@bendyline/gezel-service 1.0.4 → 1.0.6

package/dist/index-store/static-index-worker.js

@@ -9,6 +9,36 @@ import { extname as extname3, join as join10 } from "path";
 import { nowIso as nowIso2 } from "@bendyline/gezel";
 import { PROJECT_SHADOW_DIR_NAME as PROJECT_SHADOW_DIR_NAME2, projectLocalIndexDbFile } from "@bendyline/gezel/paths";
 
+// src/fs/sync-junk.ts
+import { isOutsideInInternalPath } from "@bendyline/gezel";
+var EXACT_NAMES = /* @__PURE__ */ new Set([
+  ".DS_Store",
+  "Thumbs.db",
+  "desktop.ini",
+  "Icon\r",
+  ".dropbox",
+  ".dropbox.attr"
+]);
+var SUFFIXES = [".tmp", ".temp", ".partial", ".crdownload", ".download", ".lnk"];
+function isOfficeLock(name) {
+  return name.startsWith("~$") || name.startsWith(".~lock.") && name.endsWith("#");
+}
+var SYNC_STAGING_DIRS = /* @__PURE__ */ new Set([".tmp.drivedownload", ".tmp.driveupload", ".dropbox.cache"]);
+function isSyncJunkName(name) {
+  if (EXACT_NAMES.has(name) || SYNC_STAGING_DIRS.has(name)) return true;
+  if (isOfficeLock(name)) return true;
+  const lower = name.toLowerCase();
+  return SUFFIXES.some((suffix) => lower.endsWith(suffix));
+}
+function isSyncJunkPath(relPath) {
+  return relPath.split("/").some((segment) => segment.length > 0 && isSyncJunkName(segment));
+}
+function isLibraryInternalPath(relPath) {
+  if (isOutsideInInternalPath(relPath)) return true;
+  if (isSyncJunkPath(relPath)) return true;
+  return relPath.split("/").includes(".versions");
+}
+
 // src/security/code-patterns.ts
 var EVIDENCE_CAP = 160;
 var CODE_PATTERNS = [
@@ -309,7 +339,7 @@ async function runGit(args, opts = {}) {
     "credential.interactive=false",
     ...cleanedArgs
   ];
-  return new Promise((resolve2, reject) => {
+  return new Promise((resolve3, reject) => {
     const hasStdinPayload = opts.stdin !== void 0;
     const child = hasStdinPayload ? spawn("git", gitArgs, {
       cwd: opts.cwd,
@@ -364,11 +394,11 @@ async function runGit(args, opts = {}) {
         return;
       }
       if (code === 0) {
-        resolve2({ stdout, stderr });
+        resolve3({ stdout, stderr });
         return;
       }
       if (code !== null && opts.acceptExitCodes?.includes(code)) {
-        resolve2({ stdout, stderr, code });
+        resolve3({ stdout, stderr, code });
         return;
       }
       reject(
@@ -427,13 +457,18 @@ var ALWAYS_SKIP_DIRS = /* @__PURE__ */ new Set([
   ".DS_Store"
 ]);
 async function discoverWorkspaceFiles(workspaceDir, opts) {
-  const gitPaths = await listGitVisiblePaths(workspaceDir, opts.maxFiles);
+  const gitPaths = await listGitVisiblePaths(
+    workspaceDir,
+    opts.maxFiles,
+    opts.gitTimeoutMs,
+    opts.ignorePath
+  );
   if (gitPaths) return statListedFiles(workspaceDir, gitPaths, opts.maxFiles);
   const out = [];
-  await walkFilesystem(workspaceDir, workspaceDir, out, opts.maxFiles);
+  await walkFilesystem(workspaceDir, workspaceDir, out, opts.maxFiles, opts.ignorePath);
   return out;
 }
-async function listGitVisiblePaths(workspaceDir, maxFiles) {
+async function listGitVisiblePaths(workspaceDir, maxFiles, timeoutMs = 3e4, ignorePath) {
   try {
     const { stdout } = await runGit(
       [
@@ -447,7 +482,7 @@ async function listGitVisiblePaths(workspaceDir, maxFiles) {
         "--",
         "."
