1GAWK(1) Utility Commands GAWK(1)
2
6 gawk - pattern scanning and processing language
7
9 gawk [ POSIX or GNU style options ] -f program-file [ -- ] file ...
10 gawk [ POSIX or GNU style options ] [ -- ] program-text file ...
11 pgawk [ POSIX or GNU style options ] -f program-file [ -- ] file ...
13 pgawk [ POSIX or GNU style options ] [ -- ] program-text file ...
14
16 Gawk is the GNU Project's implementation of the AWK programming lan‐
17 guage. It conforms to the definition of the language in the POSIX
18 1003.1 Standard. This version in turn is based on the description in
19 The AWK Programming Language, by Aho, Kernighan, and Weinberger, with
20 the additional features found in the System V Release 4 version of UNIX
21 awk. Gawk also provides more recent Bell Laboratories awk extensions,
22 and a number of GNU-specific extensions.
23 Pgawk is the profiling version of gawk. It is identical in every way
25 to gawk, except that programs run more slowly, and it automatically
26 produces an execution profile in the file awkprof.out when done. See
27 the --profile option, below.
28 The command line consists of options to gawk itself, the AWK program
30 text (if not supplied via the -f or --file options), and values to be
31 made available in the ARGC and ARGV pre-defined AWK variables.
32
34 Gawk options may be either traditional POSIX one letter options, or
35 GNU-style long options. POSIX options start with a single “-”, while
36 long options start with “--”. Long options are provided for both GNU-
37 specific features and for POSIX-mandated features.
38 Following the POSIX standard, gawk-specific options are supplied via
40 arguments to the -W option. Multiple -W options may be supplied Each
41 -W option has a corresponding long option, as detailed below. Argu‐
42 ments to long options are either joined with the option by an = sign,
43 with no intervening spaces, or they may be provided in the next command
44 line argument. Long options may be abbreviated, as long as the abbre‐
45 viation remains unique.
46
48 Gawk accepts the following options, listed by frequency.
49 -F fs
51 --field-separator fs
52 Use fs for the input field separator (the value of the FS prede‐
53 fined variable).
54 -v var=val
56 --assign var=val
57 Assign the value val to the variable var, before execution of
58 the program begins. Such variable values are available to the
59 BEGIN block of an AWK program.
60 -f program-file
62 --file program-file
63 Read the AWK program source from the file program-file, instead
64 of from the first command line argument. Multiple -f (or
65 --file) options may be used.
66 -mf NNN
68 -mr NNN
69 Set various memory limits to the value NNN. The f flag sets the
70 maximum number of fields, and the r flag sets the maximum record
71 size. These two flags and the -m option are from an earlier
72 version of the Bell Laboratories research version of UNIX awk.
73 They are ignored by gawk, since gawk has no pre-defined limits.
74 (Current versions of the Bell Laboratories awk no longer accept
75 them.)
76 -O
78 --optimize
79 Enable optimizations upon the internal representation of the
80 program. Currently, this includes just simple constant-folding.
81 The gawk maintainer hopes to add additional optimizations over
82 time.
83 -W compat
85 -W traditional
86 --compat
87 --traditional
88 Run in compatibility mode. In compatibility mode, gawk behaves
89 identically to UNIX awk; none of the GNU-specific extensions are
90 recognized. The use of --traditional is preferred over the
91 other forms of this option. See GNU EXTENSIONS, below, for more
92 information.
93 -W copyleft
95 -W copyright
96 --copyleft
97 --copyright
98 Print the short version of the GNU copyright information message
99 on the standard output and exit successfully.
100 -W dump-variables[=file]
102 --dump-variables[=file]
103 Print a sorted list of global variables, their types and final
104 values to file. If no file is provided, gawk uses a file named
105 awkvars.out in the current directory.
106 Having a list of all the global variables is a good way to look
107 for typographical errors in your programs. You would also use
108 this option if you have a large program with a lot of functions,
109 and you want to be sure that your functions don't inadvertently
110 use global variables that you meant to be local. (This is a
111 particularly easy mistake to make with simple variable names
112 like i, j, and so on.)
113 -W exec file
115 --exec file
116 Similar to -f, however, this is option is the last one pro‐
117 cessed. This should be used with #! scripts, particularly for
118 CGI applications, to avoid passing in options or source code (!)
119 on the command line from a URL. This option disables command-
120 line variable assignments.
121 -W gen-po
123 --gen-po
124 Scan and parse the AWK program, and generate a GNU .po format
125 file on standard output with entries for all localizable strings
126 in the program. The program itself is not executed. See the
127 GNU gettext distribution for more information on .po files.
128 -W help
130 -W usage
131 --help
132 --usage
133 Print a relatively short summary of the available options on the
134 standard output. (Per the GNU Coding Standards, these options
135 cause an immediate, successful exit.)
136 -W lint[=value]
138 --lint[=value]
139 Provide warnings about constructs that are dubious or non-porta‐
140 ble to other AWK implementations. With an optional argument of
141 fatal, lint warnings become fatal errors. This may be drastic,
142 but its use will certainly encourage the development of cleaner
143 AWK programs. With an optional argument of invalid, only warn‐
144 ings about things that are actually invalid are issued. (This is
145 not fully implemented yet.)
146 -W lint-old
148 --lint-old
149 Provide warnings about constructs that are not portable to the
150 original version of Unix awk.
151 -W non-decimal-data
153 --non-decimal-data
154 Recognize octal and hexadecimal values in input data. Use this
155 option with great caution!
156 -W posix
158 --posix
159 This turns on compatibility mode, with the following additional
160 restrictions:
161 · \x escape sequences are not recognized.
163 · Only space and tab act as field separators when FS is set to a
165 single space, newline does not.
166 · You cannot continue lines after ? and :.
168 · The synonym func for the keyword function is not recognized.
170 · The operators ** and **= cannot be used in place of ^ and ^=.
172 · The fflush() function is not available.
174 -W profile[=prof_file]
176 --profile[=prof_file]
177 Send profiling data to prof_file. The default is awkprof.out.
178 When run with gawk, the profile is just a “pretty printed” ver‐
179 sion of the program. When run with pgawk, the profile contains
180 execution counts of each statement in the program in the left
181 margin and function call counts for each user-defined function.
182 -W re-interval
184 --re-interval
185 Enable the use of interval expressions in regular expression
186 matching (see Regular Expressions, below). Interval expressions
187 were not traditionally available in the AWK language. The POSIX
188 standard added them, to make awk and egrep consistent with each
189 other. However, their use is likely to break old AWK programs,
190 so gawk only provides them if they are requested with this
191 option, or when --posix is specified.
192 -W source program-text
194 --source program-text
195 Use program-text as AWK program source code. This option allows
196 the easy intermixing of library functions (used via the -f and
197 --file options) with source code entered on the command line.
198 It is intended primarily for medium to large AWK programs used
199 in shell scripts.
200 -W use-lc-numeric
202 --use-lc-numeric
203 This forces gawk to use the locale's decimal point character
204 when parsing input data. Although the POSIX standard requires
205 this behavior, and gawk does so when --posix is in effect, the
206 default is to follow traditional behavior and use a period as
207 the decimal point, even in locales where the period is not the
208 decimal point character. This option overrides the default
209 behavior, without the full draconian strictness of the --posix
210 option.
211 -W version
213 --version
214 Print version information for this particular copy of gawk on
215 the standard output. This is useful mainly for knowing if the
216 current copy of gawk on your system is up to date with respect
217 to whatever the Free Software Foundation is distributing. This
218 is also useful when reporting bugs. (Per the GNU Coding Stan‐
219 dards, these options cause an immediate, successful exit.)
220 -- Signal the end of options. This is useful to allow further argu‐
222 ments to the AWK program itself to start with a “-”. This pro‐
223 vides consistency with the argument parsing convention used by
224 most other POSIX programs.
225 In compatibility mode, any other options are flagged as invalid, but
227 are otherwise ignored. In normal operation, as long as program text
228 has been supplied, unknown options are passed on to the AWK program in
229 the ARGV array for processing. This is particularly useful for running
230 AWK programs via the “#!” executable interpreter mechanism.
231
233 An AWK program consists of a sequence of pattern-action statements and
234 optional function definitions.
