Skip to main content

awk command man page

GAWK(1)       Utility Commands       GAWK(1)

NAME
       gawk - pattern scanning and processing language
SYNOPSIS
       gawk [ POSIX or GNU style options ] -f program-file [ -- ] file ...
       gawk [ POSIX or GNU style options ] [ -- ] program-text file ...

       pgawk [ POSIX or GNU style options ] -f program-file [ -- ] file ...
       pgawk [ POSIX or GNU style options ] [ -- ] program-text file ...

DESCRIPTION
       Gawk  is the  GNU Project’s implementation of the AWK programming lan-
       guage.  It conforms to the definition of the  language in  the POSIX
       1003.2  Command Language And Utilities Standard.  This version in turn
       is based on the description in The AWK Programming  Language,  by  Aho,
       Kernighan,  and Weinberger,  with the additional features found in the
       System V Release 4 version of UNIX awk. Gawk also provides more recent
       Bell  Laboratories  awk extensions, and a number of GNU-specific exten-
       sions.

       Pgawk is the profiling version of gawk. It is identical in  every  way
       to  gawk,  except  that programs run more slowly, and it automatically
       produces an execution profile in the file awkprof.out when  done.   See
       the --profile option, below.

       The  command  line  consists of options to gawk itself, the AWK program
       text (if not supplied via the -f or --file options), and values to  be
       made available in the ARGC and ARGV pre-defined AWK variables.

OPTION FORMAT
       Gawk options may be either traditional POSIX one letter options, or GNU
       style long options.  POSIX options start with a single “-”, while  long
       options start  with “--”.  Long options are provided for both GNU-spe-
       cific features and for POSIX-mandated features.

       Following the POSIX standard, gawk-specific options  are supplied  via
       arguments  to  the -W option.  Multiple -W options may be supplied Each
       -W option has a corresponding long option, as  detailed below. Argu-
       ments  to  long options are either joined with the option by an = sign,
       with no intervening spaces, or they may be provided in the next command
       line  argument. Long options may be abbreviated, as long as the abbre-
       viation remains unique.

OPTIONS
       Gawk accepts the following options, listed alphabetically.

       -F fs
       --field-separator fs
     Use fs for the input field separator (the value of the FS prede-
     fined variable).

       -v var=val
       --assign var=val
     Assign  the  value  val to the variable var, before execution of
     the program begins.  Such variable values are available  to  the
     BEGIN block of an AWK program.

       -f program-file
       --file program-file
     Read  the AWK program source from the file program-file, instead
     of from the  first  command  line argument.   Multiple  -f  (or
     --file) options may be used.

       -mf NNN
       -mr NNN
     Set various memory limits to the value NNN.  The f flag sets the
     maximum number of fields, and the r flag sets the maximum record
     size.  These two flags and the -m option are from the Bell Labo-
     ratories research version of UNIX awk. They  are  ignored  by
     gawk, since gawk has no pre-defined limits.

       -W compat
       -W traditional
       --compat
       --traditional
     Run  in compatibility mode.  In compatibility mode, gawk behaves
     identically to UNIX awk; none of the GNU-specific extensions are
     recognized.   The use  of  --traditional is preferred over the
     other forms of this option.  See GNU EXTENSIONS, below, for more
     information.

       -W copyleft
       -W copyright
       --copyleft
       --copyright
     Print the short version of the GNU copyright information message
     on the standard output and exit successfully.

       -W dump-variables[=file]
       --dump-variables[=file]
     Print a sorted list of global variables, their types  and final
     values  to file. If no file is provided, gawk uses a file named
     awkvars.out in the current directory.
     Having a list of all the global variables is a good way to  look
     for  typographical  errors in your programs.  You would also use
     this option if you have a large program with a lot of functions,
     and  you want to be sure that your functions don’t inadvertently
     use global variables that you meant to be local.  (This is  a
     particularly  easy  mistake  to  make with simple variable names
     like i, j, and so on.)

       -W exec file
       --exec file
     Similar to -f, however, this is option  is  the  last  one  pro-
     cessed. This should be used with #!  scripts, particularly for
     CGI applications, to avoid passing in options or source code (!)
     on  the  command line from a URL. This option disables command-
     line variable assignments.

       -W gen-po
       --gen-po
     Scan and parse the AWK program, and generate a  GNU  .po format
     file on standard output with entries for all localizable strings
     in the program.  The program itself is not  executed.   See  the
     GNU gettext distribution for more information on .po files.

       -W help
       -W usage
       --help
       --usage
     Print a relatively short summary of the available options on the
     standard output. (Per the GNU Coding Standards, these  options
     cause an immediate, successful exit.)

       -W lint[=value]
       --lint[=value]
     Provide  warnings about constructs  that  are  dubious or non-
     portable to  other  AWK implementations.   With an   optional
     argument of fatal, lint warnings become fatal errors.  This may
     be drastic, but its use will certainly encourage the development
     of  cleaner AWK programs. With an optional argument of invalid,
     only warnings about things that are actually invalid are issued.
     (This is not fully implemented yet.)

       -W lint-old
       --lint-old
     Provide  warnings about constructs that are not portable to the
     original version of Unix awk.

       -W non-decimal-data
       --non-decimal-data
     Recognize octal and hexadecimal values in input data.  Use  this
     option with great caution!

       -W posix
       --posix
     This  turns on compatibility mode, with the following additional
     restrictions:

     · \x escape sequences are not recognized.

     · Only space and tab act as field separators when FS is set to a
single space, newline does not.

     · You cannot continue lines after ?  and :.

     · The synonym func for the keyword function is not recognized.

     · The  operators ** and **= cannot be used in place of ^ and ^=.

     · The fflush() function is not available.

       -W profile[=prof_file]
       --profile[=prof_file]
     Send profiling data to prof_file. The default  is  awkprof.out.
     When  run with gawk, the profile is just a “pretty printed” ver-
     sion of the program.  When run with pgawk, the profile  contains
     execution counts of  each statement in the program in the left
     margin and function call counts for each user-defined  function.

       -W re-interval
       --re-interval
     Enable  the  use of  interval expressions in regular expression
     matching (see Regular Expressions, below).  Interval expressions
     were not traditionally available in the AWK language.  The POSIX
     standard added them, to make awk and egrep consistent with  each
     other.   However, their use is likely to break old AWK programs,
     so gawk only provides them  if  they  are requested  with  this
     option, or when --posix is specified.

       -W source program-text
       --source program-text
     Use program-text as AWK program source code.  This option allows
     the easy intermixing of library functions (used via the  -f  and
     --file  options) with  source code entered on the command line.
     It is intended primarily for medium to large AWK programs  used
     in shell scripts.

       -W version
       --version
     Print  version  information  for this particular copy of gawk on
     the standard output.  This is useful mainly for knowing  if  the
     current  copy  of gawk on your system is up to date with respect
     to whatever the Free Software Foundation is distributing.  This
     is  also useful when reporting bugs.  (Per the GNU Coding Stan-
     dards, these options cause an immediate, successful exit.)

