skipped without being processed. This applies to all files,
295 whether listed on the command line, obtained from --file-
296 list, or by scanning a directory. The pattern is a PCRE2 reg‐
297 ular expression, and is matched against the final component
298 of the file name, not the entire path. The -F, -w, and -x op‐
299 tions do not apply to this pattern. The option may be given
300 any number of times in order to specify multiple patterns. If
301 a file name matches both an --include and an --exclude pat‐
302 tern, it is excluded. There is no short form for this option.
303 --exclude-from=filename
305 Treat each non-empty line of the file as the data for an
306 --exclude option. What constitutes a newline when reading the
307 file is the operating system's default. The --newline option
308 has no effect on this option. This option may be given more
309 than once in order to specify a number of files to read.
310 --exclude-dir=pattern
312 Directories whose names match the pattern are skipped without
313 being processed, whatever the setting of the --recursive op‐
314 tion. This applies to all directories, whether listed on the
315 command line, obtained from --file-list, or by scanning a
316 parent directory. The pattern is a PCRE2 regular expression,
317 and is matched against the final component of the directory
318 name, not the entire path. The -F, -w, and -x options do not
319 apply to this pattern. The option may be given any number of
320 times in order to specify more than one pattern. If a direc‐
321 tory matches both --include-dir and --exclude-dir, it is ex‐
322 cluded. There is no short form for this option.
323 -F, --fixed-strings
325 Interpret each data-matching pattern as a list of fixed
326 strings, separated by newlines, instead of as a regular ex‐
327 pression. What constitutes a newline for this purpose is con‐
328 trolled by the --newline option. The -w (match as a word) and
329 -x (match whole line) options can be used with -F. They ap‐
330 ply to each of the fixed strings. A line is selected if any
331 of the fixed strings are found in it (subject to -w or -x, if
332 present). This option applies only to the patterns that are
333 matched against the contents of files; it does not apply to
334 patterns specified by any of the --include or --exclude op‐
335 tions.
336 -f filename, --file=filename
338 Read patterns from the file, one per line, and match them
339 against each line of input. As is the case with patterns on
340 the command line, no delimiters should be used. What consti‐
341 tutes a newline when reading the file is the operating sys‐
342 tem's default interpretation of \n. The --newline option has
343 no effect on this option. Trailing white space is removed
344 from each line, and blank lines are ignored. An empty file
345 contains no patterns and therefore matches nothing. Patterns
346 read from a file in this way may contain binary zeros, which
347 are treated as ordinary data characters. See also the com‐
348 ments about multiple patterns versus a single pattern with
349 alternatives in the description of -e above.
350 If this option is given more than once, all the specified
352 files are read. A data line is output if any of the patterns
353 match it. A file name can be given as "-" to refer to the
354 standard input. When -f is used, patterns specified on the
355 command line using -e may also be present; they are tested
356 before the file's patterns. However, no other pattern is
357 taken from the command line; all arguments are treated as the
358 names of paths to be searched.
359 --file-list=filename
361 Read a list of files and/or directories that are to be
362 scanned from the given file, one per line. What constitutes a
363 newline when reading the file is the operating system's de‐
364 fault. Trailing white space is removed from each line, and
365 blank lines are ignored. These paths are processed before any
366 that are listed on the command line. The file name can be
367 given as "-" to refer to the standard input. If --file and
368 --file-list are both specified as "-", patterns are read
369 first. This is useful only when the standard input is a ter‐
370 minal, from which further lines (the list of files) can be
371 read after an end-of-file indication. If this option is given
372 more than once, all the specified files are read.
373 --file-offsets
375 Instead of showing lines or parts of lines that match, show
376 each match as an offset from the start of the file and a
377 length, separated by a comma. In this mode, no context is
378 shown. That is, the -A, -B, and -C options are ignored. If
379 there is more than one match in a line, each of them is shown
380 separately. This option is mutually exclusive with --output,
381 --line-offsets, and --only-matching.
382 -H, --with-filename
384 Force the inclusion of the file name at the start of output
385 lines when searching a single file. By default, the file name
386 is not shown in this case. For matching lines, the file name
387 is followed by a colon; for context lines, a hyphen separator
388 is used. If a line number is also being output, it follows
389 the file name. When the -M option causes a pattern to match
390 more than one line, only the first is preceded by the file
391 name. This option overrides any previous -h, -l, or -L op‐
392 tions.
