fix(queryLanguage): allow parenthesized regex alternation in filter values

[brendan-kellam]

Summary

• Queries like file:(test|spec), -file:(test|spec), repo:(org1|org2), and sym:(Foo|Bar) previously failed with a parse error because the word tokenizer unconditionally deferred to parenToken whenever a token started with balanced parentheses, even in filter value contexts
• Fixed by detecting value context (preceding non-whitespace char is :) and using depth-tracking to consume the entire (...) as a word rather than deferring
• Added 8 test cases covering filter alternation for file:, repo:, sym:, content:, negated alternation, and combined queries

Test plan

• file:(test|spec) parses as PrefixExpr(FileExpr) ✓
• -file:(test|spec) parses as NegateExpr(PrefixExpr(FileExpr)) ✓
• chat lang:TypeScript -file:(test|spec) parses successfully ✓
• Existing paren grouping (foo bar) still works as ParenExpr ✓
• -(file:test or file:spec) still works as NegateExpr(ParenExpr(...)) ✓
• All 248 existing tests pass ✓

🤖 Generated with Claude Code

Summary by CodeRabbit

• Bug Fixes

  • Search query parser now accepts parenthesized regex alternation in filter values (e.g., file:(test|spec), -file:(test|spec), repo:(org1|org2)).

• Tests

  • Added test cases for alternation in prefix values, negated prefixes, and combined prefix scenarios.

• Chores

  • Updated CI workflow definitions and job naming.

[coderabbitai]

Note

Currently processing new changes in this PR. This may take a few minutes, please wait...

📥 Commits

Reviewing files that changed from the base of the PR and between ff4fa79 and aa44ab9.

📒 Files selected for processing (2)

• .github/workflows/test-backend.yml
• .github/workflows/test.yml

_{✏️ Tip: You can disable in-progress messages and the fortune message in your review settings.}

Walkthrough

Adds support for parenthesized regex alternation in search filter values (e.g., file:(test|spec), -repo:(org1|org2)). The tokenizer distinguishes parentheses used for value contexts (after :) from grouping, and tests and changelog entries were added.

Changes

Cohort / File(s)	Summary
Changelog CHANGELOG.md	Adds a Fixed entry documenting acceptance of parenthesized regex alternation in filter values.
Tokenizer Logic packages/queryLanguage/src/tokens.ts	Updates wordToken to detect an opening ( in value contexts (preceded by :), consume balanced parentheses with depth tracking as part of the word token, and otherwise defer to existing paren tokenization.
Tests packages/queryLanguage/test/prefixes.txt, packages/queryLanguage/test/negation.txt	Adds test cases covering parenthesized alternation in prefix values (file, repo, sym, content), combined prefix scenarios, and negated-prefix examples like `-file:(test

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~22 minutes

Possibly related PRs

• fix(parser): Allow parenthesis in query and filter terms #788: Modifies tokenizer handling of parentheses and word/paren tokenization; directly related to context-aware parentheses handling.
• feat(web): Streamed code search #623: Changes packages/queryLanguage/src/tokens.ts (negation/tokenization) which may interact with the new parenthesized value handling.

🚥 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 accurately and concisely summarizes the main change: enabling parenthesized regex alternation in filter values.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

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

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment
• Commit unit tests in branch brendan/fix-query-parser-paren-value

Tip

Try Coding Plans. Let us write the prompt for your AI agent so you can ship faster (with fewer bugs).
Share your feedback on Discord.

---

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]

🧹 Nitpick comments (1)

packages/queryLanguage/src/tokens.ts (1)

337-352: Consider handling escaped parentheses for consistency.

The depth-tracking loop doesn't skip escaped parentheses (e.g., \( or \)), while hasBalancedParensAt() at lines 124-127 does handle escapes. This inconsistency could cause unexpected behavior for patterns containing escaped parens like file:(test\)|spec).

If escaped parens in filter values are not a supported use case, this is fine. Otherwise, consider adding escape handling:

♻️ Proposed fix to handle escaped parens

         let depth = 0;
         while (input.next !== EOF) {
             const ch = input.next;
             if (isWhitespace(ch)) break;
+            // Handle escaped characters - skip the next character after a backslash
+            if (ch === 92 /* backslash */) {
+                input.advance();
+                if (input.next !== EOF) input.advance();
+                continue;
+            }
             if (ch === OPEN_PAREN) {
                 depth++;
             } else if (ch === CLOSE_PAREN) {
                 if (depth === 0) break; // outer ParenExpr closing — don't consume
                 depth--;
             }
             input.advance();
         }

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@packages/queryLanguage/src/tokens.ts` around lines 337 - 352, The
depth-tracking loop inside the inValueParenContext branch doesn't account for
escaped parens, causing mismatched depth when encountering \(' or \)'; update
the loop (the code using input.next, input.advance, OPEN_PAREN and CLOSE_PAREN)
to detect and skip escaped characters (e.g., when a backslash precedes a paren)
so escaped '(' or ')' do not change depth or terminate the loop, and ensure this
behavior mirrors hasBalancedParensAt's escape handling for consistency.

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Nitpick comments:
In `@packages/queryLanguage/src/tokens.ts`:
- Around line 337-352: The depth-tracking loop inside the inValueParenContext
branch doesn't account for escaped parens, causing mismatched depth when
encountering \(' or \)'; update the loop (the code using input.next,
input.advance, OPEN_PAREN and CLOSE_PAREN) to detect and skip escaped characters
(e.g., when a backslash precedes a paren) so escaped '(' or ')' do not change
depth or terminate the loop, and ensure this behavior mirrors
hasBalancedParensAt's escape handling for consistency.

---

ℹ️ Review info

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 3e3e0a6 and 3f7be64.

📒 Files selected for processing (4)

• CHANGELOG.md
• packages/queryLanguage/src/tokens.ts
• packages/queryLanguage/test/negation.txt
• packages/queryLanguage/test/prefixes.txt