       --     Signal the end of options. This is useful to allow further argu-
     ments  to the  AWK program itself to start with a “-”.  This is
     mainly for consistency with the argument parsing convention used
     by most other POSIX programs.
       In  compatibility  mode, any other options are flagged as invalid, but
       are otherwise ignored.  In normal operation, as long  as program  text
       has  been supplied, unknown options are passed on to the AWK program in
       the ARGV array for processing.  This is particularly useful for running
       AWK programs via the “#!” executable interpreter mechanism.
AWK PROGRAM EXECUTION
       An  AWK program consists of a sequence of pattern-action statements and
       optional function definitions.
     pattern { action statements }
     function name(parameter list) { statements }
       Gawk first reads the program source from the program-file(s) if speci-
       fied, from arguments to --source, or from the first non-option argument
       on the command line.  The -f and --source options may be used  multiple
       times  on  the command line.  Gawk reads the program text as if all the
       program-files and command  line source texts  had  been  concatenated
       together.   This is  useful  for  building libraries of AWK functions,
       without having to include them in each new AWK program that uses them.
       It also provides the ability to mix library functions with command line
       programs.
       The environment variable AWKPATH specifies a search path to  use  when
       finding source files named with the -f option. If this variable does
       not exist, the default path is ".:/usr/local/share/awk".  (The actual
       directory  may  vary, depending upon how gawk was built and installed.)
       If a file name given to the -f option contains a “/” character, no path
       search is performed.
       Gawk executes AWK programs in the following order.  First, all variable
       assignments specified via the -v option are performed.  Next, gawk com-
       piles  the program into an internal form.  Then, gawk executes the code
       in the BEGIN block(s) (if any), and then proceeds  to  read  each  file
       named  in  the  ARGV array.  If there are no files named on the command
       line, gawk reads the standard input.
       If a filename on the command line has the form var=val it is treated as
       a  variable  assignment.  The  variable var will be assigned the value
       val.  (This happens after any BEGIN block(s) have been  run.)   Command
       line  variable assignment is most useful for dynamically assigning val-
       ues to the variables AWK uses to control  how  input  is  broken  into
       fields  and records.  It is also useful for controlling state if multi-
       ple passes are needed over a single data file.
       If the value of a particular element of ARGV is empty (""), gawk skips
       over it.
       For  each record in the input, gawk tests to see if it matches any pat-
       tern in the AWK program. For each pattern that the record matches, the
       associated  action  is  executed.  The patterns are tested in the order
       they occur in the program.
       Finally, after all the input is exhausted, gawk executes the  code  in
       the END block(s) (if any).
VARIABLES, RECORDS AND FIELDS
       AWK variables are dynamic; they come into existence when they are first
       used.  Their values are either floating-point numbers  or  strings,  or
       both,  depending upon how they are used.  AWK also has one dimensional
       arrays; arrays with multiple dimensions may be simulated.  Several pre-
       defined variables are set as a program runs; these will be described as
       needed and summarized below.
   Records
       Normally, records are separated by newline characters.  You can control
       how  records are separated by assigning values to the built-in variable
       RS.  If RS is any single character, that character  separates  records.
       Otherwise,  RS is a regular expression. Text in the input that matches
       this  regular   expression   separates the   record. However,   in
       compatibility  mode,  only  the first character of its string value is
       used for separating records.  If RS is set to  the  null string,  then
       records are  separated by  blank  lines.   When RS is set to the null
       string, the newline character always acts  as  a field separator,  in
       addition to whatever value FS may have.
   Fields
       As each input record is read, gawk splits the record into fields, using
       the value of the FS variable as the field separator.  If FS is a single
       character,  fields  are separated by that character.  If FS is the null
       string, then each individual character becomes a separate field.  Oth-
       erwise, FS is expected to be a full regular expression. In the special
       case that FS is a single space, fields are separated by runs of spaces
       and/or  tabs  and/or  newlines. (But  see  the discussion of --posix,
       below). NOTE: The value of IGNORECASE (see  below)  also  affects  how
       fields  are  split when FS is a regular expression, and how records are
       separated when RS is a regular expression.
       If the FIELDWIDTHS variable is set to a space separated list  of  num-
       bers,  each  field  is expected to have fixed width, and gawk splits up
       the record using the specified widths.  The value  of  FS  is  ignored.
       Assigning  a  new  value to  FS overrides the use of FIELDWIDTHS, and
       restores the default behavior.
       Each field in the input record may be referenced by its position,  $1,
       $2,  and so on. $0 is the whole record. Fields need not be referenced
       by constants:
     n = 5
     print $n
       prints the fifth field in the input record.
       The variable NF is set to the total  number  of fields in  the input
       record.
       References  to  non-existent fields (i.e. fields after $NF) produce the
       null-string.  However, assigning to a non-existent field (e.g., $(NF+2)
       = 5) increases the value of NF, creates any intervening fields with the
       null string as their value, and causes the value of  $0 to  be recom-
       puted, with the fields being separated by the value of OFS.  References
       to negative numbered fields  cause  a  fatal  error.   Decrementing  NF
       causes  the  values  of fields past the new value to be lost, and the
       value of $0 to be recomputed, with the fields being  separated  by  the
       value of OFS.
       Assigning  a  value  to an existing field causes the whole record to be
       rebuilt when $0 is referenced.  Similarly,  assigning  a value to  $0
       causes the record to be resplit, creating new values for the fields.
   Built-in Variables
       Gawk’s built-in variables are:
       ARGC   The number of  command  line  arguments (does not include
  options to gawk, or the program source).
       ARGIND   The index in ARGV of the current file being processed.
       ARGV   Array of command line arguments.  The array is indexed from
  0  to  ARGC - 1.  Dynamically changing the contents of ARGV
  can control the files used for data.
       BINMODE   On non-POSIX systems, specifies use of  “binary”  mode  for
  all file  I/O.  Numeric values of 1, 2, or 3, specify that
  input files, output files, or  all files, respectively,
  should  use binary I/O.  String values of "r", or "w" spec-
  ify that input files, or output files, respectively, should
  use binary I/O.  String values of "rw" or "wr" specify that
  all files should use binary I/O.  Any other string value is
  treated as "rw", but generates a warning message.
       CONVFMT   The conversion format for numbers, "%.6g", by default.
       ENVIRON   An  array containing the values of the current environment.
  The array is indexed by  the environment  variables,  each
  element  being  the value  of  that variable (e.g., ENVI-
  RON["HOME"] might be /home/arnold). Changing  this array
  does not affect the environment seen by programs which gawk
  spawns via redirection or the system() function.
       ERRNO   If a system error occurs either  doing  a  redirection  for
  getline,  during  a read for getline, or during a close(),
  then ERRNO will contain a string describing the error.  The
  value is subject to translation in non-English locales.
       FIELDWIDTHS A  white-space  separated  list  of fieldwidths.  When set,
  gawk parses the input into fields of fixed  width,  instead
  of  using the value of the FS variable as the field separa-
  tor.
       FILENAME   The name of the current input file. If no files are speci-
  fied on  the  command  line, the value of FILENAME is “-”.
  However, FILENAME  is  undefined  inside  the  BEGIN block
  (unless set by getline).
       FNR   The input record number in the current input file.
       FS   The input field separator, a space by default.  See Fields,
  above.
       IGNORECASE  Controls the case-sensitivity of all regular expression and
  string  operations. If  IGNORECASE has a non-zero value,
  then string comparisons  and pattern  matching  in rules,
  field splitting with FS, record separating with RS, regular
  expression matching with  ~ and  !~,  and the  gensub(),
  gsub(), index(), match(), split(), and sub() built-in func-
  tions all ignore case when doing regular expression opera-
  tions.  NOTE: Array subscripting is not affected.  However,
  the asort() and asorti() functions are affected.
  Thus, if IGNORECASE is not equal to zero, /aB/ matches  all
  of the strings "ab", "aB", "Ab", and "AB".  As with all AWK
  variables, the initial value of IGNORECASE is zero, so  all
  regular expression and string operations are normally case-
  sensitive.  Under Unix, the full ISO 8859-1 Latin-1 charac-
  ter set is used when ignoring case. As of gawk 3.1.4, the
  case equivalencies are fully locale-aware, based on the  C
  <ctype.h> facilities such as isalpha(), and tolupper().
       LINT   Provides  dynamic  control of the --lint option from within
  an AWK program.  When true, gawk prints lint warnings. When
  false,  it  does  not.   When  assigned  the string value
  "fatal", lint warnings become fatal errors, exactly  like
  --lint=fatal.  Any other true value just prints warnings.
       NF   The number of fields in the current input record.
       NR   The total number of input records seen so far.
       OFMT   The output format for numbers, "%.6g", by default.
       OFS   The output field separator, a space by default.
       ORS   The output record separator, by default a newline.
       PROCINFO   The elements  of  this array provide access to information
  about the running AWK program.  On some systems, there  may
  be  elements in  the  array, "group1" through "groupn" for
  some n, which is the number of  supplementary  groups  that
  the process has. Use  the in operator to test for these
  elements.  The following  elements  are  guaranteed to  be
  available:
  PROCINFO["egid"]   the value of the getegid(2) system call.