       ],
-      { cwd: workspaceDir, timeoutMs: 3e4 }
+      { cwd: workspaceDir, timeoutMs }
     );
     const seen = /* @__PURE__ */ new Set();
     const paths = [];
@@ -455,6 +490,7 @@ async function listGitVisiblePaths(workspaceDir, maxFiles) {
       if (!raw) continue;
       const path = raw.replaceAll("\\", "/").replace(/^\.\//, "");
       if (!path || seen.has(path) || containsAlwaysSkippedDir(path)) continue;
+      if (ignorePath?.(path)) continue;
       seen.add(path);
       paths.push(path);
       if (paths.length >= maxFiles) break;
@@ -488,7 +524,7 @@ async function statListedFiles(workspaceDir, paths, maxFiles) {
   }
   return out;
 }
-async function walkFilesystem(root, dir, out, maxFiles) {
+async function walkFilesystem(root, dir, out, maxFiles, ignorePath) {
   if (out.length >= maxFiles) return;
   let entries;
   try {
@@ -500,13 +536,15 @@ async function walkFilesystem(root, dir, out, maxFiles) {
     if (out.length >= maxFiles) return;
     if (entry.isDirectory() && ALWAYS_SKIP_DIRS.has(entry.name)) continue;
     const abs = join(dir, entry.name);
+    const rel = relative(root, abs).replaceAll("\\", "/");
+    if (ignorePath?.(rel)) continue;
     if (entry.isDirectory()) {
-      await walkFilesystem(root, abs, out, maxFiles);
+      await walkFilesystem(root, abs, out, maxFiles, ignorePath);
     } else if (entry.isFile()) {
       try {
         const st = await lstat(abs);
         out.push({
-          path: relative(root, abs).replaceAll("\\", "/"),
+          path: rel,
           abs,
           size: st.size,
           mtimeMs: st.mtimeMs
@@ -750,7 +788,7 @@ async function renameWithWindowsRetry(from, to) {
 
 // src/fs/safe-paths.ts
 import { realpath as realpath2 } from "fs/promises";
-import { basename as basename2, dirname as dirname2, isAbsolute, join as join2, normalize, sep } from "path";
+import { basename as basename2, dirname as dirname2, isAbsolute, join as join2, normalize, resolve, sep } from "path";
 function pathEq(a, b) {
   if (process.platform === "win32" || process.platform === "darwin") {
     return a.toLowerCase() === b.toLowerCase();
@@ -1160,7 +1198,7 @@ async function resolveNodePermissionFlag(nodeBin) {
   }
 }
 async function readNodeHelp(nodeBin) {
-  return await new Promise((resolve2, reject) => {
+  return await new Promise((resolve3, reject) => {
     const child = spawn2(nodeBin, ["--help"], {
       stdio: ["ignore", "pipe", "pipe"],
       env: sandboxEnv(process.env),
@@ -1178,13 +1216,13 @@ async function readNodeHelp(nodeBin) {
     });
     child.once("error", reject);
     child.once("close", (code) => {
-      if (code === 0) resolve2(stdout);
+      if (code === 0) resolve3(stdout);
       else reject(new Error(`failed to inspect Node permission support (${code}): ${stderr}`));
     });
   });
 }
 async function runSandboxChild(command, args, opts) {
-  return await new Promise((resolve2) => {
+  return await new Promise((resolve3) => {
     const child = spawn2(command, args, {
       cwd: opts.cwd,
       stdio: opts.rpcChannel ? ["pipe", "pipe", "pipe", "pipe"] : ["pipe", "pipe", "pipe"],
@@ -1262,7 +1300,7 @@ async function runSandboxChild(command, args, opts) {
       if (timer) clearTimeout(timer);
       const signalSuffix = code === null && signal ? `[process closed by signal ${signal}]
 ` : "";
-      resolve2({
+      resolve3({
         exitCode: code ?? -1,
         signal,
         stdout,
@@ -1393,12 +1431,12 @@ async function wrapForPlatform(command, args, ctx) {
 }
 var macSandboxProbe;
 function canApplyMacSandbox() {
