Author image 歌代 和正 (Kaz Utashiro)
and 3 contributors


greple - extensible grep with lexical expression and region handling


Version 8.4004


greple [-Mmodule] [ -options ] pattern [ file... ]

    pattern              'and +must -not ?alternative &function'
    -e pattern           pattern match across line boundary
    -r pattern           pattern cannot be compromised
    -v pattern           pattern not to be matched
    --le pattern         lexical expression (same as bare pattern)
    --re pattern         regular expression
    --fe pattern         fixed expression
    -f, --file file      file contains search pattern
    --select index       select indexed pattern from -f file
    -i                   ignore case
    --need=[+-]n         required positive match count
    --allow=[+-]n        acceptable negative match count
    --matchcount=n[,m]   specify required match count for each block
    -l                   list filename only
    -c                   print count of matched block only
    -n                   print line number
    -H, -h               do or do not display filenames
    -o                   print only the matching part
    -m n[,m]             max count of blocks to be shown
    -A,-B,-C [n]         after/before/both match context
    --join               delete newline in the matched part
    --joinby=string      replace newline in the matched text by string
    --nonewline          do not add newline character at block end
    --filestyle=style    how filename printed (once, separate, line)
    --linestyle=style    how line number printed (separate, line)
    --separate           set filestyle and linestyle both "separate"
    --format LABEL=...   define line number and file name format
    --glob=glob          glob target files
    --chdir=dir          change directory before search
    --readlist           get filenames from stdin
    --color=when         use terminal color (auto, always, never)
    --nocolor            same as --color=never
    --colormap=color     R, G, B, C, M, Y etc.
    --colorful           use default multiple colors
    --colorindex=flags   color index method: Ascend/Descend/Block/Random
    --ansicolor=s        ANSI color 16, 256 or 24bit
    --[no]256            same as --ansicolor 256 or 16
    --regioncolor        use different color for inside/outside regions
    --uniqcolor          use different color for unique string
    --random             use random color each time
    --face               set/unset visual effects
    -p                   paragraph mode
    --all                print whole data
    --border=pattern     specify the border pattern
    --block=pattern      specify the block of records
    --blockend=s         specify the block end mark (Default: "--\n")
    --inside=pattern     select matches inside of pattern
    --outside=pattern    select matches outside of pattern
    --include=pattern    reduce matches to the area
    --exclude=pattern    reduce matches to outside of the area
    --strict             strict mode for --inside/outside --block
    --icode=name         specify file encoding
    --ocode=name         specify output encoding
    --if,--of=filter     input/output filter command
    --pf=filter          post process filter command
    --noif               disable default input filter
    --print=func         print function
    --continue           continue after print function
    --callback=func      callback function for matched string
    --begin=func         call function before search
    --end=func           call function after search
    --prologue=func      call function before command execution
    --epilogue=func      call function after command execution
    --usage[=expand]     show this message
    --norc               skip reading startup file
    --man                display command or module manual page
    --show               display module file
    --require=file       include perl program
    --conceal=type       conceal run time errors
    --persist            continue even after encoding error
    -d flags             display info (f:file d:dir c:color m:misc s:stat)



    $ cpanm App::Greple
    $ curl -sL | perl - App::Greple



greple has almost same function as Unix command egrep(1) but search is done in a manner similar to Internet search engine. For example, next command print lines those contain all of `foo' and bar' and `baz'.

    greple 'foo bar baz' ...

Each word can appear in any order and/or any place in the string. So this command find all of following texts.

    foo bar baz
    baz bar foo
    the foo, bar and baz

If you want to use OR syntax, prepend question mark (`?') on each token, or use regular expression.

    greple 'foo bar baz ?yabba ?dabba ?doo'
    greple 'foo bar baz yabba|dabba|doo'

This command will print lines those contains all of `foo', `bar' and `baz' and one or more of `yabba', `dabba' or `doo'.

NOT operator can be specified by prefixing the token by minus sign (`-'). Next example will show lines those contain both `foo' and bar' but none of `yabba', `dabba' or `doo'.

    greple 'foo bar -yabba -dabba -doo'

This can be written as this using -e and -v option.

    greple -e foo -e bar -v yabba -v dabba -v doo
    greple -e foo -e bar -v 'yabba|dabba|doo'

If `+' is placed to positive matching pattern, that pattern is marked as required, and required match count is automatically set to the number of required patterns. So

    greple '+foo bar baz'

commands implicitly set the option --need 1, and consequently print all lines including `foo'. In other words, it makes other patterns optional. If you want to search lines which includes `foo' and either or both of `bar' and `baz', use like this:

    greple '+foo bar baz' --need 2
    greple '+foo bar baz' --need +1


Default data block greple search and print is a line. Using --paragraph (or -p in short) option, series of text separated by empty line is taken as a record block. So next command will print whole paragraph which contains the word `foo', `bar' and `baz'.

    greple -p 'foo bar baz'

Option --all takes whole file as a single block. So next command find files which contains these strings, and print the all contents.

    greple --all 'foo bar baz'

Block also can be defined as a pattern. Next command search and print mail header, ignoring mail body text.

    greple --block '\A(.+\n)+'

You can also define arbitrary complex blocks by writing script.

    greple --block '&your_original_function' ...