  PROCINFO["euid"]   the value of the geteuid(2) system call.
  PROCINFO["FS"]     "FS"  if field  splitting with FS is in
     effect, or "FIELDWIDTHS" if field split-
     ting with FIELDWIDTHS is in effect.
  PROCINFO["gid"]    the  value of the getgid(2) system call.
  PROCINFO["pgrpid"] the process group ID of the current pro-
     cess.
  PROCINFO["pid"]    the process ID of the current process.
  PROCINFO["ppid"]   the  parent  process  ID of the current
     process.
  PROCINFO["uid"]    the value of the getuid(2) system call.
  PROCINFO["version"]
     The  version of gawk.  This is available
     from version 3.1.4 and later.
       RS   The input record separator, by default a newline.
       RT   The record terminator.  Gawk sets RT to the input text that
  matched  the character  or regular expression specified by
  RS.
       RSTART   The index of the first character matched by match(); 0  if
  no  match. (This  implies that character indices start at
  one.)
       RLENGTH   The length of the string  matched  by  match();  -1 if  no
  match.
       SUBSEP   The character used to separate multiple subscripts in array
  elements, by default "\034".
       TEXTDOMAIN  The text domain of the AWK program; used to find the local-
  ized translations for the program’s strings.
   Arrays
       Arrays  are  subscripted with an expression between square brackets ([
       and ]). If the expression is an expression list (expr, expr ...)  then
       the  array subscript is a string consisting of the concatenation of the
       (string) value of each expression, separated by the value of the SUBSEP
       variable.   This facility  is  used  to simulate multiply dimensioned
       arrays. For example:
     i = "A"; j = "B"; k = "C"
     x[i, j, k] = "hello, world\n"
       assigns the string "hello, world\n" to the element of the array x which
       is indexed by the string "A\034B\034C". All arrays in AWK are associa-
       tive, i.e. indexed by string values.
       The special operator in may be used in an if or while statement to  see
       if an array has an index consisting of a particular value.
     if (val in array)
  print array[val]
       If the array has multiple subscripts, use (i, j) in array.
       The in construct may also be used in a for loop to iterate over all the
       elements of an array.
       An element may be deleted from an array using  the  delete  statement.
       The  delete statement may also be used to delete the entire contents of
       an array, just by specifying the array name without a subscript.
   Variable Typing And Conversion
       Variables and fields may be (floating point) numbers,  or  strings,  or
       both.  How the value of a variable is interpreted depends upon its con-
       text.  If used in a numeric expression, it will be treated as a number,
       if used as a string it will be treated as a string.
       To force a variable to be treated as a number, add 0 to it; to force it
       to be treated as a string, concatenate it with the null string.
       When a string must be converted to a number, the conversion  is accom-
       plished using  strtod(3).   A number is converted to a string by using
       the value of CONVFMT as a  format  string  for sprintf(3),  with  the
       numeric value  of  the variable as the argument.  However, even though
       all numbers in AWK are floating-point, integral values are always  con-
       verted as integers.  Thus, given
     CONVFMT = "%2.2f"
     a = 12
     b = a ""
       the variable b has a string value of "12" and not "12.00".
       Gawk  performs  comparisons  as follows: If two variables are numeric,
       they are compared numerically.  If one value is numeric and  the other
       has  a  string  value  that is a “numeric string,” then comparisons are
       also done numerically.  Otherwise, the numeric value is converted to  a
       string and a string comparison is performed.  Two strings are compared,
       of course, as strings.  Note that the POSIX standard applies  the  con-
       cept  of “numeric  string” everywhere, even to string constants.  How-
       ever, this is clearly incorrect, and gawk does not  do  this.   (Fortu-
       nately, this is fixed in the next version of the standard.)
       Note that string constants, such as "57", are not numeric strings, they
       are string constants.  The idea of “numeric  string”  only  applies  to
       fields, getline input, FILENAME, ARGV elements, ENVIRON elements and
       the elements of an array created by split() that are  numeric  strings.
       The  basic  idea is  that  user input, and only user input, that looks
       numeric, should be treated that way.
       Uninitialized variables have the numeric value 0 and the string value
       "" (the null, or empty, string).
   Octal and Hexadecimal Constants
       Starting with version 3.1 of gawk , you may use C-style octal and hex-
       adecimal constants in your AWK program source code.  For example,  the
       octal  value  011 is equal to decimal 9, and the hexadecimal value 0x11
       is equal to decimal 17.
   String Constants
       String constants in AWK are sequences of characters  enclosed  between
       double quotes (").  Within strings, certain escape sequences are recog-
       nized, as in C. These are:
       \\   A literal backslash.
       \a   The “alert” character; usually the ASCII BEL character.
       \b   backspace.
       \f   form-feed.
       \n   newline.
       \r   carriage return.
       \t   horizontal tab.
       \v   vertical tab.
       \xhex digits
   The character represented by the string of hexadecimal digits fol-
   lowing the \x.  As in ANSI C, all following hexadecimal digits are
   considered part of the escape sequence.  (This feature should tell
   us something about language design by committee.)  E.g., "\x1B" is
   the ASCII ESC (escape) character.
       \ddd The character represented by the 1-, 2-, or 3-digit  sequence  of
   octal digits.  E.g., "\033" is the ASCII ESC (escape) character.
       \c   The literal character c.
       The  escape  sequences may also be used inside constant regular expres-
       sions (e.g., /[ \t\f\n\r\v]/ matches whitespace characters).
       In compatibility mode, the characters represented by octal and hexadec-
       imal  escape  sequences are  treated  literally when  used in regular
       expression constants.  Thus, /a\52b/ is equivalent to /a\*b/.
PATTERNS AND ACTIONS
       AWK is a line-oriented language. The pattern comes first, and then the
       action. Action statements are enclosed in { and }.  Either the pattern
       may be missing, or the action may be missing, but, of course, not both.
       If  the pattern is  missing,  the action is executed for every single
       record of input. A missing action is equivalent to
     { print }
       which prints the entire record.
       Comments begin with the “#” character, and continue until  the  end  of
       the line.  Blank lines may be used to separate statements.  Normally, a
       statement ends with a newline, however, this is not the case for lines
       ending  in  a “,”, {, ?, :, &&, or ||.  Lines ending in do or else also
       have their statements automatically continued on the  following line.
       In  other  cases,  a  line can be continued by ending it with a “\”, in
       which case the newline will be ignored.
       Multiple statements may be put on one line by separating them  with  a
       “;”.   This  applies to both the statements within the action part of a
       pattern-action pair (the usual case), and to the pattern-action state-
       ments themselves.
   Patterns
       AWK patterns may be one of the following:
     BEGIN
     END
     /regular expression/
     relational expression
     pattern && pattern
     pattern || pattern
     pattern ? pattern : pattern
     (pattern)
     ! pattern
     pattern1, pattern2
       BEGIN  and  END are two special kinds of patterns which are not tested
       against the input.  The action parts of all BEGIN patterns  are merged
       as  if  all  the statements  had been written in a single BEGIN block.
       They are executed before any of the input is read.  Similarly, all  the
       END blocks are merged, and executed when all the input is exhausted (or
       when an exit statement is executed).  BEGIN and END patterns cannot  be
       combined with  other  patterns in pattern expressions. BEGIN and END
       patterns cannot have missing action parts.
       For /regular expression/ patterns, the associated statement is executed
       for  each  input record that matches the regular expression.  Regular
       expressions are the same as  those  in  egrep(1),  and  are  summarized
       below.
       A  relational  expression may use any of the operators defined below in
       the section on actions. These generally test  whether  certain fields
       match certain regular expressions.
       The  &&, ||, and !  operators are logical AND, logical OR, and logical
       NOT, respectively, as in C.  They do short-circuit evaluation, also  as
       in  C,  and  are used for combining more primitive pattern expressions.
       As in most languages, parentheses may be used to change the  order  of
       evaluation.
       The  ?: operator is like the same operator in C.  If the first pattern
       is true then the pattern used for testing is the second pattern, other-
       wise  it is  the  third.  Only one of the second and third patterns is
       evaluated.
       The pattern1, pattern2 form of an expression is called a range pattern.
       It  matches  all input records starting with a record that matches pat-
       tern1, and continuing until a record that matches pattern2,  inclusive.
       It does not combine with any other sort of pattern expression.
   Regular Expressions
       Regular expressions  are  the  extended kind found in egrep.  They are
       composed of characters as follows:
       c  matches the non-metacharacter c.
       \c  matches the literal character c.
       .  matches any character including newline.
       ^  matches the beginning of a string.
       $  matches the end of a string.
       [abc...]  character list, matches any of the characters abc....
       [^abc...]  negated character list, matches any character except abc....
       r1|r2  alternation: matches either r1 or r2.
       r1r2  concatenation: matches r1, and then r2.
       r+  matches one or more r’s.
       r*  matches zero or more r’s.
       r?  matches zero or one r’s.
       (r)  grouping: matches r.
       r{n}
       r{n,}
       r{n,m}  One  or two numbers inside braces denote an interval expres-
 sion. If there is one number in the braces, the  preceding
 regular  expression r is repeated n times.  If there are two
 numbers separated by a comma, r is repeated n to  m times.
 If  there  is one  number  followed by  a comma, then r is
 repeated at least n times.
 Interval expressions are only available if either --posix or
 --re-interval is specified on the command line.

