@maxgfr/codeindex 2.11.1 → 2.13.0

package/src/types.ts

@@ -1,7 +1,7 @@
 // Single source of truth for the engine version the bundle reports. Kept in
 // lockstep with package.json by the release pipeline. Do not edit by hand
 // outside a release.
-export const ENGINE_VERSION = "2.11.1";
+export const ENGINE_VERSION = "2.13.0";
 
 // Bumped whenever the on-disk artifact shape changes, so a consumer can reject
 // an index written by an incompatible engine instead of misreading it. The
@@ -30,9 +30,18 @@ export const SCHEMA_VERSION = 4;
 // "reexport" when it resolves in-file. v8 fixes the C/C++ regex tier
 // reporting a function DEFINITION as a call to itself (`void load(void) {`
 // yielding a spurious call `load@<defline>`, found during the ultrasec
-// consumer migration) by excluding call candidates whose name+line match one
-// of the file's own extracted symbols.
-export const EXTRACTOR_VERSION = 8;
+// consumer migration) by excluding, from call candidates, only the
+// definition's own token on its definition line — not every occurrence of
+// that name+line pair, which would also drop genuine same-line calls (dense
+// one-liners packing several definitions, one-line recursion); v9 makes a
+// resolved export-alias symbol also cite the original declaration's own line
+// (and endLine, when the AST tier populated one) instead of the export
+// statement's line — a citation-precision fix (issue #9) for consumers
+// (ultradoc) that use file:line as evidence; v10 stops the regex tier
+// emitting a spurious exported class symbol named "extends" for
+// `export default class extends Base {}` (issue #11) — the file-stem
+// default symbol is unchanged.
+export const EXTRACTOR_VERSION = 10;
 
 // How a file is classified. `code` gets symbol/import extraction; `doc` gets
 // link/heading extraction; the rest are catalogued but not deeply parsed.

package/src/walk.ts

@@ -4,12 +4,15 @@ import { parseGitignore, isIgnored, type IgnoreRule } from "./ignore.js";
 
 // Directories that never carry signal for a documentation/code question and
 // would bloat the index (dependencies, build output, VCS internals, caches).
+// .codeindex is the engine's OWN output (index artifacts, pulled models, MCP
+// memories) — indexing it would feed memories into search and churn the scan
+// fingerprint on every write_memory (issue #12).
 // Exported so grep.ts can align ripgrep's universe with the walker's.
 export const IGNORE_DIRS = new Set([
   ".git", "node_modules", ".pnpm", "bower_components", "vendor", "dist", "build", "out",
   "target", ".next", ".nuxt", ".svelte-kit", ".turbo", "coverage", "__pycache__", ".venv",
   "venv", ".tox", ".mypy_cache", ".pytest_cache", ".gradle", ".idea", ".vscode", ".cache",
-  "tmp", ".ultraindex", "Pods", "DerivedData", ".terraform", "elm-stuff", ".dart_tool",
+  "tmp", ".ultraindex", ".codeindex", "Pods", "DerivedData", ".terraform", "elm-stuff", ".dart_tool",
 ]);
 
 // Lockfiles: huge, machine-generated, and pure noise for a code/docs question —
@@ -36,6 +39,13 @@ export interface WalkOptions {
   // semantics — see ignore.ts). Default TRUE: an ignored file is noise for
   // every consumer; pass false to index generated/ignored trees deliberately.
   gitignore?: boolean;
+  // Directory names to skip, REPLACING the default set entirely (not merging
+  // with it). IGNORE_DIRS is a public export, so consumers compose
+  // `[...IGNORE_DIRS, "extra"]` — or filter it — themselves; replace is the
+  // simplest contract. Deliberate scope boundary: grep.ts (the ripgrep
+  // universe) and the MCP server keep the DEFAULT set — recall consumers
+  // (e.g. ultrasec) consume scan/extract, not grep.
+  ignoreDirs?: string[];
 }
 
 export interface WalkedFile {
@@ -65,6 +75,9 @@ export function walk(root: string, opts: WalkOptions = {}): WalkResult {
   const maxFileBytes = opts.maxFileBytes ?? 1024 * 1024;
   const maxFiles = opts.maxFiles ?? DEFAULT_MAX_FILES;
   const useGitignore = opts.gitignore !== false;
+  // Effective ignored-directory set, built once: the caller's replacement when
+  // given (see WalkOptions.ignoreDirs — replace, never merge), else the default.
+  const ignoreDirs = opts.ignoreDirs ? new Set(opts.ignoreDirs) : IGNORE_DIRS;
   const out: WalkedFile[] = [];
   let capped = false;
   let excluded = 0;
@@ -125,7 +138,7 @@ export function walk(root: string, opts: WalkOptions = {}): WalkResult {
         continue;
       }
       if (st.isDirectory()) {
-        if (IGNORE_DIRS.has(name)) continue;
+        if (ignoreDirs.has(name)) continue;
         // An in-repo DIRECTORY symlink is skipped entirely: its target is (or
         // will be) walked under its canonical name, and letting both paths race
         // through the cycle guard would keep whichever readdir served first —