Text Recipes

Source: text.spq

Field-extraction functions that stand in for awk and cut when you are processing plain text with SuperDB. See the Plain Text Without the Bashisms tutorial for the pipelines these are meant to slot into.

Two things these smooth over:

• awk and cut number fields from 1; SuperDB arrays index from 0. These take the 1-based number you already know from the shell, so translating a pipeline is mechanical instead of an off-by-one hunt.
• An out-of-range field returns "", the way awk treats $9 on a 3-field line, rather than raising error("missing") and poisoning the record.

---

sk_fields

Splits a line into fields on runs of whitespace, the way awk does. Leading and trailing whitespace is ignored, and repeated spaces or tabs collapse into a single separator. Returns an empty array for a blank line.

Use this instead of split(s, " "), which treats every space as its own separator and hands back empty strings between them.

Type: function

Argument	Description
s	The line to split

join(sk_fields('  alpha   beta\tgamma  '), '|')
-- => 'alpha|beta|gamma'

len(sk_fields('   '))
-- => 0

Implementation:

fn sk_fields(s): (
  trim(s) == "" ? [] : split(regexp_replace(trim(s), "\\s+", " "), " ")
)

---

sk_nf

Counts whitespace-separated fields in a line, the equivalent of awk’s NF. Blank lines count as zero.

Type: function

Argument	Description
s	The line to count fields in

sk_nf('a b  c')
-- => 3

sk_nf('')
-- => 0

Implementation:

fn sk_nf(s): (len(sk_fields(s)))

---

sk_field

Extracts a single whitespace-separated field by 1-based position, the equivalent of awk’s $n. Returns "" when n is out of range or less than 1, so a short line does not raise error("missing").

Type: function

Argument	Description
s	The line to extract from
n	1-based field number, as in awk

sk_field('alpha beta gamma', 2)
-- => 'beta'

sk_field('alpha beta gamma', 9)
-- => ''

Implementation:

fn sk_field(s, n): (
  n >= 1 and n <= len(sk_fields(s)) ? sk_fields(s)[n - 1] : ""
)

---

sk_cut

Extracts a single delimited field by 1-based position, the equivalent of cut -d DELIM -f N. Unlike sk_field, the delimiter is literal and repeats are not collapsed, so consecutive delimiters yield empty fields — matching cut. Returns "" when n is out of range.

Type: function

Argument	Description
s	The line to extract from
delim	Literal field delimiter, not a regular expression
n	1-based field number, as in cut -f

sk_cut('root:x:0:0', ':', 1)
-- => 'root'

sk_cut('root:x:0:0', ':', 3)
-- => '0'

Implementation:

fn sk_cut(s, delim, n): (
  n >= 1 and n <= len(split(s, delim)) ? split(s, delim)[n - 1] : ""
)

---

Picking between them

sk_field and sk_cut look interchangeable but split on different rules, the same way awk and cut do:

Input	sk_field(s, 2)	sk_cut(s, ' ', 2)
'a b c'	'b'	'b'
'a b c'	'b'	''
' a b'	'b'	''

Whitespace-aligned output — ps, ls -l, df — wants sk_field. Genuinely delimited data — /etc/passwd, a : or , separated file — wants sk_cut, because there an empty field is real data rather than padding.