       \y  matches  the empty string at either the beginning or the end
 of a word.

       \B  matches the empty string within a word.

       \<  matches the empty string at the beginning of a word.

       \>  matches the empty string at the end of a word.

       \w  matches any word-constituent character  (letter,  digit,  or
 underscore).

       \W  matches any character that is not word-constituent.

       \‘  matches  the empty  string  at  the beginning  of a buffer
 (string).

       \’  matches the empty string at the end of a buffer.

       The escape sequences that are valid in string constants (see below) are
       also valid in regular expressions.

       Character  classes  are a new feature introduced in the POSIX standard.
       A character class is a special notation for describing lists of charac-
       ters  that  have a specific attribute, but where the actual characters
       themselves can vary from country to country and/or from character  set
       to  character  set.   For  example, the notion of what is an alphabetic
       character differs in the USA and in France.

       A character class is only valid in  a  regular  expression  inside  the
       brackets of a character list.  Character classes consist of [:, a key-
       word denoting the class, and :]. The character classes defined by  the
       POSIX standard are:

       [:alnum:]  Alphanumeric characters.

       [:alpha:]  Alphabetic characters.

       [:blank:]  Space or tab characters.

       [:cntrl:]  Control characters.

       [:digit:]  Numeric characters.

       [:graph:]  Characters that are both printable and visible.  (A space is
 printable, but not visible, while an a is both.)

       [:lower:]  Lower-case alphabetic characters.

       [:print:]  Printable characters (characters that are not control char-
 acters.)

       [:punct:]  Punctuation characters (characters that are not letter, dig-
 its, control characters, or space characters).

       [:space:]  Space characters (such as space, tab, and formfeed, to  name
 a few).

       [:upper:]  Upper-case alphabetic characters.

       [:xdigit:] Characters that are hexadecimal digits.

       For  example,  before the POSIX standard, to match alphanumeric charac-
       ters, you would have had to write /[A-Za-z0-9]/. If your character set
       had  other  alphabetic characters in it, this would not match them, and
       if your character set collated differently from ASCII, this  might  not
       even match the ASCII alphanumeric characters.  With the POSIX character
       classes, you can write /[[:alnum:]]/, and this matches  the  alphabetic
       and numeric characters in your character set.

       Two  additional special sequences can appear in character lists. These
       apply to non-ASCII  character  sets,  which  can have  single  symbols
       (called collating  elements)  that  are represented with more than one
       character, as well as several characters that are equivalent  for  col-
       lating, or  sorting,  purposes.  (E.g.,  in French, a plain “e” and a
       grave-accented e` are equivalent.)

       Collating Symbols
     A collating  symbol  is a  multi-character  collating  element
     enclosed in [.  and .]. For example, if ch is a collating ele-
     ment, then [[.ch.]]  is a regular expression that matches  this
     collating element,  while  [ch] is  a  regular expression that
     matches either c or h.

       Equivalence Classes
     An equivalence class is a locale-specific name  for  a  list  of
     characters  that are equivalent. The name is enclosed in [= and
     =].  For example, the name e might be used to represent  all  of
     “e,”  “´,”  and “`.”  In this case, [[=e=]] is a regular expres-
     sion that matches any of e, ´, or `.

       These features are very valuable in non-English speaking locales.   The
       library functions  that gawk uses for regular expression matching cur-
       rently only recognize POSIX character classes; they  do not  recognize
       collating symbols or equivalence classes.

       The  \y, \B, \<, \>, \w, \W, \‘, and \’ operators are specific to gawk;
       they are extensions based on facilities in the GNU  regular  expression
       libraries.

       The various command line options control how gawk interprets characters
       in regular expressions.

       No options
     In the default case, gawk provide all the facilities  of POSIX
     regular  expressions  and the  GNU regular expression operators
     described above. However, interval  expressions are  not  sup-
     ported.

       --posix
     Only  POSIX regular expressions are supported, the GNU operators
     are not special. (E.g., \w  matches  a  literal w).   Interval
     expressions are allowed.

       --traditional
     Traditional  Unix awk regular expressions are matched.  The GNU
     operators are not special, interval expressions are  not avail-
     able,  and  neither are the POSIX character classes ([[:alnum:]]
     and so on).   Characters described  by  octal  and  hexadecimal
     escape  sequences are treated literally, even if they represent
     regular expression metacharacters.

       --re-interval
     Allow interval  expressions  in  regular expressions,  even  if
     --traditional has been provided.

   Actions
       Action  statements  are enclosed in braces, { and }.  Action statements
       consist of the usual assignment, conditional,  and  looping  statements
       found  in  most languages.   The  operators,  control  statements, and
       input/output statements available are patterned after those in C.

   Operators
       The operators in AWK, in order of decreasing precedence, are


       (...)   Grouping

       $   Field reference.

       ++ --   Increment and decrement, both prefix and postfix.

       ^   Exponentiation (** may  also be  used,  and **=  for  the
  assignment operator).

       + - !   Unary plus, unary minus, and logical negation.

       * / %   Multiplication, division, and modulus.

       + -   Addition and subtraction.

       space   String concatenation.

       < >
       <= >=
       != ==   The regular relational operators.

       ~ !~   Regular  expression match, negated match.  NOTE: Do not use
  a constant regular expression (/foo/) on the left-hand side
  of  a  ~  or !~.  Only use one on the right-hand side.  The
  expression /foo/ ~ exp has  the  same  meaning  as  (($0  ~
  /foo/) ~ exp).  This is usually not what was intended.

       in   Array membership.

       &&   Logical AND.

       ||   Logical OR.

