Warn against semicolons in categorisation rules editor

[AbdulGoodLooks]

Workaround to attempt to prevent creating rules that break aw-server like this https://forum.activitywatch.net/t/invalid-syntax-for-value-to-assign-out-of-nowhere/4627/3

---

Important

Update validateRegex to flag semicolons in regex as invalid to prevent server errors.

• Behavior:
  • Update validateRegex in validate.ts to return false if regex contains semicolons, preventing server errors.
• Validation:
  • Semicolons in regex patterns are now flagged as invalid to avoid breaking aw-server.

^{This description was created by for 93f4484. You can customize this summary. It will automatically update as commits are pushed.}

[ellipsis-dev]

Caution

Changes requested ❌

Reviewed everything up to 93f4484 in 1 minute and 31 seconds. Click for details.

• Reviewed 15 lines of code in 1 files
• Skipped 0 files when reviewing.
• Skipped posting 1 draft comments. View those below.
• Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.

1. src/util/validate.ts:6

• Draft comment:
  Document the rationale further: consider adding the forum post URL in the comment so future maintainers understand why semicolons break aw-server.
• Reason this comment was not posted:
  Comment did not seem useful. Confidence is useful = 0% <= threshold 50% The comment is asking the author to document the rationale further by adding a URL to a forum post. This is a request for additional documentation, which is not allowed according to the rules. The comment does not provide a specific code suggestion or ask for a test to be written, so it should be removed.

Workflow ID: wflow_E2xAh4bMlQ1LhDNM

^{You can customize by changing your verbosity settings, reacting with 👍 or 👎, replying to comments, or adding code review rules.}

[codecov]

Codecov Report

❌ Patch coverage is 0% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 26.29%. Comparing base (c766e86) to head (3d1e04f).
⚠️ Report is 2 commits behind head on master.

Files with missing lines	Patch %	Lines
src/util/validate.ts	0.00%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #724      +/-   ##
==========================================
- Coverage   26.30%   26.29%   -0.02%     
==========================================
  Files          29       29              
  Lines        1684     1685       +1     
  Branches      301      302       +1     
==========================================
  Hits          443      443              
+ Misses       1219     1180      -39     
- Partials       22       62      +40     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[ErikBjare]

@TimeToBuildBob Investigate this, shouldn't it get escaped in the query and be supported? Does it break aw-server-python or aw-server-rust or both?

[TimeToBuildBob]

Investigated both servers.

Short answer: semicolons should be supported and don't need to be blocked. Both servers handle them correctly when the regex is embedded as a string literal in the AQL query (which is what aw-webui does via JSON.stringify).

aw-server-python: Was broken — the old query2.py split statements on all semicolons without tracking string context. Fixed in ActivityWatch/aw-core#131 (merged 2026-02-27). The _split_query_statements() function now tracks quote state and skips semicolons inside string literals.

aw-server-rust: Never broken. The plex-based lexer in aw-query/src/lexer.rs tokenizes double-quoted strings first ("([^"]|(\\"))*" regex → Token::String), so a semicolon inside "abc;def" is consumed as part of the String token and the Token::Semi pattern never fires for it.

How aw-webui builds queries: queries.ts serializes categories via JSON.stringify(params.categories), which wraps regex values in double quotes. So a regex abc;def appears in the AQL query as "abc;def" — safely inside a quoted string. No escaping is needed on top of that.

Recommendation: This PR isn't needed. The original bug (forum post) was in an older aw-server-python before the aw-core fix. The right fix was aw-core#131, which is already merged. Blocking semicolons in the UI unnecessarily restricts valid regex patterns (e.g. alternation with ; characters in paths on some systems).

Could close this PR as not needed, or the contributor could be asked to add a test to aw-core instead verifying that semicolons in categorization rules work end-to-end.

[TimeToBuildBob]

Investigation complete — root cause found, fix in #918.

Short answer: yes, semicolons should be supported. Both servers handle them correctly — the bug is in the webui's client-side query splitter, not in aw-server-python or aw-server-rust.

What breaks: querystr_to_array() in queries.ts naively does .split(';') on the whole query string. Category rules with semicolons in their regex get embedded as JSON string literals:

events = categorize(events, [["Work", {"type":"regex","regex":"foo;bar"}]]);

A plain .split(';') shreds this at the ; inside the JSON string before the query ever reaches either server.

Both servers are fine: I checked aw-query/src/lexer.rs — the Rust lexer's string-literal rule ("([^\"]|(\\\"))*") captures the full quoted string including any ; inside it, so semicolons in string literals never become Token::Semi. The Python parser works the same way. Neither server sees the malformed query because the client breaks it first.

Fix: PR #918 replaces the naive split with a string-context-aware parser that only treats ; as a statement separator when it's outside a double-quoted string literal. 4 unit tests added including the regression case.