ctxloom-pro 1.5.4 → 1.5.6

package/dist/{chunk-DVI2RWJR.js → chunk-JULFFD7O.js}

@@ -13,8 +13,19 @@ var VectorStore = class {
   db = null;
   table = null;
   initialized = false;
-  constructor(dbPath) {
+  /**
+   * Upserts since the last compaction. LanceDB writes 2 transactions per
+   * upsert (delete + add); without periodic compact_files() + cleanup, a
+   * long-lived MCP server accumulates tens of thousands of fragment FDs
+   * (observed: ~60k FDs / process after 18h of watcher-driven reindex).
+   */
+  upsertsSinceCompact = 0;
+  compactEvery;
+  cleanupOlderThanMs;
+  constructor(dbPath, options = {}) {
     this.dbPath = dbPath ?? path.join(process.cwd(), ".ctxloom", "vectors.lancedb");
+    this.compactEvery = options.compactEvery ?? 200;
+    this.cleanupOlderThanMs = options.cleanupOlderThanMs ?? 60 * 60 * 1e3;
   }
   async init() {
     if (this.initialized) return;
@@ -84,6 +95,41 @@ var VectorStore = class {
       content: content.slice(0, 512)
     };
     await this.table.add([record]);
+    this.upsertsSinceCompact++;
+    if (this.upsertsSinceCompact >= this.compactEvery) {
+      this.upsertsSinceCompact = 0;
+      await this.compact();
+    }
+  }
+  /**
+   * Merge fragments and prune old LanceDB versions. Idempotent and safe to
+   * call mid-flight; the Table API serializes writes internally. Called
+   * automatically every `compactEvery` upserts (default 200) to bound FD
+   * growth in long-lived MCP server processes.
+   *
+   * Uses optional-chaining so older `@lancedb/lancedb` builds without
+   * `optimize()` degrade to a no-op instead of crashing.
+   */
+  async compact() {
+    if (!this.table) return;
+    try {
+      const optimizable = this.table;
+      const result = await optimizable.optimize?.({
+        cleanupOlderThan: new Date(Date.now() - this.cleanupOlderThanMs)
+      });
+      if (result) {
+        logger.info("VectorStore.compact", {
+          fragmentsRemoved: result.compaction?.fragmentsRemoved ?? 0,
+          fragmentsAdded: result.compaction?.fragmentsAdded ?? 0,
+          oldVersionsRemoved: result.prune?.oldVersionsRemoved ?? 0,
+          bytesRemoved: result.prune?.bytesRemoved ?? 0
+        });
+      }
+    } catch (err) {
+      logger.warn("VectorStore.compact failed (non-fatal)", {
+        detail: err instanceof Error ? err.message : String(err)
+      });
+    }
   }
   /**
    * Search for the top-K most similar code records using vector search.
@@ -141,4 +187,4 @@ var VectorStore = class {
 export {
   VectorStore
 };
-//# sourceMappingURL=chunk-DVI2RWJR.js.map
+//# sourceMappingURL=chunk-JULFFD7O.js.map

package/dist/{chunk-UVR65QBJ.js → chunk-WDX4PJGL.js}

@@ -158,7 +158,7 @@ function collectFiles(dir, results = []) {
   return results;
 }
 async function indexDirectory(rootDir, onProgress) {
-  const { VectorStore } = await import("./VectorStore-XYLGD37W.js");
+  const { VectorStore } = await import("./VectorStore-2LVECRTY.js");
   const store = new VectorStore(path.join(rootDir, ".ctxloom", "vectors.lancedb"));
   await store.init();
   const files = collectFiles(rootDir);
@@ -211,4 +211,4 @@ export {
   collectFiles,
   indexDirectory
 };
-//# sourceMappingURL=chunk-UVR65QBJ.js.map
+//# sourceMappingURL=chunk-WDX4PJGL.js.map

package/dist/{embedder-R4KCXSGO.js → embedder-3AE4CSR7.js}

@@ -3,7 +3,7 @@ import {
   collectFiles,
   generateEmbedding,
   indexDirectory
-} from "./chunk-UVR65QBJ.js";
+} from "./chunk-WDX4PJGL.js";
 import "./chunk-TYDMSHV7.js";
 export {
   EMBEDDING_DIMENSION,
@@ -11,4 +11,4 @@ export {
   generateEmbedding,
   indexDirectory
 };
-//# sourceMappingURL=embedder-R4KCXSGO.js.map
+//# sourceMappingURL=embedder-3AE4CSR7.js.map