       ?:   The C  conditional expression.  This has the form expr1 ?
  expr2 : expr3.  If expr1 is true, the value of the  expres-
  sion is  expr2,  otherwise it is expr3.  Only one of expr2
  and expr3 is evaluated.

       = += -=
       *= /= %= ^= Assignment. Both absolute assignment  (var =  value)  and
  operator-assignment (the other forms) are supported.

   Control Statements
       The control statements are as follows:

     if (condition) statement [ else statement ]
     while (condition) statement
     do statement while (condition)
     for (expr1; expr2; expr3) statement
     for (var in array) statement
     break
     continue
     delete array[index]
     delete array
     exit [ expression ]
     { statements }

   I/O Statements
       The input/output statements are as follows:


       close(file [, how])   Close file, pipe or co-process.  The optional how
    should only be used when closing  one  end of  a
    two-way  pipe  to a  co-process. It  must be a
    string value, either "to" or "from".

       getline     Set $0 from next input record; set NF, NR, FNR.

       getline <file     Set $0 from next record of file; set NF.

       getline var     Set var from next input record; set NR, FNR.

       getline var <file     Set var from next record of file.

       command | getline [var]
    Run command piping the output either into $0  or
    var, as above.

       command |& getline [var]
    Run  command  as  a  co-process piping the output
    either into $0 or var,  as above.  Co-processes
    are a gawk extension.

       next     Stop  processing  the  current input record.  The
    next input record is read and  processing starts
    over  with the first pattern in the AWK program.
    If the end of the input data is reached, the  END
    block(s), if any, are executed.

       nextfile     Stop processing the current input file.  The next
    input record read comes from the next input file.
    FILENAME  and ARGIND are updated, FNR is reset to
    1, and processing starts over with the first pat-
    tern  in the AWK program. If the end of the input
    data is reached, the END block(s), if  any,  are
    executed.

       print     Prints  the current record.  The output record is
    terminated with the value of the ORS variable.

       print expr-list     Prints expressions.  Each expression is separated
    by the  value  of the OFS variable.  The output
    record is terminated with the value  of  the  ORS
    variable.

       print expr-list >file Prints  expressions  on file.  Each expression is
    separated by the value of the OFS variable.   The
    output record is terminated with the value of the
    ORS variable.

       printf fmt, expr-list Format and print.

       printf fmt, expr-list >file
    Format and print on file.

       system(cmd-line)     Execute the command cmd-line, and return the exit
    status.   (This may not be available on non-POSIX
    systems.)

       fflush([file])     Flush any buffers associated with the open output
    file  or  pipe  file.   If file is missing, then
    standard output is flushed.  If file is the  null
    string, then all open output files and pipes have
    their buffers flushed.

       Additional output redirections are allowed for print and printf.

       print ... >> file
     appends output to the file.

       print ... | command
     writes on a pipe.

       print ... |& command
     sends data to a co-process.

       The getline command returns 0 on end of file and -1 on an error.  Upon
       an error, ERRNO contains a string describing the problem.

       NOTE: If using a pipe or co-process to getline, or from print or printf
       within a loop, you must use close() to create new instances of the com-
       mand.  AWK does not automatically close pipes or co-processes when they
       return EOF.

   The printf Statement
       The AWK versions of the printf statement and  sprintf() function  (see
       below) accept the following conversion specification formats:

       %c      An ASCII character.  If the argument used for %c is numeric, it
      is treated as a character and printed.  Otherwise, the argument
      is assumed to be a string, and the only first character of that
      string is printed.

       %d, %i  A decimal number (the integer part).

       %e ,  %E
      A floating point number of the form [-]d.dddddde[+-]dd. The %E
      format uses E instead of e.

       %f      A floating point number of the form [-]ddd.dddddd.

       %g ,  %G
      Use %e or %f conversion, whichever is shorter, with nonsignifi-
      cant zeros suppressed.  The %G format uses %E instead of %e.

       %o      An unsigned octal number (also an integer).

       %u      An unsigned decimal number (again, an integer).

       %s      A character string.

       %x ,  %X
      An unsigned hexadecimal number (an  integer).   The  %X format
      uses ABCDEF instead of abcdef.

       %%      A single % character; no argument is converted.

       NOTE: When using the integer format-control letters for values that are
       outside the range of a C long integer, gawk switches to the  %g format
       specifier.  If  --lint is provided on the command line gawk warns about
       this.  Other versions of awk may print invalid values or do  something
       else entirely.

       Optional,  additional  parameters may lie between the % and the control
       letter:

       count$ Use the count’th argument at this point in the formatting.  This
     is  called  a positional specifier and is intended primarily for
     use in translated versions of format strings, not in the origi-
     nal text of an AWK program.  It is a gawk extension.

       -      The expression should be left-justified within its field.

       space  For  numeric  conversions,  prefix positive values with a space,
     and negative values with a minus sign.

       +      The plus sign, used before the width modifier (see below),  says
     to  always  supply  a  sign for numeric conversions, even if the
     data to be formatted is positive. The  + overrides  the space
     modifier.

       #      Use  an  “alternate  form” for certain control letters.  For %o,
     supply a leading zero.  For %x, and %X, supply a leading 0x  or
     0X  for a nonzero result. For %e, %E, and %f, the result always
     contains a decimal point. For %g, and %G,  trailing  zeros  are
     not removed from the result.

       0      A leading 0 (zero) acts as a flag, that indicates output should
     be padded with zeroes instead of spaces. This applies  even  to
     non-numeric  output  formats.  This flag only has an effect when
     the field width is wider than the value to be printed.

       width  The field should be padded to this width. The field is normally
     padded  with  spaces.  If the 0 flag has been used, it is padded
     with zeroes.

       .prec  A number that specifies the precision to use when printing.  For
     the  %e, %E, and %f formats, this specifies the number of digits
     you want printed to the right of the decimal point.  For the %g,
     and  %G  formats, it specifies the maximum number of significant
     digits.  For the %d, %o, %i, %u, %x, and %X formats,  it speci-
     fies  the minimum number of digits to print.  For %s, it speci-
     fies the maximum number  of  characters  from  the  string  that
     should be printed.

       The dynamic width and prec capabilities of the ANSI C printf() routines
       are supported.  A * in place of either the width or prec specifications
       causes  their  values  to  be taken from the argument list to printf or
       sprintf().  To use a positional specifier with a dynamic width or  pre-
       cision, supply the  count$  after  the *  in the format string.  For
       example, "%3$*2$.*1$s".

   Special File Names
       When doing I/O redirection from either print or printf into a file,  or
       via  getline  from  a  file,  gawk recognizes certain special filenames
       internally.  These filenames allow  access  to  open  file  descriptors
       inherited  from gawk’s parent process (usually the shell).  These file
       names may also be used on the command line to  name  data  files.   The
       filenames are:

       /dev/stdin  The standard input.

       /dev/stdout The standard output.

       /dev/stderr The standard error output.

       /dev/fd/n   The file associated with the open file descriptor n.

       These are particularly useful for error messages.  For example:

     print "You blew it!" > "/dev/stderr"

       whereas you would otherwise have to use

     print "You blew it!" | "cat 1>&2"

       The  following  special filenames  may be used with the |& co-process
       operator for creating TCP/IP network connections.

       /inet/tcp/lport/rhost/rport  File for TCP/IP connection on  local  port
   lport  to remote host rhost on remote port
   rport.  Use a port of 0 to have the system
   pick a port.

       /inet/udp/lport/rhost/rport  Similar, but use UDP/IP instead of TCP/IP.

       /inet/raw/lport/rhost/rport  Reserved for future use.

       Other special filenames provide access to information about the running
       gawk  process. These  filenames  are  now obsolete.  Use the PROCINFO
       array to obtain the information they provide.  The filenames are:

       /dev/pid   Reading this file returns the process  ID  of  the  current
  process, in decimal, terminated with a newline.

