f-strings & Raw Strings

f-strings

f"..." substitutes expressions into a string. It is clearer than sprintf or concatenation when you build SQL, JSON, URLs, log lines or LLM prompts.

zawk '{ print f"user {$1} logged in at line {NR}" }' login.txt
zawk -v table=users '{ print f"SELECT * FROM {table} WHERE id = {$1};" }' ids.txt
zawk '{ print f"{{\"id\": {$1}}}" }' ids.txt            # {{ and }} are literal braces
zawk '{ print f"{NR+1}: {toupper($2)} {$3 * 1.1:.2f} {$1:>10}" }' data.txt

• A placeholder is any expression: {name}, {$1}, {NR+1}, {toupper($2)}, {a[\"k\"]}, {x > 0 ? \"pos\" : \"neg\"}. Write \" for a quote inside a placeholder (the f-string ends at a bare ").
• An invalid expression is a syntax error; {} stays as literal text.
• Escapes (\t, \n, \") work as in normal strings.
• Values are not escaped: use escape("sql", v) or escape("json", v) first when the value comes from input (better: use query parameters for SQL).

Format specs

A format spec follows a : inside the placeholder, Python style: [[fill]align][sign][#][0][width][.precision][type].

Placeholder	Value	Result
{p:.2f}	3.14159	3.14
{n:5d}	42	42
{n:<5}	42	42
{n:05}	42	00042
{s:*^6}	ab	**ab**
{n:x}	42	2a
{r:.1%}	0.256	25.6%
{s:.2}	hello	he
{n:,}	1234567	1,234,567
{n:_d}	1234567	1_234_567
{p:,.2f}	1234.5	1,234.50

Without a type the value is formatted as a string; values are right-aligned by default. printf/sprintf also accept the ' and _ flags for grouping: printf "%'d\n", 1234567 prints 1,234,567.

caution

A misspelled name such as {nmae} silently becomes an empty string, like any uninitialized AWK variable. zawk lint reports variables that are read but never set (W201).

Raw strings

Raw string literals work like Rust's: backslashes are not escapes. They are handy for Windows paths and for dynamic regular expressions, where you would otherwise write "\\d+".

path = r"C:\Program Files\Rust"
if ($0 ~ r"\d+\.\d+") print "version-like"
split($0, parts, r"\s*,\s*")

To include a ", put the same number of #s around the quotes:

print r#"say "hi""#          # say "hi"
print r##"a"#b"##            # a"#b

A raw string can span lines, which is convenient for embedded SQL or templates:

BEGIN {
    sql = r"
        select name, age
        from people
        where age > 18
    "
}

note

r"..." (no space) is always a raw string, never the variable r concatenated with a string. With r = 5, gawk prints 5x for print r"x" while zawk prints x. Write r "x" or rename the variable.

Unicode-aware strings

Text is UTF-8: length, substr, index and match count characters, and strlen counts bytes.

$ zawk 'BEGIN { print length("你好Hello"), substr("你好Hello", 1, 2), strlen("你好") }'
7 你好 6

Input that is not valid UTF-8 (GBK, Latin-1, binary) is kept byte for byte; use --utf8 to validate input.