Using option --inside and --outside, you can specify text area where the match should be occurred. Next commands search only in mail header and body area respectively. In these cases, data block is not changed, so print lines which contains the pattern in the specified area.

    greple --inside '\A(.+\n)+' pattern

    greple --outside '\A(.+\n)+' pattern

Option --inside/--outside can be used repeatedly to enhance the area to be matched. There are similar option --include/--exclude, but they are used to trim down the area.

Those four options also takes user defined function and any complex region can be used.


greple search a given pattern across line boundaries. This is especially useful to handle Asian multi-byte text, more specifically Japanese. Japanese text can be separated by newline almost any place in the text. So the search pattern may spread out onto multiple lines.

As for ascii word list, space character in the pattern matches any kind of space including newline. Next example will search the word sequence of `foo', `bar' and 'baz', even they spread out to multiple lines.

    greple -e 'foo bar baz'

Option -e is necessary because space is taken as a token separator in the bare or --le pattern.


User can define default and original options in ~/.greplerc. Next example enables colored output always, and define new option using macro processing.

    option default --color=always

    define :re1 complex-regex-1
    define :re2 complex-regex-2
    define :re3 complex-regex-3
    option --newopt --inside :re1 --exclude :re2 --re :re3

Specific set of function and option interface can be implemented as module. Modules are invoked by -M option immediately after command name.

For example, greple does not have recursive search option, but it can be implemented by --readlist option which accept target file list from standard input. Using find module, it can be written like this:

    greple -Mfind . -type f -- pattern

Also dig module implements more complex search. It can be used as simple as this:

    greple -Mdig pattern --dig .

but this command is finally translated into following option list.

    greple -Mfind . ( -name .git -o -name .svn -o -name RCS ) -prune -o 
        -type f ! -name .* ! -name *,v ! -name *~ 
        ! -iname *.jpg ! -iname *.jpeg ! -iname *.gif ! -iname *.png 
        ! -iname *.tar ! -iname *.tbz  ! -iname *.tgz ! -iname *.pdf 
        -print -- pattern



If no specific option is given, greple takes the first argument as a search pattern specified by --le option. All of these patterns can be specified multiple times.

Command itself is written in Perl, and any kind of Perl style regular expression can be used in patterns. See perlre(1) for detail.

Note that multiple line modifier (m) is set when executed, so put (?-m) at the beginning of regex if you want to explicitly disable it.

Order of capture group in the pattern is not guaranteed. Please avoid to use direct index, and use relative or named capture group instead. For example, repeated character can be written as (\w)\g{-1} or (?<c>\w)\g{c}.


Treat the string as a collection of tokens separated by spaces. Each token is interpreted by the first character. Token start with `-' means negative pattern, `?' means alternative, and `+' does required.

Next example print lines which contains `foo' and `bar', and one or more of `yabba' and 'dabba', and none of `baz' and `doo'.

    greple --le='foo bar -baz ?yabba ?dabba -doo'

Multiple `?' preceded tokens are treated all mixed together. That means `?A|B ?C|D' is equivalent to `?A|B|C|D'. If you want to mean `(A or B) and (C or D)', use AND syntax instead: `A|B C|D'.

This is the summary of start character for --le option:

    +  Required pattern
    -  Negative match pattern
    ?  Alternative pattern
    &  Function call (see next section)

If the pattern start with ampersand (`&'), it is treated as a function, and the function is called instead of searching pattern. Function call interface is same as the one for block/region options.

If you have a definition of odd_line function in you .greplerc, which is described in this manual later, you can print odd number lines like this:

    greple -n '&odd_line' file

Required (+) and negative (-) mark can be used for function pattern.

This version experimentally support callback function for each pattern. Region list returned by function can have two extra element besides start/end position. Third element is index. Fourth element is callback function pointer which is called to produce string to be shown in command output. Callback function takes four argument (start position, end position, index, matched string) and returns replacement string.