       /dev/ppid   Reading this file returns the parent process ID of the cur-
  rent process, in decimal, terminated with a newline.

       /dev/pgrpid Reading this file returns the process group ID of the  cur-
  rent process, in decimal, terminated with a newline.

       /dev/user   Reading this file returns a single record terminated with a
  newline.  The fields are separated with spaces.  $1 is  the
  value  of the getuid(2) system call, $2 is the value of the
  geteuid(2) system call, $3 is the value  of the  getgid(2)
  system  call,  and $4 is the value of the getegid(2) system
  call.  If there are any additional  fields, they  are  the
  group  IDs  returned by getgroups(2).  Multiple groups may
  not be supported on all systems.

   Numeric Functions
       AWK has the following built-in arithmetic functions:


       atan2(y, x)   Returns the arctangent of y/x in radians.

       cos(expr)     Returns the cosine of expr, which is in radians.

       exp(expr)     The exponential function.

       int(expr)     Truncates to integer.

       log(expr)     The natural logarithm function.

       rand()     Returns a random number N, between 0 and 1, such that 0 ≤
    N < 1.

       sin(expr)     Returns the sine of expr, which is in radians.

       sqrt(expr)    The square root function.

       srand([expr]) Uses  expr as a new seed for the random number generator.
    If no expr is provided, the time of  day  is  used.   The
    return  value  is the previous seed for the random number
    generator.

   String Functions
       Gawk has the following built-in string functions:


       asort(s [, d])       Returns the number of elements  in  the source
      array  s.   The contents of s are sorted using
      gawk’s normal rules for comparing  values,  and
      the  indexes  of the  sorted  values  of s are
      replaced with sequential integers starting with
      1. If the optional destination array d is spec-
      ified, then s is first duplicated into  d,  and
      then  d is  sorted, leaving the indexes of the
      source array s unchanged.

       asorti(s [, d])       Returns the number of elements  in  the source
      array  s.   The behavior is the same as that of
      asort(), except that the array indices are used
      for  sorting, not the array values.  When done,
      the array is indexed numerically, and the  val-
      ues  are those of  the original indices.  The
      original values are lost; thus provide a second
      array if you wish to preserve the original.

       gensub(r, s, h [, t])   Search  the  target string t for matches of the
      regular expression r.  If h is a string begin-
      ning with g or G, then replace all matches of r
      with s. Otherwise, h is a  number  indicating
      which  match of r to replace.  If t is not sup-
      plied, $0 is used instead.  Within the replace-
      ment  text  s,  the  sequence  \n, where n is a
      digit from 1 to 9, may be used to indicate just
      the  text  that matched the n’th parenthesized
      subexpression.  The sequence \0 represents  the
      entire  matched text, as does the character &.
      Unlike sub() and gsub(), the modified string is
      returned as the result of the function, and the
      original target string is not changed.

       gsub(r, s [, t])       For each substring matching the regular expres-
      sion  r in the string t, substitute the string
      s, and return the number of substitutions.   If
      t  is  not  supplied,  use  $0. An  & in the
      replacement text is replaced with the text that
      was  actually matched.  Use \& to get a literal
      &.  (This must be typed as  "\\&";  see GAWK:
      Effective  AWK Programming for a fuller discus-
      sion of the rules for &’s  and  backslashes  in
      the replacement text of sub(), gsub(), and gen-
      sub().)

       index(s, t)       Returns the index of the string t in the string
      s,  or  0  if  t is not present. (This implies
      that character indices start at one.)

       length([s])       Returns the length of  the  string  s,  or  the
      length  of  $0  if s is not supplied.  Starting
      with version 3.1.5, as  a  non-standard exten-
      sion,  with an array argument, length() returns
      the number of elements in the array.

       match(s, r [, a])       Returns the position in s  where  the  regular
      expression  r occurs, or 0 if r is not present,
      and sets the  values  of RSTART and  RLENGTH.
      Note that the argument order is the same as for
      the ~ operator: str ~ re.  If array a  is  pro-
      vided, a is cleared and then elements 1 through
      n are filled with the portions of s that match
      the  corresponding  parenthesized subexpression
      in r.  The 0’th element of a contains the  por-
      tion of s matched by the entire regular expres-
      sion r. Subscripts  a[n,  "start"],  and  a[n,
      "length"]  provide  the starting  index in the
      string and length respectively, of each match-
      ing substring.

       split(s, a [, r])       Splits  the  string  s  into the array a on the
      regular expression r, and returns the number of
      fields. If  r is omitted, FS is used instead.
      The  array  a  is  cleared  first.    Splitting
      behaves identically   to   field   splitting,
      described above.

       sprintf(fmt, expr-list) Prints expr-list according to fmt, and  returns
      the resulting string.

       strtonum(str)       Examines str,  and  returns its numeric value.
      If str begins  with  a  leading 0,  strtonum()
      assumes that  str  is an octal number. If str
      begins with a  leading  0x  or  0X,  strtonum()
      assumes that str is a hexadecimal number.

       sub(r, s [, t])       Just  like  gsub(), but only the first matching
      substring is replaced.

       substr(s, i [, n])      Returns the at most n-character substring of  s
      starting at i. If n is omitted, the rest of s
      is used.

       tolower(str)       Returns a copy of the string str, with all  the
      upper-case  characters  in  str translated  to
      their  corresponding  lower-case counterparts.
      Non-alphabetic characters are left unchanged.

       toupper(str)       Returns a copy of the string str, with all the
      lower-case  characters  in  str translated  to
      their  corresponding  upper-case counterparts.
      Non-alphabetic characters are left unchanged.

   Time Functions
       Since one of the primary uses of AWK programs is processing  log files
       that  contain time stamp information, gawk provides the following func-
       tions for obtaining time stamps and formatting them.


       mktime(datespec)
Turns datespec into a time stamp of the same form as returned
by  systime().  The datespec is a string of the form YYYY MM
DD HH MM SS[ DST].  The contents of the  string  are  six  or
seven numbers representing respectively the full year includ-
ing century, the month from 1 to 12, the  day of  the month
from  1  to  31, the hour of the day from 0 to 23, the minute
from 0 to 59, and the second from 0 to 60,  and  an  optional
daylight  saving  flag.  The values of these numbers need not
be within the ranges specified; for example, an  hour of  -1
means 1 hour before midnight. The origin-zero Gregorian cal-
endar is assumed, with year 0 preceding year 1 and  year  -1
preceding  year  0.   The  time is assumed to be in the local
timezone.  If the daylight saving flag is positive, the  time
is  assumed  to be daylight saving time; if zero, the time is
assumed to be standard time; and if negative  (the  default),
mktime()  attempts  to determine whether daylight saving time
is in effect for the specified time.  If  datespec  does  not
contain  enough  elements  or if the resulting time is out of
range, mktime() returns -1.

       strftime([format [, timestamp]])
Formats timestamp according to the specification  in  format.
The  timestamp should be of the same form as returned by sys-
time().  If timestamp is missing, the current time of day  is
used. If  format is missing, a default format equivalent to
the output of date(1) is used. See the specification for the
strftime() function in ANSI C for the format conversions that
are guaranteed to be available.  A public-domain  version  of
strftime(3)  and  a  man  page for it come with gawk; if that
version was used to build gawk, then all of  the  conversions
described in that man page are available to gawk.

       systime() Returns  the  current time  of  day as the number of seconds
since the Epoch (1970-01-01 00:00:00 UTC on POSIX systems).

   Bit Manipulations Functions
       Starting with version 3.1 of gawk, the following bit manipulation func-
       tions are available.  They work by converting double-precision floating
       point values to unsigned long integers, doing the operation,  and  then
       converting the result back to floating point.  The functions are:

       and(v1, v2)   Return the bitwise AND of the values provided by v1
  and v2.

       compl(val)   Return the bitwise complement of val.

       lshift(val, count)  Return the value of val,  shifted  left  by count
  bits.

       or(v1, v2)   Return  the bitwise OR of the values provided by v1
  and v2.

       rshift(val, count)  Return the value of val,  shifted  right  by count
  bits.

       xor(v1, v2)   Return the bitwise XOR of the values provided by v1
  and v2.