393 -h, --no-filename
395 Suppress the output file names when searching multiple files.
396 By default, file names are shown when multiple files are
397 searched. For matching lines, the file name is followed by a
398 colon; for context lines, a hyphen separator is used. If a
399 line number is also being output, it follows the file name.
400 This option overrides any previous -H, -L, or -l options.
401 --heap-limit=number
403 See --match-limit below.
404 --help Output a help message, giving brief details of the command
406 options and file type support, and then exit. Anything else
407 on the command line is ignored.
408 -I Ignore binary files. This is equivalent to --binary-
410 files=without-match.
411 -i, --ignore-case
413 Ignore upper/lower case distinctions during comparisons.
414 --include=pattern
416 If any --include patterns are specified, the only files that
417 are processed are those whose names match one of the patterns
418 and do not match an --exclude pattern. This option does not
419 affect directories, but it applies to all files, whether
420 listed on the command line, obtained from --file-list, or by
421 scanning a directory. The pattern is a PCRE2 regular expres‐
422 sion, and is matched against the final component of the file
423 name, not the entire path. The -F, -w, and -x options do not
424 apply to this pattern. The option may be given any number of
425 times. If a file name matches both an --include and an --ex‐
426 clude pattern, it is excluded. There is no short form for
427 this option.
428 --include-from=filename
430 Treat each non-empty line of the file as the data for an
431 --include option. What constitutes a newline for this purpose
432 is the operating system's default. The --newline option has
433 no effect on this option. This option may be given any number
434 of times; all the files are read.
435 --include-dir=pattern
437 If any --include-dir patterns are specified, the only direc‐
438 tories that are processed are those whose names match one of
439 the patterns and do not match an --exclude-dir pattern. This
440 applies to all directories, whether listed on the command
441 line, obtained from --file-list, or by scanning a parent di‐
442 rectory. The pattern is a PCRE2 regular expression, and is
443 matched against the final component of the directory name,
444 not the entire path. The -F, -w, and -x options do not apply
445 to this pattern. The option may be given any number of times.
446 If a directory matches both --include-dir and --exclude-dir,
447 it is excluded. There is no short form for this option.
448 -L, --files-without-match
450 Instead of outputting lines from the files, just output the
451 names of the files that do not contain any lines that would
452 have been output. Each file name is output once, on a sepa‐
453 rate line. This option overrides any previous -H, -h, or -l
454 options.
455 -l, --files-with-matches
457 Instead of outputting lines from the files, just output the
458 names of the files containing lines that would have been out‐
459 put. Each file name is output once, on a separate line.
460 Searching normally stops as soon as a matching line is found
461 in a file. However, if the -c (count) option is also used,
462 matching continues in order to obtain the correct count, and
463 those files that have at least one match are listed along
464 with their counts. Using this option with -c is a way of sup‐
465 pressing the listing of files with no matches that occurs
466 with -c on its own. This option overrides any previous -H,
467 -h, or -L options.
468 --label=name
470 This option supplies a name to be used for the standard input
471 when file names are being output. If not supplied, "(standard
472 input)" is used. There is no short form for this option.
473 --line-buffered
475 When this option is given, non-compressed input is read and
476 processed line by line, and the output is flushed after each
477 write. By default, input is read in large chunks, unless
478 pcre2grep can determine that it is reading from a terminal,
479 which is currently possible only in Unix-like environments or
480 Windows. Output to terminal is normally automatically flushed
481 by the operating system. This option can be useful when the
482 input or output is attached to a pipe and you do not want
483 pcre2grep to buffer up large amounts of data. However, its
484 use will affect performance, and the -M (multiline) option
485 ceases to work. When input is from a compressed .gz or .bz2
486 file, --line-buffered is ignored.
487 --line-offsets
489 Instead of showing lines or parts of lines that match, show
490 each match as a line number, the offset from the start of the
491 line, and a length. The line number is terminated by a colon
492 (as usual; see the -n option), and the offset and length are
493 separated by a comma. In this mode, no context is shown.
494 That is, the -A, -B, and -C options are ignored. If there is
495 more than one match in a line, each of them is shown sepa‐
496 rately. This option is mutually exclusive with --output,
497 --file-offsets, and --only-matching.