235 pattern { action statements }
237 function name(parameter list) { statements }
238 Gawk first reads the program source from the program-file(s) if speci‐
240 fied, from arguments to --source, or from the first non-option argument
241 on the command line. The -f and --source options may be used multiple
242 times on the command line. Gawk reads the program text as if all the
243 program-files and command line source texts had been concatenated
244 together. This is useful for building libraries of AWK functions,
245 without having to include them in each new AWK program that uses them.
246 It also provides the ability to mix library functions with command line
247 programs.
248 The environment variable AWKPATH specifies a search path to use when
250 finding source files named with the -f option. If this variable does
251 not exist, the default path is ".:/usr/local/share/awk". (The actual
252 directory may vary, depending upon how gawk was built and installed.)
253 If a file name given to the -f option contains a “/” character, no path
254 search is performed.
255 Gawk executes AWK programs in the following order. First, all variable
257 assignments specified via the -v option are performed. Next, gawk com‐
258 piles the program into an internal form. Then, gawk executes the code
259 in the BEGIN block(s) (if any), and then proceeds to read each file
260 named in the ARGV array. If there are no files named on the command
261 line, gawk reads the standard input.
262 If a filename on the command line has the form var=val it is treated as
264 a variable assignment. The variable var will be assigned the value
265 val. (This happens after any BEGIN block(s) have been run.) Command
266 line variable assignment is most useful for dynamically assigning val‐
267 ues to the variables AWK uses to control how input is broken into
268 fields and records. It is also useful for controlling state if multi‐
269 ple passes are needed over a single data file.
270 If the value of a particular element of ARGV is empty (""), gawk skips
272 over it.
273 For each record in the input, gawk tests to see if it matches any pat‐
275 tern in the AWK program. For each pattern that the record matches, the
276 associated action is executed. The patterns are tested in the order
277 they occur in the program.
278 Finally, after all the input is exhausted, gawk executes the code in
280 the END block(s) (if any).
281
283 AWK variables are dynamic; they come into existence when they are first
284 used. Their values are either floating-point numbers or strings, or
285 both, depending upon how they are used. AWK also has one dimensional
286 arrays; arrays with multiple dimensions may be simulated. Several pre-
287 defined variables are set as a program runs; these are described as
288 needed and summarized below.
289 Records
291 Normally, records are separated by newline characters. You can control
292 how records are separated by assigning values to the built-in variable
293 RS. If RS is any single character, that character separates records.
294 Otherwise, RS is a regular expression. Text in the input that matches
295 this regular expression separates the record. However, in compatibil‐
296 ity mode, only the first character of its string value is used for sep‐
297 arating records. If RS is set to the null string, then records are
298 separated by blank lines. When RS is set to the null string, the new‐
299 line character always acts as a field separator, in addition to what‐
300 ever value FS may have.
301 Fields
303 As each input record is read, gawk splits the record into fields, using
304 the value of the FS variable as the field separator. If FS is a single
305 character, fields are separated by that character. If FS is the null
306 string, then each individual character becomes a separate field. Oth‐
307 erwise, FS is expected to be a full regular expression. In the special
308 case that FS is a single space, fields are separated by runs of spaces
309 and/or tabs and/or newlines. (But see the section POSIX COMPATIBILITY,
310 below). NOTE: The value of IGNORECASE (see below) also affects how
311 fields are split when FS is a regular expression, and how records are
312 separated when RS is a regular expression.
313 If the FIELDWIDTHS variable is set to a space separated list of num‐
315 bers, each field is expected to have fixed width, and gawk splits up
316 the record using the specified widths. The value of FS is ignored.
317 Assigning a new value to FS overrides the use of FIELDWIDTHS, and
318 restores the default behavior.
319 Each field in the input record may be referenced by its position, $1,
321 $2, and so on. $0 is the whole record. Fields need not be referenced
322 by constants:
323 n = 5
325 print $n
326 prints the fifth field in the input record.
328 The variable NF is set to the total number of fields in the input
330 record.
331 References to non-existent fields (i.e. fields after $NF) produce the
333 null-string. However, assigning to a non-existent field (e.g., $(NF+2)
334 = 5) increases the value of NF, creates any intervening fields with the
335 null string as their value, and causes the value of $0 to be recom‐
336 puted, with the fields being separated by the value of OFS. References
337 to negative numbered fields cause a fatal error. Decrementing NF
338 causes the values of fields past the new value to be lost, and the
339 value of $0 to be recomputed, with the fields being separated by the
340 value of OFS.
341 Assigning a value to an existing field causes the whole record to be
343 rebuilt when $0 is referenced. Similarly, assigning a value to $0
344 causes the record to be resplit, creating new values for the fields.
345 Built-in Variables
347 Gawk's built-in variables are:
348 ARGC The number of command line arguments (does not include
350 options to gawk, or the program source).
351 ARGIND The index in ARGV of the current file being processed.
353 ARGV Array of command line arguments. The array is indexed from
355 0 to ARGC - 1. Dynamically changing the contents of ARGV
356 can control the files used for data.
357 BINMODE On non-POSIX systems, specifies use of “binary” mode for
359 all file I/O. Numeric values of 1, 2, or 3, specify that
360 input files, output files, or all files, respectively,
361 should use binary I/O. String values of "r", or "w" spec‐
362 ify that input files, or output files, respectively, should
363 use binary I/O. String values of "rw" or "wr" specify that
364 all files should use binary I/O. Any other string value is
365 treated as "rw", but generates a warning message.
366 CONVFMT The conversion format for numbers, "%.6g", by default.
368 ENVIRON An array containing the values of the current environment.
370 The array is indexed by the environment variables, each
371 element being the value of that variable (e.g., ENVI‐
372 RON["HOME"] might be /home/arnold). Changing this array
373 does not affect the environment seen by programs which gawk
374 spawns via redirection or the system() function.
375 ERRNO If a system error occurs either doing a redirection for
377 getline, during a read for getline, or during a close(),
378 then ERRNO will contain a string describing the error. The
379 value is subject to translation in non-English locales.
380 FIELDWIDTHS A white-space separated list of fieldwidths. When set,
382 gawk parses the input into fields of fixed width, instead
383 of using the value of the FS variable as the field separa‐
384 tor.
385 FILENAME The name of the current input file. If no files are speci‐
387 fied on the command line, the value of FILENAME is “-”.
388 However, FILENAME is undefined inside the BEGIN block
389 (unless set by getline).
390 FNR The input record number in the current input file.
392 FS The input field separator, a space by default. See Fields,
394 above.
395 IGNORECASE Controls the case-sensitivity of all regular expression and
397 string operations. If IGNORECASE has a non-zero value,
398 then string comparisons and pattern matching in rules,
399 field splitting with FS, record separating with RS, regular
400 expression matching with ~ and !~, and the gensub(),
401 gsub(), index(), match(), split(), and sub() built-in func‐
402 tions all ignore case when doing regular expression opera‐
403 tions. NOTE: Array subscripting is not affected. However,
404 the asort() and asorti() functions are affected.
405 Thus, if IGNORECASE is not equal to zero, /aB/ matches all
406 of the strings "ab", "aB", "Ab", and "AB". As with all AWK
407 variables, the initial value of IGNORECASE is zero, so all
408 regular expression and string operations are normally case-
409 sensitive. Under Unix, the full ISO 8859-1 Latin-1 charac‐
410 ter set is used when ignoring case. As of gawk 3.1.4, the
411 case equivalencies are fully locale-aware, based on the C
412 <ctype.h> facilities such as isalpha(), and toupper().
413 LINT Provides dynamic control of the --lint option from within
415 an AWK program. When true, gawk prints lint warnings. When
416 false, it does not. When assigned the string value
417 "fatal", lint warnings become fatal errors, exactly like
418 --lint=fatal. Any other true value just prints warnings.
419 NF The number of fields in the current input record.
421 NR The total number of input records seen so far.
423 OFMT The output format for numbers, "%.6g", by default.
425 OFS The output field separator, a space by default.
427 ORS The output record separator, by default a newline.
429 PROCINFO The elements of this array provide access to information
431 about the running AWK program. On some systems, there may
432 be elements in the array, "group1" through "groupn" for
433 some n, which is the number of supplementary groups that
434 the process has. Use the in operator to test for these
435 elements. The following elements are guaranteed to be
436 available:
437 PROCINFO["egid"] the value of the getegid(2) system
439 call.