-e pattern, --and=pattern

Specify positive match token. Next two commands are equivalent.

    greple 'foo bar baz'
    greple -e foo -e bar -e baz

First character is not interpreted, so next commands will search the pattern `-baz'.

    greple -e -baz

Space characters are treated specially by -e and -v options. They are replaced by the pattern which matches any number of white spaces including newline. So the pattern can be expand to multiple lines. Next commands search the series of word `foo', `bar' and `baz' even if they are separated by newlines.

    greple -e 'foo bar baz'
-r pattern, --must=pattern

Specify required match token. Next two commands are equivalent.

    greple '+foo bar baz'
    greple -r foo -e bar -e baz
-v pattern, --not=pattern

Specify negative match token. Because it does not affect to the bare pattern argument, you can narrow down the search result like this.

    greple foo file
    greple foo file -v bar
    greple foo file -v bar -v baz

Specify regular expression. No special treatment for space and wide characters.


Specify fixed string pattern, like fgrep.

-i, --ignore-case

Ignore case.


Option to compromise matching condition. Option --need specifies the required match count, and --allow the number of negative condition to be overlooked.

    greple --need=2 --allow=1 'foo bar baz -yabba -dabba -doo'

Above command prints the line which contains two or more from `foo', `bar' and `baz', and does not include more than one of `yabba', `dabba' or `doo'.

Using option --need=1, greple produces same result as grep command.

    grep   -e foo -e bar -e baz
    greple -e foo -e bar -e baz --need=1

When the count n is negative value, it is subtracted from default value.

If the option --need=0 is specified and no pattern was found, entire data is printed. This is true even for required pattern.

--matchcount=min[,max], --mc=min[,max]

When option --matchcount is specified, only blocks which have given match count will be shown. Count min and max can be omitted, and single number is taken as min. Next commands print lines including semicolons; 3 or more, exactly 3, and 3 or less, respectively.

    greple --matchcount=3 ';' file

    greple --matchcount=3,3 ';' file

    greple --matchcount=,3 ';' file
-f file, --file=file

Specify the file which contains search pattern. When file contains multiple lines, patterns on each lines are mixed together by OR context.

Blank line and the line starting with sharp (#) character is ignored. Two slashes (//) and following string are taken as a comment and removed with preceding spaces.

When multiple files specified, each file produces individual pattern.

If the file name is followed by `[index]` string, it is treated as specified by --select option. Next two commands are equivalent.

    greple -f pattern_file'[1,5:7]'

    greple -f pattern_file --select 1,5:7

See App::Greple::subst module.


When you want to choose specific pattern in the pattern file provided by -f option, use --select option. index is number list separated by comma (,) character and each numbers are interpreted by Getopt::EX::Numbers module. Take a look at the module document for detail.

Next command use 1st and 5,6,7th pattern in the file.

    greple -f pattern_file --select 1,5:7



List filename only.

-c, --count

Print count of matched block.

-n, --line-number

Show line number.

-h, --no-filename

Do not display filename.


Display filename always.

-o, --only-matching

Print matched string only. Newline character is printed after matched string if it does not end with newline. Use --no-newline option if you don't need extra newline.

-m n[,m], --max-count=n[,m]

Set the maximum count of blocks to be shown to n.

Actually n and m are simply passed to perl splice function as offset and length. Works like this:

    greple -m  10      # get first 10 blocks
    greple -m   0,-10  # get last 10 blocks
    greple -m   0,10   # remove first 10 blocks
    greple -m -10      # remove last 10 blocks
    greple -m  10,10   # remove 10 blocks from 10th (10-19)

This option does not affect to search performance and command exit status.

Note that grep command also has same option, but it's behavior is different when invoked to multiple files. greple produces given number of output for each files, while grep takes it as a total number of output.

-A[n], --after-context[=n]
-B[n], --before-context[=n]
-C[n], --context[=n]

Print n-blocks before/after matched string. The value n can be omitted and the default is 2. When used with --paragraph or --block option, n means number of paragraph or block.

Actually, these options expand the area of logical operation. It means

    greple -C1 'foo bar baz'

matches following text.



    greple -C1 'foo baz'

also matches this text, because matching blocks around `foo' and `bar' overlaps each other and makes single block.


Convert newline character found in matched string to empty or specified string. Using --join with -o (only-matching) option, you can collect searching sentence list in one per line form. This is sometimes useful for Japanese text processing. For example, next command prints the list of KATAKANA words, including those spread across multiple lines.

    greple -ho --join '\p{InKatakana}+(\n\p{InKatakana}+)*'

Space separated word sequence can be processed with --joinby option. Next example prints all `for *something*' pattern in pod documents within Perl script.

    greple -Mperl --pod -ioe '\bfor \w+' --joinby ' '

Since greple can handle arbitrary blocks other than normal text lines, they sometimes do not end with newline character. Option -o makes similar situation. In that case, extra newline is appended at the end of block to be shown. Option --no-newline disables this behavior.