498 --locale=locale-name
500 This option specifies a locale to be used for pattern match‐
501 ing. It overrides the value in the LC_ALL or LC_CTYPE envi‐
502 ronment variables. If no locale is specified, the PCRE2 li‐
503 brary's default (usually the "C" locale) is used. There is no
504 short form for this option.
505 -M, --multiline
507 Allow patterns to match more than one line. When this option
508 is set, the PCRE2 library is called in "multiline" mode. This
509 allows a matched string to extend past the end of a line and
510 continue on one or more subsequent lines. Patterns used with
511 -M may usefully contain literal newline characters and inter‐
512 nal occurrences of ^ and $ characters. The output for a suc‐
513 cessful match may consist of more than one line. The first
514 line is the line in which the match started, and the last
515 line is the line in which the match ended. If the matched
516 string ends with a newline sequence, the output ends at the
517 end of that line. If -v is set, none of the lines in a
518 multi-line match are output. Once a match has been handled,
519 scanning restarts at the beginning of the line after the one
520 in which the match ended.
521 The newline sequence that separates multiple lines must be
523 matched as part of the pattern. For example, to find the
524 phrase "regular expression" in a file where "regular" might
525 be at the end of a line and "expression" at the start of the
526 next line, you could use this command:
527 pcre2grep -M 'regular\s+expression' <file>
529 The \s escape sequence matches any white space character, in‐
531 cluding newlines, and is followed by + so as to match trail‐
532 ing white space on the first line as well as possibly han‐
533 dling a two-character newline sequence.
534 There is a limit to the number of lines that can be matched,
536 imposed by the way that pcre2grep buffers the input file as
537 it scans it. With a sufficiently large processing buffer,
538 this should not be a problem, but the -M option does not work
539 when input is read line by line (see --line-buffered.)
540 -m number, --max-count=number
542 Stop processing after finding number matching lines, or non-
543 matching lines if -v is also set. Any trailing context lines
544 are output after the final match. In multiline mode, each
545 multiline match counts as just one line for this purpose. If
546 this limit is reached when reading the standard input from a
547 regular file, the file is left positioned just after the last
548 matching line. If -c is also set, the count that is output
549 is never greater than number. This option has no effect if
550 used with -L, -l, or -q, or when just checking for a match in
551 a binary file.
552 --match-limit=number
554 Processing some regular expression patterns may take a very
555 long time to search for all possible matching strings. Others
556 may require a very large amount of memory. There are three
557 options that set resource limits for matching.
558 The --match-limit option provides a means of limiting comput‐
560 ing resource usage when processing patterns that are not go‐
561 ing to match, but which have a very large number of possibil‐
562 ities in their search trees. The classic example is a pattern
563 that uses nested unlimited repeats. Internally, PCRE2 has a
564 counter that is incremented each time around its main pro‐
565 cessing loop. If the value set by --match-limit is reached,
566 an error occurs.
567 The --heap-limit option specifies, as a number of kibibytes
569 (units of 1024 bytes), the amount of heap memory that may be
570 used for matching. Heap memory is needed only if matching the
571 pattern requires a significant number of nested backtracking
572 points to be remembered. This parameter can be set to zero to
573 forbid the use of heap memory altogether.
574 The --depth-limit option limits the depth of nested back‐
576 tracking points, which indirectly limits the amount of memory
577 that is used. The amount of memory needed for each backtrack‐
578 ing point depends on the number of capturing parentheses in
579 the pattern, so the amount of memory that is used before this
580 limit acts varies from pattern to pattern. This limit is of
581 use only if it is set smaller than --match-limit.
582 There are no short forms for these options. The default lim‐
584 its can be set when the PCRE2 library is compiled; if they
585 are not specified, the defaults are very large and so effec‐
586 tively unlimited.
587 --max-buffer-size=number
589 This limits the expansion of the processing buffer, whose
590 initial size can be set by --buffer-size. The maximum buffer
591 size is silently forced to be no smaller than the starting
592 buffer size.