   Internationalization Functions
       Starting with version 3.1 of gawk, the following functions may be  used
       from  within your AWK program for translating strings at run-time.  For
       full details, see GAWK: Effective AWK Programming.

       bindtextdomain(directory [, domain])
     Specifies the directory where gawk looks for the .mo  files,  in
     case they will not or cannot be placed in the ‘‘standard’’ loca-
     tions (e.g., during testing).  It returns the  directory where
     domain is ‘‘bound.’’
     The  default domain is the value of TEXTDOMAIN.  If directory is
     the null string (""), then bindtextdomain() returns the  current
     binding for the given domain.

       dcgettext(string [, domain [, category]])
     Returns  the  translation of  string  in text domain domain for
     locale category category. The default value for domain  is  the
     current  value of TEXTDOMAIN.  The default value for category is
     "LC_MESSAGES".
     If you supply a value for category, it must be a string equal to
     one  of the known locale categories described in GAWK: Effective
     AWK Programming. You must  also supply a  text domain.   Use
     TEXTDOMAIN if you want to use the current domain.

       dcngettext(string1 , string2 , number [, domain [, category]])
     Returns  the  plural  form used for number of the translation of
     string1 and string2 in text domain domain for  locale  category
     category.  The default value for domain is the current value of
     TEXTDOMAIN.  The default value for category is "LC_MESSAGES".
     If you supply a value for category, it must be a string equal to
     one  of the known locale categories described in GAWK: Effective
     AWK Programming. You must  also supply a  text domain.   Use
     TEXTDOMAIN if you want to use the current domain.

USER-DEFINED FUNCTIONS
       Functions in AWK are defined as follows:

     function name(parameter list) { statements }

       Functions  are executed when they are called from within expressions in
       either patterns or actions.  Actual parameters supplied in the function
       call  are  used to  instantiate the formal parameters declared in the
       function.  Arrays are passed by reference, other variables  are passed
       by value.

       Since  functions were not originally part of the AWK language, the pro-
       vision for local variables is rather clumsy: They are declared as extra
       parameters  in the parameter list.  The convention is to separate local
       variables from real parameters by extra spaces in the  parameter list.
       For example:

     function f(p, q,    a, b)   # a and b are local
     {
  ...
     }

     /abc/ { ... ; f(1, 2) ; ... }

       The left parenthesis in a function call is required to immediately fol-
       low the function name, without any intervening white space.  This is to
       avoid  a syntactic  ambiguity  with  the concatenation operator.  This
       restriction does not apply to the built-in functions listed above.

       Functions may call each other and may be recursive.   Function  parame-
       ters used as local variables are initialized to the null string and the
       number zero upon function invocation.

       Use return expr to return a value from a function.  The return value is
       undefined  if  no  value is  provided, or  if the function returns by
       “falling off” the end.

       If --lint has been provided, gawk warns about calls to undefined func-
       tions  at  parse time, instead of at run time.  Calling an undefined
       function at run time is a fatal error.

       The word func may be used in place of function.

DYNAMICALLY LOADING NEW FUNCTIONS
       Beginning with version 3.1 of gawk, you can dynamically add new built-
       in  functions  to  the  running gawk interpreter.  The full details are
       beyond the scope of this manual page; see GAWK: Effective AWK  Program-
       ming for the details.

       extension(object, function)
      Dynamically  link  the  shared object file named by object, and
      invoke function in  that object,  to  perform  initialization.
      These  should  both  be provided as strings.  Returns the value
      returned by function.

       This function is provided and documented in GAWK:  Effective  AWK  Pro-
       gramming,  but everything about this feature is likely to change in the
       next release.  We STRONGLY recommend that you do not use this  feature
       for anything that you aren’t willing to redo.

SIGNALS
       pgawk  accepts  two  signals.   SIGUSR1 causes it to dump a profile and
       function call stack to the profile file, which is  either  awkprof.out,
       or  whatever file was named with the --profile option.  It then contin-
       ues to run.  SIGHUP causes it to dump the  profile  and function  call
       stack and then exit.

EXAMPLES
       Print and sort the login names of all users:

   BEGIN     { FS = ":" }
{ print $1 | "sort" }

       Count lines in a file:

{ nlines++ }
   END { print nlines }

       Precede each line by its number in the file:

   { print FNR, $0 }

       Concatenate and line number (a variation on a theme):

   { print NR, $0 }
       Run an external command for particular lines of data:

   tail -f access_log |
   awk ’/myhome.html/ { system("nmap " $1 ">> logdir/myhome.html") }’

INTERNATIONALIZATION
       String constants are sequences of characters enclosed in double quotes.
       In non-English speaking environments, it is possible to mark strings in
       the  AWK program  as  requiring translation to the native natural lan-
       guage. Such strings are marked in the AWK program with a leading under-
       score (“_”).  For example,

     gawk ’BEGIN { print "hello, world" }’

       always prints hello, world.  But,

     gawk ’BEGIN { print _"hello, world" }’

       might print bonjour, monde in France.

       There are several steps involved in producing and running a localizable
       AWK program.

       1.  Add a BEGIN action to assign a value to the TEXTDOMAIN variable  to
  set the text domain to a name associated with your program.

BEGIN { TEXTDOMAIN = "myprog" }

  This allows gawk to find the .mo file associated with your program.
  Without this step, gawk uses the messages text domain, which likely
  does not contain translations for your program.

       2.  Mark all  strings  that  should  be translated with leading under-
  scores.

       3.  If necessary, use the dcgettext() and/or bindtextdomain() functions
  in your program, as appropriate.

       4.  Run gawk --gen-po -f myprog.awk > myprog.po to generate a .po file
  for your program.

       5.  Provide appropriate translations, and build and  install  a corre-
  sponding .mo file.

       The internationalization features are described in full detail in GAWK:
       Effective AWK Programming.

POSIX COMPATIBILITY
       A primary goal for gawk is compatibility with the  POSIX standard,  as
       well  as with the latest version of UNIX awk.  To this end, gawk incor-
       porates the following user visible features which are not described  in
       the AWK book, but are part of the Bell Laboratories version of awk, and
       are in the POSIX standard.

       The book indicates that command line variable assignment happens  when
       awk  would  otherwise  open  the argument as a file, which is after the
       BEGIN block is executed. However,  in  earlier implementations,  when
       such an assignment appeared before any file names, the assignment would
       happen before the BEGIN block was run.  Applications came to depend  on
       this  “feature.”  When awk was changed to match its documentation, the
       -v option for assigning variables before program execution was added to
       accommodate  applications  that depended upon the old behavior. (This
       feature was agreed upon by both the  Bell  Laboratories and  the  GNU
       developers.)

       The  -W option for implementation specific features is from the POSIX
       standard.

       When processing arguments, gawk uses the special option “--” to signal
       the end of arguments.  In compatibility mode, it warns about but other-
       wise ignores undefined options. In normal  operation,  such  arguments
       are passed on to the AWK program for it to process.

       The  AWK book  does not define the return value of srand().  The POSIX
       standard has it return the seed it was using, to allow keeping track of
       random  number  sequences.   Therefore srand() in gawk also returns its
       current seed.

       Other new features are: The use of multiple -f options (from MKS awk);
       the  ENVIRON array; the \a, and \v escape sequences (done originally in
       gawk and fed back into the Bell Laboratories  version); the  tolower()
       and  toupper() built-in functions (from the Bell Laboratories version);
       and the ANSI C conversion specifications in printf (done first  in  the
       Bell Laboratories version).