-  macSandboxProbe ??= new Promise((resolve2) => {
+  macSandboxProbe ??= new Promise((resolve3) => {
     let settled = false;
     const finish = (ok) => {
       if (settled) return;
       settled = true;
-      resolve2(ok);
+      resolve3(ok);
     };
     const child = spawn2("sandbox-exec", ["-p", "(version 1)\n(allow default)", "/usr/bin/true"], {
       stdio: ["ignore", "ignore", "ignore"],
@@ -1420,12 +1458,12 @@ function canApplyMacSandbox() {
 var linuxSystemdSandboxProbe;
 function canApplyLinuxSystemdSandbox() {
   if (process.platform !== "linux") return Promise.resolve(false);
-  linuxSystemdSandboxProbe ??= new Promise((resolve2) => {
+  linuxSystemdSandboxProbe ??= new Promise((resolve3) => {
     let settled = false;
     const finish = (ok) => {
       if (settled) return;
       settled = true;
-      resolve2(ok);
+      resolve3(ok);
     };
     const probe = "const net=require('node:net');const s=net.connect({host:'127.0.0.1',port:9});s.once('connect',()=>process.exit(9));s.once('error',e=>process.exit(['EAFNOSUPPORT','EPERM','EACCES'].includes(e.code)?0:2));setTimeout(()=>process.exit(3),1000);";
     const child = spawn2(
@@ -1541,7 +1579,7 @@ syncBuiltinESMExports();
 
 // src/utils/service-worker-entry.ts
 import { existsSync } from "fs";
-import { dirname as dirname3, join as join4, resolve } from "path";
+import { dirname as dirname3, join as join4, resolve as resolve2 } from "path";
 import { fileURLToPath } from "url";
 var SERVICE_WORKER_ENTRIES = {
   embeddings: { built: "memory/embed-worker.js" },
@@ -1559,7 +1597,7 @@ function findServiceBundleResource(moduleUrl, relativePaths) {
   for (let depth = 0; depth < MAX_ANCESTOR_DEPTH; depth++) {
     for (const root of [dir, join4(dir, "dist")]) {
       for (const relativePath of relativePaths) {
-        const candidate = resolve(root, relativePath);
+        const candidate = resolve2(root, relativePath);
         if (visited.has(candidate)) continue;
         visited.add(candidate);
         if (existsSync(candidate)) return candidate;
@@ -2851,7 +2889,8 @@ var IndexStore = class _IndexStore {
     if (!this.caps.fts) return [];
     return this.db.prepare(
       `SELECT file_path, name, kind, line_start, line_end, signature
-         FROM fts_symbols WHERE fts_symbols MATCH ? AND collection_id = ? LIMIT ?`
+         FROM fts_symbols WHERE fts_symbols MATCH ? AND collection_id = ?
+         ORDER BY bm25(fts_symbols) LIMIT ?`
     ).all(ftsPhrase(query), this.collectionId, limit).map((r) => ({
       id: symbolId(r.file_path, r.name),
       filePath: r.file_path,
@@ -3040,7 +3079,8 @@ var IndexStore = class _IndexStore {
     return this.db.prepare(
       `SELECT f.path AS file_path, s.snip FROM (
            SELECT content_hash, collection_id, snippet(fts_summaries, 0, '', '', '\u2026', 12) AS snip
-           FROM fts_summaries WHERE fts_summaries MATCH ? AND collection_id = ? LIMIT ?
+           FROM fts_summaries WHERE fts_summaries MATCH ? AND collection_id = ?
+           ORDER BY bm25(fts_summaries) LIMIT ?
          ) s
          JOIN files f ON f.hash = s.content_hash AND f.collection_id = s.collection_id
          LIMIT ?`
@@ -3163,8 +3203,12 @@ var IndexStore = class _IndexStore {
   searchDocs(query, limit = 20) {
     if (!this.caps.fts) return [];
     return this.db.prepare(
+      // FTS5 bm25() is lower-is-better, so plain ascending order is the
+      // best-first order. Without it FTS5 yields rowid (insertion) order and
+      // a LIMIT returns the oldest-indexed matches, not the closest ones.
