karajan-code 4.49.4 → 4.49.5

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "karajan-code",
-  "version": "4.49.4",
+  "version": "4.49.5",
   "description": "Local multi-agent coding orchestrator with TDD, SonarQube, and code review pipeline",
   "type": "module",
   "license": "AGPL-3.0",

package/src/commands/rag.js

@@ -10,6 +10,7 @@ import { indexFile, indexProject, indexProjectDelta, sourceKind } from "../rag/i
 import { indexableReason, ragExclude } from "../rag/indexable.js";
 import { fileIndexState } from "../rag/coverage.js";
 import { query } from "../rag/retriever.js";
+import { pathNamedHits, withPathNamedHits } from "../rag/path-hits.js";
 import { installPostMergeHook, maybeAutoUpdate } from "../rag/auto-update.js";
 import { indexLibrary, LIBRARY_PROJECT } from "../rag/library.js";
 import { loadGoldenQueries, runEval } from "../rag/eval.js";
@@ -166,7 +167,8 @@ export async function ragQueryCommand({ text, config, logger, flags = {} }) {
     let where = flags.where || null;
     if (library) where = flags.where ? `kind=library AND ${flags.where}` : "kind=library";
     const rerankOpts = flags.rerank ? { model: flags.rerankModel } : null;
-    const hits = await query(db, makeGovernedEmbedder(config), text, { topK, scope, project, mode, alpha, where, rerankOpts });
+    const ranked = await query(db, makeGovernedEmbedder(config), text, { topK, scope, project, mode, alpha, where, rerankOpts });
+    const hits = library ? ranked : withPathNamedHits(pathNamedHits(db, text, { project, projectDir: config?.projectDir || process.cwd() }), ranked, topK);
     if (flags.json) {
       process.stdout.write(`${JSON.stringify(hits)}\n`);
     } else {

package/src/mcp/handlers/rag-handler.js

@@ -9,6 +9,7 @@ import { makeGovernedEmbedder } from "../../rag/governed-embedder.js";
 import { indexProject } from "../../rag/indexer.js";
 import { ragExclude } from "../../rag/indexable.js";
 import { query } from "../../rag/retriever.js";
+import { pathNamedHits, withPathNamedHits } from "../../rag/path-hits.js";
 import { getKarajanHome } from "../../utils/paths.js";
 import { resolveProjectDir, buildConfig, responseText, failPayload } from "../shared-helpers.js";
 
@@ -40,7 +41,9 @@ export async function handleRagQuery(args, server) {
         const remedy = emptyIndexRemedy({ slug: project, legacyPath: join(getKarajanHome(), "rag.db"), dim });
         return responseText({ hits: [], empty: true, topK, scope, project, remedy });
       }
-      const hits = await query(db, makeGovernedEmbedder(config), text, { topK, scope, project });
+      const ranked = await query(db, makeGovernedEmbedder(config), text, { topK, scope, project });
+      // KJC-BUG-0348: a file the question names comes first, as in the CLI.
+      const hits = withPathNamedHits(pathNamedHits(db, text, { project, projectDir }), ranked, topK);
       return responseText({ hits, empty: false, topK, scope, project });
     } finally { db.close(); }
   } catch (err) {

package/src/rag/path-hits.js

@@ -0,0 +1,36 @@
+/**
+ * KJC-BUG-0348 (#2115): ranking alone decides what a query returns, and in a
+ * large index a freshly indexed file (an .astro page) never made the top hits,
+ * neither by its path nor by its text: the rag-first gate kept blocking it.
+ * A query that NAMES an indexed file returns that file first, deterministically.
+ */
+import { isAbsolute, relative } from "node:path";
+
+// A path-like token: segments joined by "/" ending in a file extension.
+const PATH_TOKEN = /[\w@.-]+(?:\/[\w@.-]+)*\.[A-Za-z0-9]+/g;
+
+const parseMetadata = (raw) => {
+  try { return raw ? JSON.parse(raw) : {}; } catch { return {}; }
+};
+
+/** @returns {object[]} the first chunk of each indexed file of `project` the text names */
+export function pathNamedHits(db, text, { project, projectDir }) {
+  const tokens = [...new Set(String(text || "").match(PATH_TOKEN) ?? [])];
+  if (!project || tokens.length === 0) return [];
+  const sources = db.prepare("SELECT DISTINCT source FROM chunks WHERE project_slug = ?").all(project).map((r) => r.source);
+  const named = sources.filter((source) => {
+    const rel = (isAbsolute(source) ? relative(projectDir, source) : source).split("\\").join("/");
+    return tokens.some((t) => rel === t || rel.endsWith(`/${t}`));
+  });
+  const first = db.prepare("SELECT id, source, kind, text, metadata, project_slug FROM chunks WHERE project_slug = ? AND source = ? ORDER BY id LIMIT 1");
+  return named.map((source) => {
+    const row = first.get(project, source);
+    return { ...row, metadata: parseMetadata(row.metadata), distance: 0, score: 0, matchedBy: "path" };
+  });
+}
+
+/** The named files first, then the ranked hits, never the same chunk twice, at most topK. */
+export function withPathNamedHits(named, ranked, topK) {
+  const ids = new Set(named.map((h) => h.id));
+  return [...named, ...ranked.filter((h) => !ids.has(h.id))].slice(0, topK);
+}