440 PROCINFO["euid"] the value of the geteuid(2) system
442 call.
443 PROCINFO["FS"] "FS" if field splitting with FS is in
445 effect, or "FIELDWIDTHS" if field
446 splitting with FIELDWIDTHS is in
447 effect.
448 PROCINFO["gid"] the value of the getgid(2) system call.
450 PROCINFO["pgrpid"] the process group ID of the current
452 process.
453 PROCINFO["pid"] the process ID of the current process.
455 PROCINFO["ppid"] the parent process ID of the current
457 process.
458 PROCINFO["uid"] the value of the getuid(2) system call.
460 PROCINFO["version"] the version of gawk. This is available
462 from version 3.1.4 and later.
463 RS The input record separator, by default a newline.
465 RT The record terminator. Gawk sets RT to the input text that
467 matched the character or regular expression specified by
468 RS.
469 RSTART The index of the first character matched by match(); 0 if
471 no match. (This implies that character indices start at
472 one.)
473 RLENGTH The length of the string matched by match(); -1 if no
475 match.
476 SUBSEP The character used to separate multiple subscripts in array
478 elements, by default "\034".
479 TEXTDOMAIN The text domain of the AWK program; used to find the local‐
481 ized translations for the program's strings.
482 Arrays
484 Arrays are subscripted with an expression between square brackets ([
485 and ]). If the expression is an expression list (expr, expr ...) then
486 the array subscript is a string consisting of the concatenation of the
487 (string) value of each expression, separated by the value of the SUBSEP
488 variable. This facility is used to simulate multiply dimensioned
489 arrays. For example:
490 i = "A"; j = "B"; k = "C"
492 x[i, j, k] = "hello, world\n"
493 assigns the string "hello, world\n" to the element of the array x which
495 is indexed by the string "A\034B\034C". All arrays in AWK are associa‐
496 tive, i.e. indexed by string values.
497 The special operator in may be used to test if an array has an index
499 consisting of a particular value.
500 if (val in array)
502 print array[val]
503 If the array has multiple subscripts, use (i, j) in array.
505 The in construct may also be used in a for loop to iterate over all the
507 elements of an array.
508 An element may be deleted from an array using the delete statement.
510 The delete statement may also be used to delete the entire contents of
511 an array, just by specifying the array name without a subscript.
512 Variable Typing And Conversion
514 Variables and fields may be (floating point) numbers, or strings, or
515 both. How the value of a variable is interpreted depends upon its con‐
516 text. If used in a numeric expression, it will be treated as a number;
517 if used as a string it will be treated as a string.
518 To force a variable to be treated as a number, add 0 to it; to force it
520 to be treated as a string, concatenate it with the null string.
521 When a string must be converted to a number, the conversion is accom‐
523 plished using strtod(3). A number is converted to a string by using
524 the value of CONVFMT as a format string for sprintf(3), with the
525 numeric value of the variable as the argument. However, even though
526 all numbers in AWK are floating-point, integral values are always con‐
527 verted as integers. Thus, given
528 CONVFMT = "%2.2f"
530 a = 12
531 b = a ""
532 the variable b has a string value of "12" and not "12.00".
534 When operating in POSIX mode (such as with the --posix command line
536 option), beware that locale settings may interfere with the way decimal
537 numbers are treated: the decimal separator of the numbers you are feed‐
538 ing to gawk must conform to what your locale would expect, be it a
539 comma (,) or a period (.).
540 Gawk performs comparisons as follows: If two variables are numeric,
542 they are compared numerically. If one value is numeric and the other
543 has a string value that is a “numeric string,” then comparisons are
544 also done numerically. Otherwise, the numeric value is converted to a
545 string and a string comparison is performed. Two strings are compared,
546 of course, as strings.
547 Note that string constants, such as "57", are not numeric strings, they
549 are string constants. The idea of “numeric string” only applies to
550 fields, getline input, FILENAME, ARGV elements, ENVIRON elements and
551 the elements of an array created by split() that are numeric strings.
552 The basic idea is that user input, and only user input, that looks
553 numeric, should be treated that way.
554 Uninitialized variables have the numeric value 0 and the string value
556 "" (the null, or empty, string).
557 Octal and Hexadecimal Constants
559 Starting with version 3.1 of gawk , you may use C-style octal and hexa‐
560 decimal constants in your AWK program source code. For example, the
561 octal value 011 is equal to decimal 9, and the hexadecimal value 0x11
562 is equal to decimal 17.
563 String Constants
565 String constants in AWK are sequences of characters enclosed between
566 double quotes ("). Within strings, certain escape sequences are recog‐
567 nized, as in C. These are:
568 \\ A literal backslash.
570 \a The “alert” character; usually the ASCII BEL character.
572 \b backspace.
574 \f form-feed.
576 \n newline.
578 \r carriage return.
580 \t horizontal tab.
582 \v vertical tab.
584 \xhex digits
586 The character represented by the string of hexadecimal digits fol‐
587 lowing the \x. As in ANSI C, all following hexadecimal digits are
588 considered part of the escape sequence. (This feature should tell
589 us something about language design by committee.) E.g., "\x1B" is
590 the ASCII ESC (escape) character.
591 \ddd The character represented by the 1-, 2-, or 3-digit sequence of
593 octal digits. E.g., "\033" is the ASCII ESC (escape) character.
594 \c The literal character c.
596 The escape sequences may also be used inside constant regular expres‐
598 sions (e.g., /[ \t\f\n\r\v]/ matches whitespace characters).
599 In compatibility mode, the characters represented by octal and hexadec‐
601 imal escape sequences are treated literally when used in regular
602 expression constants. Thus, /a\52b/ is equivalent to /a\*b/.
603
605 AWK is a line-oriented language. The pattern comes first, and then the
606 action. Action statements are enclosed in { and }. Either the pattern
607 may be missing, or the action may be missing, but, of course, not both.
608 If the pattern is missing, the action is executed for every single
609 record of input. A missing action is equivalent to
610 { print }
612 which prints the entire record.
614 Comments begin with the “#” character, and continue until the end of
616 the line. Blank lines may be used to separate statements. Normally, a
617 statement ends with a newline, however, this is not the case for lines
618 ending in a “,”, {, ?, :, &&, or ||. Lines ending in do or else also
619 have their statements automatically continued on the following line.
620 In other cases, a line can be continued by ending it with a “\”, in
621 which case the newline will be ignored.
622 Multiple statements may be put on one line by separating them with a
624 “;”. This applies to both the statements within the action part of a
625 pattern-action pair (the usual case), and to the pattern-action state‐
626 ments themselves.
627 Patterns
629 AWK patterns may be one of the following:
630 BEGIN
632 END
633 /regular expression/
634 relational expression
635 pattern && pattern
636 pattern || pattern
637 pattern ? pattern : pattern
638 (pattern)
639 ! pattern
640 pattern1, pattern2
641 BEGIN and END are two special kinds of patterns which are not tested
643 against the input. The action parts of all BEGIN patterns are merged
644 as if all the statements had been written in a single BEGIN block.
645 They are executed before any of the input is read. Similarly, all the
646 END blocks are merged, and executed when all the input is exhausted (or
647 when an exit statement is executed). BEGIN and END patterns cannot be
648 combined with other patterns in pattern expressions. BEGIN and END
649 patterns cannot have missing action parts.
650 For /regular expression/ patterns, the associated statement is executed
652 for each input record that matches the regular expression. Regular
653 expressions are the same as those in egrep(1), and are summarized
654 below.
655 A relational expression may use any of the operators defined below in
657 the section on actions. These generally test whether certain fields
658 match certain regular expressions.
659 The &&, ||, and ! operators are logical AND, logical OR, and logical
661 NOT, respectively, as in C. They do short-circuit evaluation, also as
662 in C, and are used for combining more primitive pattern expressions.
663 As in most languages, parentheses may be used to change the order of
664 evaluation.
665 The ?: operator is like the same operator in C. If the first pattern
667 is true then the pattern used for testing is the second pattern, other‐
668 wise it is the third. Only one of the second and third patterns is
669 evaluated.