593 -N newline-type, --newline=newline-type
595 Six different conventions for indicating the ends of lines in
596 scanned files are supported. For example:
597 pcre2grep -N CRLF 'some pattern' <file>
599 The newline type may be specified in upper, lower, or mixed
601 case. If the newline type is NUL, lines are separated by bi‐
602 nary zero characters. The other types are the single-charac‐
603 ter sequences CR (carriage return) and LF (linefeed), the
604 two-character sequence CRLF, an "anycrlf" type, which recog‐
605 nizes any of the preceding three types, and an "any" type,
606 for which any Unicode line ending sequence is assumed to end
607 a line. The Unicode sequences are the three just mentioned,
608 plus VT (vertical tab, U+000B), FF (form feed, U+000C), NEL
609 (next line, U+0085), LS (line separator, U+2028), and PS
610 (paragraph separator, U+2029).
611 When the PCRE2 library is built, a default line-ending se‐
613 quence is specified. This is normally the standard sequence
614 for the operating system. Unless otherwise specified by this
615 option, pcre2grep uses the library's default.
616 This option makes it possible to use pcre2grep to scan files
618 that have come from other environments without having to mod‐
619 ify their line endings. If the data that is being scanned
620 does not agree with the convention set by this option,
621 pcre2grep may behave in strange ways. Note that this option
622 does not apply to files specified by the -f, --exclude-from,
623 or --include-from options, which are expected to use the op‐
624 erating system's standard newline sequence.
625 -n, --line-number
627 Precede each output line by its line number in the file, fol‐
628 lowed by a colon for matching lines or a hyphen for context
629 lines. If the file name is also being output, it precedes the
630 line number. When the -M option causes a pattern to match
631 more than one line, only the first is preceded by its line
632 number. This option is forced if --line-offsets is used.
633 --no-jit If the PCRE2 library is built with support for just-in-time
635 compiling (which speeds up matching), pcre2grep automatically
636 makes use of this, unless it was explicitly disabled at build
637 time. This option can be used to disable the use of JIT at
638 run time. It is provided for testing and working round prob‐
639 lems. It should never be needed in normal use.
640 -O text, --output=text
642 When there is a match, instead of outputting the line that
643 matched, output just the text specified in this option, fol‐
644 lowed by an operating-system standard newline. In this mode,
645 no context is shown. That is, the -A, -B, and -C options are
646 ignored. The --newline option has no effect on this option,
647 which is mutually exclusive with --only-matching, --file-off‐
648 sets, and --line-offsets. However, like --only-matching, if
649 there is more than one match in a line, each of them causes a
650 line of output.
651 Escape sequences starting with a dollar character may be used
653 to insert the contents of the matched part of the line and/or
654 captured substrings into the text.
655 $<digits> or ${<digits>} is replaced by the captured sub‐
657 string of the given decimal number; zero substitutes the
658 whole match. If the number is greater than the number of cap‐
659 turing substrings, or if the capture is unset, the replace‐
660 ment is empty.
661 $a is replaced by bell; $b by backspace; $e by escape; $f by
663 form feed; $n by newline; $r by carriage return; $t by tab;
664 $v by vertical tab.
665 $o<digits> or $o{<digits>} is replaced by the character whose
667 code point is the given octal number. In the first form, up
668 to three octal digits are processed. When more digits are
669 needed in Unicode mode to specify a wide character, the sec‐
670 ond form must be used.
671 $x<digits> or $x{<digits>} is replaced by the character rep‐
673 resented by the given hexadecimal number. In the first form,
674 up to two hexadecimal digits are processed. When more digits
675 are needed in Unicode mode to specify a wide character, the
676 second form must be used.
677 Any other character is substituted by itself. In particular,
679 $$ is replaced by a single dollar.
680 -o, --only-matching
682 Show only the part of the line that matched a pattern instead
683 of the whole line. In this mode, no context is shown. That
684 is, the -A, -B, and -C options are ignored. If there is more
685 than one match in a line, each of them is shown separately,
686 on a separate line of output. If -o is combined with -v (in‐
687 vert the sense of the match to find non-matching lines), no
688 output is generated, but the return code is set appropri‐
689 ately. If the matched portion of the line is empty, nothing
690 is output unless the file name or line number are being
691 printed, in which case they are shown on an otherwise empty
692 line. This option is mutually exclusive with --output,
693 --file-offsets and --line-offsets.
