man/cat1/pgawk.1.txt

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 pro-
       gramming language. It conforms to the definition of the
       language in the POSIX 1003.1 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 ver-
       sion of UNIX awk. Gawk also provides more recent Bell
       Laboratories awk extensions, and a number of GNU-spe-
       cific extensions.
 
       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 pro-
       file in the file awkprof.out when done. See the --pro-
       file 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-specific 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 correspond-
       ing long option, as detailed below. Arguments 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 abbreviation remains unique.
 
OPTIONS
       Gawk accepts the following options, listed by frequency.
 
       -F fs
       --field-separator fs
              Use fs for the input field separator (the value
              of the FS predefined 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 pro-
              gram-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 an earlier ver-
              sion of the Bell Laboratories 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 mis-
              take 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 processed. 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 dis-
              ables 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 pro-
              gram. 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 immedi-
              ate, successful exit.)
 
       -W lint[=value]
       --lint[=value]
              Provide warnings about constructs that are dubi-
              ous 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 argu-
              ment 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 fol-
              lowing 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" version of the program.
              When run with pgawk, the profile contains execu-
              tion 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 tradition-
              ally available in the AWK language. The POSIX
              standard added them, to make awk and egrep con-
              sistent 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 use-lc-numeric
       --use-lc-numeric
              This forces gawk to use the locale's decimal
              point character when parsing input data.
              Although the POSIX standard requires this behav-
              ior, and gawk does so when --posix is in effect,
              the default is to follow traditional behavior and
              use a period as the decimal point, even in
              locales where the period is not the decimal point
              character. This option overrides the default
              behavior, without the full draconian strictness
              of the --posix option.
 
       -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 dis-
              tributing. This is also useful when reporting
              bugs. (Per the GNU Coding Standards, these
              options cause an immediate, successful exit.)
 
       -- Signal the end of options. This is useful to
              allow further arguments to the AWK program itself
              to start with a "-". This provides 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 opera-
       tion, 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 specified, 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 com-
       mand 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 con-
       tains 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 compiles 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 vari-
       able assignment is most useful for dynamically assigning
       values to the variables AWK uses to control how input is
       broken into fields and records. It is also useful for
       controlling state if multiple 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 pattern in the AWK program. For each pat-
       tern 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. Sev-
       eral pre-defined variables are set as a program runs;
       these are 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 sepa-
       rate field. Otherwise, FS is expected to be a full reg-
       ular expression. In the special case that FS is a sin-
       gle space, fields are separated by runs of spaces and/or
       tabs and/or newlines. (But see the section POSIX COM-
       PATIBILITY, 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 numbers, 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 recomputed, 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, respec-
                   tively, should use binary I/O. String val-
                   ues of "r", or "w" specify 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 cur-
                   rent environment. The array is indexed by
                   the environment variables, each element
                   being the value of that variable (e.g., ENV-
                   IRON["HOME"] might be /home/arnold). Chang-
                   ing this array does not affect the environ-
                   ment 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 separator.
       FILENAME The name of the current input file. If no
                   files are specified 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 operations. NOTE: Array sub-
                   scripting 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 character 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 toupper().
       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 gete-
                                      gid(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
                                      splitting with FIELD-
                                      WIDTHS is in effect.
                   PROCINFO["gid"] the value of the get-
                                      gid(2) system call.
                   PROCINFO["pgrpid"] the process group ID of
                                      the current process.
                   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 sub-
                   scripts in array elements, by default
                   "\034".
       TEXTDOMAIN The text domain of the AWK program; used to
                   find the localized translations for the pro-
                   gram'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 mul-
       tiply 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 associative, i.e.
       indexed by string values.
       The special operator in may be used to test 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 iter-
       ate 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 specify-
       ing 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 inter-
       preted depends upon its context. 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 conver-
       sion is accomplished using strtod(3). A number is con-
       verted 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 converted as integers. Thus, given
              CONVFMT = "%2.2f"
              a = 12
              b = a ""
       the variable b has a string value of "12" and not
       "12.00".
       When operating in POSIX mode (such as with the --posix
       command line option), beware that locale settings may
       interfere with the way decimal numbers are treated: the
       decimal separator of the numbers you are feeding to gawk
       must conform to what your locale would expect, be it a
       comma (,) or a period (.).
       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 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 ele-
       ments 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 hexadecimal 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 recognized, as in C. These
       are:
       \\ A literal backslash.
       \a The "alert" character; usually the ASCII BEL char-
            acter.
       \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 hexadec-
            imal digits following 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 expressions (e.g., /[ \t\f\n\r\v]/ matches
       whitespace characters).
       In compatibility mode, the characters represented by
       octal and hexadecimal escape sequences are treated lit-
       erally 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 statements them-
       selves.
   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 exe-
       cuted 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 exe-
       cuted). 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 state-
       ment 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 gener-
       ally 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, otherwise 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 pattern1, and continuing
       until a record that matches pattern2, inclusive. It
       does not combine with any other sort of pattern expres-
       sion.
   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 expression. If there is one number
                  in the braces, the preceding regular expres-
                  sion 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 begin-
                  ning 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 (let-
                  ter, digit, or underscore).
 
