Add try_parse_tag_token that parses {%...%} tag tokens using
getbyte/byteslice + ByteTables lookup arrays instead of the
FullToken regex with 4 capture groups. Allocates only the 2
strings needed (tag_name, markup) vs 4+ from regex captures.
Uses ByteTables::WORD (no hyphen) for tag name scanning,
matching TagName = /#|\w+/ exactly. Falls back to FullToken
regex when the fast path returns nil.
Skip the expensive recursive VariableParser regex for simple
lookups like 'product.title' (~90% of real-world cases).
SIMPLE_LOOKUP_RE validates the input is a plain a.b.c chain
(no brackets, no quotes). On match, byte-walks on dots to split
segments instead of invoking the regex engine. Falls through to
the original VariableParser scan for complex inputs.
parse_number: replace INTEGER_REGEX/FLOAT_REGEX matching and
StringScanner loop with a single byte-walking pass using
ByteTables::DIGIT. Avoids MatchData allocation and StringScanner
reset on every call.
Expression.parse: only call String#strip when leading/trailing
whitespace is actually present (checked via ByteTables::WHITESPACE).
Avoids allocating a new String on ~4,464 calls per compile.
ByteTables provides pre-computed 256-entry boolean lookup arrays for
byte classification (IDENT_START, IDENT_CONT, WORD, DIGIT, WHITESPACE)
and named constants for delimiter bytes (NEWLINE, DASH, DOT, HASH).
bench_quick.rb measures parse µs, render µs, and object allocations
for the theme benchmark suite.
Chris Pak