670 The pattern1, pattern2 form of an expression is called a range pattern.
672 It matches all input records starting with a record that matches pat‐
673 tern1, and continuing until a record that matches pattern2, inclusive.
674 It does not combine with any other sort of pattern expression.
675 Regular Expressions
677 Regular expressions are the extended kind found in egrep. They are
678 composed of characters as follows:
679 c matches the non-metacharacter c.
681 \c matches the literal character c.
683 . matches any character including newline.
685 ^ matches the beginning of a string.
687 $ matches the end of a string.
689 [abc...] character list, matches any of the characters abc....
691 [^abc...] negated character list, matches any character except abc....
693 r1|r2 alternation: matches either r1 or r2.
695 r1r2 concatenation: matches r1, and then r2.
697 r+ matches one or more r's.
699 r* matches zero or more r's.
701 r? matches zero or one r's.
703 (r) grouping: matches r.
705 r{n}
707 r{n,}
708 r{n,m} One or two numbers inside braces denote an interval expres‐
709 sion. If there is one number in the braces, the preceding
710 regular expression r is repeated n times. If there are two
711 numbers separated by a comma, r is repeated n to m times.
712 If there is one number followed by a comma, then r is
713 repeated at least n times.
714 Interval expressions are only available if either --posix or
715 --re-interval is specified on the command line.
716 \y matches the empty string at either the beginning or the end
718 of a word.
719 \B matches the empty string within a word.
721 \< matches the empty string at the beginning of a word.
723 \> matches the empty string at the end of a word.
725 \w matches any word-constituent character (letter, digit, or
727 underscore).
728 \W matches any character that is not word-constituent.
730 \` matches the empty string at the beginning of a buffer
732 (string).
733 \' matches the empty string at the end of a buffer.
735 The escape sequences that are valid in string constants (see below) are
737 also valid in regular expressions.
738 Character classes are a feature introduced in the POSIX standard. A
740 character class is a special notation for describing lists of charac‐
741 ters that have a specific attribute, but where the actual characters
742 themselves can vary from country to country and/or from character set
743 to character set. For example, the notion of what is an alphabetic
744 character differs in the USA and in France.
745 A character class is only valid in a regular expression inside the
747 brackets of a character list. Character classes consist of [:, a key‐
748 word denoting the class, and :]. The character classes defined by the
749 POSIX standard are:
750 [:alnum:] Alphanumeric characters.
752 [:alpha:] Alphabetic characters.
754 [:blank:] Space or tab characters.
756 [:cntrl:] Control characters.
758 [:digit:] Numeric characters.
760 [:graph:] Characters that are both printable and visible. (A space is
762 printable, but not visible, while an a is both.)
763 [:lower:] Lower-case alphabetic characters.
765 [:print:] Printable characters (characters that are not control char‐
767 acters.)
768 [:punct:] Punctuation characters (characters that are not letter, dig‐
770 its, control characters, or space characters).
771 [:space:] Space characters (such as space, tab, and formfeed, to name
773 a few).
774 [:upper:] Upper-case alphabetic characters.
776 [:xdigit:] Characters that are hexadecimal digits.
778 For example, before the POSIX standard, to match alphanumeric charac‐
780 ters, you would have had to write /[A-Za-z0-9]/. If your character set
781 had other alphabetic characters in it, this would not match them, and
782 if your character set collated differently from ASCII, this might not
783 even match the ASCII alphanumeric characters. With the POSIX character
784 classes, you can write /[[:alnum:]]/, and this matches the alphabetic
785 and numeric characters in your character set, no matter what it is.
786 Two additional special sequences can appear in character lists. These
788 apply to non-ASCII character sets, which can have single symbols
789 (called collating elements) that are represented with more than one
790 character, as well as several characters that are equivalent for col‐
791 lating, or sorting, purposes. (E.g., in French, a plain “e” and a
792 grave-accented “e`” are equivalent.)
793 Collating Symbols
795 A collating symbol is a multi-character collating element
796 enclosed in [. and .]. For example, if ch is a collating ele‐
797 ment, then [[.ch.]] is a regular expression that matches this
798 collating element, while [ch] is a regular expression that
799 matches either c or h.
800 Equivalence Classes
802 An equivalence class is a locale-specific name for a list of
803 characters that are equivalent. The name is enclosed in [= and
804 =]. For example, the name e might be used to represent all of
805 “e,” “e´,” and “e`.” In this case, [[=e=]] is a regular expres‐
806 sion that matches any of e, e´, or e`.
807 These features are very valuable in non-English speaking locales. The
809 library functions that gawk uses for regular expression matching cur‐
810 rently only recognize POSIX character classes; they do not recognize
811 collating symbols or equivalence classes.
812 The \y, \B, \<, \>, \w, \W, \`, and \' operators are specific to gawk;
814 they are extensions based on facilities in the GNU regular expression
815 libraries.
816 The various command line options control how gawk interprets characters
818 in regular expressions.
819 No options
821 In the default case, gawk provide all the facilities of POSIX
822 regular expressions and the GNU regular expression operators
823 described above. However, interval expressions are not sup‐
824 ported.
825 --posix
827 Only POSIX regular expressions are supported, the GNU operators
828 are not special. (E.g., \w matches a literal w). Interval
829 expressions are allowed.
830 --traditional
832 Traditional Unix awk regular expressions are matched. The GNU
833 operators are not special, interval expressions are not avail‐
834 able, and neither are the POSIX character classes ([[:alnum:]]
835 and so on). Characters described by octal and hexadecimal
836 escape sequences are treated literally, even if they represent
837 regular expression metacharacters.
838 --re-interval
840 Allow interval expressions in regular expressions, even if
841 --traditional has been provided.
842 Actions
844 Action statements are enclosed in braces, { and }. Action statements
845 consist of the usual assignment, conditional, and looping statements
846 found in most languages. The operators, control statements, and
847 input/output statements available are patterned after those in C.
848 Operators
850 The operators in AWK, in order of decreasing precedence, are
851 (...) Grouping
853 $ Field reference.
855 ++ -- Increment and decrement, both prefix and postfix.
857 ^ Exponentiation (** may also be used, and **= for the
859 assignment operator).
860 + - ! Unary plus, unary minus, and logical negation.
862 * / % Multiplication, division, and modulus.
864 + - Addition and subtraction.
866 space String concatenation.
868 | |& Piped I/O for getline, print, and printf.
870 < > <= >= != ==
872 The regular relational operators.
873 ~ !~ Regular expression match, negated match. NOTE: Do not use
875 a constant regular expression (/foo/) on the left-hand side
876 of a ~ or !~. Only use one on the right-hand side. The
877 expression /foo/ ~ exp has the same meaning as (($0 ~
878 /foo/) ~ exp). This is usually not what was intended.
879 in Array membership.
881 && Logical AND.
883 || Logical OR.
885 ?: The C conditional expression. This has the form expr1 ?
887 expr2 : expr3. If expr1 is true, the value of the expres‐
888 sion is expr2, otherwise it is expr3. Only one of expr2
889 and expr3 is evaluated.
890 = += -= *= /= %= ^=
892 Assignment. Both absolute assignment (var = value) and
893 operator-assignment (the other forms) are supported.
894 Control Statements
896 The control statements are as follows:
897 if (condition) statement [ else statement ]
899 while (condition) statement
900 do statement while (condition)
901 for (expr1; expr2; expr3) statement
902 for (var in array) statement
903 break
904 continue
905 delete array[index]
906 delete array
907 exit [ expression ]
908 { statements }
909 I/O Statements
911 The input/output statements are as follows:
912 close(file [, how]) Close file, pipe or co-process. The optional how
914 should only be used when closing one end of a
915 two-way pipe to a co-process. It must be a
916 string value, either "to" or "from".
917 getline Set $0 from next input record; set NF, NR, FNR.
919 getline <file Set $0 from next record of file; set NF.
921 getline var Set var from next input record; set NR, FNR.
923 getline var <file Set var from next record of file.
925 command | getline [var]
927 Run command piping the output either into $0 or
928 var, as above.
929 command |& getline [var]
931 Run command as a co-process piping the output
932 either into $0 or var, as above. Co-processes
933 are a gawk extension. (command can also be a
934 socket. See the subsection Special File Names,
935 below.)