package/dist/index.js

@@ -22,6 +22,7 @@ import {
   activateLicense,
   aliasNotFoundError,
   captureError,
+  cleanupVectors,
   createToolRegistry,
   deactivateLicense,
   ensureVectorsInitialized,
@@ -29,6 +30,7 @@ import {
   getLicenseInfo,
   getTelemetryLevel,
   hashProjectRoot,
+  inspectVectorsDb,
   isActive,
   loadReviewConfig,
   noDefaultProjectError,
@@ -45,18 +47,18 @@ import {
   validateDefaultRoot,
   wrapWithIndexingEnvelope,
   writeCODEOWNERS
-} from "./chunk-4ZXFJSDG.js";
+} from "./chunk-FPMNXF4D.js";
 import {
   addCtxloomToConfig,
   detectInstalledClients
 } from "./chunk-II2DPYRJ.js";
 import {
   VectorStore
-} from "./chunk-DVI2RWJR.js";
+} from "./chunk-JULFFD7O.js";
 import {
   generateEmbedding,
   indexDirectory
-} from "./chunk-UVR65QBJ.js";
+} from "./chunk-WDX4PJGL.js";
 import "./chunk-5I6CJITG.js";
 import {
   logger
@@ -535,19 +537,36 @@ async function startServer(opts = {}) {
     });
     watcher.start();
     logger.info("File watcher active");
-    process.on("SIGINT", () => {
-      if (overlayRefreshTimer) clearTimeout(overlayRefreshTimer);
-      watcher.stop();
-      process.exit(0);
-    });
-    process.on("SIGTERM", () => {
-      if (overlayRefreshTimer) clearTimeout(overlayRefreshTimer);
-      watcher.stop();
+    const shutdown = (signal) => {
+      const forceExit = setTimeout(() => {
+        logger.warn("Shutdown timeout reached, force-exiting", { signal });
+        process.exit(1);
+      }, 5e3);
+      forceExit.unref();
+      try {
+        if (overlayRefreshTimer) clearTimeout(overlayRefreshTimer);
+        watcher.stop();
+      } catch (err) {
+        logger.warn("Shutdown step failed (non-fatal)", {
+          signal,
+          detail: err instanceof Error ? err.message : String(err)
+        });
+      }
       process.exit(0);
-    });
+    };
+    process.on("SIGINT", () => shutdown("SIGINT"));
+    process.on("SIGTERM", () => shutdown("SIGTERM"));
   } else {
-    process.on("SIGINT", () => process.exit(0));
-    process.on("SIGTERM", () => process.exit(0));
+    const shutdown = (signal) => {
+      const forceExit = setTimeout(() => {
+        logger.warn("Shutdown timeout reached, force-exiting", { signal });
+        process.exit(1);
+      }, 5e3);
+      forceExit.unref();
+      process.exit(0);
+    };
+    process.on("SIGINT", () => shutdown("SIGINT"));
+    process.on("SIGTERM", () => shutdown("SIGTERM"));
   }
 }
 
