@threadbase-sh/scanner 0.10.2 → 0.11.1

package/dist/cli.js

@@ -5,7 +5,7 @@ import { Command } from "commander";
 import pino2 from "pino";
 
 // package.json
-var version = "0.10.2";
+var version = "0.11.1";
 
 // src/logger.ts
 import pino from "pino";
@@ -62,7 +62,7 @@ async function saveProfiles(profiles, configPath) {
 import { EventEmitter } from "events";
 import { closeSync as closeSync2, openSync as openSync2, readSync as readSync2, statSync as statSync3 } from "fs";
 import { homedir as homedir2 } from "os";
-import { join as join4 } from "path";
+import { join as join5 } from "path";
 
 // src/cache.ts
 var LRUCache = class {
@@ -100,6 +100,12 @@ var LRUCache = class {
   }
 };
 
+// src/canonical-path.ts
+import { sep } from "path";
+function canonicalPath(p) {
+  return sep === "\\" ? p.replace(/\//g, "\\") : p;
+}
+
 // src/discovery.ts
 import fg from "fast-glob";
 import { stat } from "fs/promises";
@@ -141,7 +147,7 @@ async function discoverJsonlFiles(dirs, onProgress) {
         if (size < 0) {
           skippedInaccessible++;
         } else if (size > 0) {
-          results.push({ filePath, account });
+          results.push({ filePath: canonicalPath(filePath), account });
           kept++;
         } else {
           skippedEmpty++;
@@ -1173,7 +1179,7 @@ var CodexCliProvider = class {
       for (const filePath of paths) {
         try {
           const s = await stat2(filePath);
-          if (s.size > 0) results.push({ filePath, account: "codex" });
+          if (s.size > 0) results.push({ filePath: canonicalPath(filePath), account: "codex" });
         } catch (err) {
           log.warn({ filePath, err }, "codex discovery: stat failed");
         }
@@ -1632,6 +1638,7 @@ function openDatabase(dbPath) {
 
 // src/persistent/dir-watermark.ts
 import { readdir, stat as stat3 } from "fs/promises";
+import { dirname as dirname4, join as join4 } from "path";
 var FULL_RECONCILE_EVERY_N_SCANS = 20;
 async function discoverJsonlFilesGated(dirs, files, scannedDirs, options = {}) {
   const log = getLogger();
@@ -1641,7 +1648,7 @@ async function discoverJsonlFilesGated(dirs, files, scannedDirs, options = {}) {
   const results = [];
   for (const rawDir of dirs) {
     const { account } = rawDir;
-    const projectsDir = rawDir.projectsDir.replace(/\\/g, "/");
+    const projectsDir = canonicalPath(rawDir.projectsDir);
     const resolved = await resolveProjectDirs(projectsDir, scannedDirs);
     if (resolved === null) continue;
     for (const projectDir of resolved.entries) {
@@ -1702,11 +1709,10 @@ async function resolveProjectDirs(projectsDir, scannedDirs) {
   return { entries, commitRoot: { mtimeMs: rootStat.mtimeMs } };
 }
 function dirnameOf(filePath) {
-  const idx = filePath.lastIndexOf("/");
-  return idx === -1 ? filePath : filePath.slice(0, idx);
+  return dirname4(filePath);
 }
 function joinPath(dir, name) {
-  return dir.endsWith("/") ? `${dir}${name}` : `${dir}/${name}`;
+  return join4(dir, name);
 }
 