936 next Stop processing the current input record. The
938 next input record is read and processing starts
939 over with the first pattern in the AWK program.
940 If the end of the input data is reached, the END
941 block(s), if any, are executed.
942 nextfile Stop processing the current input file. The next
944 input record read comes from the next input file.
945 FILENAME and ARGIND are updated, FNR is reset to
946 1, and processing starts over with the first pat‐
947 tern in the AWK program. If the end of the input
948 data is reached, the END block(s), if any, are
949 executed.
950 print Prints the current record. The output record is
952 terminated with the value of the ORS variable.
953 print expr-list Prints expressions. Each expression is separated
955 by the value of the OFS variable. The output
956 record is terminated with the value of the ORS
957 variable.
958 print expr-list >file Prints expressions on file. Each expression is
960 separated by the value of the OFS variable. The
961 output record is terminated with the value of the
962 ORS variable.
963 printf fmt, expr-list Format and print.
965 printf fmt, expr-list >file
967 Format and print on file.
968 system(cmd-line) Execute the command cmd-line, and return the exit
970 status. (This may not be available on non-POSIX
971 systems.)
972 fflush([file]) Flush any buffers associated with the open output
974 file or pipe file. If file is missing, then
975 standard output is flushed. If file is the null
976 string, then all open output files and pipes have
977 their buffers flushed.
978 Additional output redirections are allowed for print and printf.
980 print ... >> file
982 Appends output to the file.
983 print ... | command
985 Writes on a pipe.
986 print ... |& command
988 Sends data to a co-process or socket. (See also the subsection
989 Special File Names, below.)
990 The getline command returns 1 on success, 0 on end of file, and -1 on
992 an error. Upon an error, ERRNO contains a string describing the prob‐
993 lem.
994 NOTE: Failure in opening a two-way socket will result in a non-fatal
996 error being returned to the calling function. If using a pipe, co-
997 process, or socket to getline, or from print or printf within a loop,
998 you must use close() to create new instances of the command or socket.
999 AWK does not automatically close pipes, sockets, or co-processes when
1000 they return EOF.
1001 The printf Statement
1003 The AWK versions of the printf statement and sprintf() function (see
1004 below) accept the following conversion specification formats:
1005 %c An ASCII character. If the argument used for %c is numeric, it
1007 is treated as a character and printed. Otherwise, the argument
1008 is assumed to be a string, and the only first character of that
1009 string is printed.
1010 %d, %i A decimal number (the integer part).
1012 %e, %E A floating point number of the form [-]d.dddddde[+-]dd. The %E
1014 format uses E instead of e.
1015 %f, %F A floating point number of the form [-]ddd.dddddd. If the sys‐
1017 tem library supports it, %F is available as well. This is like
1018 %f, but uses capital letters for special “not a number” and
1019 “infinity” values. If %F is not available, gawk uses %f.
1020 %g, %G Use %e or %f conversion, whichever is shorter, with nonsignifi‐
1022 cant zeros suppressed. The %G format uses %E instead of %e.
1023 %o An unsigned octal number (also an integer).
1025 %u An unsigned decimal number (again, an integer).
1027 %s A character string.
1029 %x, %X An unsigned hexadecimal number (an integer). The %X format
1031 uses ABCDEF instead of abcdef.
1032 %% A single % character; no argument is converted.
1034 Optional, additional parameters may lie between the % and the control
1036 letter:
1037 count$ Use the count'th argument at this point in the formatting. This
1039 is called a positional specifier and is intended primarily for
1040 use in translated versions of format strings, not in the origi‐
1041 nal text of an AWK program. It is a gawk extension.
1042 - The expression should be left-justified within its field.
1044 space For numeric conversions, prefix positive values with a space,
1046 and negative values with a minus sign.
1047 + The plus sign, used before the width modifier (see below), says
1049 to always supply a sign for numeric conversions, even if the
1050 data to be formatted is positive. The + overrides the space
1051 modifier.
1052 # Use an “alternate form” for certain control letters. For %o,
1054 supply a leading zero. For %x, and %X, supply a leading 0x or
1055 0X for a nonzero result. For %e, %E, %f and %F, the result
1056 always contains a decimal point. For %g, and %G, trailing zeros
1057 are not removed from the result.
1058 0 A leading 0 (zero) acts as a flag, that indicates output should
1060 be padded with zeroes instead of spaces. This applies only to
1061 the numeric output formats. This flag only has an effect when
1062 the field width is wider than the value to be printed.
1063 width The field should be padded to this width. The field is normally
1065 padded with spaces. If the 0 flag has been used, it is padded
1066 with zeroes.
1067 .prec A number that specifies the precision to use when printing. For
1069 the %e, %E, %f and %F, formats, this specifies the number of
1070 digits you want printed to the right of the decimal point. For
1071 the %g, and %G formats, it specifies the maximum number of sig‐
1072 nificant digits. For the %d, %o, %i, %u, %x, and %X formats, it
1073 specifies the minimum number of digits to print. For %s, it
1074 specifies the maximum number of characters from the string that
1075 should be printed.
1076 The dynamic width and prec capabilities of the ANSI C printf() routines
1078 are supported. A * in place of either the width or prec specifications
1079 causes their values to be taken from the argument list to printf or
1080 sprintf(). To use a positional specifier with a dynamic width or pre‐
1081 cision, supply the count$ after the * in the format string. For exam‐
1082 ple, "%3$*2$.*1$s".
1083 Special File Names
1085 When doing I/O redirection from either print or printf into a file, or
1086 via getline from a file, gawk recognizes certain special filenames
1087 internally. These filenames allow access to open file descriptors
1088 inherited from gawk's parent process (usually the shell). These file
1089 names may also be used on the command line to name data files. The
1090 filenames are:
1091 /dev/stdin The standard input.
1093 /dev/stdout The standard output.
1095 /dev/stderr The standard error output.
1097 /dev/fd/n The file associated with the open file descriptor n.
1099 These are particularly useful for error messages. For example:
1101 print "You blew it!" > "/dev/stderr"
1103 whereas you would otherwise have to use
1105 print "You blew it!" | "cat 1>&2"
1107 The following special filenames may be used with the |& co-process
1109 operator for creating TCP/IP network connections.
1110 /inet/tcp/lport/rhost/rport File for TCP/IP connection on local port
1112 lport to remote host rhost on remote port
1113 rport. Use a port of 0 to have the system
1114 pick a port.
1115 /inet/udp/lport/rhost/rport Similar, but use UDP/IP instead of TCP/IP.
1117 /inet/raw/lport/rhost/rport Reserved for future use.
1119 Other special filenames provide access to information about the running
1121 gawk process. These filenames are now obsolete. Use the PROCINFO
1122 array to obtain the information they provide. The filenames are:
1123 /dev/pid Reading this file returns the process ID of the current
1125 process, in decimal, terminated with a newline.
1126 /dev/ppid Reading this file returns the parent process ID of the cur‐
1128 rent process, in decimal, terminated with a newline.
1129 /dev/pgrpid Reading this file returns the process group ID of the cur‐
1131 rent process, in decimal, terminated with a newline.
1132 /dev/user Reading this file returns a single record terminated with a
1134 newline. The fields are separated with spaces. $1 is the
1135 value of the getuid(2) system call, $2 is the value of the
1136 geteuid(2) system call, $3 is the value of the getgid(2)
1137 system call, and $4 is the value of the getegid(2) system
1138 call. If there are any additional fields, they are the
1139 group IDs returned by getgroups(2). Multiple groups may
1140 not be supported on all systems.
1141 Numeric Functions
1143 AWK has the following built-in arithmetic functions:
1144 atan2(y, x) Returns the arctangent of y/x in radians.
1146 cos(expr) Returns the cosine of expr, which is in radians.
1148 exp(expr) The exponential function.
1150 int(expr) Truncates to integer.
1152 log(expr) The natural logarithm function.
1154 rand() Returns a random number N, between 0 and 1, such that 0 ≤
1156 N < 1.
1157 sin(expr) Returns the sine of expr, which is in radians.
1159 sqrt(expr) The square root function.
1161 srand([expr]) Uses expr as a new seed for the random number generator.
