Skip to main content

grep command man page

GREP(1)       GREP(1)

       grep, egrep, fgrep - print lines matching a pattern
       grep [options] PATTERN [FILE...]
       grep [options] [-e PATTERN | -f FILE] [FILE...]

       Grep  searches the named input FILEs (or standard input if no files are
       named, or the file name - is given) for lines containing a match to the
       given PATTERN.  By default, grep prints the matching lines.

       In addition, two variant programs egrep and fgrep are available. Egrep
       is the same as grep -E. Fgrep is the same as grep -F.

       -A NUM, --after-context=NUM
     Print NUM lines of  trailing  context  after  matching lines.
     Places  a line  containing  --  between contiguous  groups  of

       -a, --text
     Process a binary file as if it were text; this is equivalent  to
     the --binary-files=text option.

       -B NUM, --before-context=NUM
     Print  NUM  lines of  leading  context  before  matching lines.
     Places  a line  containing  --  between contiguous  groups  of

       -C NUM, --context=NUM
     Print  NUM lines of output context.  Places a line containing --
     between contiguous groups of matches.

       -b, --byte-offset
     Print the byte offset within the input file before each line  of

     If the first few bytes of a file indicate that the file contains
     binary data, assume that the file is of type TYPE.  By  default,
     TYPE is binary, and grep normally outputs either a one-line mes-
     sage saying that a binary file matches, or no message  if there
     is  no  match.   If  TYPE is without-match, grep assumes that a
     binary file does not match; this is equivalent to the -I option.
     If  TYPE is  text,  grep processes a binary file as if it were
     text; this is  equivalent to  the  -a  option. Warning:  grep
     --binary-files=text  might output binary garbage, which can have
     nasty side effects if the output is a terminal and if the termi-
     nal driver interprets some of it as commands.

       --colour[=WHEN], --color[=WHEN]
     Surround the matching string with the marker find in GREP_COLOR
     environment variable. WHEN may be ‘never’, ‘always’, or ‘auto’

       -c, --count
     Suppress normal output; instead print a count of matching lines
     for  each input file.  With the -v, --invert-match option (see
     below), count non-matching lines.

       -D ACTION, --devices=ACTION
     If an input file is a device, FIFO or socket, use ACTION to pro-
     cess  it.  By default, ACTION is read, which means that devices
     are read just as if they were  ordinary  files. If  ACTION  is
     skip, devices are silently skipped.

       -d ACTION, --directories=ACTION
     If  an  input file is a directory, use ACTION to process it.  By
     default, ACTION is read, which means that directories  are  read
     just  as if they were ordinary files.  If ACTION is skip, direc-
     tories are silently skipped.  If ACTION is recurse,  grep reads
     all  files under each directory, recursively; this is equivalent
     to the -r option.

       -E, --extended-regexp
     Interpret PATTERN as an extended regular expression (see below).

       -e PATTERN, --regexp=PATTERN
     Use PATTERN as the pattern; useful to protect patterns beginning
     with -.

       -F, --fixed-strings
     Interpret PATTERN as a list of fixed strings, separated by  new-
     lines, any of which is to be matched.

       -P, --perl-regexp
     Interpret PATTERN as a Perl regular expression.

       -f FILE, --file=FILE
     Obtain  patterns from  FILE, one per line.  The empty file con-
     tains zero patterns, and therefore matches nothing.

       -G, --basic-regexp
     Interpret PATTERN as a basic  regular  expression (see  below).
     This is the default.

       -H, --with-filename
     Print the filename for each match.

       -h, --no-filename
     Suppress the  prefixing of  filenames  on output when multiple
     files are searched.

       --help Output a brief help message.

       -I     Process a binary file as if it did not  contain  matching data;
     this is equivalent to the --binary-files=without-match option.

       -i, --ignore-case
     Ignore  case  distinctions  in  both  the PATTERN and the input

       -L, --files-without-match
     Suppress normal output; instead print the name  of  each input
     file from which no output would normally have been printed.  The
     scanning will stop on the first match.