--filestyle=line|once|separate, --fs

Default style is line, and greple prints filename at the beginning of each line. Style once prints the filename only once at the first time. Style separate prints filename in the separate line before each line or block.

--linestyle=line|separate, --ls

Default style is line, and greple prints line numbers at the beginning of each line. Style separate prints line number in the separate line before each line or block.


Shortcut for --filestyle=separate --linestyle=separate. This is convenient to use block mode search and visiting each location from supporting tool, such as Emacs.

--format LABEL=format

Define the format string of line number (LINE) and file name (FILE) to be displayed. Default is:

    --format LINE='%d:'

    --format FILE='%s:'

Format string is passed to sprintf function. Tab character can be expressed as \t.

Next example will show line numbers in five digits with tab space:

    --format LINE='%05d\t'



Get files matches to specified pattern and use them as a target files. Using --chdir and --glob makes easy to use greple for fixed common job.


Change directory before processing files. When multiple directories are specified in --chdir option, by using wildcard form or repeating option, --glob file expansion will be done for every directories.

    greple --chdir '/usr/man/man?' --glob '*.[0-9]' ...

Get filenames from standard input. Read standard input and use each line as a filename for searching. You can feed the output from other command like find(1) for greple with this option. Next example searches string from files modified within 7 days:

    find . -mtime -7 -print | greple --readlist pattern

Using find module, this can be done like:

    greple -Mfind . -mtime -7 -- pattern


--color=auto|always|never, --nocolor

Use terminal color capability to emphasize the matched text. Default is `auto': effective when STDOUT is a terminal and option -o is not given, not otherwise. Option value `always' and `never' will work as expected.

Option --nocolor is alias for --color=never.


Specify color map. Because this option is mostly implemented by Getopt::EX::Colormap module, consult its document for detail and up-to-date specification.

Color specification is combination of single uppercase character representing basic colors, and (usually brighter) alternative colors in lowercase:

    R  r   Red
    G  g   Green
    B  b   Blue
    C  c   Cyan
    M  m   Magenta
    Y  y   Yellow
    K  k   Black
    W  w   White

or RGB value and 24 grey levels if using ANSI 256 color terminal:

    (255,255,255)      : 24bit decimal RGB colors
    #000000 .. #FFFFFF : 24bit hex RGB colors
    #000    .. #FFF    : 12bit hex RGB 4096 colors
    000 .. 555         : 6x6x6 RGB 216 colors
    L00 .. L25         : Black (L00), 24 grey levels, White (L25)

    Beginning # can be omitted in 24bit RGB notation.

    When values are all same in 24bit or 12bit RGB, it is converted to 24 grey level, otherwise 6x6x6 216 color.

or color names enclosed by angle bracket:

    <red> <blue> <green> <cyan> <magenta> <yellow>
    <aliceblue> <honeydue> <hotpink> <mooccasin>

with other special effects:

    N    None
    Z  0 Zero (reset)
    D  1 Double-struck (boldface)
    P  2 Pale (dark)
    I  3 Italic
    U  4 Underline
    F  5 Flash (blink: slow)
    Q  6 Quick (blink: rapid)
    S  7 Stand-out (reverse video)
    V  8 Vanish (concealed)
    X  9 Crossed out
    E    Erase Line

    ;    No effect
    /    Toggle foreground/background
    ^    Reset to foreground

If the spec includes /, left side is considered as foreground color and right side as background. If multiple colors are given in same spec, all indicators are produced in the order of their presence. As a result, the last one takes effect.

Effect characters are case insensitive, and can be found anywhere and in any order in color spec string. Character ; does nothing and can be used just for readability, like SD;K/544.


    RGB  6x6x6    12bit      24bit           color name
    ===  =======  =========  =============  ==================
    B    005      #00F       (0,0,255)      <blue>
     /M     /505      /#F0F   /(255,0,255)  /<magenta>
    K/W  000/555  #000/#FFF  000000/FFFFFF  <black>/<white>
    R/G  500/050  #F00/#0F0  FF0000/00FF00  <red>/<green>
    W/w  L03/L20  #333/#ccc  303030/c6c6c6  <dimgrey>/<lightgrey>

Multiple colors can be specified separating by white space or comma, or by repeating options. Those colors will be applied for each pattern keywords. Next command will show word `foo' in red, `bar' in green and `baz' in blue.

    greple --colormap='R G B' 'foo bar baz'

    greple --cm R -e foo --cm G -e bar --cm B -e baz

Coloring capability is implemented in Getopt::EX::Colormap module.