@@ -1020,7 +1039,7 @@ try {
 } catch {
 }
 var args = process.argv.slice(2);
-var ctxloomVersion = "1.5.4".length > 0 ? "1.5.4" : "dev";
+var ctxloomVersion = "1.5.6".length > 0 ? "1.5.6" : "dev";
 if (args.includes("--version") || args.includes("-v")) {
   process.stdout.write(`ctxloom ${ctxloomVersion}
 `);
@@ -1093,7 +1112,7 @@ async function checkLicense() {
   if (command !== void 0 && LICENSE_GATE_BYPASS_COMMANDS.has(command)) return;
   const ciKey = process.env["CTXLOOM_LICENSE_KEY"];
   if (ciKey) {
-    const { ApiClient } = await import("./src-AN6JHJMC.js");
+    const { ApiClient } = await import("./src-DL44T55H.js");
     const client = new ApiClient(process.env["CTXLOOM_API_BASE"]);
     try {
       const result = await client.validate(ciKey, "ci-ephemeral");
@@ -1471,7 +1490,7 @@ async function main() {
         }
         if (!skipHarness) {
           process.stdout.write("\n");
-          const { installHarness } = await import("./src-AN6JHJMC.js");
+          const { installHarness } = await import("./src-DL44T55H.js");
           const h = installHarness({ cwd: initRoot, dryRun, force, extraHosts });
           const harnessFiles = [
             h.claudeMd,
@@ -1534,7 +1553,7 @@ async function main() {
         process.exit(1);
       }
       if (alias !== void 0) {
-        const { validateAlias } = await import("./src-AN6JHJMC.js");
+        const { validateAlias } = await import("./src-DL44T55H.js");
         const v = validateAlias(alias);
         if (!v.ok) {
           console.error(`[ctxloom] Invalid alias: ${v.reason}`);
@@ -1607,6 +1626,101 @@ async function main() {
       }
       break;
     }
+    case "vectors-cleanup": {
+      const root = process.cwd();
+      const dryRun = hasFlag("--dry-run");
+      const force = hasFlag("--force");
+      const before = inspectVectorsDb(root);
+      if (before.txn + before.manifest + before.lance === 0) {
+        process.stdout.write(`  ${success("No vectors.lancedb to clean up \u2014 nothing to do.")}
+`);
+        break;
+      }
+      const mb = (before.totalBytes / (1024 * 1024)).toFixed(1);
+      process.stdout.write(header("Vectors cleanup"));
+      process.stdout.write(`  ${style.dim("Root:")} ${root}
+`);
+      process.stdout.write(`  ${style.dim("On-disk state:")}
+`);
+      process.stdout.write(`    ${style.bold(String(before.txn).padStart(6))} .txn files
+`);
+      process.stdout.write(`    ${style.bold(String(before.manifest).padStart(6))} .manifest files
+`);
+      process.stdout.write(`    ${style.bold(String(before.lance).padStart(6))} .lance fragments
+`);
+      process.stdout.write(`    ${style.bold(mb.padStart(6))} MB total
+
+`);
+      const activePids = [];
+      if (!force) {
+        try {
+          const dbPath = `${root}/.ctxloom/vectors.lancedb`;
+          const out = execSync(`lsof +D "${dbPath}" -F p 2>/dev/null || true`, {
+            encoding: "utf-8"
+          });
+          for (const line of out.split("\n")) {
+            if (line.startsWith("p")) {
+              const pid = parseInt(line.slice(1), 10);
+              if (Number.isFinite(pid) && pid !== process.pid) activePids.push(pid);
+            }
+          }
+        } catch {
+        }
+      }
+      if (activePids.length > 0) {
+        const uniq = [...new Set(activePids)];
+        process.stdout.write(
+          `  ${warn(`Refusing to clean \u2014 ${uniq.length} process(es) have files open:`)}
+`
+        );
+        for (const pid of uniq) {
+          process.stdout.write(`    PID ${pid}
+`);
+        }
+        process.stdout.write(
+          `
+  ${style.dim("Stop those ctxloom MCP servers first (close Claude Code windows or `kill <pid>`),")}
+`
+        );
+        process.stdout.write(
+          `  ${style.dim("then re-run. Use --force to override (not recommended \u2014 may corrupt the DB).")}
+`
+        );
+        process.exitCode = 1;
+        break;
+      }
+      const result = cleanupVectors({ rootDir: root, dryRun }, force ? [] : activePids);
+      if (!result.cleaned) {
+        process.stdout.write(`  ${warn(`Cleanup skipped: ${result.reason ?? "unknown"}`)}
+`);
+        break;
+      }
+      if (dryRun) {
+        process.stdout.write(
+          `  ${success(`Dry run \u2014 would have freed ${mb} MB across ${before.txn + before.manifest + before.lance} files.`)}
+`
+        );
+        process.stdout.write(`  ${style.dim("Re-run without --dry-run to actually clean up.")}
+`);
+      } else {
+        process.stdout.write(`  ${success(`Cleanup complete \u2014 freed ${mb} MB.`)}
+`);
+        if (result.backupPath) {
+          process.stdout.write(`  ${style.dim(`Backup: ${result.backupPath}`)}
+`);
+          process.stdout.write(
+            `  ${style.dim("Delete the backup with `rm -rf` once you confirm the next index works.")}
+`
+          );
+        }
+        process.stdout.write(
+          `
+  ${style.dim("Next ctxloom run will rebuild embeddings (~30-60s on a mid-sized repo).")}
+`
+        );
+      }
+      break;
+    }
     case "budget-stats": {
       const windowArg = args.find((a) => a.startsWith("--window="))?.split("=")[1] ?? "14d";
       const toolArg = args.find((a) => a.startsWith("--tool="))?.split("=")[1];
@@ -1798,7 +1912,7 @@ Suggested reviewers for ${files.length} file(s):`);
         process.stderr.write("[ctxloom] --limit must be a non-negative integer (0 for unlimited)\n");
         process.exit(2);
       }
-      const { loadRulesConfig, RulesChecker, formatText, formatJson, RulesConfigError } = await import("./src-AN6JHJMC.js");
+      const { loadRulesConfig, RulesChecker, formatText, formatJson, RulesConfigError } = await import("./src-DL44T55H.js");
       let config;
       try {
         config = await loadRulesConfig(root);
@@ -1822,7 +1936,7 @@ Suggested reviewers for ${files.length} file(s):`);
       }
       let graph;
       if (useSnapshot) {
-        const { DependencyGraph: DG } = await import("./src-AN6JHJMC.js");
+        const { DependencyGraph: DG } = await import("./src-DL44T55H.js");
         graph = new DG();
         const loaded = await graph.loadSnapshotOnly(root);
         if (!loaded) {
@@ -1831,7 +1945,7 @@ Suggested reviewers for ${files.length} file(s):`);
         }
       } else {
         process.stderr.write("[ctxloom] Building dependency graph...\n");
-        const { ASTParser: ASTParser2, DependencyGraph: DependencyGraph2 } = await import("./src-AN6JHJMC.js");
+        const { ASTParser: ASTParser2, DependencyGraph: DependencyGraph2 } = await import("./src-DL44T55H.js");
         let parser;
         try {
           parser = new ASTParser2();
@@ -1892,6 +2006,9 @@ Usage:
   ctxloom rules check --json       Output violations as JSON
   ctxloom rules check --use-snapshot  Fast mode: use existing graph snapshot
   ctxloom rules check --limit=N   Show first N violations (default 50, 0=unlimited)
+  ctxloom vectors-cleanup      Clear accumulated LanceDB version state to free FDs
+                                (use --dry-run to preview, --force to skip the
+                                 active-process safety check)
   ctxloom --version            Print installed version and exit
   ctxloom --help               Show this help
 

