Code review fixes: SQL params, symlinks, FTS scaling, typing, tests#13
Open
aausch wants to merge 15 commits intoglommer:mainfrom
Open
Code review fixes: SQL params, symlinks, FTS scaling, typing, tests#13aausch wants to merge 15 commits intoglommer:mainfrom
aausch wants to merge 15 commits intoglommer:mainfrom
Conversation
Replace template literal interpolation with a prepared statement ? parameter, consistent with all other queries in the file. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The previous args pointed to the original developer's local machine path. Replace with just 'codemogger mcp' so the config works for anyone who has the package installed globally. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Grammars were scattered across ^0.23.x, ^0.24.x, and ^0.25.x against web-tree-sitter@0.26.5. Mismatched WASM ABI versions can cause silent parse failures or wrong ASTs. Align everything to ^0.25.0. Run `bun install` to apply. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Directory symlinks could recurse outside the intended tree (and into cycles). File symlinks could index files the user expects to be excluded. Skip both explicitly. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Previously an error from the embedder would propagate up and abort the entire index operation, leaving no record of what failed. Wrap each batch in try/catch: record the failure in IndexResult.errors and print a warning to stderr so the user sees a partial index rather than a crash. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
An empty catch was treating permission errors the same as 'no .gitignore exists'. Now only ENOENT is silently ignored; any other error (EACCES, EISDIR, etc.) is added to the scan error list. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Define private row types (CodebaseRow, StaleEmbeddingRow, etc.) for each query result and replace all 'as any[]' / 'as any' casts. Column renames are now caught at compile time. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Previously every index run dropped and recreated the entire FTS table. Now: - ensureFtsTable() creates the table+index once (idempotent) - populateFtsForFiles() inserts FTS entries per file after chunk upsert (the ON DELETE CASCADE on fts.chunk_id handles cleanup automatically when old chunks are deleted by batchUpsertAllFileChunks) - optimizeFts() runs OPTIMIZE INDEX once at the end Re-indexing a large codebase now only rewrites the changed files' FTS entries instead of rebuilding everything from scratch. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Previously ftsSearch did: - SELECT id FROM codebases (1 query) - N * SELECT from sqlite_master to check if each FTS table exists - M individual SELECT from chunks per FTS result row Now: - One SELECT from sqlite_master GLOB 'fts_*' discovers all tables - JOIN chunks inline in the FTS query eliminates per-row lookups Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add a note in README explaining why the pre-release version is used and flagging it for upgrade once a stable 0.5.x is released. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add inline comments explaining the rationale behind the two undocumented numeric constants. The RRF k=60 constant in rank.ts was already documented in its JSDoc comment. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
0.25.x versions have not been published for: tree-sitter-c, c-sharp, cpp, java, php, ruby, rust, scala, or typescript. Revert those to the latest available (0.23.x or 0.24.x). The version mismatch with web-tree-sitter@0.26.5 remains a known issue. It can be addressed when upstream grammar packages publish 0.25.x compatible WASM builds. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
walker.test.ts covers: - TypeScript file discovery - 1MB size limit - .gitignore pattern filtering - Symlink skipping (directory and file) - Unreadable .gitignore reported as error - Language filter index.test.ts covers: - Basic chunk-count assertion - Embedding batch failures captured in errors (not a crash) - Re-indexing unchanged files skips them (hash check) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Moves from the pre-release 0.5.0-pre.14 to the first stable 0.5.x release. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
boolean was pulled in via: @huggingface/transformers -> onnxruntime-node -> global-agent@3 -> boolean global-agent@4.x dropped boolean (and roarr) entirely. Added an overrides entry to force global-agent to ^4.1.3 since onnxruntime-node pins ^3.0.0. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Addresses issues found in a Claude-powered code review of the codebase.
Fixes
LIMITingetStaleEmbeddings(was interpolated via template literal).mcp.jsonwith justcodemogger mcp.gitignore(ENOENT, silent) from an unreadable one (other errors, now reported)IndexResult.errorsinstead of crashingRefactors
as any[]/as anyDB casts with typed row interfaces (CodebaseRow,StaleEmbeddingRow, etc.) so column renames are caught at compile timeensureFtsTable+ per-filepopulateFtsForFiles+optimizeFts. The existingON DELETE CASCADEonfts.chunk_idhandles cleanup automatically when chunks are replacedsqlite_master GLOB 'fts_*'query, and JOIN chunk data inline instead of per-row lookupsDependencies
@tursodatabase/databasefrom pre-release0.5.0-pre.14to stable0.5.1boolean@3.2.0transitive dep by overridingglobal-agentto^4.1.3(which dropped it)^0.25.0packages don't exist yet for most grammars; documented as a known issueDocs / chore
MAX_CHUNK_LINES,BATCH, and pre-release DB dependency@tursodatabase/databasepre-release statusTests
test/walker.test.ts: 7 new tests covering file discovery, size limit,.gitignore, symlink skipping (dir + file), unreadable.gitignore, and language filtertest/index.test.ts: 3 new tests covering chunk count, embed failure capture, and re-index deduplicationTest plan
bun test— all 22 tests passbun run build— CLI builds cleanlycodemogger index .on a real repo, verify FTS search still works.mcp.jsonworks after installing the package globally🤖 Generated with Claude Code