Standard Library

AWK ships with about 20 builtin functions. zawk adds a standard library of ~190 functions, inspired by the standard libraries of DuckDB, ClickHouse, Go, Rust, Deno, PHP and Java's commons-lang, so that the common work of data processing does not need a pipeline of other tools.

Categories

Category	Functions
Text	trim, truncate, pad, mask, starts_with, contains, camel_case, snake_case, title_case, similar, wrap, words, parse, rparse, format_bytes, figlet, ...
Parsers	url, data_url, shlex, path, semver, pairs, record, message, func, tuple, variant, parse_array, flags
JSON / XML / HTML / CSV	from_json, to_json, json_flatten, json_value, json_query, json_set, xml_value, html_query, from_csv, to_csv
Array & Math	seq, uniq, asort, asorti, _sum, _mean, _median, _quantile, _stddev, _top, bloom filter, abs, round, eval, fend, min, max
Date/Time	systime, strftime, mktime, datetime, duration
Crypto & Encoding	digest, hmac, encrypt, decrypt, jwt, dejwt, encode, decode, uuid, ulid, tsid, snowflake
Network	http_get, http_post, send_mail, s3_get, s3_put, publish, local_ip, jev*
Database & KV	sqlite_*, mysql_*, pg_*, kv_get, kv_put, kv_delete, kv_clear
OS, I/O & Misc	getenv, system2, read_all, write_all, read_config, var_dump, log_info, typeof, version, fake, hex2rgb

Discover functions from the command line

Don't guess function names or argument orders: ask zawk. It is never out of date.

zawk functions                 # all functions, grouped by category: signature, summary, example
zawk functions json            # functions whose name contains "json"
zawk functions Crypto          # one category
zawk functions --json trim     # machine readable, one function per line

{"name":"trim","category":"Text","signature":"trim(s: str[, chars: str]) -> str",
 "params":[{"name":"s","type":"str","optional":false},{"name":"chars","type":"str","optional":true}],
 "returns":"str","summary":"Remove whitespace, or any of chars, from both ends of s.","example":"trim(\"[hello]\", \"[]\")"}

There is also a printable cheat sheet.

Conventions

Arrays as results. Functions documented as returning map[...] return an AWK array. Index the call directly (url($1)["host"]) or assign it (u = url($1)). Use var_dump(x) to see the structure on stderr.

Failures are warnings, not crashes. Functions working with external services or user-supplied input (network, KV, databases, crypto, parsers) print a warning to stderr and return an empty value ("", an empty array, or 0 for an HTTP status). Each distinct warning is printed once. The failure also sets ERRNO:

ERRNO = ""
rows = sqlite_query("app.db", "select name from users")
if (ERRNO != "") { print "query failed: " ERRNO > "/dev/stderr"; exit 1 }

# or, in one expression and without the warning:
rows = try sqlite_query("app.db", "select name from users") else fail("query failed: " ERRNO)

Your functions win. A user-defined function with the same name as a stdlib function (such as trim, max or abs) takes precedence over it in the whole program, so existing AWK programs that define these helpers keep working. The POSIX/gawk builtins (length, substr, split, gsub, sprintf...) cannot be redefined.

Parentheses are required. Builtin functions must be followed directly by (: length($1), not length ($1).

gawk functions. gensub, patsplit, strtonum, asort, asorti, systime, strftime, mktime, typeof, isarray and the bitwise functions behave as in gawk.