1163 If no expr is provided, the time of day is used. The
1164 return value is the previous seed for the random number
1165 generator.
1166 String Functions
1168 Gawk has the following built-in string functions:
1169 asort(s [, d]) Returns the number of elements in the source
1171 array s. The contents of s are sorted using
1172 gawk's normal rules for comparing values, and
1173 the indices of the sorted values of s are
1174 replaced with sequential integers starting with
1175 1. If the optional destination array d is spec‐
1176 ified, then s is first duplicated into d, and
1177 then d is sorted, leaving the indices of the
1178 source array s unchanged.
1179 asorti(s [, d]) Returns the number of elements in the source
1181 array s. The behavior is the same as that of
1182 asort(), except that the array indices are used
1183 for sorting, not the array values. When done,
1184 the array is indexed numerically, and the val‐
1185 ues are those of the original indices. The
1186 original values are lost; thus provide a second
1187 array if you wish to preserve the original.
1188 gensub(r, s, h [, t]) Search the target string t for matches of the
1190 regular expression r. If h is a string begin‐
1191 ning with g or G, then replace all matches of r
1192 with s. Otherwise, h is a number indicating
1193 which match of r to replace. If t is not sup‐
1194 plied, $0 is used instead. Within the replace‐
1195 ment text s, the sequence \n, where n is a
1196 digit from 1 to 9, may be used to indicate just
1197 the text that matched the n'th parenthesized
1198 subexpression. The sequence \0 represents the
1199 entire matched text, as does the character &.
1200 Unlike sub() and gsub(), the modified string is
1201 returned as the result of the function, and the
1202 original target string is not changed.
1203 gsub(r, s [, t]) For each substring matching the regular expres‐
1205 sion r in the string t, substitute the string
1206 s, and return the number of substitutions. If
1207 t is not supplied, use $0. An & in the
1208 replacement text is replaced with the text that
1209 was actually matched. Use \& to get a literal
1210 &. (This must be typed as "\\&"; see GAWK:
1211 Effective AWK Programming for a fuller discus‐
1212 sion of the rules for &'s and backslashes in
1213 the replacement text of sub(), gsub(), and gen‐
1214 sub().)
1215 index(s, t) Returns the index of the string t in the string
1217 s, or 0 if t is not present. (This implies
1218 that character indices start at one.)
1219 length([s]) Returns the length of the string s, or the
1221 length of $0 if s is not supplied. Starting
1222 with version 3.1.5, as a non-standard exten‐
1223 sion, with an array argument, length() returns
1224 the number of elements in the array.
1225 match(s, r [, a]) Returns the position in s where the regular
1227 expression r occurs, or 0 if r is not present,
1228 and sets the values of RSTART and RLENGTH.
1229 Note that the argument order is the same as for
1230 the ~ operator: str ~ re. If array a is pro‐
1231 vided, a is cleared and then elements 1 through
1232 n are filled with the portions of s that match
1233 the corresponding parenthesized subexpression
1234 in r. The 0'th element of a contains the por‐
1235 tion of s matched by the entire regular expres‐
1236 sion r. Subscripts a[n, "start"], and a[n,
1237 "length"] provide the starting index in the
1238 string and length respectively, of each match‐
1239 ing substring.
1240 split(s, a [, r]) Splits the string s into the array a on the
1242 regular expression r, and returns the number of
1243 fields. If r is omitted, FS is used instead.
1244 The array a is cleared first. Splitting
1245 behaves identically to field splitting,
1246 described above.
1247 sprintf(fmt, expr-list) Prints expr-list according to fmt, and returns
1249 the resulting string.
1250 strtonum(str) Examines str, and returns its numeric value.
1252 If str begins with a leading 0, strtonum()
1253 assumes that str is an octal number. If str
1254 begins with a leading 0x or 0X, strtonum()
1255 assumes that str is a hexadecimal number.
1256 sub(r, s [, t]) Just like gsub(), but only the first matching
1258 substring is replaced.
1259 substr(s, i [, n]) Returns the at most n-character substring of s
1261 starting at i. If n is omitted, the rest of s
1262 is used.
1263 tolower(str) Returns a copy of the string str, with all the
1265 upper-case characters in str translated to
1266 their corresponding lower-case counterparts.
1267 Non-alphabetic characters are left unchanged.
1268 toupper(str) Returns a copy of the string str, with all the
1270 lower-case characters in str translated to
1271 their corresponding upper-case counterparts.
1272 Non-alphabetic characters are left unchanged.
1273 As of version 3.1.5, gawk is multibyte aware. This means that index(),
1275 length(), substr() and match() all work in terms of characters, not
1276 bytes.
1277 Time Functions
1279 Since one of the primary uses of AWK programs is processing log files
1280 that contain time stamp information, gawk provides the following func‐
1281 tions for obtaining time stamps and formatting them.
1282 mktime(datespec)
1284 Turns datespec into a time stamp of the same form as returned
1285 by systime(). The datespec is a string of the form YYYY MM
1286 DD HH MM SS[ DST]. The contents of the string are six or
1287 seven numbers representing respectively the full year includ‐
1288 ing century, the month from 1 to 12, the day of the month
1289 from 1 to 31, the hour of the day from 0 to 23, the minute
1290 from 0 to 59, and the second from 0 to 60, and an optional
1291 daylight saving flag. The values of these numbers need not
1292 be within the ranges specified; for example, an hour of -1
1293 means 1 hour before midnight. The origin-zero Gregorian cal‐
1294 endar is assumed, with year 0 preceding year 1 and year -1
1295 preceding year 0. The time is assumed to be in the local
1296 timezone. If the daylight saving flag is positive, the time
1297 is assumed to be daylight saving time; if zero, the time is
1298 assumed to be standard time; and if negative (the default),
1299 mktime() attempts to determine whether daylight saving time
1300 is in effect for the specified time. If datespec does not
1301 contain enough elements or if the resulting time is out of
1302 range, mktime() returns -1.
1303 strftime([format [, timestamp[, utc-flag]]])
1305 Formats timestamp according to the specification in format.
1306 If utc-flag is present and is non-zero or non-null, the
1307 result is in UTC, otherwise the result is in local time. The
1308 timestamp should be of the same form as returned by sys‐
1309 time(). If timestamp is missing, the current time of day is
1310 used. If format is missing, a default format equivalent to
1311 the output of date(1) is used. See the specification for the
1312 strftime() function in ANSI C for the format conversions that
1313 are guaranteed to be available.
1314 systime() Returns the current time of day as the number of seconds
1316 since the Epoch (1970-01-01 00:00:00 UTC on POSIX systems).
1317 Bit Manipulations Functions
1319 Starting with version 3.1 of gawk, the following bit manipulation func‐
1320 tions are available. They work by converting double-precision floating
1321 point values to uintmax_t integers, doing the operation, and then con‐
1322 verting the result back to floating point. The functions are:
1323 and(v1, v2) Return the bitwise AND of the values provided by v1
1325 and v2.
1326 compl(val) Return the bitwise complement of val.
1328 lshift(val, count) Return the value of val, shifted left by count
1330 bits.
1331 or(v1, v2) Return the bitwise OR of the values provided by v1
1333 and v2.
1334 rshift(val, count) Return the value of val, shifted right by count
1336 bits.
1337 xor(v1, v2) Return the bitwise XOR of the values provided by v1
1339 and v2.
1340 Internationalization Functions
1342 Starting with version 3.1 of gawk, the following functions may be used
1343 from within your AWK program for translating strings at run-time. For
1344 full details, see GAWK: Effective AWK Programming.
1345 bindtextdomain(directory [, domain])
1347 Specifies the directory where gawk looks for the .mo files, in
1348 case they will not or cannot be placed in the ``standard'' loca‐
1349 tions (e.g., during testing). It returns the directory where
1350 domain is ``bound.''
1351 The default domain is the value of TEXTDOMAIN. If directory is
1352 the null string (""), then bindtextdomain() returns the current
1353 binding for the given domain.
1354 dcgettext(string [, domain [, category]])
1356 Returns the translation of string in text domain domain for
1357 locale category category. The default value for domain is the
1358 current value of TEXTDOMAIN. The default value for category is
1359 "LC_MESSAGES".