Another form of colormap option to specify the color for fields:

    FILE      File name
    LINE      Line number
    TEXT      Unmatched normal text
    BLOCKEND  Block end mark

In current release, BLOCKEND mark is colored with E effect recently implemented in Getopt::EX module, which allows to fill up the line with background color. This effect uses irregular escape sequence, and you may need to define LESSANSIENDCHARS environment as "mK" to see the result with less command.


You can also set the name of perl subroutine name or definition to be called handling matched words. Target word is passed as variable $_, and the return value of the subroutine will be displayed.

Next command convert all words in C comment to upper case.

    greple --all '/\*(?s:.*?)\*/' --cm 'sub{uc}'

You can quote matched string instead of coloring (this emulates deprecated option --quote):

    greple --cm 'sub{"<".$_.">"}' ...

It is possible to use this definition with field names. Next example print line numbers in seven digits.

    greple -n --cm 'LINE=sub{s/(\d+)/sprintf("%07d",$1)/e;$_}'

Experimentally, function can be combined with other normal color specifications. Also the form &func; can be repeated.

    greple --cm 'BF/544;sub{uc}'

    greple --cm 'R;&func1;&func2;&func3'

When color for 'TEXT' field is specified, whole text including matched part is passed to the function, exceptionally. It is not recommended to use user defined function for 'TEXT' field.


Shortcut for --colormap='RD GD BD CD MD YD' in ANSI 16 colors mode, and --colormap='D/544 D/454 D/445 D/455 D/454 D/554' and other combination of 3, 4, 5 for 256 colors mode. Enabled by default.

When single pattern is specified, first color in colormap is used for the pattern. If multiple patterns and multiple colors are specified, each patterns are colored with corresponding colors cyclically.

Option --regioncolor, --uniqcolor and --colorindex change this behavior.

--colorindex=spec, --ci=spec

Specify color index method by combination of spec characters. Meaningful combinations are A, D, AB, DB and R.

A (Ascending)

Apply different color sequentially according to the order of appearance.

D (Descending)

Apply different color sequentially according to the reversed order of appearance.

B (Block)

Reset sequential index on every block.

R (Random)

Apply random color.

N (Normal)

Reset to normal behavior. Because the last option takes effect, --ci=N can be used to reset the behavior set by previous options.


Shortcut for --colorindex=R.


If set as 16, use ANSI 16 colors as a default color set, otherwise ANSI 256 colors. When set as 24bit, 6 hex digits notation produces 24bit color sequence. Default is 256.


Shortcut for --ansicolor=256 or 16.

--[no]regioncolor, --[no]rc

Use different colors for each --inside/outside regions.

Disabled by default, but automatically enabled when only single search pattern is specified. Use --no-regioncolor to cancel automatic action.

--[no]uniqcolor, --[no]uc

Use different colors for different string matched. Disabled by default.

Next example prints all words start by `color' and display them all in different colors.

    greple --uniqcolor 'colou?r\w*'

When used with option -i, color is selected in case-insensitive fashion. If you want case-insensitive match and case-sensitive color selection, indicate insensitiveness in the pattern rather than command option (e.g. (?i)pattern).


Set or unset specified effect for all color specs. Use `+' (optional) to set, and `-' to unset. Effect is a single character expressing: S (Stand-out), U (Underline), D (Double-struck), F (Flash) or E (Erase Line).

Next example remove D (double-struck) effect.

    greple --face -D

Multiple effects can be set/unset at once.

    greple --face SF-D

Use `/' to set effect to background. Only `E' makes sense to use in background, though.

    greple --face /E


-p, --paragraph

Print a paragraph which contains the pattern. Each paragraph is delimited by two or more successive newlines by default. Be aware that an empty line is not a paragraph delimiter if which contains space characters. Example:

    greple -np 'setuid script' /usr/man/catl/perl.l

    greple -pe '^struct sockaddr' /usr/include/sys/socket.h

It changes the unit of context specified by -A, -B, -C options. Space gap between paragraphs are also treated as block unit. Thus, option -pC2 will print with previous and next paragraph, while -pC1 will print with just surrounding spaces.

You can create original paragraph pattern by --border option.


Treat entire file contents as a single block. This is almost identical to following command.

    greple --block='(?s).*'

Specify record block border pattern. Default block is a single line and use /^/m as a pattern. Paragraph mode uses /(?:\A|\R)\K\R+/, which means continuous newlines at the beginning of text or following another newline (\R means more general linebreaks including \r\n; consult perlrebackslash for detail).

