[oyi77]

Summary

Optimize database queries by replacing SELECT * with selective column fetching in multiple files.

Changes

1. src/lib/usage/callLogs.ts

• Add CALL_LOGS_COLUMNS constant with needed columns
• Replace SELECT * with selective column query
• Avoids fetching unused columns (connection_id, artifact_relpath, etc.)

2. src/lib/db/webhooks.ts

• Add WEBHOOKS_COLUMNS constant
• Optimize getWebhooks(), getWebhook(), getEnabledWebhooks()
• Replaces SELECT * in all webhook queries

3. scripts/benchmark-call-logs.mjs (NEW)

• Benchmark script to verify performance improvements
• Tests against production database
• Verifies data integrity

Performance Impact

Tested with production database (113,051 call_logs records):

Metric	Before	After	Improvement
Query Time	5.48ms	4.63ms	15.5% faster
Columns Fetched	24	21	12.5% reduction

Complementary to PR #1038

This PR optimizes callLogs.ts and webhooks.ts while PR #1038 optimizes usageHistory.ts:

• PR perf: reduce analytics page load — DB filter, parallel costs, merged queries #1038: Date filtering, parallel costs, merged queries
• This PR: Selective column fetching

Zero conflicts between the two PRs.

Test Plan

• All 1440+ unit tests pass
• Benchmark verified on production database (~/.omniroute/storage.sqlite)
• No API changes (backward compatible)
• Query results verified to match previous behavior
• All pre-commit hooks pass (prettier, eslint)

Benchmark

Run node scripts/benchmark-call-logs.mjs to verify:

• Compares SELECT * vs selective columns
• Tests on actual production database
• Verifies data integrity

---

Related to: Performance optimizations (complements PR #1038)

[gemini-code-assist]

Code Review

This pull request introduces a performance optimization issue template, a benchmark script for evaluating call log queries, and refactors the getCallLogs function to use selective column fetching instead of SELECT *. The review feedback suggests further optimizing database I/O by using SUBSTR for large text columns where only existence is verified, improving the maintainability of the issue template by removing specific line numbers, and cleaning up a redundant dynamic import in the benchmark script.

[gemini-code-assist]

The query fetches the full content of the request_body and response_body columns, which can be up to 8KB each. However, the mapping logic in getCallLogs (lines 575-576) only uses these columns to determine if they are non-empty (setting hasRequestBody and hasResponseBody). Fetching the full text for every row in a list query is a significant and unnecessary I/O and memory overhead. Since the consumer only checks for string length, you can optimize this by fetching only the first character to verify existence without transferring the entire payload.

Suggested change

	"request_body",
	"response_body",
	"SUBSTR(request_body, 1, 1) AS request_body",
	"SUBSTR(response_body, 1, 1) AS response_body",

[gemini-code-assist]

Including specific line numbers in an issue template is generally discouraged as they will quickly become inaccurate as the codebase evolves. Consider using function names or general file paths instead to keep the template maintainable and relevant over time.

[gemini-code-assist]

This dynamic import is redundant because better-sqlite3 is already imported at the top of the file (line 7). Removing it simplifies the code and avoids unnecessary overhead.

Suggested change

	const { default: DatabaseCtor } = await import("better-sqlite3");
	const db = new DatabaseCtor(DB_PATH, { readonly: true });
	const db = new DatabaseCtor(DB_PATH, { readonly: true });

[diegosouzapw]

Thanks @oyi77 for this great contribution! 🎉 The PR has been integrated into the release/v3.5.5 branch and will be part of the next release. We appreciate your effort!