694 -onumber, --only-matching=number
696 Show only the part of the line that matched the capturing
697 parentheses of the given number. Up to 50 capturing parenthe‐
698 ses are supported by default. This limit can be changed via
699 the --om-capture option. A pattern may contain any number of
700 capturing parentheses, but only those whose number is within
701 the limit can be accessed by -o. An error occurs if the num‐
702 ber specified by -o is greater than the limit.
703 -o0 is the same as -o without a number. Because these options
705 can be given without an argument (see above), if an argument
706 is present, it must be given in the same shell item, for ex‐
707 ample, -o3 or --only-matching=2. The comments given for the
708 non-argument case above also apply to this option. If the
709 specified capturing parentheses do not exist in the pattern,
710 or were not set in the match, nothing is output unless the
711 file name or line number are being output.
712 If this option is given multiple times, multiple substrings
714 are output for each match, in the order the options are
715 given, and all on one line. For example, -o3 -o1 -o3 causes
716 the substrings matched by capturing parentheses 3 and 1 and
717 then 3 again to be output. By default, there is no separator
718 (but see the next but one option).
719 --om-capture=number
721 Set the number of capturing parentheses that can be accessed
722 by -o. The default is 50.
723 --om-separator=text
725 Specify a separating string for multiple occurrences of -o.
726 The default is an empty string. Separating strings are never
727 coloured.
728 -q, --quiet
730 Work quietly, that is, display nothing except error messages.
731 The exit status indicates whether or not any matches were
732 found.
733 -r, --recursive
735 If any given path is a directory, recursively scan the files
736 it contains, taking note of any --include and --exclude set‐
737 tings. By default, a directory is read as a normal file; in
738 some operating systems this gives an immediate end-of-file.
739 This option is a shorthand for setting the -d option to "re‐
740 curse".
741 --recursion-limit=number
743 This is an obsolete synonym for --depth-limit. See --match-
744 limit above for details.
745 -s, --no-messages
747 Suppress error messages about non-existent or unreadable
748 files. Such files are quietly skipped. However, the return
749 code is still 2, even if matches were found in other files.
750 -t, --total-count
752 This option is useful when scanning more than one file. If
753 used on its own, -t suppresses all output except for a grand
754 total number of matching lines (or non-matching lines if -v
755 is used) in all the files. If -t is used with -c, a grand to‐
756 tal is output except when the previous output is just one
757 line. In other words, it is not output when just one file's
758 count is listed. If file names are being output, the grand
759 total is preceded by "TOTAL:". Otherwise, it appears as just
760 another number. The -t option is ignored when used with -L
761 (list files without matches), because the grand total would
762 always be zero.
763 -u, --utf Operate in UTF-8 mode. This option is available only if PCRE2
765 has been compiled with UTF-8 support. All patterns (including
766 those for any --exclude and --include options) and all lines
767 that are scanned must be valid strings of UTF-8 characters.
768 If an invalid UTF-8 string is encountered, an error occurs.
769 -U, --utf-allow-invalid
771 As --utf, but in addition subject lines may contain invalid
772 UTF-8 code unit sequences. These can never form part of any
773 pattern match. Patterns themselves, however, must still be
774 valid UTF-8 strings. This facility allows valid UTF-8 strings
775 to be sought within arbitrary byte sequences in executable or
776 other binary files. For more details about matching in non-
777 valid UTF-8 strings, see the pcre2unicode(3) documentation.
778 -V, --version
780 Write the version numbers of pcre2grep and the PCRE2 library
781 to the standard output and then exit. Anything else on the
782 command line is ignored.
783 -v, --invert-match
785 Invert the sense of the match, so that lines which do not
786 match any of the patterns are the ones that are found. When
787 this option is set, options such as --only-matching and
788 --output, which specify parts of a match that are to be out‐
789 put, are ignored.
790 -w, --word-regex, --word-regexp
792 Force the patterns only to match "words". That is, there must
793 be a word boundary at the start and end of each matched
794 string. This is equivalent to having "\b(?:" at the start of
795 each pattern, and ")\b" at the end. This option applies only
796 to the patterns that are matched against the contents of
797 files; it does not apply to patterns specified by any of the
798 --include or --exclude options.
799 -x, --line-regex, --line-regexp
801 Force the patterns to start matching only at the beginnings
802 of lines, and in addition, require them to match entire
803 lines. In multiline mode the match may be more than one line.