Next command treat the data as a series of 10-line unit.

    greple -n --border='(.*\n){1,10}'

Contrary to the next --block option, --border never produce disjoint records.


Specify the record block to display. Default block is a single line.

When blocks are not continuous and there are gaps between them, the match occurred outside blocks are ignored.

If multiple block options are supplied, overlapping blocks are merged into single block.

Please be aware that this option is sometimes quite time consuming, because it finds all blocks before processing.


Change the end mark displayed after -pABC or --block options. Default value is "--\n".



Option --inside and --outside limit the text area to be matched. For simple example, if you want to find string `and' not in the word `command', it can be done like this.

    greple --outside=command and

The block can be larger and expand to multiple lines. Next command searches from C source, excluding comment part.

    greple --outside '(?s)/\*.*?\*/'

Next command searches only from POD part of the perl script.

    greple --inside='(?s)^=.*?(^=cut|\Z)'

When multiple inside and outside regions are specified, those regions are mixed up in union way.

In multiple color environment, and if single keyword is specified, matches in each --inside/outside regions are printed in different colors. Forcing this operation with multiple keywords, use --regioncolor option.


If the pattern name begins by ampersand (&) character, it is treated as a name of subroutine which returns a list of blocks. Using this option, user can use arbitrary function to determine from what part of the text they want to search. User defined function can be defined in .greplerc file or by module option.


--include/exclude option behave exactly same as --inside/outside when used alone.

When used in combination, --include/exclude are mixed in AND manner, while --inside/outside are in OR.

Thus, in the next example, first line prints all matches, and second does none.

    greple --inside PATTERN --outside PATTERN

    greple --include PATTERN --exclude PATTERN

You can make up desired matches using --inside/outside option, then remove unnecessary part by --include/exclude


Limit the match area strictly.

By default, --block, --inside/outside, --include/exclude option allows partial match within the specified area. For instance,

    greple --inside and command

matches pattern command because the part of matched string is included in specified inside-area. Partial match fails when option --strict provided, and longer string never matches within shorter area.

Interestingly enough, above example

    greple --include PATTERN --exclude PATTERN

produces output, as a matter of fact. Think of the situation searching, say, ' PATTERN ' with this condition. Matched area includes surrounding spaces, and satisfies both conditions partially. This match does not occur when option --strict is given, either.



Target file is assumed to be encoded in utf8 by default. Use this option to set specific encoding. When handling Japanese text, you may choose from 7bit-jis (jis), euc-jp or shiftjis (sjis). Multiple code can be supplied using multiple option or combined code names with space or comma, then file encoding is guessed from those code sets. Use encoding name `guess' for automatic recognition from default code list which is euc-jp and 7bit-jis. Following commands are all equivalent.

    greple --icode=guess ...
    greple --icode=euc-jp,7bit-jis ...
    greple --icode=euc-jp --icode=7bit-jis ...

Default code set are always included suspect code list. If you have just one code adding to suspect list, put + mark before the code name. Next example does automatic code detection from euc-kr, ascii, utf8 and UTF-16/32.

    greple --icode=+euc-kr ...

If the string "binary" is given as encoding name, no character encoding is expected and all files are processed as binary data.


Specify output code. Default is utf8.


--if=filter, --if=EXP:filter

You can specify filter command which is applied to each files before search. If only one filter command is specified, it is applied to all files. If filter information include colon, first field will be perl expression to check the filename saved in variable $_. If it successes, next filter command is pushed.

    greple --if=rev perg
    greple --if='/\.tar$/:tar tvf -'

If the command doesn't accept standard input as processing data, you may be able to use special device:

    greple --if='nm /dev/stdin' crypt /usr/lib/lib*

Filters for compressed and gzipped file is set by default unless --noif option is given. Default action is like this:

    greple --if='s/\.Z$//:zcat' --if='s/\.g?z$//:gunzip -c'

File with .gpg suffix is filtered by gpg command. In that case, pass-phrase is asked for each file. If you want to input pass-phrase only once to find from multiple files, use -Mpgp module.

If the filter start with &, perl subroutine is called instead of external command. You can define the subroutine in .greplerc or modules. Greple simply call the subroutine, so it should be responsible for process control.


Disable default input filter. Which means compressed files will not be decompressed automatically.


Specify output filter which process the output of greple command. Filter command can be specified in multiple times, and they are invoked for each file to be processed. So next command reset the line number for each files.

    greple --of 'cat -n' string file1 file2 ...