1360 If you supply a value for category, it must be a string equal to
1361 one of the known locale categories described in GAWK: Effective
1362 AWK Programming. You must also supply a text domain. Use
1363 TEXTDOMAIN if you want to use the current domain.
1364 dcngettext(string1 , string2 , number [, domain [, category]])
1366 Returns the plural form used for number of the translation of
1367 string1 and string2 in text domain domain for locale category
1368 category. The default value for domain is the current value of
1369 TEXTDOMAIN. The default value for category is "LC_MESSAGES".
1370 If you supply a value for category, it must be a string equal to
1371 one of the known locale categories described in GAWK: Effective
1372 AWK Programming. You must also supply a text domain. Use
1373 TEXTDOMAIN if you want to use the current domain.
1374
1376 Functions in AWK are defined as follows:
1377 function name(parameter list) { statements }
1379 Functions are executed when they are called from within expressions in
1381 either patterns or actions. Actual parameters supplied in the function
1382 call are used to instantiate the formal parameters declared in the
1383 function. Arrays are passed by reference, other variables are passed
1384 by value.
1385 Since functions were not originally part of the AWK language, the pro‐
1387 vision for local variables is rather clumsy: They are declared as extra
1388 parameters in the parameter list. The convention is to separate local
1389 variables from real parameters by extra spaces in the parameter list.
1390 For example:
1391 function f(p, q, a, b) # a and b are local
1393 {
1394 ...
1395 }
1396 /abc/ { ... ; f(1, 2) ; ... }
1398 The left parenthesis in a function call is required to immediately fol‐
1400 low the function name, without any intervening white space. This
1401 avoids a syntactic ambiguity with the concatenation operator. This
1402 restriction does not apply to the built-in functions listed above.
1403 Functions may call each other and may be recursive. Function parame‐
1405 ters used as local variables are initialized to the null string and the
1406 number zero upon function invocation.
1407 Use return expr to return a value from a function. The return value is
1409 undefined if no value is provided, or if the function returns by “fall‐
1410 ing off” the end.
1411 If --lint has been provided, gawk warns about calls to undefined func‐
1413 tions at parse time, instead of at run time. Calling an undefined
1414 function at run time is a fatal error.
1415 The word func may be used in place of function.
1417
1419 Beginning with version 3.1 of gawk, you can dynamically add new built-
1420 in functions to the running gawk interpreter. The full details are
1421 beyond the scope of this manual page; see GAWK: Effective AWK Program‐
1422 ming for the details.
1423 extension(object, function)
1425 Dynamically link the shared object file named by object, and
1426 invoke function in that object, to perform initialization.
1427 These should both be provided as strings. Returns the value
1428 returned by function.
1429 This function is provided and documented in GAWK: Effective AWK Pro‐
1431 gramming, but everything about this feature is likely to change eventu‐
1432 ally. We STRONGLY recommend that you do not use this feature for any‐
1433 thing that you aren't willing to redo.
1434
1436 pgawk accepts two signals. SIGUSR1 causes it to dump a profile and
1437 function call stack to the profile file, which is either awkprof.out,
1438 or whatever file was named with the --profile option. It then contin‐
1439 ues to run. SIGHUP causes pgawk to dump the profile and function call
1440 stack and then exit.
1441
1443 Print and sort the login names of all users:
1444 BEGIN { FS = ":" }
1446 { print $1 | "sort" }
1447 Count lines in a file:
1449 { nlines++ }
1451 END { print nlines }
1452 Precede each line by its number in the file:
1454 { print FNR, $0 }
1456 Concatenate and line number (a variation on a theme):
1458 { print NR, $0 }
1460 Run an external command for particular lines of data:
1461 tail -f access_log |
1463 awk '/myhome.html/ { system("nmap " $1 ">> logdir/myhome.html") }'
1464
1466 String constants are sequences of characters enclosed in double quotes.
1467 In non-English speaking environments, it is possible to mark strings in
1468 the AWK program as requiring translation to the native natural lan‐
1469 guage. Such strings are marked in the AWK program with a leading under‐
1470 score (“_”). For example,
1471 gawk 'BEGIN { print "hello, world" }'
1473 always prints hello, world. But,
1475 gawk 'BEGIN { print _"hello, world" }'
1477 might print bonjour, monde in France.
1479 There are several steps involved in producing and running a localizable
1481 AWK program.
1482 1. Add a BEGIN action to assign a value to the TEXTDOMAIN variable to
1484 set the text domain to a name associated with your program.
1485 BEGIN { TEXTDOMAIN = "myprog" }
1487 This allows gawk to find the .mo file associated with your program.
1489 Without this step, gawk uses the messages text domain, which likely
1490 does not contain translations for your program.
1491 2. Mark all strings that should be translated with leading under‐
1493 scores.
1494 3. If necessary, use the dcgettext() and/or bindtextdomain() functions
1496 in your program, as appropriate.
1497 4. Run gawk --gen-po -f myprog.awk > myprog.po to generate a .po file
1499 for your program.
1500 5. Provide appropriate translations, and build and install the corre‐
1502 sponding .mo files.
1503 The internationalization features are described in full detail in GAWK:
1505 Effective AWK Programming.
1506
1508 A primary goal for gawk is compatibility with the POSIX standard, as
1509 well as with the latest version of UNIX awk. To this end, gawk incor‐
1510 porates the following user visible features which are not described in
1511 the AWK book, but are part of the Bell Laboratories version of awk, and
1512 are in the POSIX standard.
1513 The book indicates that command line variable assignment happens when
1515 awk would otherwise open the argument as a file, which is after the
1516 BEGIN block is executed. However, in earlier implementations, when
1517 such an assignment appeared before any file names, the assignment would
1518 happen before the BEGIN block was run. Applications came to depend on
1519 this “feature.” When awk was changed to match its documentation, the
1520 -v option for assigning variables before program execution was added to
1521 accommodate applications that depended upon the old behavior. (This
1522 feature was agreed upon by both the Bell Laboratories and the GNU
1523 developers.)
1524 The -W option for implementation specific features is from the POSIX
1526 standard.
1527 When processing arguments, gawk uses the special option “--” to signal
1529 the end of arguments. In compatibility mode, it warns about but other‐
1530 wise ignores undefined options. In normal operation, such arguments
1531 are passed on to the AWK program for it to process.
1532 The AWK book does not define the return value of srand(). The POSIX
1534 standard has it return the seed it was using, to allow keeping track of
1535 random number sequences. Therefore srand() in gawk also returns its
1536 current seed.
1537 Other new features are: The use of multiple -f options (from MKS awk);
1539 the ENVIRON array; the \a, and \v escape sequences (done originally in
1540 gawk and fed back into the Bell Laboratories version); the tolower()
1541 and toupper() built-in functions (from the Bell Laboratories version);
1542 and the ANSI C conversion specifications in printf (done first in the
1543 Bell Laboratories version).
1544
1546 There are two features of historical AWK implementations that gawk sup‐
1547 ports. First, it is possible to call the length() built-in function
1548 not only with no argument, but even without parentheses! Thus,
1549 a = length # Holy Algol 60, Batman!
1551 is the same as either of
1553 a = length()
1555 a = length($0)
1556 This feature is marked as “deprecated” in the POSIX standard, and gawk
1558 issues a warning about its use if --lint is specified on the command
1559 line.
1560 The other feature is the use of either the continue or the break state‐
1562 ments outside the body of a while, for, or do loop. Traditional AWK
1563 implementations have treated such usage as equivalent to the next
1564 statement. Gawk supports this usage if --traditional has been speci‐
1565 fied.
1566
1568 Gawk has a number of extensions to POSIX awk. They are described in
1569 this section. All the extensions described here can be disabled by
1570 invoking gawk with the --traditional or --posix options.
1571 The following features of gawk are not available in POSIX awk.
1573 · No path search is performed for files named via the -f option.
1575 Therefore the AWKPATH environment variable is not special.
1576 · The \x escape sequence. (Disabled with --posix.)
1578 · The fflush() function. (Disabled with --posix.)
1580 · The ability to continue lines after ? and :. (Disabled with
1582 --posix.)
1583 · Octal and hexadecimal constants in AWK programs.
