tired-labs / tiredize / 22610705118

99%

push

github

Issue/parser greedy regex (#21)

* Fix 5 greedy regex bugs across markdown parser

1. InlineImage/InlineLink URL pattern: \S+ -> [^\s)]+ to stop at
   closing paren instead of consuming adjacent syntax.
2. Table rows: require | in each data row per GFM, preventing
   non-table lines from being consumed.
3. Start-of-line anchor: replace (?<![^|\n]) with (?:(?<=\n)|(?:^))
   in CodeBlock, Header, QuoteBlock, ReferenceDefinition to stop
   treating | as valid start-of-line.
4. BareLink: add ../ as explicit alternative before ./ to prevent
   false partial match of ./ inside ../ paths.
5. ImageReference: remove dead lookbehind (?<!(\])) that always
   passed because ! is never ].

* Update issue files and parser spec for greedy regex fixes

Checks off all acceptance criteria. Documents start-of-line anchor,
URL pattern, and BareLink path matching in the parser spec. Updates
gfm-parity issue with balanced parentheses acceptance criterion.

* Fix stale comments and document all regex patterns in parser spec

Update stale comments in test_header.py and test_reference.py to
reflect the new start-of-line anchor pattern. Update anchor pattern
references in quoteblock-over-sanitization.md issue. Add comprehensive
Pattern Reference section to the parser specification documenting all
13 RE_* regex constants with component descriptions, named capture
groups, and design rationale.

* Fix spec claim about VERBOSE lookbehinds and misleading regex comments

Correct inaccurate specification claim that negative lookbehinds
cannot be used in re.VERBOSE mode. The actual issue was the negated
character class including pipe. Fix misleading inline comments in
LinkReference and ImageReference regexes that said "Capture the URL"
for reference label groups.

* Mark parser-greedy-regex issue as completed

* Fix IndexError on pipe-only table data rows

Guard the leading-pipe strip in row parsing against empty string
after trailing-pipe strip. A data row containing only | no longer... (continued)

15 of 15 new or added lines in 7 files covered. (100.0%)

4634 of 4867 relevant lines covered (95.21%)

1.9 hits per line