       -l, --files-with-matches
     Suppress normal output; instead print the name  of  each input
     file  from  which output would normally have been printed.  The
     scanning will stop on the first match.

       -m NUM, --max-count=NUM
     Stop reading a file after NUM matching lines.  If the  input  is
     standard input  from a regular file, and NUM matching lines are
     output, grep ensures that the standard input  is positioned  to
     just  after the last matching line before exiting, regardless of
     the presence of trailing context lines.  This enables a  calling
     process  to resume a search.  When grep stops after NUM matching
     lines, it outputs any trailing context lines.  When  the -c  or
     --count  option  is  also used, grep  does  not output a count
     greater than NUM. When the -v or --invert-match option is  also
     used, grep stops after outputting NUM non-matching lines.

       --mmap If  possible, use the mmap(2) system call to read input, instead
     of the default read(2) system call.  In some situations, --mmap
     yields  better performance.  However, --mmap can cause undefined
     behavior (including core dumps) if an input file shrinks while
     grep is operating, or if an I/O error occurs.

       -n, --line-number
     Prefix each line of output with the line number within its input

       -o, --only-matching
     Show only the part of a matching line that matches PATTERN.

     Displays input actually coming from standard input as input com-
     ing  from file LABEL.  This is especially useful for tools like
     zgrep, e.g.  gzip -cd foo.gz |grep -H --label=foo something

     Use line buffering, it can be a performance penality.

       -q, --quiet, --silent
     Quiet; do not write anything to standard output. Exit  immedi-
     ately  with  zero status if any match is found, even if an error
     was detected.  Also see the -s or --no-messages option.

       -R, -r, --recursive
     Read all files under each directory, recursively; this is equiv-
     alent to the -d recurse option.

     Recurse in directories only searching file matching PATTERN.

     Recurse in directories skip file matching PATTERN.

       -s, --no-messages
     Suppress error  messages about nonexistent or unreadable files.
     Portability note: unlike GNU grep, traditional grep did not con-
     form to POSIX.2, because traditional grep lacked a -q option and
     its -s option behaved like GNU grep’s -q option. Shell  scripts
     intended to be portable to traditional grep should avoid both -q
     and -s and should redirect output to /dev/null instead.

       -U, --binary
     Treat the file(s) as binary.  By default, under MS-DOS  and  MS-
     Windows, grep  guesses the file type by looking at the contents
     of the first 32KB read from the file.  If grep decides the  file
     is  a  text  file, it strips the CR characters from the original
     file contents (to make regular expressions with  ^  and  $  work
     correctly).  Specifying -U overrules this guesswork, causing all
     files to be read and passed to the matching mechanism  verbatim;
     if  the  file is a text file with CR/LF pairs at the end of each
     line, this will cause some regular expressions  to  fail.  This
     option  has no effect on platforms other than MS-DOS and MS-Win-

       -u, --unix-byte-offsets
     Report Unix-style byte offsets. This  switch  causes  grep  to
     report  byte  offsets  as if the file were Unix-style text file,
     i.e. with CR characters stripped off.  This will produce results
     identical to running grep on a Unix machine.  This option has no
     effect unless -b option is  also used;  it  has no  effect  on
     platforms other than MS-DOS and MS-Windows.

       -V, --version
     Print  the  version number of grep to standard error.  This ver-
     sion number should be included in all bug reports (see below).

       -v, --invert-match
     Invert the sense of matching, to select non-matching lines.

       -w, --word-regexp
     Select only those lines containing  matches  that  form whole
     words.   The  test is that the matching substring must either be
     at the beginning of the line, or preceded by  a non-word  con-
     stituent character.  Similarly, it must be either at the end of
     the line or followed by a non-word constituent character. Word-
     constituent  characters are letters, digits, and the underscore.

       -x, --line-regexp
     Select only those matches that exactly match the whole line.

       -y     Obsolete synonym for -i.