1585 · The ARGIND, BINMODE, ERRNO, LINT, RT and TEXTDOMAIN variables are not
1587 special.
1588 · The IGNORECASE variable and its side-effects are not available.
1590 · The FIELDWIDTHS variable and fixed-width field splitting.
1592 · The PROCINFO array is not available.
1594 · The use of RS as a regular expression.
1596 · The special file names available for I/O redirection are not recog‐
1598 nized.
1599 · The |& operator for creating co-processes.
1601 · The ability to split out individual characters using the null string
1603 as the value of FS, and as the third argument to split().
1604 · The optional second argument to the close() function.
1606 · The optional third argument to the match() function.
1608 · The ability to use positional specifiers with printf and sprintf().
1610 · The ability to pass an array to length().
1612 · The use of delete array to delete the entire contents of an array.
1614 · The use of nextfile to abandon processing of the current input file.
1616 · The and(), asort(), asorti(), bindtextdomain(), compl(), dcgettext(),
1618 dcngettext(), gensub(), lshift(), mktime(), or(), rshift(), strf‐
1619 time(), strtonum(), systime() and xor() functions.
1620 · Localizable strings.
1622 · Adding new built-in functions dynamically with the extension() func‐
1624 tion.
1625 The AWK book does not define the return value of the close() function.
1627 Gawk's close() returns the value from fclose(3), or pclose(3), when
1628 closing an output file or pipe, respectively. It returns the process's
1629 exit status when closing an input pipe. The return value is -1 if the
1630 named file, pipe or co-process was not opened with a redirection.
1631 When gawk is invoked with the --traditional option, if the fs argument
1633 to the -F option is “t”, then FS is set to the tab character. Note
1634 that typing gawk -F\t ... simply causes the shell to quote the “t,”
1635 and does not pass “\t” to the -F option. Since this is a rather ugly
1636 special case, it is not the default behavior. This behavior also does
1637 not occur if --posix has been specified. To really get a tab character
1638 as the field separator, it is best to use single quotes: gawk -F'\t'
1639 ....
1640 If gawk is configured with the --enable-switch option to the configure
1642 command, then it accepts an additional control-flow statement:
1643 switch (expression) {
1644 case value|regex : statement
1645 ...
1646 [ default: statement ]
1647 }
1648 If gawk is configured with the --disable-directories-fatal option, then
1650 it will silently skip directories named on the command line. Other‐
1651 wise, it will do so only if invoked with the --traditional option.
1652
1654 The AWKPATH environment variable can be used to provide a list of
1655 directories that gawk searches when looking for files named via the -f
1656 and --file options.
1657 For socket communication, two special environment variables can be used
1659 to control the number of retries (GAWK_SOCK_RETRIES), and the interval
1660 between retries (GAWK_MSEC_SLEEP). The interval is in milliseconds. On
1661 systems that do not support usleep(3), the value is rounded up to an
1662 integral number of seconds.
1663 If POSIXLY_CORRECT exists in the environment, then gawk behaves exactly
1665 as if --posix had been specified on the command line. If --lint has
1666 been specified, gawk issues a warning message to this effect.
1667
1669 If the exit statement is used with a value, then gawk exits with the
1670 numeric value given to it.
1671 Otherwise, if there were no problems during execution, gawk exits with
1673 the value of the C constant EXIT_SUCCESS. This is usually zero.
1674 If an error occurs, gawk exits with the value of the C constant
1676 EXIT_FAILURE. This is usually one.
1677 If gawk exits because of a fatal error, the exit status is 2. On non-
1679 POSIX systems, this value may be mapped to EXIT_FAILURE.
1680
1682 egrep(1), getpid(2), getppid(2), getpgrp(2), getuid(2), geteuid(2),
1683 getgid(2), getegid(2), getgroups(2)
1684 The AWK Programming Language, Alfred V. Aho, Brian W. Kernighan, Peter
1686 J. Weinberger, Addison-Wesley, 1988. ISBN 0-201-07981-X.
1687 GAWK: Effective AWK Programming, Edition 3.0, published by the Free
1689 Software Foundation, 2001. The current version of this document is
1690 available online at http://www.gnu.org/software/gawk/manual.
1691
1693 The -F option is not necessary given the command line variable assign‐
1694 ment feature; it remains only for backwards compatibility.
1695 Syntactically invalid single character programs tend to overflow the
1697 parse stack, generating a rather unhelpful message. Such programs are
1698 surprisingly difficult to diagnose in the completely general case, and
1699 the effort to do so really is not worth it.
1700
1702 The original version of UNIX awk was designed and implemented by Alfred
1703 Aho, Peter Weinberger, and Brian Kernighan of Bell Laboratories. Brian
1704 Kernighan continues to maintain and enhance it.
1705 Paul Rubin and Jay Fenlason, of the Free Software Foundation, wrote
1707 gawk, to be compatible with the original version of awk distributed in
1708 Seventh Edition UNIX. John Woods contributed a number of bug fixes.
1709 David Trueman, with contributions from Arnold Robbins, made gawk com‐
1710 patible with the new version of UNIX awk. Arnold Robbins is the cur‐
1711 rent maintainer.
1712 The initial DOS port was done by Conrad Kwok and Scott Garfinkle.
1714 Scott Deifik is the current DOS maintainer. Pat Rankin did the port to
1715 VMS, and Michal Jaegermann did the port to the Atari ST. The port to
1716 OS/2 was done by Kai Uwe Rommel, with contributions and help from Dar‐
1717 rel Hankerson. Andreas Buening now maintains the OS/2 port. Fred Fish
1718 supplied support for the Amiga, and Martin Brown provided the BeOS
1719 port. Stephen Davies provided the original Tandem port, and Matthew
1720 Woehlke provided changes for Tandem's POSIX-compliant systems. Ralf
1721 Wildenhues now maintains that port.
1722 See the README file in the gawk distribution for current information
1724 about maintainers and which ports are currently supported.
1725
1727 This man page documents gawk, version 3.1.8.
1728
1730 If you find a bug in gawk, please send electronic mail to bug-
1731 gawk@gnu.org. Please include your operating system and its revision,
1732 the version of gawk (from gawk --version), what C compiler you used to
1733 compile it, and a test program and data that are as small as possible
1734 for reproducing the problem.
1735 Before sending a bug report, please do the following things. First,
1737 verify that you have the latest version of gawk. Many bugs (usually
1738 subtle ones) are fixed at each release, and if yours is out of date,
1739 the problem may already have been solved. Second, please see if set‐
1740 ting the environment variable LC_ALL to LC_ALL=C causes things to
1741 behave as you expect. If so, it's a locale issue, and may or may not
1742 really be a bug. Finally, please read this man page and the reference
1743 manual carefully to be sure that what you think is a bug really is,
1744 instead of just a quirk in the language.
1745 Whatever you do, do NOT post a bug report in comp.lang.awk. While the
1747 gawk developers occasionally read this newsgroup, posting bug reports
1748 there is an unreliable way to report bugs. Instead, please use the
1749 electronic mail addresses given above.
1750 If you're using a GNU/Linux system or BSD-based system, you may wish to
1752 submit a bug report to the vendor of your distribution. That's fine,
1753 but please send a copy to the official email address as well, since
1754 there's no guarantee that the bug will be forwarded to the gawk main‐
1755 tainer.
1756
1758 Brian Kernighan of Bell Laboratories provided valuable assistance dur‐
1759 ing testing and debugging. We thank him.
1760
1762 Copyright © 1989, 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
1763 2001, 2002, 2003, 2004, 2005, 2007, 2009, 2010 Free Software Founda‐
1764 tion, Inc.
1765 Permission is granted to make and distribute verbatim copies of this
1767 manual page provided the copyright notice and this permission notice
1768 are preserved on all copies.
1769 Permission is granted to copy and distribute modified versions of this
1771 manual page under the conditions for verbatim copying, provided that
1772 the entire resulting derived work is distributed under the terms of a
1773 permission notice identical to this one.
1774 Permission is granted to copy and distribute translations of this man‐
1776 ual page into another language, under the above conditions for modified
1777 versions, except that this permission notice may be stated in a trans‐
1778 lation approved by the Foundation.
1779Free Software Foundation Apr 20 2010 GAWK(1)