 // src/persistent/repositories/checkpoints.repo.ts
@@ -1716,15 +1722,16 @@ var CheckpointsRepo = class {
   }
   db;
   replaceAll(sourcePath, checkpoints) {
+    const path = canonicalPath(sourcePath);
     const tx = this.db.transaction(() => {
-      this.db.prepare("DELETE FROM message_checkpoints WHERE source_path = ?").run(sourcePath);
+      this.db.prepare("DELETE FROM message_checkpoints WHERE source_path = ?").run(path);
       const insert = this.db.prepare(
         `INSERT INTO message_checkpoints
            (source_path, message_index, byte_offset, line_number, parser_state)
          VALUES (?, ?, ?, ?, ?)`
       );
       for (const c of checkpoints) {
-        insert.run(sourcePath, c.messageIndex, c.byteOffset, c.lineNumber, JSON.stringify(c.state));
+        insert.run(path, c.messageIndex, c.byteOffset, c.lineNumber, JSON.stringify(c.state));
       }
     });
     tx();
@@ -1733,6 +1740,7 @@ var CheckpointsRepo = class {
   // the chain covering the immutable prefix (Kafka sparse-index style); rows are
   // only ever removed on truncation/replace or deletion.
   append(sourcePath, checkpoints) {
+    const path = canonicalPath(sourcePath);
     const tx = this.db.transaction(() => {
       const insert = this.db.prepare(
         `INSERT INTO message_checkpoints
@@ -1740,7 +1748,7 @@ var CheckpointsRepo = class {
          VALUES (?, ?, ?, ?, ?)`
       );
       for (const c of checkpoints) {
-        insert.run(sourcePath, c.messageIndex, c.byteOffset, c.lineNumber, JSON.stringify(c.state));
+        insert.run(path, c.messageIndex, c.byteOffset, c.lineNumber, JSON.stringify(c.state));
       }
     });
     tx();
@@ -1753,7 +1761,7 @@ var CheckpointsRepo = class {
          FROM message_checkpoints
          WHERE source_path = ? AND message_index <= ?
          ORDER BY message_index DESC LIMIT 1`
-    ).get(sourcePath, messageIndex);
+    ).get(canonicalPath(sourcePath), messageIndex);
     return row ? toCheckpoint(row) : null;
   }
   // The highest-index checkpoint for a file, or null if none. The resume point
@@ -1764,14 +1772,14 @@ var CheckpointsRepo = class {
          FROM message_checkpoints
          WHERE source_path = ?
          ORDER BY message_index DESC LIMIT 1`
-    ).get(sourcePath);
+    ).get(canonicalPath(sourcePath));
     return row ? toCheckpoint(row) : null;
   }
   count(sourcePath) {
-    return this.db.prepare("SELECT COUNT(*) AS n FROM message_checkpoints WHERE source_path = ?").get(sourcePath).n;
+    return this.db.prepare("SELECT COUNT(*) AS n FROM message_checkpoints WHERE source_path = ?").get(canonicalPath(sourcePath)).n;
   }
   remove(sourcePath) {
-    this.db.prepare("DELETE FROM message_checkpoints WHERE source_path = ?").run(sourcePath);
+    this.db.prepare("DELETE FROM message_checkpoints WHERE source_path = ?").run(canonicalPath(sourcePath));
   }
 };
 function toCheckpoint(row) {
@@ -1784,24 +1792,30 @@ function toCheckpoint(row) {
 }
 
 // src/persistent/repositories/conversation-files.repo.ts
-import { basename as basename5, dirname as dirname4 } from "path";
+import { basename as basename5, dirname as dirname5 } from "path";
 var ConversationFilesRepo = class {
   constructor(db) {
     this.db = db;
   }
   db;
   getByPath(absolutePath) {
-    return this.db.prepare("SELECT * FROM conversation_files WHERE absolute_path = ?").get(absolutePath);
+    return this.db.prepare("SELECT * FROM conversation_files WHERE absolute_path = ?").get(canonicalPath(absolutePath));
   }
   // Insert a freshly-discovered file at offset 0; returns its row id. Existing
   // path is left untouched (returns the existing id) so a re-discovery is safe.
+  //
+  // absolute_path/parent_dir are stored canonicalized so a caller that built the
+  // path with the native separator (watcher, path.join) and one that got it from
+  // fast-glob (forward slashes, even on Windows) land on the same row instead of
+  // inserting a duplicate.
   ensure(absolutePath, account) {
-    const existing = this.getByPath(absolutePath);
+    const canonical = canonicalPath(absolutePath);
+    const existing = this.getByPath(canonical);
     if (existing) return existing.id;
     const info = this.db.prepare(
       `INSERT INTO conversation_files (absolute_path, parent_dir, file_name, account)
          VALUES (?, ?, ?, ?)`
-    ).run(absolutePath, dirname4(absolutePath), basename5(absolutePath), account);
+    ).run(canonical, dirname5(canonical), basename5(canonical), account);
     return Number(info.lastInsertRowid);
   }
   // Advance the cursor + persisted reducer state after a successful index pass.
@@ -1863,7 +1877,7 @@ var ConversationFilesRepo = class {
   activePathsByParentDir(parentDir) {
     return this.db.prepare(
       "SELECT absolute_path, account FROM conversation_files WHERE parent_dir = ? AND status != 'deleted'"
-    ).all(parentDir);
+    ).all(canonicalPath(parentDir));
   }
 };
 
@@ -1958,7 +1972,10 @@ var ConversationsRepo = class {
            updated_at = CURRENT_TIMESTAMP`
     ).run({
       file_id: fileId,
-      source_path: meta.id,
+      // Canonical form, matching conversation_files.absolute_path — otherwise
+      // a native-separator caller writes a second spelling of the same file
+      // that every by-path lookup (and the deletion-reconcile) then misses.
+      source_path: canonicalPath(meta.id),
       provider: meta.provider ?? CLAUDE_CODE_PROVIDER,
       kind: meta.kind ?? null,
       external_session_id: meta.externalSessionId ?? null,
@@ -1988,7 +2005,7 @@ var ConversationsRepo = class {
     });
   }
   getBySourcePath(sourcePath) {
-    const row = this.db.prepare("SELECT * FROM conversations WHERE source_path = ? AND status = 'active'").get(sourcePath);
+    const row = this.db.prepare("SELECT * FROM conversations WHERE source_path = ? AND status = 'active'").get(canonicalPath(sourcePath));
     return row ? rowToMeta(row) : null;
   }
   // Dual lookup matching scanner.getConversation: resolve by source_path (the
@@ -2052,7 +2069,7 @@ var ConversationsRepo = class {
   pageMessageCount(sourcePath) {
     const row = this.db.prepare(
       "SELECT page_message_count AS n FROM conversations WHERE source_path = ? AND status = 'active'"
-    ).get(sourcePath);
+    ).get(canonicalPath(sourcePath));
     return row?.n ?? 0;
   }
   count() {
@@ -2067,14 +2084,15 @@ var FtsRepo = class {
   }
   db;
   upsert(meta) {
+    const sourcePath = canonicalPath(meta.id);
     const tx = this.db.transaction(() => {
-      this.db.prepare("DELETE FROM conversation_messages_fts WHERE source_path = ?").run(meta.id);
+      this.db.prepare("DELETE FROM conversation_messages_fts WHERE source_path = ?").run(sourcePath);
       this.db.prepare(
         `INSERT INTO conversation_messages_fts
              (source_path, content, project_name, session_id, session_name, account, model, branch, tool_names)
            VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)`
       ).run(
-        meta.id,
+        sourcePath,
         meta.contentSnippet ?? "",
         meta.projectName ?? "",
         meta.sessionId ?? "",
@@ -2088,7 +2106,7 @@ var FtsRepo = class {
     tx();
   }
   remove(sourcePath) {
-    this.db.prepare("DELETE FROM conversation_messages_fts WHERE source_path = ?").run(sourcePath);
+    this.db.prepare("DELETE FROM conversation_messages_fts WHERE source_path = ?").run(canonicalPath(sourcePath));
   }
   // Ranked source_paths matching the query, best first. Returns [] on an empty
   // query (callers fall back to a recency listing).
@@ -2120,11 +2138,11 @@ var ScannedDirsRepo = class {
   }
   db;
   get(path) {
-    return this.db.prepare("SELECT * FROM scanned_dirs WHERE path = ?").get(path);
+    return this.db.prepare("SELECT * FROM scanned_dirs WHERE path = ?").get(canonicalPath(path));
   }
   // Known project subdirectories under a root, path ascending (stable order).
   childrenOf(parentRoot) {
-    return this.db.prepare("SELECT * FROM scanned_dirs WHERE parent_root = ? ORDER BY path ASC").all(parentRoot);
+    return this.db.prepare("SELECT * FROM scanned_dirs WHERE parent_root = ? ORDER BY path ASC").all(canonicalPath(parentRoot));
   }
   upsert(path, parentRoot, mtimeMs, hasNested) {
     this.db.prepare(
@@ -2135,11 +2153,16 @@ var ScannedDirsRepo = class {
            mtime_ms = excluded.mtime_ms,
            has_nested = excluded.has_nested,
            updated_at = CURRENT_TIMESTAMP`
-    ).run(path, parentRoot, mtimeMs, hasNested ? 1 : 0);
+    ).run(
+      canonicalPath(path),
+      parentRoot === null ? null : canonicalPath(parentRoot),
+      mtimeMs,
+      hasNested ? 1 : 0
+    );
   }
   // Drop a project subdirectory's watermark row (it vanished from disk).
   remove(path) {
-    this.db.prepare("DELETE FROM scanned_dirs WHERE path = ?").run(path);
+    this.db.prepare("DELETE FROM scanned_dirs WHERE path = ?").run(canonicalPath(path));
   }
 };
 
@@ -2272,7 +2295,7 @@ var PersistentEngine = class {
     if (enabled.includes(CODEX_CLI_PROVIDER) && (options.codexRoots?.length ?? 0) > 0) {
       coveredAccounts.add("codex");
     }
-    const seen = new Set(discovered.map((d) => d.filePath));
+    const seen = new Set(discovered.map((d) => canonicalPath(d.filePath)));
     for (const path of this.files.activePathsByAccounts([...coveredAccounts])) {
       if (!seen.has(path)) this.markDeleted(path);
     }
@@ -2286,9 +2309,10 @@ var PersistentEngine = class {
   // half-written row or an over-advanced cursor. Returns the classification
   // alongside the meta so callers (refreshFile) can keep, extend, or evict
   // their own per-file caches without re-stat'ing the file (racy) themselves.
-  async indexFile(filePath, account, tierName, customTiers, resolveGitBranch, force = false, provider) {
+  async indexFile(rawFilePath, account, tierName, customTiers, resolveGitBranch, force = false, provider) {
     const log = getLogger();
     const tier = resolveTier(tierName, customTiers);
+    const filePath = canonicalPath(rawFilePath);
     const existing = this.files.getByPath(filePath);
     const { change, stat: stat4 } = classify(filePath, existing);
     if (change === "vanished" || !stat4) {
@@ -2635,7 +2659,7 @@ var IndexQueue = class {
 var BATCH_SIZE2 = 12;
 var DEFAULT_CONFIG_PATH = "~/.config/threadbase-scanner";
 function defaultDbPath() {
-  return process.env.TB_SCANNER_DB ?? join4(homedir2(), ".config", "threadbase-scanner", "index.db");
+  return process.env.TB_SCANNER_DB ?? join5(homedir2(), ".config", "threadbase-scanner", "index.db");
 }
 var ConversationScanner = class {
   metadataCache = /* @__PURE__ */ new Map();
@@ -2928,12 +2952,13 @@ var ConversationScanner = class {
   }
   async getConversation(id, _options) {
     const log = getLogger();
-    const cached = this.conversationLRU.get(id);
+    const cid = canonicalPath(id);
+    const cached = this.conversationLRU.get(cid);
     if (cached) {
       log.debug({ id }, "getConversation: cache hit");
       return cached.conversation;
     }
-    const meta = this.persistent ? this.engine().getByIdOrSession(id) : this.metadataCache.get(id) ?? this.resolveSessionId(id);
+    const meta = this.persistent ? this.engine().getByIdOrSession(cid) : this.metadataCache.get(cid) ?? this.resolveSessionId(cid);
     if (!meta) {
       log.debug({ id }, "getConversation: not found in metadata");
       return null;
@@ -2942,12 +2967,12 @@ var ConversationScanner = class {
     try {
       if (this.persistent && meta.provider !== CODEX_CLI_PROVIDER) {
         const parsed = await parseConversationResumable(meta.filePath, meta.account);
-        if (parsed) this.conversationLRU.set(id, parsed);
+        if (parsed) this.conversationLRU.set(cid, parsed);
         return parsed?.conversation ?? null;
       }
       const conversation = meta.provider === CODEX_CLI_PROVIDER ? await parseCodexConversation(meta.filePath, meta.account) : await parseConversation(meta.filePath, meta.account);
       if (conversation) {
-        this.conversationLRU.set(id, { conversation });
+        this.conversationLRU.set(cid, { conversation });
       }
       return conversation;
     } catch (err) {
@@ -2975,7 +3000,7 @@ var ConversationScanner = class {
   // which is identical by construction.
   async getConversationPage(id, options) {
     if (this.persistent) {
-      return this.engine().getPage(id, options);
+      return this.engine().getPage(canonicalPath(id), options);
     }
     const conversation = await this.getConversation(id);
     if (!conversation) return null;
@@ -3025,14 +3050,15 @@ var ConversationScanner = class {
   // each re-reading the file.
   refreshesInFlight = /* @__PURE__ */ new Map();
   refreshFile(filePath, account) {
-    const inFlight = this.refreshesInFlight.get(filePath);
+    const cpath = canonicalPath(filePath);
+    const inFlight = this.refreshesInFlight.get(cpath);
     if (inFlight) return inFlight;
-    const refresh = this.doRefreshFile(filePath, account).finally(() => {
-      if (this.refreshesInFlight.get(filePath) === refresh) {
-        this.refreshesInFlight.delete(filePath);
+    const refresh = this.doRefreshFile(cpath, account).finally(() => {
+      if (this.refreshesInFlight.get(cpath) === refresh) {
+        this.refreshesInFlight.delete(cpath);
       }
     });
-    this.refreshesInFlight.set(filePath, refresh);
+    this.refreshesInFlight.set(cpath, refresh);
     return refresh;
   }
   async doRefreshFile(filePath, account) {