HISTORICAL FEATURES
       There are two features of historical AWK implementations that gawk sup-
       ports.  First, it is possible to call the  length()  built-in  function
       not only with no argument, but even without parentheses! Thus,

     a = length     # Holy Algol 60, Batman!

       is the same as either of

     a = length()
     a = length($0)

       This  feature is marked as “deprecated” in the POSIX standard, and gawk
       issues a warning about its use if --lint is specified  on  the  command
       line.

       The other feature is the use of either the continue or the break state-
       ments outside the body of a while, for, or do  loop.   Traditional  AWK
       implementations have  treated  such  usage  as equivalent to the next
       statement.  Gawk supports this usage if --traditional has  been speci-
       fied.

GNU EXTENSIONS
       Gawk  has  a  number of extensions to POSIX awk. They are described in
       this section.  All the extensions described here can  be  disabled  by
       invoking gawk with the --traditional option.

       The following features of gawk are not available in POSIX awk.

       · No  path  search  is  performed  for  files  named via the -f option.
Therefore the AWKPATH environment variable is not special.

       · The \x escape sequence.  (Disabled with --posix.)

       · The fflush() function. (Disabled with --posix.)

       · The ability to continue  lines  after ?   and  :.   (Disabled  with
--posix.)

       · Octal and hexadecimal constants in AWK programs.

       · The ARGIND, BINMODE, ERRNO, LINT, RT and TEXTDOMAIN variables are not
special.

       · The IGNORECASE variable and its side-effects are not available.

       · The FIELDWIDTHS variable and fixed-width field splitting.

       · The PROCINFO array is not available.

       · The use of RS as a regular expression.

       · The special file names available for I/O redirection are  not recog-
nized.

       · The |& operator for creating co-processes.

       · The  ability to split out individual characters using the null string
as the value of FS, and as the third argument to split().

       · The optional second argument to the close() function.

       · The optional third argument to the match() function.

       · The ability to use positional specifiers with printf and sprintf().

       · The use of delete array to delete the entire contents of an array.

       · The use of nextfile to abandon processing of the current input file.

       · The and(), asort(), asorti(), bindtextdomain(), compl(), dcgettext(),
dcngettext(), gensub(), lshift(),  mktime(),  or(),  rshift(), strf-
time(), strtonum(), systime() and xor() functions.

       · Localizable strings.

       · Adding new built-in functions dynamically with the extension() func-
tion.

       The AWK book does not define the return value of the close()  function.
       Gawk’s  close() returns the  value from fclose(3), or pclose(3), when
       closing an output file or pipe, respectively.  It returns the process’s
       exit  status when closing an input pipe. The return value is -1 if the
       named file, pipe or co-process was not opened with a redirection.

       When gawk is invoked with the --traditional option, if the fs  argument
       to  the -F  option  is “t”, then FS is set to the tab character.  Note
       that typing gawk -F\t ...  simply causes the shell to quote  the “t,”,
       and  does  not pass “\t” to the -F option.  Since this is a rather ugly
       special case, it is not the default behavior.  This behavior also  does
       not occur if --posix has been specified. To really get a tab character
       as the field separator, it is best to use single quotes:  gawk -F’\t’
       ....

       If  gawk is configured with the --enable-switch option to the configure
       command, then it accepts an additional control-flow statement:
     switch (expression) {
     case value|regex : statement
     ...
     [ default: statement ]
     }

ENVIRONMENT VARIABLES
       The AWKPATH environment variable can be used  to  provide  a  list  of
       directories  that gawk searches when looking for files named via the -f
       and --file options.

       If POSIXLY_CORRECT exists in the environment, then gawk behaves exactly
       as  if  --posix had been specified on the command line. If --lint has
       been specified, gawk issues a warning message to this effect.

SEE ALSO
       egrep(1), getpid(2),  getppid(2),  getpgrp(2),  getuid(2),  geteuid(2),
       getgid(2), getegid(2), getgroups(2)

       The  AWK Programming Language, Alfred V. Aho, Brian W. Kernighan, Peter
       J. Weinberger, Addison-Wesley, 1988.  ISBN 0-201-07981-X.

       GAWK: Effective AWK Programming, Edition 3.0,  published by  the  Free
       Software Foundation, 2001.

BUGS
       The  -F option is not necessary given the command line variable assign-
       ment feature; it remains only for backwards compatibility.

       Syntactically invalid single character programs tend  to overflow  the
       parse  stack, generating a rather unhelpful message.  Such programs are
       surprisingly difficult to diagnose in the completely general case,  and
       the effort to do so really is not worth it.

AUTHORS
       The original version of UNIX awk was designed and implemented by Alfred
       Aho, Peter Weinberger, and Brian Kernighan of Bell Laboratories. Brian
       Kernighan continues to maintain and enhance it.

       Paul  Rubin  and Jay  Fenlason, of the Free Software Foundation, wrote
       gawk, to be compatible with the original version of awk distributed  in
       Seventh Edition UNIX. John Woods contributed a number of bug fixes.
       David Trueman, with contributions from Arnold Robbins, made  gawk  com-
       patible with  the new version of UNIX awk.  Arnold Robbins is the cur-
       rent maintainer.

       The initial DOS port was done  by  Conrad  Kwok and  Scott  Garfinkle.
       Scott Deifik is the current DOS maintainer.  Pat Rankin did the port to
       VMS, and Michal Jaegermann did the port to the Atari ST. The  port  to
       OS/2  was done by Kai Uwe Rommel, with contributions and help from Dar-
       rel Hankerson.  Fred Fish  supplied  support  for  the  Amiga,  Stephen
       Davies  provided the  Tandem  port, and Martin Brown provided the BeOS
       port.

VERSION INFORMATION
       This man page documents gawk, version 3.1.5.

BUG REPORTS
       If you find a  bug  in  gawk,  please  send  electronic mail  to  bug-
       gawk@gnu.org.   Please  include your operating system and its revision,
       the version of gawk (from gawk --version), what C compiler you used  to
       compile it,  and a test program and data that are as small as possible
       for reproducing the problem.

       Before sending a bug report, please do two things.  First, verify  that
       you  have  the latest version of gawk.  Many bugs (usually subtle ones)
       are fixed at each release, and if yours is out of date, the problem may
       already have  been  solved.  Second, please read this man page and the
       reference manual carefully to be sure that what you  think  is a  bug
       really is, instead of just a quirk in the language.

       Whatever you do, do NOT post a bug report in comp.lang.awk.  While the
       gawk developers occasionally read this newsgroup, posting  bug  reports
       there  is  an  unreliable  way to report bugs.  Instead, please use the
       electronic mail addresses given above.

       If you’re using a GNU/Linux system or BSD-based system, you may wish to
       submit  a  bug report to the vendor of your distribution.  That’s fine,
       but please send a copy to the official email  address  as  well, since
       there’s no  guarantee that the bug will be forwarded to the gawk main-
       tainer.

ACKNOWLEDGEMENTS
       Brian Kernighan of Bell Laboratories provided valuable assistance  dur-
       ing testing and debugging.  We thank him.

COPYING PERMISSIONS
       Copyright © 1989, 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
       2001, 2002, 2003, 2004, 2005 Free Software Foundation, Inc.

       Permission is granted to make and distribute verbatim  copies  of  this
       manual  page  provided  the copyright notice and this permission notice
       are preserved on all copies.

       Permission is granted to copy and distribute modified versions of  this
       manual  page  under  the conditions for verbatim copying, provided that
       the entire resulting derived work is distributed under the terms of  a
       permission notice identical to this one.

       Permission  is granted to copy and distribute translations of this man-
       ual page into another language, under the above conditions for modified
       versions,  except that this permission notice may be stated in a trans-
       lation approved by the Foundation.

Free Software Foundation June 26 2005       GAWK(1)

Comments

Popular posts from this blog