       `SELECT file_path, line_start, chunk_id, snippet(fts_docs, 0, '', '', '\u2026', 12) AS snip
-         FROM fts_docs WHERE fts_docs MATCH ? AND collection_id = ? LIMIT ?`
+         FROM fts_docs WHERE fts_docs MATCH ? AND collection_id = ?
+         ORDER BY bm25(fts_docs) LIMIT ?`
     ).all(
       ftsPhrase(query),
       this.collectionId,
@@ -3406,6 +3450,42 @@ var IndexStore = class _IndexStore {
     ).get(this.collectionId, kind, rubricHash, rubricHash, maxAttempts);
     return row?.n ?? 0;
   }
+  /**
+   * How much indexing work landed in `[since, until)`, read off the tiers'
+   * own success stamps rather than any separate bookkeeping — which is what
+   * makes a Night Shift tally survive a daemon restart mid-window. Bounds
+   * are ISO-UTC strings, compared lexicographically (`nowIso()` is the only
+   * writer of these columns, so the format is uniform).
+   *
+   * Counted per content hash, so a file re-summarized after an edit counts
+   * again — that is real work done in the period, not double-counting.
+   */
+  workCountsSince(since, until) {
+    const count = (sql, ...params) => Number(this.db.prepare(sql).get(...params)?.n ?? 0);
+    return {
+      summarized: count(
+        `SELECT COUNT(*) AS n FROM summaries
+         WHERE collection_id = ? AND created_at >= ? AND created_at < ?`,
+        this.collectionId,
+        since,
+        until
+      ),
+      reviewed: count(
+        `SELECT COUNT(*) AS n FROM file_reviews
+         WHERE collection_id = ? AND reviewed_at >= ? AND reviewed_at < ?`,
+        this.collectionId,
+        since,
+        until
+      ),
+      described: count(
+        `SELECT COUNT(*) AS n FROM shadow_state
+         WHERE collection_id = ? AND state = 'ok' AND updated_at >= ? AND updated_at < ?`,
+        this.collectionId,
+        since,
+        until
+      )
+    };
+  }
   /**
    * Review coverage across the given rubric set. `reviewed` counts ANY
    * successful review on a current file hash regardless of rubric — coverage
@@ -4737,7 +4817,7 @@ function embeddablePngText(text) {
   }
   return parts.length > 0 ? ` ${parts.join(" ")}` : "";
 }
-async function indexWorkspaceContent(store, workspaceDir, artifactsDir) {
+async function indexWorkspaceContent(store, workspaceDir, artifactsDir, opts = {}) {
   const stats = {
     scanned: 0,
     changed: 0,
@@ -4749,7 +4829,15 @@ async function indexWorkspaceContent(store, workspaceDir, artifactsDir) {
   const seen = /* @__PURE__ */ new Set();
   const indexedAt = nowIso2();
   const forceCode = store.getMeta("extractor_version") !== String(EXTRACTOR_VERSION);
-  const walkedFiles = await discoverWorkspaceFiles(workspaceDir, { maxFiles: MAX_FILES });
+  const walkedFiles = await discoverWorkspaceFiles(workspaceDir, {
+    maxFiles: MAX_FILES,
+    // The shared document library is a user folder, not a source tree: it
+    // holds outside-in companion twins (a derived markdown view of a
+    // document already in the listing) and whatever a cloud-sync client
+    // leaves behind. Indexing either produces a second, worse hit for the
+    // same document.
+    ...opts.scope === "library" ? { ignorePath: isLibraryInternalPath } : {}
+  });
   for (const file of walkedFiles) {
     stats.scanned++;
     seen.add(file.path);
@@ -5046,7 +5134,8 @@ port.on("message", (request) => {
         const stats = await indexWorkspaceContent(
           index,
           request.workspaceDir,
-          request.artifactsDir
+          request.artifactsDir,
+          { ...request.scope ? { scope: request.scope } : {} }
         );
         port.postMessage({ id: request.id, stats });
       } finally {