@cccarv82/freya 3.1.0 → 3.2.0

package/cli/web-ui.js

@@ -2498,6 +2498,15 @@
         msg += 'Contexto registrado no log diário. Nenhuma tarefa ou blocker identificado.\n';
       }
 
+      // Show semantic duplicates detected
+      if (summary && Array.isArray(summary.semanticDups) && summary.semanticDups.length > 0) {
+        msg += '\n⚠️ **Duplicatas detectadas** (não criadas):\n';
+        for (var di = 0; di < summary.semanticDups.length; di++) {
+          var dup = summary.semanticDups[di];
+          msg += '- "' + dup.newDesc + '" → já existe: "' + dup.existingDesc + '" (' + dup.similarity + ' similar)\n';
+        }
+      }
+
       if (summary && Array.isArray(summary.reportsSuggested) && summary.reportsSuggested.length) {
         msg += '\n**Relatórios sugeridos:** ' + summary.reportsSuggested.join(', ');
         msg += '\n\nUse: **Rodar relatórios sugeridos** (barra lateral)';

package/cli/web.js

@@ -3491,6 +3491,49 @@ async function cmdWeb({ port, dir, open, dev }) {
           const insertTask = dl.db.prepare(`INSERT INTO tasks (id, project_slug, description, category, status, metadata) VALUES (?, ?, ?, ?, ?, ?)`);
           const insertBlocker = dl.db.prepare(`INSERT INTO blockers (id, project_slug, title, severity, status, owner, next_action, metadata) VALUES (?, ?, ?, ?, ?, ?, ?, ?)`);
 
+          // ── Semantic deduplication (async, runs BEFORE the sync transaction) ──
+          // Load pending tasks for cosine similarity comparison
+          const pendingTasks = dl.db.prepare("SELECT id, description, project_slug FROM tasks WHERE status = 'PENDING'").all();
+          let pendingEmbeddings = []; // [{ id, description, project_slug, vector }]
+          try {
+            const { defaultEmbedder } = require(path.join(workspaceDir, 'scripts', 'lib', 'Embedder.js'));
+            for (const pt of pendingTasks) {
+              try {
+                const vec = await defaultEmbedder.embedText(pt.description);
+                pendingEmbeddings.push({ id: pt.id, description: pt.description, project_slug: pt.project_slug, vector: vec });
+              } catch { /* skip if embedding fails for individual task */ }
+            }
+          } catch (embErr) {
+            // Embedder not available — fall back to exact-match only
+            console.error('[dedup] Semantic dedup unavailable:', embErr.message);
+          }
+
+          // Pre-compute semantic duplicates for each action
+          const semanticDupMap = new Map(); // action index → existing task id (if duplicate)
+          const SIMILARITY_THRESHOLD = 0.78;
+          if (pendingEmbeddings.length > 0) {
+            try {
+              const { defaultEmbedder } = require(path.join(workspaceDir, 'scripts', 'lib', 'Embedder.js'));
+              for (let ai = 0; ai < actions.length; ai++) {
+                const a = actions[ai];
+                if (!a || a.type !== 'create_task') continue;
+                const desc = normalizeWhitespace(a.description);
+                if (!desc) continue;
+                try {
+                  const newVec = await defaultEmbedder.embedText(desc);
+                  let bestScore = 0, bestMatch = null;
+                  for (const pe of pendingEmbeddings) {
+                    const score = defaultEmbedder.cosineSimilarity(newVec, pe.vector);
+                    if (score > bestScore) { bestScore = score; bestMatch = pe; }
+                  }
+                  if (bestScore >= SIMILARITY_THRESHOLD && bestMatch) {
+                    semanticDupMap.set(ai, { existingId: bestMatch.id, existingDesc: bestMatch.description, score: bestScore });
+                  }
+                } catch { /* skip */ }
+              }
+            } catch { /* embedder not available */ }
+          }
+
           // BUG-31: Move deduplication queries INSIDE the transaction to eliminate TOCTOU race
           const applyTx = dl.db.transaction((actionsToApply) => {
             // Query for existing keys inside the transaction for atomicity
@@ -3499,7 +3542,8 @@ async function cmdWeb({ port, dir, open, dev }) {
             const recentBlockers = dl.db.prepare("SELECT title FROM blockers WHERE created_at >= datetime('now', '-1 day')").all();
             const existingBlockerKeys24h = new Set(recentBlockers.map(b => sha1(normalizeTextForKey(b.title))));
 
-            for (const a of actionsToApply) {
+            for (let ai = 0; ai < actionsToApply.length; ai++) {
+              const a = actionsToApply[ai];
               if (!a || typeof a !== 'object') continue;
               const type = String(a.type || '').trim();
 
@@ -3509,8 +3553,25 @@ async function cmdWeb({ port, dir, open, dev }) {
                 if (!description) continue;
                 const projectSlug = String(a.projectSlug || '').trim() || inferProjectSlug(description, slugMap);
                 const streamSlug = String(a.streamSlug || '').trim();
+
+                // Exact-match dedup (24h window)
                 const key = sha1(normalizeTextForKey((projectSlug ? projectSlug + ' ' : '') + description));
                 if (existingTaskKeys24h.has(key)) { applied.tasksSkipped++; continue; }
+
+                // Semantic dedup (all pending tasks)
+                if (semanticDupMap.has(ai)) {
+                  const dup = semanticDupMap.get(ai);
+                  applied.tasksSkipped++;
+                  if (!applied.semanticDups) applied.semanticDups = [];
+                  applied.semanticDups.push({
+                    newDesc: description,
+                    existingId: dup.existingId,
+                    existingDesc: dup.existingDesc,
+                    similarity: Math.round(dup.score * 100) + '%'
+                  });
+                  continue;
+                }
+
                 const category = validTaskCats.has(String(a.category || '').trim()) ? String(a.category).trim() : 'DO_NOW';
                 const priority = normPriority(a.priority);
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@cccarv82/freya",
-  "version": "3.1.0",
+  "version": "3.2.0",
   "description": "Personal AI Assistant with local-first persistence",
   "scripts": {
     "health": "node scripts/validate-data.js && node scripts/validate-structure.js",