       \W matches any character that is not word-con-
                  stituent.
 
       \` 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 feature introduced in the POSIX
       standard. A character class is a special notation for
       describing lists of characters 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 keyword 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 visi-
                  ble. (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 characters.)
 
       [:punct:] Punctuation characters (characters that are
                  not letter, digits, 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 characters, you would have had to write
       /[A-Za-z0-9]/. If your character set had other alpha-
       betic 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 charac-
       ters. With the POSIX character classes, you can write
       /[[:alnum:]]/, and this matches the alphabetic and
       numeric characters in your character set, no matter what
       it is.
 
       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 collating, or
       sorting, purposes. (E.g., in French, a plain "e" and a
       grave-accented "`" are equivalent.)
 
       Collating Symbols
              A collating symbol is a multi-character collating
              element enclosed in [. and .]. For example, if
              ch is a collating element, 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
              expression that matches any of e, �, or `.
 
       These features are very valuable in non-English speaking
       locales. The library functions that gawk uses for regu-
       lar expression matching currently only recognize POSIX
       character classes; they do not recognize collating sym-
       bols or equivalence classes.
 
       The \y, \B, \<, \>, \w, \W, \`, and \' operators are
       specific to gawk; they are extensions based on facili-
       ties in the GNU regular expression libraries.
 
       The various command line options control how gawk inter-
       prets characters in regular expressions.
 
       No options
              In the default case, gawk provide all the facili-
              ties of POSIX regular expressions and the GNU
              regular expression operators described above.
              However, interval expressions are not supported.
 
       --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 available, and nei-
              ther are the POSIX character classes ([[:alnum:]]
              and so on). Characters described by octal and
              hexadecimal escape sequences are treated liter-
              ally, even if they represent regular expression
              metacharacters.
 
       --re-interval
              Allow interval expressions in regular expres-
              sions, even if --traditional has been provided.
 
   Actions
       Action statements are enclosed in braces, { and }.
       Action statements consist of the usual assignment, con-
       ditional, and looping statements found in most lan-
       guages. 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 nega-
                   tion.
 
       * / % Multiplication, division, and modulus.
 
       + - Addition and subtraction.
 
       space String concatenation.
 
       | |& Piped I/O for getline, print, and printf.
 
       < >
       <= >=
       != == The regular relational operators.
 
       ~ !~ Regular expression match, negated match.
                   NOTE: Do not use a constant regular expres-
                   sion (/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 expression 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. (command can also be a
                             socket. See the subsection Spe-
                             cial File Names, below.)
 
       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 pattern
                             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 expres-
                             sion 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 sys-
                             tems.)
 
       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 or socket. (See also
              the subsection Special File Names, below.)
 
       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, co-process, or socket to getline,
       or from print or printf within a loop, you must use
       close() to create new instances of the command or
       socket. AWK does not automatically close pipes, sock-
       ets, 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, %F A floating point number of the form
               [-]ddd.dddddd. If the system library supports
               it, %F is available as well. This is like %f,
               but uses capital letters for special "not a num-
               ber" and "infinity" values. If %F is not avail-
               able, gawk uses %f.
 
       %g, %G Use %e or %f conversion, whichever is shorter,
               with nonsignificant 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 %0f 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 speci-
              fier and is intended primarily for use in trans-
              lated versions of format strings, not in the
              original 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 for-
              matted is positive. The + overrides the space
              modifier.
 