package/dist/{src-AN6JHJMC.js → src-DL44T55H.js}

@@ -61,6 +61,7 @@ import {
   buildBlastRadiusXml,
   buildCodeownersBlock,
   captureError,
+  cleanupVectors,
   computeBlockHmac,
   computeRiskBreakdown,
   computeRiskCaps,
@@ -89,6 +90,7 @@ import {
   getTaskBudgetTracker,
   getTelemetryLevel,
   hashProjectRoot,
+  inspectVectorsDb,
   installHarness,
   isActive,
   isSiloed,
@@ -129,16 +131,16 @@ import {
   wrapBlock,
   wrapWithIndexingEnvelope,
   writeCODEOWNERS
-} from "./chunk-4ZXFJSDG.js";
+} from "./chunk-FPMNXF4D.js";
 import {
   VectorStore
-} from "./chunk-DVI2RWJR.js";
+} from "./chunk-JULFFD7O.js";
 import {
   EMBEDDING_DIMENSION,
   collectFiles,
   generateEmbedding,
   indexDirectory
-} from "./chunk-UVR65QBJ.js";
+} from "./chunk-WDX4PJGL.js";
 import {
   filenameForDate,
   readEvents,
@@ -216,6 +218,7 @@ export {
   buildBlastRadiusXml,
   buildCodeownersBlock,
   captureError,
+  cleanupVectors,
   collectFiles,
   computeBlockHmac,
   computeRiskBreakdown,
@@ -248,6 +251,7 @@ export {
   getTelemetryLevel,
   hashProjectRoot,
   indexDirectory,
+  inspectVectorsDb,
   installHarness,
   isActive,
   isSiloed,
@@ -294,4 +298,4 @@ export {
   wrapWithIndexingEnvelope,
   writeCODEOWNERS
 };
-//# sourceMappingURL=src-AN6JHJMC.js.map
+//# sourceMappingURL=src-DL44T55H.js.map

package/dist/workers/indexerWorker.js

@@ -1,9 +1,9 @@
 import {
   VectorStore
-} from "../chunk-DVI2RWJR.js";
+} from "../chunk-JULFFD7O.js";
 import {
   generateEmbedding
-} from "../chunk-UVR65QBJ.js";
+} from "../chunk-WDX4PJGL.js";
 import "../chunk-TYDMSHV7.js";
 
 // packages/core/src/workers/indexerWorker.ts

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ctxloom-pro",
-  "version": "1.5.4",
+  "version": "1.5.6",
   "description": "ctxloom — The Universal Code Context Engine. A local-first MCP server providing intelligent code context via hybrid Vector + AST + Graph search with Skeletonization (92% token reduction).",
   "type": "module",
   "main": "dist/index.js",