804 This is equivalent to having "^(?:" at the start of each pat‐
805 tern and ")$" at the end. This option applies only to the
806 patterns that are matched against the contents of files; it
807 does not apply to patterns specified by any of the --include
808 or --exclude options.
809
811 The environment variables LC_ALL and LC_CTYPE are examined, in that or‐
813 der, for a locale. The first one that is set is used. This can be over‐
814 ridden by the --locale option. If no locale is set, the PCRE2 library's
815 default (usually the "C" locale) is used.
816
818 The -N (--newline) option allows pcre2grep to scan files with newline
820 conventions that differ from the default. This option affects only the
821 way scanned files are processed. It does not affect the interpretation
822 of files specified by the -f, --file-list, --exclude-from, or --in‐
823 clude-from options.
824 Any parts of the scanned input files that are written to the standard
826 output are copied with whatever newline sequences they have in the in‐
827 put. However, if the final line of a file is output, and it does not
828 end with a newline sequence, a newline sequence is added. If the new‐
829 line setting is CR, LF, CRLF or NUL, that line ending is output; for
830 the other settings (ANYCRLF or ANY) a single NL is used.
831 The newline setting does not affect the way in which pcre2grep writes
833 newlines in informational messages to the standard output and error
834 streams. Under Windows, the standard output is set to be binary, so
835 that "\r\n" at the ends of output lines that are copied from the input
836 is not converted to "\r\r\n" by the C I/O library. This means that any
837 messages written to the standard output must end with "\r\n". For all
838 other operating systems, and for all messages to the standard error
839 stream, "\n" is used.
840
842 Many of the short and long forms of pcre2grep's options are the same as
844 in the GNU grep program. Any long option of the form --xxx-regexp (GNU
845 terminology) is also available as --xxx-regex (PCRE2 terminology). How‐
846 ever, the --depth-limit, --file-list, --file-offsets, --heap-limit,
847 --include-dir, --line-offsets, --locale, --match-limit, -M, --multi‐
848 line, -N, --newline, --om-separator, --output, -u, --utf, -U, and
849 --utf-allow-invalid options are specific to pcre2grep, as is the use of
850 the --only-matching option with a capturing parentheses number.
851 Although most of the common options work the same way, a few are dif‐
853 ferent in pcre2grep. For example, the --include option's argument is a
854 glob for GNU grep, but a regular expression for pcre2grep. If both the
855 -c and -l options are given, GNU grep lists only file names, without
856 counts, but pcre2grep gives the counts as well.
857
859 There are four different ways in which an option with data can be spec‐
861 ified. If a short form option is used, the data may follow immedi‐
862 ately, or (with one exception) in the next command line item. For exam‐
863 ple:
864 -f/some/file
866 -f /some/file
867 The exception is the -o option, which may appear with or without data.
869 Because of this, if data is present, it must follow immediately in the
870 same item, for example -o3.
871 If a long form option is used, the data may appear in the same command
873 line item, separated by an equals character, or (with two exceptions)
874 it may appear in the next command line item. For example:
875 --file=/some/file
877 --file /some/file
878 Note, however, that if you want to supply a file name beginning with ~
880 as data in a shell command, and have the shell expand ~ to a home di‐
881 rectory, you must separate the file name from the option, because the
882 shell does not treat ~ specially unless it is at the start of an item.
883 The exceptions to the above are the --colour (or --color) and --only-
885 matching options, for which the data is optional. If one of these op‐
886 tions does have data, it must be given in the first form, using an
887 equals character. Otherwise pcre2grep will assume that it has no data.
888
890 pcre2grep has, by default, support for calling external programs or
892 scripts or echoing specific strings during matching by making use of
893 PCRE2's callout facility. However, this support can be completely or
894 partially disabled when pcre2grep is built. You can find out whether
895 your binary has support for callouts by running it with the --help op‐
896 tion. If callout support is completely disabled, all callouts in pat‐
897 terns are ignored by pcre2grep. If the facility is partially disabled,
898 calling external programs is not supported, and callouts that request
899 it are ignored.
900 A callout in a PCRE2 pattern is of the form (?C<arg>) where the argu‐
902 ment is either a number or a quoted string (see the pcre2callout docu‐
903 mentation for details). Numbered callouts are ignored by pcre2grep;
904 only callouts with string arguments are useful.