       -Z, --null
     Output a zero byte (the ASCII  NUL  character)  instead  of  the
     character that normally follows a file name.  For example, grep
     -lZ outputs a zero byte after each  file name  instead  of  the
     usual  newline. This option makes the output unambiguous, even
     in the presence of file names containing unusual characters like
     newlines.  This option can  be used  with commands like find
     -print0, perl -0, sort -z, and xargs  -0 to  process  arbitrary
     file names, even those that contain newline characters.

       A  regular  expression  is  a  pattern that describes a set of strings.
       Regular expressions are constructed analogously to  arithmetic  expres-
       sions, by using various operators to combine smaller expressions.

       Grep  understands  two different versions of regular expression syntax:
       “basic” and “extended.” In GNU grep, there is no difference in avail-
       able  functionality  using  either  syntax.   In other implementations,
       basic regular expressions are less powerful.  The following description
       applies to extended regular expressions; differences for basic regular
       expressions are summarized afterwards.

       The fundamental building blocks are the regular expressions that match
       a single character.  Most characters, including all letters and digits,
       are regular expressions that match themselves.  Any metacharacter  with
       special meaning may be quoted by preceding it with a backslash.

       A  bracket  expression is a list of characters enclosed by [ and ].  It
       matches any single character in that list; if the  first character  of
       the  list is the caret ^ then it matches any character not in the list.
       For example, the regular expression  [0123456789]  matches  any single

       Within a bracket expression, a range expression consists of two charac-
       ters separated by a hyphen.  It matches any single character that sorts
       between the  two  characters,  inclusive, using the locale’s collating
       sequence and character set.  For example,  in  the  default  C  locale,
       [a-d] is equivalent to [abcd].  Many locales sort characters in dictio-
       nary order, and in these locales [a-d] is typically not equivalent  to
       [abcd]; it  might  be equivalent to [aBbCcDd], for example.  To obtain
       the traditional interpretation of bracket expressions, you can use  the
       C locale by setting the LC_ALL environment variable to the value C.

       Finally, certain  named classes  of  characters are predefined within
       bracket expressions, as follows. Their names are self explanatory, and
       they   are   [:alnum:], [:alpha:],  [:cntrl:], [:digit:],  [:graph:],
       [:lower:], [:print:], [:punct:], [:space:], [:upper:], and  [:xdigit:].
       For  example,  [[:alnum:]]  means  [0-9A-Za-z], except the latter form
       depends upon the C locale and the ASCII character encoding, whereas the
       former  is  independent of  locale  and character set. (Note that the
       brackets in these class names are part of the symbolic names, and  must
       be  included  in addition to the brackets delimiting the bracket list.)
       Most metacharacters  lose  their special  meaning  inside  lists.   To
       include a literal ] place it first in the list. Similarly, to include
       a literal ^ place it anywhere but first. Finally, to include a literal
       - place it last.

       The period .  matches any single character.  The symbol \w is a synonym
       for [[:alnum:]] and \W is a synonym for [^[:alnum]].

       The caret ^ and the dollar sign $ are metacharacters that  respectively
       match the empty string at the beginning and end of a line.  The symbols
       \< and \> respectively match the empty string at the beginning and  end
       of  a  word.   The  symbol \b matches the empty string at the edge of a
       word, and \B matches the empty string provided it’s not at the edge  of
       a word.

       A regular expression may be followed by one of several repetition oper-
       ?      The preceding item is optional and matched at most once.
       *      The preceding item will be matched zero or more times.
       +      The preceding item will be matched one or more times.
       {n}    The preceding item is matched exactly n times.
       {n,}   The preceding item is matched n or more times.
       {n,m}  The preceding item is matched at least n times, but  not  more
     than m times.

       Two  regular  expressions  may  be  concatenated; the resulting regular
       expression matches any string formed by concatenating  two  substrings
       that respectively match the concatenated subexpressions.

