test: remove unused Ripgrep.search coverage

[thdxr]

Summary

• remove the Ripgrep.search metadata test from packages/opencode/test/file/ripgrep.test.ts
• keep coverage focused on the Ripgrep.files behavior and the empty search case that are still exercised here

Verification

• /tmp/bun-1.3.10 test test/file/ripgrep.test.ts
• pre-push bun turbo typecheck

[greptile-apps]

Greptile Summary

This PR removes the "search returns match metadata" test from packages/opencode/test/file/ripgrep.test.ts, leaving Ripgrep.search covered only by a no-match negative case.

• Coverage gap introduced: The deleted test was the sole test exercising Ripgrep.search with a real match. The function's pipeline — shell command construction, JSON.parse, Zod schema validation, .filter(r => r.type === "match"), and .map(r => r.data) — is now untested for the positive path.
• No indirect coverage: GrepTool (tested in test/tool/grep.test.ts) calls Ripgrep.filepath() directly and does its own parsing; it does not call Ripgrep.search, so the gap is real.
• Misleading framing: The PR description labels the removed test as "unused coverage," but the test was actively exercising the core match-returning behavior of search(), not a redundant or dead code path.

Confidence Score: 3/5

• Safe to merge if the author intentionally accepts reduced coverage for Ripgrep.search, but the positive-match path is now completely untested.
• The change itself is low-risk (deleting a test cannot break production code), but it removes the only test validating that Ripgrep.search correctly returns match metadata, leaving a meaningful coverage gap in a non-trivial parsing pipeline.
• packages/opencode/test/file/ripgrep.test.ts — only file changed; the deletion of the positive-path test deserves review.

Important Files Changed

Filename	Overview
packages/opencode/test/file/ripgrep.test.ts	Removes the only positive-case test for Ripgrep.search, leaving the match-parsing pipeline (JSON parsing, Zod validation, type filtering) without any coverage from a real match result.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
A["Ripgrep.search()"] --> B["Build shell command\nargs.join(' ')"]
B --> C["Execute via Bun ${{raw}}"]
C --> D{"exitCode !== 0?"}
D -- "yes" --> E["return []"]
D -- "no" --> F["Split output lines"]
F --> G["JSON.parse each line"]
G --> H["Result.parse (Zod)"]
H --> I["filter type === 'match'"]
I --> J["map r => r.data"]
J --> K["Return match array"]
style E fill:#90EE90,stroke:#228B22
style K fill:#FFB6C1,stroke:#CC0000
E:::tested
K:::untested
classDef tested fill:#90EE90,stroke:#228B22,color:#000
classDef untested fill:#FFB6C1,stroke:#CC0000,color:#000
subgraph Legend
L1["✅ Tested (empty-result path)"]:::tested
L2["❌ No longer tested (match-result path)"]:::untested
end

Loading

Comments Outside Diff (1)

1. packages/opencode/test/file/ripgrep.test.ts, line 40-53 (link)

   Positive-path coverage for Ripgrep.search now gone

   The removed test was the only test that exercised Ripgrep.search with an actual match. After this change the only remaining coverage for search is the negative (no-match) path, leaving the following non-trivial pipeline completely untested with real results:

   1. Shell command construction via args.join(" ") and ${{ raw: command }}
   2. JSON output parsing (JSON.parse(line))
   3. Zod schema validation (Result.parse(parsed))
   4. Type filtering (.filter((r) => r.type === "match"))
   5. Data extraction (.map((r) => r.data))

   GrepTool in src/tool/grep.ts uses Ripgrep.filepath() directly and does its own output parsing — it does not call Ripgrep.search — so test/tool/grep.test.ts provides no indirect coverage here.

   A regression in any of the steps above (e.g. a change to the Zod schema, the filter predicate, or the shell quoting) will now silently pass the test suite. Consider keeping this test, or at minimum replacing it with an equivalent that validates the returned path, line_number, and lines.text fields from a known match.

_{Last reviewed commit: 5f2f2f9}