Field Separators and Record Separators

Analogy categoryAnalogies

Splitting Input into Fields with FS

FS, the field separator, tells AWK how to break each record into $1, $2, and so on. By default FS is a single space, which AWK treats specially: it splits on any run of whitespace and ignores leading and trailing blanks. Set FS to a single character like a colon for /etc/passwd, or to a regular expression such as [,:] to split on either commas or colons. You can set it on the command line with -F',' or inside a BEGIN block with FS=",". A crucial subtlety: the default whitespace behavior disappears the moment you set FS to a literal space " " versus leaving it at default, and setting FS to a single tab makes empty fields between consecutive tabs significant.

🏏

Cricket analogy: FS is like deciding what counts as a boundary line on the field — change it and the same shot is scored differently, just as changing the separator re-carves a line into different fields.

Records and RS: Beyond Line-by-Line

RS, the record separator, defines where one record ends and the next begins; its default is the newline, so AWK is naturally line-oriented. Setting RS to an empty string "" enables paragraph mode, where records are separated by one or more blank lines and newlines within a paragraph become field separators automatically. In GNU AWK (gawk), RS can be a full regular expression, letting you split on multi-character or variable delimiters like RS="\n---\n" for Markdown-style dividers. Combined with a custom FS, this makes AWK capable of parsing structured, multi-line records that other line-based tools struggle with.

🏏

Cricket analogy: RS is like defining what ends an over — normally six balls, but switch the rule and a whole session becomes one record, changing what AWK treats as a single unit.

Reassembling Output with OFS and ORS

OFS and ORS are the output counterparts of FS and RS. OFS is placed between arguments when you use print with comma-separated expressions, defaulting to a single space; ORS is appended after each print, defaulting to a newline. Because AWK only reassembles $0 from fields when a field is modified, changing OFS has no visible effect until you touch a field or reassign $1=$1. Setting ORS to something like ";" lets you join all records onto one line, and setting OFS to a tab or comma is the standard way to convert between delimited formats. These four variables together form a complete, symmetric parsing-and-emitting pipeline.

🏏

Cricket analogy: OFS and ORS are like how the commentary team reassembles the scorecard for broadcast — the delimiter between runs and the break between overs, mirroring how the innings was originally recorded.

awk

# Parse a colon-delimited file, output as CSV
awk -F: 'BEGIN{OFS=","} { $1=$1; print }' /etc/passwd

# Paragraph mode: count words in each blank-line-separated block
awk 'BEGIN{RS=""; FS="\n"} { print "Block", NR, "has", NF, "lines" }' notes.txt

# gawk regex RS: split on '---' dividers, print first line of each record
gawk 'BEGIN{RS="\n---\n"} { print $1 }' document.md

# Join all records onto one semicolon-separated line
awk 'BEGIN{ORS=";"} { print $1 }' names.txt

In paragraph mode (RS=""), newlines always act as field separators in addition to whatever FS you set, so a paragraph's individual lines become fields even if FS is a comma. This is a deliberate convenience for processing multi-line records.

Regex-valued RS is a GNU AWK (gawk) extension; POSIX AWK and mawk only accept a single-character RS. Scripts relying on RS="\n---\n" will behave differently or fail on non-gawk interpreters, so check your target environment before depending on it.

FS splits records into fields; its default (single space) collapses whitespace runs specially.
FS can be a single character or a full regular expression, set via -F or FS=.
RS defines record boundaries; the default is newline, making AWK line-oriented.
RS="" enables paragraph mode where blank lines separate records and newlines split fields.
gawk allows a regex RS for multi-character delimiters; POSIX AWK does not.
OFS and ORS reassemble output; OFS changes need a field rebuild to show.
Together FS/RS/OFS/ORS convert between delimited and multi-line formats.