       Two  regular  expressions  may  be  joined by the infix operator |; the
       resulting regular expression matches any string matching either subex-

       Repetition  takes  precedence  over  concatenation, which in turn takes
       precedence over alternation.  A whole subexpression may be enclosed  in
       parentheses to override these precedence rules.

       The  backreference \n, where n is a single digit, matches the substring
       previously matched by the nth parenthesized subexpression of the regu-
       lar expression.

       In  basic  regular  expressions the metacharacters ?, +, {, |, (, and )
       lose their special meaning; instead use the  backslashed versions  \?,
       \+, \{, \|, \(, and \).

       Traditional  egrep  did not support the { metacharacter, and some egrep
       implementations support \{ instead, so portable scripts should avoid  {
       in egrep patterns and should use [{] to match a literal {.

       GNU  egrep  attempts to support traditional usage by assuming that { is
       not special if it would be the start of an invalid interval  specifica-
       tion.   For example, the shell command egrep ’{1’ searches for the two-
       character string {1 instead of reporting a syntax error in the  regular
       expression.  POSIX.2 allows this behavior as an extension, but portable
       scripts should avoid it.

       Grep’s behavior is affected by the following environment variables.

       A locale LC_foo is  specified by  examining  the  three  environment
       variables  LC_ALL,  LC_foo,  LANG,  in  that order.  The first of these
       variables that is set specifies the locale.  For example, if LC_ALL  is
       not  set, but LC_MESSAGES is set to pt_BR, then Brazilian Portuguese is
       used for the LC_MESSAGES locale. The C locale is used if none of these
       environment  variables  are  set,  or  if  the  locale  catalog is not
       installed, or if grep was not compiled with national  language  support

     This variable specifies default options to be placed in front of
     any  explicit  options. For  example, if   GREP_OPTIONS   is
     ’--binary-files=without-match  --directories=skip’, grep behaves
     as if the two options --binary-files=without-match and  --direc-
     tories=skip  had been  specified before any explicit options.
     Option specifications are separated by whitespace.  A  backslash
     escapes  the  next  character,  so  it can be used to specify an
     option containing whitespace or a backslash.

     Specifies the marker for highlighting.

     These variables specify the LC_COLLATE locale, which  determines
     the  collating sequence used to interpret range expressions like

     These variables specify the LC_CTYPE  locale,  which  determines
     the type of characters, e.g., which characters are whitespace.

     These variables specify the LC_MESSAGES locale, which determines
     the language that grep uses for messages. The default C locale
     uses American English messages.

     If  set, grep  behaves  as  POSIX.2  requires;  otherwise, grep
     behaves more like other GNU  programs.   POSIX.2 requires  that
     options that follow file names must be treated as file names; by
     default, such options are permuted to the front of  the  operand
     list  and are  treated as options.  Also, POSIX.2 requires that
     unrecognized options be diagnosed as “illegal”, but  since  they
     are  not really against the law the default is to diagnose them
     as  “invalid”.   POSIXLY_CORRECT also  disables _N_GNU_nonop-
     tion_argv_flags_, described below.

     (Here  N is grep’s numeric process ID.)  If the ith character of
     this environment variable’s value is 1, do not consider the  ith
     operand  of  grep to be an option, even if it appears to be one.
     A shell can put this variable in the environment for  each  com-
     mand  it runs, specifying which operands are the results of file
     name wildcard expansion and therefore should not be  treated  as
     options. This  behavior is  available only  with  the GNU C
     library, and only when POSIXLY_CORRECT is not set.

       Normally, exit status is 0 if selected lines are found and 1 otherwise.
       But the exit status is 2 if an error occurred, unless the -q or --quiet
       or --silent option is used and a selected line is found.

       Email bug reports to

       Large repetition counts in the {n,m} construct may cause grep  to  use
       lots of memory. In addition, certain other obscure regular expressions
       require exponential time and space, and may cause grep to  run  out  of

       Backreferences are very slow, and may require exponential time.

GNU Project  2002/01/22       GREP(1)


Popular posts from this blog