       # Use an "alternate form" for certain control let-
              ters. For %o, supply a leading zero. For %x,
              and %X, supply a leading 0x or 0X for a nonzero
              result. For %e, %E, %f and %F, the result always
              contains a decimal point. For %g, and %G, trail-
              ing 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, %f 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 specifies the min-
              imum number of digits to print. For %s, it spec-
              ifies 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
       precision, 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 connec-
       tions.
 
       /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 informa-
       tion 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 current process, in decimal, ter-
                   minated with a newline.
 
       /dev/pgrpid Reading this file returns the process group
                   ID of the current process, in decimal, ter-
                   minated 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 con-
                               tents of s are sorted using
                               gawk's normal rules for compar-
                               ing values, and the indices of
                               the sorted values of s are
                               replaced with sequential inte-
                               gers starting with 1. If the
                               optional destination array d is
                               specified, then s is first
                               duplicated into d, and then d is
                               sorted, leaving the indices 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 values are
                               those of the original indices.
                               The original values are lost;
                               thus provide a second array if
                               you wish to preserve the origi-
                               nal.
 
       gensub(r, s, h [, t]) Search the target string t for
                               matches of the regular expres-
                               sion r. If h is a string begin-
                               ning with g or G, then replace
                               all matches of r with s. Other-
                               wise, h is a number indicating
                               which match of r to replace. If
                               t is not supplied, $0 is used
                               instead. Within the replacement
                               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 parenthe-
                               sized 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 tar-
                               get string is not changed.
 
       gsub(r, s [, t]) For each substring matching the
                               regular expression r in the
                               string t, substitute the string
                               s, and return the number of sub-
                               stitutions. If t is not sup-
                               plied, use $0. An & in the
                               replacement text is replaced
                               with the text that was actually
                               matched. Use \& to get a lit-
                               eral &. (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
                               extension, with an array argu-
                               ment, length() returns the num-
                               ber 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 provided, 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 con-
                               tains the portion of s matched
                               by the entire regular expression
                               r. Subscripts a[n, "start"],
                               and a[n, "length"] provide the
                               starting index in the string and
                               length respectively, of each
                               matching substring.
 
       split(s, a [, r]) Splits the string s into the
                               array a on the regular expres-
                               sion 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 hexadeci-
                               mal 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.
 
       As of version 3.1.5, gawk is multibyte aware. This
       means that index(), length(), substr() and match() all
       work in terms of characters, not bytes.
 
   Time Functions
       Since one of the primary uses of AWK programs is pro-
       cessing log files that contain time stamp information,
       gawk provides the following functions 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 including 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 Grego-
                 rian calendar is assumed, with year 0 preced-
                 ing 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[, utc-flag]]])
                 Formats timestamp according to the specifica-
                 tion in format. If utc-flag is present and is
                 non-zero or non-null, the result is in UTC,
                 otherwise the result is in local time. The
                 timestamp should be of the same form as
                 returned by systime(). If timestamp is miss-
                 ing, the current time of day is used. If for-
                 mat is missing, a default format equivalent to
                 the output of date(1) is used. See the speci-
                 fication for the strftime() function in ANSI C
                 for the format conversions that are guaranteed
                 to be available.
 
       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 functions are available. They work by con-
       verting double-precision floating point values to
       uintmax_t integers, doing the operation, and then con-
       verting the result back to floating point. The func-
       tions 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 func-
       tions 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'' locations (e.g., dur-
              ing testing). It returns the directory where
              domain is ``bound.''
              The default domain is the value of TEXTDOMAIN.
              If directory is the null string (""), then bind-
              textdomain() 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 TEXTDO-
              MAIN. The default value for category is "LC_MES-
              SAGES".
              If you supply a value for category, it must be a
              string equal to one of the known locale cate-
              gories described in GAWK: Effective AWK Program-
              ming. You must also supply a text domain. Use
              TEXTDOMAIN if you want to use the current domain.
 
       dcngettext(string1 , string2 , number [, domain [, cate-
       gory]])
              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 TEXTDO-
              MAIN. The default value for category is "LC_MES-
              SAGES".
              If you supply a value for category, it must be a
              string equal to one of the known locale cate-
              gories described in GAWK: Effective AWK Program-
              ming. 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 func-
       tion. Arrays are passed by reference, other variables
       are passed by value.
 
