Unused by the parser; will be removed in the next major version.
Matches are read through the named groups qty1, sep, qty2, and uom — never by
index. numericRegexAnywhere is inlined twice and user range separators may contain
capture groups of their own, so every index here is unstable by construction.
The uom body is uomWordSource, shared with firstWordRegEx; it is group-free, so
inlining it is index-neutral.
Unanchored and $-terminated, so every start position is tried; the lookarounds keep that
linear-ish. Only lookarounds are used, so no groups are added.
'A4' is not a quantity; CJK '大さじ2' is), nor split a
digit run (an earlier start wins).\s*(?!\s) consumes whitespace atomically; adjacent \s*s otherwise split it every way.x prefix must not end a word ('box 2' is not 'bo' × 2).qty2 must be non-empty (numericRegex can match empty when unanchored) and must not
end mid-digit-run, so uom can't re-scan its digits.
Builds a regex to capture trailing quantity and unit of measure, using the provided range separator words.
No longer used by the parser, which scans trailing quantities with
numericQuantityso that vulgar fractions and non-ASCII digits are recognized. Kept as public API; matches ASCII quantities only.