@massa-ai/tools-api 1.35.0 → 1.36.0

package/dist/index.js

@@ -116739,6 +116739,7 @@ class IndexJobTracker {
           filesIndexed: r2?.filesIndexed ?? 0,
           chunksIndexed: r2?.chunksIndexed ?? 0,
           symbolsIndexed: 0,
+          errors: r2?.errors ?? 0,
           durationMs: r2?.duration ?? 0,
           activatedGraphGenerationId: r2?.activatedGraphGenerationId ?? ""
         });
@@ -116760,6 +116761,11 @@ var init_index_job_tracker = __esm(() => {
   indexJobTracker = IndexJobTracker.getInstance();
 });
 
+// ../../packages/core/dist/kernel/sanitize/strip-nul.js
+function stripNul(content) {
+  return content.includes("\x00") ? content.replaceAll("\x00", "") : content;
+}
+
 // ../../packages/core/dist/services/etl/stages/discover.js
 import fs9 from "fs/promises";
 import path15 from "path";
@@ -116850,7 +116856,7 @@ class DiscoverStage {
     const absolutePath = path15.join(ctx.projectPath, relativePath);
     try {
       const stat2 = await fs9.stat(absolutePath);
-      const content = await fs9.readFile(absolutePath, "utf-8");
+      const content = stripNul(await fs9.readFile(absolutePath, "utf-8"));
       const contentHash = createHash5("sha256").update(content).digest("hex");
       let needsReparse = forceReindex;
       if (!forceReindex) {
@@ -120911,6 +120917,7 @@ class LoadStage {
     let chunksLoaded = 0;
     let symbolsLoaded = 0;
     let errors5 = 0;
+    const fileErrors = [];
     let processedSinceStart = 0;
     let lastEtaLogAt = 0;
     const BATCH = 10;
@@ -120981,6 +120988,12 @@ class LoadStage {
         } catch (err) {
           errors5++;
           processedSinceStart++;
+          if (fileErrors.length < MAX_FILE_ERRORS) {
+            fileErrors.push({
+              filePath: file3.file.relativePath,
+              error: err.message
+            });
+          }
           if (emitLifecycle)
             ctx.emit({
               type: "file_error",
@@ -121046,7 +121059,7 @@ class LoadStage {
       duration: formatDuration(durationMs),
       filesPerSec: filesLoaded > 0 && durationMs > 0 ? Number((filesLoaded / durationMs * 1000).toFixed(2)) : 0
     });
-    return { filesLoaded, chunksLoaded, symbolsLoaded, errors: errors5 };
+    return { filesLoaded, chunksLoaded, symbolsLoaded, errors: errors5, fileErrors };
   }
   async loadToSearchStores(ctx, file3) {
     if (file3.chunks.length === 0)
@@ -121118,6 +121131,7 @@ class LoadStage {
     return batch.definitions.length;
   }
 }
+var MAX_FILE_ERRORS = 50;
 var init_load = __esm(() => {
   init_dist();
   init_with_deadlock_retry();
@@ -121298,8 +121312,8 @@ class GraphGenerationRepositoryPg {
   async heartbeat(lease, leaseTtlMs) {
     const ttl = validateTtl2(leaseTtlMs);
     return getPrismaClient2().$transaction(async (tx) => {
-      const workspace = await lockWorkspace(tx, lease.projectId);
       const generation = await lockGeneration(tx, lease.projectId, lease.generationId);
+      const workspace = await lockWorkspace(tx, lease.projectId);
       if (!leaseMatches(workspace, generation, lease))
         return { status: "lease_lost" };
       const renewed = await tx.$queryRaw`
@@ -121322,8 +121336,8 @@ class GraphGenerationRepositoryPg {
   }
   async complete(lease) {
     return getPrismaClient2().$transaction(async (tx) => {
-      const workspace = await lockWorkspace(tx, lease.projectId);
       const generation = await lockGeneration(tx, lease.projectId, lease.generationId);
+      const workspace = await lockWorkspace(tx, lease.projectId);
       if (!leaseMatches(workspace, generation, lease))
         return { status: "lease_lost" };
       const live = await tx.$queryRaw`
@@ -121352,8 +121366,8 @@ class GraphGenerationRepositoryPg {
   }
   async activate(lease) {
     return getPrismaClient2().$transaction(async (tx) => {
-      const workspace = await lockWorkspace(tx, lease.projectId);
       const generation = await lockGeneration(tx, lease.projectId, lease.generationId);
+      const workspace = await lockWorkspace(tx, lease.projectId);
       if (!leaseMatches(workspace, generation, lease))
         return { status: "lease_lost" };
       if (workspace.active_graph_generation_id !== generation.expected_active_id) {
@@ -121403,8 +121417,8 @@ class GraphGenerationRepositoryPg {
   async abort(lease, reason) {
     const failureReason = boundedText2(reason, "reason", MAX_FAILURE_REASON);
     return getPrismaClient2().$transaction(async (tx) => {
-      const workspace = await lockWorkspace(tx, lease.projectId);
       const generation = await lockGeneration(tx, lease.projectId, lease.generationId);
+      const workspace = await lockWorkspace(tx, lease.projectId);
       if (!leaseMatches(workspace, generation, lease))
         return { status: "lease_lost" };
       const live = await tx.$queryRaw`
