fix: Deduplicate parent projects in global index

Parent projects that contain sub-projects are now automatically removed
from the global registry. Reduces noise in cross-project queries by
eliminating duplicate matches. Existing duplicates are cleaned up on
next global_init run (215 → 167 projects in test index).

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: CSCSoftware

.claude/CLAUDE.md

@@ -222,6 +222,7 @@ aidex_global_refresh()                                                 # Stats u
 - Session-Cache (5-Min TTL) für schnelle wiederholte Queries
 - Bulk-Index: ≤500 Code-Dateien automatisch, >500 werden dem User gezeigt
 - Progress-UI: SSE-basiert auf Port 3334 mit Browser-Auto-Open
+- Auto-Deduplizierung: Parent-Projekte mit Sub-Projekten werden übersprungen
 
 ### Auto-Cleanup (v1.3.1)
 `aidex_init` entfernt automatisch Dateien die jetzt excluded sind (z.B. build/).

CHANGELOG.md

@@ -23,6 +23,10 @@ All notable changes to AiDex will be documented in this file.
   - Server-Sent Events (SSE) for real-time updates
   - Shows per-project status (indexing/done/error), progress bar, scrolling log
   - Dark theme, auto-closes after completion
+- **Project deduplication**: Parent projects that contain sub-projects are automatically removed
+  - e.g., `AudioGrabber/` is skipped when `AudioGrabber/AudioGrabber/` and `AudioGrabber/AudioGrabber2/` exist
+  - Existing duplicates in global DB are cleaned up on next `global_init` run
+  - Reduced test index from 215 to 167 projects (48 parent-duplicates removed)
 - **Extended excludes**: Better handling of embedded runtimes and external code
   - `init.ts`: Added `**/site-packages/**`, `**/Lib/**`, `**/fdk-aac/**` to DEFAULT_EXCLUDE
   - `global-init.ts`: Added Python venvs, embedded Python runtimes (Python310-313), `.cargo`, `packages`, `fdk-aac` to DEFAULT_EXCLUDED_DIRS

README.md

@@ -350,6 +350,7 @@ aidex_global_signatures({ name: "Player", kind: "class" })             # Find cl
 - Results are cached in memory (5-minute TTL) for fast repeated queries
 - Projects are batched (8 at a time) to respect SQLite's attachment limit
 - Each project keeps its own `.aidex/index.db` as the single source of truth
+- **Auto-deduplication**: Parent projects that contain sub-projects are automatically skipped (e.g., `MyApp/` is removed when `MyApp/Frontend/` and `MyApp/Backend/` exist as separate indexed projects)
 
 ### Management
 

src/commands/global/global-init.ts

@@ -139,6 +139,13 @@ export async function globalInit(params: GlobalInitParams): Promise<GlobalInitRe
     const globalDb = openGlobalDatabase();
 
     try {
+        // --------------------------------------------------------
+        // Deduplicate: Remove parent projects that contain sub-projects
+        // e.g., AudioGrabber/ contains AudioGrabber/AudioGrabber/ and AudioGrabber/AudioGrabber2/
+        // → keep only the sub-projects, skip the parent
+        // --------------------------------------------------------
+        const deduplicatedProjects = deduplicateProjects(scanResult.projects);
+
         // Get existing projects for comparison
         const existingProjects = new Map(
             globalDb.getProjects().map(p => [p.path.replace(/\\/g, '/'), p])
@@ -147,8 +154,8 @@ export async function globalInit(params: GlobalInitParams): Promise<GlobalInitRe
         let newCount = 0;
         let updatedCount = 0;
 
-        // Register each found project
-        for (const project of scanResult.projects) {
+        // Register each found project (deduplicated)
+        for (const project of deduplicatedProjects) {
             const normalizedPath = project.path.replace(/\\/g, '/');
             const stats = readProjectStats(project.path);
             if (!stats) continue;
@@ -185,9 +192,27 @@ export async function globalInit(params: GlobalInitParams): Promise<GlobalInitRe
             }
         }
 
-        // Collect paths of indexed projects for exclusion
+        // Remove parent-duplicate projects already in the global DB
+        // (from previous runs before deduplication was added)
+        const allRegistered = globalDb.getProjects();
+        const allPaths = allRegistered.map(p => ({
+            id: p.id,
+            path: p.path.replace(/\\/g, '/').replace(/\/+$/, '') + '/',
+        }));
+        for (const project of allPaths) {
+            const isParent = allPaths.some(other =>
+                other.id !== project.id &&
+                other.path.startsWith(project.path)
+            );
+            if (isParent) {
+                globalDb.unregisterProject(project.path.replace(/\/+$/, ''));
+                removedCount++;
+            }
+        }
+
+        // Collect paths of indexed projects for exclusion (use deduplicated list)
         const indexedPaths = new Set(
-            scanResult.projects.map(p => p.path.replace(/\\/g, '/'))
+            deduplicatedProjects.map(p => p.path.replace(/\\/g, '/'))
         );
 
         // Find projects without .aidex/ index
@@ -290,8 +315,8 @@ export async function globalInit(params: GlobalInitParams): Promise<GlobalInitRe
             success: true,
             searchPath,
             registered: params.indexUnindexed
-                ? scanResult.projects.length + (indexedResults?.filter(r => r.success).length ?? 0)
-                : scanResult.projects.length,
+                ? deduplicatedProjects.length + (indexedResults?.filter(r => r.success).length ?? 0)
+                : deduplicatedProjects.length,
             newProjects: newCount,
             updatedProjects: updatedCount,
             removedProjects: removedCount,
@@ -477,3 +502,36 @@ function findUnindexedProjects(searchPath: string, maxDepth: number, indexedPath
     walk(searchPath, 0);
     return projects;
 }
+
+/**
+ * Deduplicate projects: If project A is a parent directory of project B,
+ * keep only B (the more specific sub-project).
+ *
+ * Example: Given paths [AudioGrabber/, AudioGrabber/AudioGrabber/, AudioGrabber/AudioGrabber2/]
+ * → Remove AudioGrabber/ because it contains sub-projects.
+ * → Keep AudioGrabber/AudioGrabber/ and AudioGrabber/AudioGrabber2/.
+ */
+function deduplicateProjects<T extends { path: string; name: string }>(projects: T[]): T[] {
+    // Normalize all paths
+    const normalized = projects.map(p => ({
+        ...p,
+        _normPath: p.path.replace(/\\/g, '/').replace(/\/+$/, '') + '/',
+    }));
+
+    // Sort by path length descending (deepest first)
+    normalized.sort((a, b) => b._normPath.length - a._normPath.length);
+
+    // A project is a "parent" if any other project's path starts with it
+    const result: typeof projects = [];
+    for (const project of normalized) {
+        const isParent = normalized.some(other =>
+            other !== project &&
+            other._normPath.startsWith(project._normPath)
+        );
+        if (!isParent) {
+            result.push(project);
+        }
+    }
+
+    return result;
+}