905 Echoing a specific string
907 Starting the callout string with a pipe character invokes an echoing
909 facility that avoids calling an external program or script. This facil‐
910 ity is always available, provided that callouts were not completely
911 disabled when pcre2grep was built. The rest of the callout string is
912 processed as a zero-terminated string, which means it should not con‐
913 tain any internal binary zeros. It is written to the output, having
914 first been passed through the same escape processing as text from the
915 --output (-O) option (see above). However, $0 cannot be used to insert
916 a matched substring because the match is still in progress. Instead,
917 the single character '0' is inserted. Any syntax errors in the string
918 (for example, a dollar not followed by another character) causes the
919 callout to be ignored. No terminator is added to the output string, so
920 if you want a newline, you must include it explicitly using the escape
921 $n. For example:
922 pcre2grep '(.)(..(.))(?C"|[$1] [$2] [$3]$n")' <some file>
924 Matching continues normally after the string is output. If you want to
926 see only the callout output but not any output from an actual match,
927 you should end the pattern with (*FAIL).
928 Calling external programs or scripts
930 This facility can be independently disabled when pcre2grep is built. It
932 is supported for Windows, where a call to _spawnvp() is used, for VMS,
933 where lib$spawn() is used, and for any Unix-like environment where
934 fork() and execv() are available.
935 If the callout string does not start with a pipe (vertical bar) charac‐
937 ter, it is parsed into a list of substrings separated by pipe charac‐
938 ters. The first substring must be an executable name, with the follow‐
939 ing substrings specifying arguments:
940 executable_name|arg1|arg2|...
942 Any substring (including the executable name) may contain escape se‐
944 quences started by a dollar character. These are the same as for the
945 --output (-O) option documented above, except that $0 cannot insert the
946 matched string because the match is still in progress. Instead, the
947 character '0' is inserted. If you need a literal dollar or pipe charac‐
948 ter in any substring, use $$ or $| respectively. Here is an example:
949 echo -e "abcde\n12345" | pcre2grep \
951 '(?x)(.)(..(.))
952 (?C"/bin/echo|Arg1: [$1] [$2] [$3]|Arg2: $|${1}$| ($4)")()' -
953 Output:
955 Arg1: [a] [bcd] [d] Arg2: |a| ()
957 abcde
958 Arg1: [1] [234] [4] Arg2: |1| ()
959 12345
960 The parameters for the system call that is used to run the program or
962 script are zero-terminated strings. This means that binary zero charac‐
963 ters in the callout argument will cause premature termination of their
964 substrings, and therefore should not be present. Any syntax errors in
965 the string (for example, a dollar not followed by another character)
966 causes the callout to be ignored. If running the program fails for any
967 reason (including the non-existence of the executable), a local match‐
968 ing failure occurs and the matcher backtracks in the normal way.
969
971 It is possible to supply a regular expression that takes a very long
973 time to fail to match certain lines. Such patterns normally involve
974 nested indefinite repeats, for example: (a+)*\d when matched against a
975 line of a's with no final digit. The PCRE2 matching function has a re‐
976 source limit that causes it to abort in these circumstances. If this
977 happens, pcre2grep outputs an error message and the line that caused
978 the problem to the standard error stream. If there are more than 20
979 such errors, pcre2grep gives up.
980 The --match-limit option of pcre2grep can be used to set the overall
982 resource limit. There are also other limits that affect the amount of
983 memory used during matching; see the discussion of --heap-limit and
984 --depth-limit above.
985
987 Exit status is 0 if any matches were found, 1 if no matches were found,
989 and 2 for syntax errors, overlong lines, non-existent or inaccessible
990 files (even if matches were found in other files) or too many matching
991 errors. Using the -s option to suppress error messages about inaccessi‐
992 ble files does not affect the return code.
993 When run under VMS, the return code is placed in the symbol
995 PCRE2GREP_RC because VMS does not distinguish between exit(0) and
996 exit(1).
997
999 pcre2pattern(3), pcre2syntax(3), pcre2callout(3), pcre2unicode(3).
1001
1003 Philip Hazel
1005 Retired from University Computing Service
1006 Cambridge, England.
1007
1009 Last updated: 31 August 2021
1011 Copyright (c) 1997-2021 University of Cambridge.
1012PCRE2 10.38 31 August 2021 PCRE2GREP(1)