@@ -121480,7 +121494,7 @@ class GraphGenerationCoordinator {
   async begin(input) {
     const deadline = Date.now() + GRAPH_GENERATION_LEASE_TTL_MS;
     for (;; ) {
-      const outcome2 = await this.repository.begin({ ...input, leaseTtlMs: GRAPH_GENERATION_LEASE_TTL_MS });
+      const outcome2 = await withDeadlockRetry(() => this.repository.begin({ ...input, leaseTtlMs: GRAPH_GENERATION_LEASE_TTL_MS }), { operation: "graph_generation.begin", maxAttempts: 5 });
       if (outcome2.status === "acquired")
         return outcome2.lease;
       if (outcome2.status === "stale_active") {
@@ -121492,12 +121506,12 @@ class GraphGenerationCoordinator {
     }
   }
   async heartbeat(lease) {
-    const outcome2 = await withDeadlockRetry(() => this.repository.heartbeat(lease, GRAPH_GENERATION_LEASE_TTL_MS), { operation: "graph_generation.heartbeat", maxAttempts: 3 });
+    const outcome2 = await withDeadlockRetry(() => this.repository.heartbeat(lease, GRAPH_GENERATION_LEASE_TTL_MS), { operation: "graph_generation.heartbeat", maxAttempts: 5 });
     if (outcome2.status !== "renewed")
       throw new Error("graph_generation_lease_lost");
   }
   async activate(lease) {
-    const completeness = await this.repository.complete(lease);
+    const completeness = await withDeadlockRetry(() => this.repository.complete(lease), { operation: "graph_generation.complete", maxAttempts: 5 });
     if (completeness.status === "incomplete") {
       throw new Error(`graph_generation_incomplete:${completeness.reasons.join(",")}`);
     }
@@ -121506,7 +121520,7 @@ class GraphGenerationCoordinator {
     if (completeness.status === "stale_active") {
       throw new Error(`graph_generation_stale_active:${completeness.activeGenerationId ?? "none"}`);
     }
-    const activation = await this.repository.activate(lease);
+    const activation = await withDeadlockRetry(() => this.repository.activate(lease), { operation: "graph_generation.activate", maxAttempts: 5 });
     if (activation.status === "activated")
       return activation;
     if (activation.status === "incomplete") {
@@ -121517,7 +121531,7 @@ class GraphGenerationCoordinator {
     throw new Error(`graph_generation_stale_active:${activation.activeGenerationId ?? "none"}`);
   }
   async abort(lease, reason) {
-    const outcome2 = await this.repository.abort(lease, reason);
+    const outcome2 = await withDeadlockRetry(() => this.repository.abort(lease, reason), { operation: "graph_generation.abort", maxAttempts: 5 });
     if (outcome2.status === "lease_lost")
       throw new Error("graph_generation_lease_lost_during_abort");
   }
@@ -121860,7 +121874,12 @@ var init_pipeline = __esm(() => {
           throw managedRunLeaseLost;
         stageTimings.load = Math.round(performance.now() - st4);
         if (loadResult.errors > 0) {
-          throw new Error(`ETL completed with ${loadResult.errors} file error${loadResult.errors === 1 ? "" : "s"}`);
+          logger.warn("EtlPipeline: completed with file errors", {
+            projectId,
+            jobId,
+            errors: loadResult.errors,
+            fileErrors: (loadResult.fileErrors ?? []).slice(0, 10)
+          });
         }
         const activationSnapshot = await abortable(this.discover.run(ctx, { forceReindex, includeTests: include_tests }), graphAbortController.signal);
         if (buildGraphInputSnapshotHash(activationSnapshot) !== graphGenerationLease.inputSnapshotHash) {
@@ -121917,6 +121936,7 @@ var init_pipeline = __esm(() => {
           filesIndexed: result.filesIndexed,
           chunksIndexed: result.chunksIndexed,
           errors: result.errors,
+          fileErrors: loadResult.fileErrors ?? [],
           duration: durationMs,
           activatedGraphGenerationId: result.activatedGraphGenerationId,
           parserDiagnostics: result.parserDiagnostics
@@ -121927,6 +121947,7 @@ var init_pipeline = __esm(() => {
           filesIndexed: result.filesIndexed,
           chunksIndexed: result.chunksIndexed,
           symbolsIndexed: result.symbolsIndexed,
+          errors: result.errors,
           durationMs,
           activatedGraphGenerationId: result.activatedGraphGenerationId
         });

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@massa-ai/tools-api",
-  "version": "1.35.0",
+  "version": "1.36.0",
   "author": "luizgmassa",
   "description": "massa-ai REST API server - Semantic code search, memory, and context compression",
   "type": "module",
@@ -21,8 +21,8 @@
     "test": "bun scripts/run-tests-isolated.ts"
   },
   "dependencies": {
-    "@massa-ai/core": "^1.35.0",
-    "@massa-ai/shared": "^1.35.0",
+    "@massa-ai/core": "^1.36.0",
+    "@massa-ai/shared": "^1.36.0",
     "elysia": "^1.2.25",
     "@elysiajs/swagger": "^1.2.0",
     "@elysiajs/cors": "^1.2.0",