If the filter start with &, perl subroutine is called instead of external command. You can define the subroutine in .greplerc or modules.

Output filter command is executed only when matched string exists to avoid invoking many unnecessary processes. No effect for option -l and -c.


Similar to --of filter but invoked just once and takes care of entire output from greple command.



Specify user defined function executed before data print. Text to be printed is replaced by the result of the function. Arbitrary function can be defined in .greplerc file or module. Matched data is placed in variable $_. Filename is passed by &FILELABEL key, as described later.

It is possible to use multiple --print options. In that case, second function will get the result of the first function. The command will print the final result of the last function.


When --print option is given, greple will immediately print the result returned from print function and finish the cycle. Option --continue forces to continue normal printing process after print function called. So please be sure that all data being consistent.


Callback function is called before printing every matched pattern with four labeled parameters: start, end, index and match, which corresponds to start and end position in the text, pattern index, and the matched string. Matched string in the text is replaced by returned string from the function.


Option --begin specify the function executed at the beginning of each file processing. This function have to be called from main package. So if you define the function in the module package, use the full package name or export properly.


Option --end is almost same as --begin, except that the function is called after the file processing.


Option --prologue and --epilogue specify functions called before and after processing. During the execution, file is not opened and therefore, file name is not given to those functions.


Function can be given with module option, following module name. In this form, the function will be called with module package name. So you don't have to export it. Because it is called only once at the beginning of command execution, before starting file processing, FILELABEL parameter is not given exceptionally.

For these run-time functions, optional argument list can be set in the form of key or key=value, connected by comma. These arguments will be passed to the function in key => value list. Sole key will have the value one. Also processing file name is passed with the key of FILELABEL constant. As a result, the option in the next form:

    --begin function(key1,key2=val2)
    --begin function=key1,key2=val2

will be transformed into following function call:

    function(&FILELABEL => "filename", key1 => 1, key2 => "val2")

As described earlier, FILELABEL parameter is not given to the function specified with module option. So


simply becomes:

    function(key1 => 1, key2 => "val2")

The function can be defined in .greplerc or modules. Assign the arguments into hash, then you can access argument list as member of the hash. It's safe to delete FILELABEL key if you expect random parameter is given. Content of the target file can be accessed by $_. Ampersand (&) is required to avoid the hash key is interpreted as a bare word.

    sub function {
        my %arg = @_;
        my $filename = delete $arg{&FILELABEL};
        $arg{key1};             # 1
        $arg{key2};             # "val2"
        $_;                     # contents



Greple print usage and exit with option --usage, or no valid parameter is not specified. In this case, module option is displayed with help information if available. If you want to see how they are expanded, supply something not empty to --usage option, like:

    greple -Mmodule --usage=expand
--man, --doc

Show manual page. Display module's manual page when used with -M option.

--show, --less

Show module file contents. Use with -M option.


Show module file path. Use with -M option.


Do not read startup file: ~/.greplerc.


Include arbitrary perl program.

--conceal type=val

Conceal runtime errors. Repeatable. Types are:


(Default 1) Errors occurred during file read. Mainly unicode related errors when reading binary or ambiguous text file.


(Default 0) File skip warnings produced when fatal error was occurred during file read. Occurs when reading binary files with automatic character code recognition.


Set same value for all types.


As greple tries to read data as a character string, sometimes fails to convert them into internal representation, and the file is skipped without processing. When option --persist is specified, command does not give up the file, and tries to read as binary data.

Next command will show strings in binary file.

    greple -o --re '(?a)\w{4,}' --persist --uc /bin/*

If you want read all files as binary data, use --icode=binary instead.

-Mdebug, -dx

Debug option is described in App::Greple::debug module.


Environment variable GREPLEOPTS is used as a default options. They are inserted before command line options.

Before starting execution, greple reads the file named .greplerc on user's home directory. Following directives can be used.

option name string

Argument name of `option' directive is user defined option name. The rest are processed by shellwords routine defined in Text::ParseWords module. Be sure that this module sometimes requires escape backslashes.

Any kind of string can be used for option name but it is not combined with other options.

    option --fromcode --outside='(?s)\/\*.*?\*\/'
    option --fromcomment --inside='(?s)\/\*.*?\*\/'

If the option named default is defined, it will be used as a default option.

For the purpose to include following arguments within replaced strings, two special notations can be used in option definition. String $<n> is replaced by the nth argument after the substituted option, where n is number start from one. String $<shift> is replaced by following command line argument and the argument is removed from option list.

