fix: detect partial spacing profanity obfuscation

[deemonic]

Summary

• Fixes partial spacing profanity obfuscation not being detected (e.g., "s hit", "f uck", "t wat")
• Updates isSpanningWordBoundary() to check surrounding context instead of relying on heuristics
• Adds 6 new test cases for partial spacing detection

Problem

Profanity obfuscation using partial spacing was not being detected:

• "s hit" - not detected as "shit"
• "f uck" - not detected as "fuck"
• "t wat" - not detected as "twat"
• "fu c k" - not detected as "fuck"
• "tw a t" - not detected as "twat"

The isSpanningWordBoundary() method had overly strict logic that rejected legitimate partial spacing patterns when the first or last part was a single character.

Solution

Modified isSpanningWordBoundary() to check surrounding context:

• If there's an alphanumeric character immediately before the match → it's embedded in a word → reject
• If there's an alphanumeric character immediately after the match → it's embedded in a word → reject
• Otherwise → standalone text, likely intentional obfuscation → allow

Test plan

• All 194 existing tests pass
• New test cases for partial spacing detection pass
• False positive prevention still works ("This musicals hit" not flagged)

🤖 Generated with Claude Code

Summary by CodeRabbit

• Bug Fixes

  • Improved profanity detection with Unicode-aware, context-sensitive word-boundary checks to better catch profanities split by spaces/characters and reduce false positives.

• Tests

  • Added tests covering multiple spaced-profane patterns and a negative case to ensure legitimate phrases are not flagged.

_{✏️ Tip: You can customize this high-level summary in your review settings.}

[coderabbitai]

📝 Walkthrough

Walkthrough

The isSpanningWordBoundary method in BlaspService now accepts fullString and matchStart and uses Unicode-aware, context-sensitive logic to analyze surrounding characters and split parts; test coverage added for multiple partial-spacing profanity cases.

Changes

Cohort / File(s)	Summary
Profanity boundary detection logic src/BlaspService.php	Signature changed to isSpanningWordBoundary(string $matchedText, string $fullString, int $matchStart): bool. Implementation rewritten: early return for non-spacing matches, Unicode-aware splitting and counting (mb_strlen, mb_substr), context checks of chars before/after match, and nuanced embedding rules to decide when a match spans word boundaries.
Partial-spacing test coverage tests/BlaspCheckTest.php	Added multiple tests for spaced/partially-spaced profanity (s hit, f uck, t wat, fu c k, tw a t) and a negative case ensuring no false positive for an embedded phrase.

Sequence Diagram(s)

(omitted — changes do not introduce multi-component sequential flows requiring a diagram)

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~25 minutes

Possibly related PRs

• fix: detect spaced profanity like "f u c k i n g" (#36) #37: Modifies BlaspService::isSpanningWordBoundary and tests for spaced/obfuscated profanity handling; closely related to these changes.

Poem

🐰
I hop through strings both wide and thin,
I sniff the gaps where sly words grin,
With Unicode ears and careful eye,
No crafty swear can slip on by.
A tidy patch — a tidy nibble!

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title 'fix: detect partial spacing profanity obfuscation' directly and specifically describes the main change: improving detection of profanity that has been obfuscated through partial spacing (e.g., 's hit' for 'shit').
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing touches

• 📝 Generate docstrings

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 1

🤖 Fix all issues with AI agents

In `@src/BlaspService.php`:
- Around line 435-452: preg_match_all returns byte offsets but the code treats
$matchStart as a character index when calling mb_substr/mb_strlen, which breaks
for multibyte strings; convert the byte offset to a character offset first (e.g.
compute $matchCharStart = mb_strlen(substr($fullString, 0, $matchStart),
'UTF-8')) and then compute $matchCharEnd = $matchCharStart +
mb_strlen($matchedText, 'UTF-8'), and use $matchCharStart/$matchCharEnd instead
of $matchStart/$matchEnd when deriving $charBefore and $charAfter and when
comparing against mb_strlen($fullString, 'UTF-8').