       Since functions were not originally part of the AWK lan-
       guage, the provision 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 follow the function name, without any inter-
       vening white space. This avoids 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 parameters used as local variables are initial-
       ized to the null string and the number zero upon func-
       tion 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 functions 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 inter-
       preter. The full details are beyond the scope of this
       manual page; see GAWK: Effective AWK Programming 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: Effec-
       tive AWK Programming, but everything about this feature
       is likely to change eventually. 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 continues to run.
       SIGHUP causes pgawk 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 language.
       Such strings are marked in the AWK program with a lead-
       ing underscore ("_"). 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 run-
       ning a localizable AWK program.
 
       1. Add a BEGIN action to assign a value to the TEXTDO-
           MAIN variable to set the text domain to a name asso-
           ciated 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 underscores.
 
       3. If necessary, use the dcgettext() and/or bindtextdo-
           main() functions in your program, as appropriate.
 
       4. Run gawk --gen-po -f myprog.awk > myprog.po to gen-
           erate a .po file for your program.
 
       5. Provide appropriate translations, and build and
           install the corresponding .mo files.
 
       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 incorporates 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. How-
       ever, in earlier implementations, when such an assign-
       ment appeared before any file names, the assignment
       would happen before the BEGIN block was run. Applica-
       tions 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 otherwise 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 supports. 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 statements 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 spec-
       ified.
 
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 --tradi-
       tional or --posix options.
 
       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 :. (Dis-
         abled 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 split-
         ting.
 
       � 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 recognized.
 
       � 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 ability to pass an array to length().
 
       � The use of delete array to delete the entire contents
         of an array.
 
       � The use of nextfile to abandon processing of the cur-
         rent input file.
 
       � The and(), asort(), asorti(), bindtextdomain(),
         compl(), dcgettext(), dcngettext(), gensub(),
         lshift(), mktime(), or(), rshift(), strftime(), str-
         tonum(), systime() and xor() functions.
 
       � Localizable strings.
 
       � Adding new built-in functions dynamically with the
         extension() function.
 
       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 sta-
       tus 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 speci-
       fied. To really get a tab character as the field sepa-
       rator, 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 ]
              }
 
       If gawk is configured with the --disable-directories-
       fatal option, then it will silently skip directories
       named on the command line. Otherwise, it will do so
       only if invoked with the --traditional option.
 
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. The current ver-
       sion of this document is available online at
       http://www.gnu.org/software/gawk/manual.
 
BUGS
       The -F option is not necessary given the command line
       variable assignment feature; it remains only for back-
       wards 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 imple-
       mented by Alfred Aho, Peter Weinberger, and Brian
       Kernighan of Bell Laboratories. Brian Kernighan contin-
       ues to maintain and enhance it.
 
       Paul Rubin and Jay Fenlason, of the Free Software Foun-
       dation, 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 compatible with the new version of UNIX awk.
       Arnold Robbins is the current 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. Juan M. Guerrero now maintains the OS/2
       port. Fred Fish supplied support for the Amiga, and
       Martin Brown provided the BeOS port. Stephen Davies
       provided the original Tandem port, and Matthew Woehlke
       provided changes for Tandem's POSIX-compliant systems.
 
VERSION INFORMATION
       This man page documents gawk, version 3.1.6.
 
BUG REPORTS
       If you find a bug in gawk, please send electronic mail
       to bug-gawk@gnu.org. Please include your operating sys-
       tem 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 the following
       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 see if
       setting the environment variable LC_ALL to LC_ALL=C
       causes things to behave as you expect. If so, it's a
       locale issue, and may or may not really be a bug.
       Finally, 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
       maintainer.
 
ACKNOWLEDGEMENTS
       Brian Kernighan of Bell Laboratories provided valuable
       assistance during testing and debugging. We thank him.
 
COPYING PERMISSIONS
       Copyright � 1989, 1991, 1992, 1993, 1994, 1995, 1996,
       1997, 1998, 1999, 2001, 2002, 2003, 2004, 2005, 2007
       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 permis-
       sion notice identical to this one.
 
       Permission is granted to copy and distribute transla-
       tions of this manual page into another language, under
       the above conditions for modified versions, except that
       this permission notice may be stated in a translation
       approved by the Foundation.
 
 
 
Free Software Foundation Oct 19 2007 GAWK(1)