For example, when

    option --line --le &line=$<shift>

is defined, command

    greple --line 10,20-30,40

will be evaluated as this:

    greple --le &line=10,20-30,40
expand name string

Define local option name. Command expand is almost same as command option in terms of its function. However, option defined by this command is expanded in, and only in, the process of definition, while option definition is expanded when command arguments are processed.

This is similar to string macro defined by following define command. But macro expansion is done by simple string replacement, so you have to use expand to define option composed by multiple arguments.

define name string

Define macro. This is similar to option, but argument is not processed by shellwords and treated just a simple text, so meta-characters can be included without escape. Macro expansion is done for option definition and other macro definition. Macro is not evaluated in command line option. Use option directive if you want to use in command line,

    define (#kana) \p{InKatakana}
    option --kanalist --nocolor -o --join --re '(#kana)+(\n(#kana)+)*'
    help   --kanalist List up Katakana string
help name

If `help' directive is used for same option name, it will be printed in usage message. If the help message is `ignore', corresponding line won't show up in the usage.

builtin spec variable

Define built-in option which should be processed by option parser. Arguments are assumed to be Getopt::Long style spec, and variable is string start with $, @ or %. They will be replaced by a reference to the object which the string represent.

See pgp module for example.

autoload module options ...

Define module which should be loaded automatically when specified option is found in the command arguments.

For example,

    autoload -Mdig --dig

replaces option "--dig" to "-Mdig --dig", and dig module is loaded before processing --dig option.

Environment variable substitution is done for string specified by `option' and `define' directives. Use Perl syntax $ENV{NAME} for this purpose. You can use this to make a portable module.

When greple found __PERL__ line in .greplerc file, the rest of the file is evaluated as a Perl program. You can define your own subroutines which can be used by --inside/outside, --include/exclude, --block options.

For those subroutines, file content will be provided by global variable $_. Expected response from the subroutine is the list of array references, which is made up by start and end offset pairs.

For example, suppose that the following function is defined in your .greplerc file. Start and end offset for each pattern match can be taken as array element $-[0] and $+[0].

    sub odd_line {
        my @list;
        my $i;
        while (/.*\n/g) {
            push(@list, [ $-[0], $+[0] ]) if ++$i % 2;

You can use next command to search pattern included in odd number lines.

    % greple --inside '&odd_line' pattern files...


You can expand the greple command using module. Module files are placed at App/Greple/ directory in Perl library, and therefor has App::Greple::module package name.

In the command line, module have to be specified preceding any other options in the form of -Mmodule. However, it also can be specified at the beginning of option expansion.

If the package name is declared properly, __DATA__ section in the module file will be interpreted same as .greplerc file content. So you can declare the module specific options there. Functions declared in the module can be used from those options, it makes highly expandable option/programming interaction possible.

Using -M without module argument will print available module list. Option --man will display module document when used with -M option. Use --show option to see the module itself. Option --path will print the path of module file.

See this sample module code. This sample defines options to search from pod, comment and other segment in Perl script. Those capability can be implemented both in function and macro.

    package App::Greple::perl;

    use Exporter 'import';
    our @EXPORT      = qw(pod comment podcomment);
    our %EXPORT_TAGS = ( );
    our @EXPORT_OK   = qw();
    use App::Greple::Common;
    use App::Greple::Regions;
    my $pod_re = qr{^=\w+(?s:.*?)(?:\Z|^=cut\s*\n)}m;
    my $comment_re = qr{^(?:[ \t]*#.*\n)+}m;
    sub pod {
        match_regions(pattern => $pod_re);
    sub comment {
        match_regions(pattern => $comment_re);
    sub podcomment {
        match_regions(pattern => qr/$pod_re|$comment_re/);
    define :comment: ^(\s*#.*\n)+
    define :pod: ^=(?s:.*?)(?:\Z|^=cut\s*\n)
    #option --pod --inside :pod:
    #option --comment --inside :comment:
    #option --code --outside :pod:|:comment:
    option --pod --inside '&pod'
    option --comment --inside '&comment'
    option --code --outside '&podcomment'

You can use the module like this:

    greple -Mperl --pod default greple

    greple -Mperl --colorful --code --comment --pod default greple

If special subroutine initialize() is defined in the module, it is called at the beginning with Getopt::EX::Module object as a first argument. Second argument is the reference to @ARGV, and you can modify actual @ARGV using it. See find module as a sample.


Most capability of greple is derived from mg command, which has been developing from early 1990's by the same author. Because modern standard grep family command becomes to have similar capabilities, it is a time to clean up entire functionalities, totally remodel the option interfaces, and change the command name. (2013.11)


grep(1), perl(1)




Kazumasa Utashiro


Copyright 1991-2020 Kazumasa Utashiro

This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself.