chainlesschain 0.162.17 → 0.162.19

package/src/commands/hub.js

@@ -25,6 +25,34 @@ function printJson(obj) {
   console.log(JSON.stringify(obj, null, 2));
 }
 
+/**
+ * Drain-then-exit. console.log → process.stdout.write is async for pipes;
+ * exiting immediately can truncate output to the parent. Use this in
+ * commands that must terminate even when getHub() left libuv handles
+ * registered (aichat-health setInterval, sidecar supervisor, etc.) — sync
+ * + read commands are all-done after the report, no reason to keep idling.
+ * Real-device repro on Xiaomi 24115RA8EC 2026-05-23.
+ */
+function jsonAndExit(obj) {
+  process.stdout.write(JSON.stringify(obj, null, 2) + "\n", () =>
+    process.exit(0),
+  );
+}
+
+/**
+ * Coerce a Commander string option to a positive integer, returning null if
+ * unset / blank / non-numeric / ≤0. Used by `--max-facts` / `--max-query-limit`
+ * on `cc hub ask` so on-device callers can pass smaller budgets without
+ * silent fallback to constructor defaults on a typo (`--max-facts abc`
+ * should not silently use 80).
+ */
+function parsePositiveInt(raw) {
+  if (raw === undefined || raw === null || raw === "") return null;
+  const n = Number.parseInt(raw, 10);
+  if (!Number.isFinite(n) || n <= 0) return null;
+  return n;
+}
+
 function fail(spinner, err, asJson) {
   if (spinner) spinner.stop();
   const msg = err && err.message ? err.message : String(err);
@@ -41,7 +69,9 @@ function fail(spinner, err, asJson) {
 async function cmdAsk(question, options) {
   const spinner = options.json ? null : ora("Asking hub...").start();
   try {
-    const hub = await getHub();
+    // _getHub injection mirrors other handlers — lets unit tests stub the
+    // engine without standing up a real vault.
+    const hub = await (options._getHub || getHub)();
     if (!hub.engine) throw new Error("Analysis engine unavailable");
     // 推文 §三道锁第二把 "默认不许问云端" — acceptNonLocal 默认 false (拒云)。
     // 优先 --accept-non-local CLI 旗，其次 env CC_HUB_ALLOW_NON_LOCAL (Android UI
@@ -50,10 +80,19 @@ async function cmdAsk(question, options) {
       process.env.CC_HUB_ALLOW_NON_LOCAL === "1" ||
       process.env.CC_HUB_ALLOW_NON_LOCAL === "true";
     const acceptNonLocal = !!options.acceptNonLocal || envAllow;
-    const result = await hub.engine.ask(question, {
+    // Per-call budget overrides for small-model callers (Android Qwen2.5-1.5B
+    // passes --max-facts 20 --max-query-limit 50 to keep prompt ~1.5K tokens
+    // and stay under the model's effective instruction-following window).
+    // Commander parses --max-facts <n> as string — coerce + drop invalid.
+    const maxFacts = parsePositiveInt(options.maxFacts);
+    const maxQueryLimit = parsePositiveInt(options.maxQueryLimit);
+    const askOptions = {
       useRag: options.useRag !== false,
       acceptNonLocal,
-    });
+    };
+    if (maxFacts !== null) askOptions.maxFacts = maxFacts;
+    if (maxQueryLimit !== null) askOptions.maxQueryLimit = maxQueryLimit;
+    const result = await hub.engine.ask(question, askOptions);
     if (spinner) spinner.stop();
     if (options.json) {
       printJson(result);
@@ -203,13 +242,34 @@ async function cmdSyncAdapter(name, options) {
     if (options.input) opts.inputPath = String(options.input);
     const report = await hub.registry.syncAdapter(name, opts);
     if (spinner) spinner.succeed(`synced ${name}`);
+    // 2026-05-24 in-APK Android exit + flush-race fix. Real-device repro on
+    // Xiaomi 24115RA8EC:
+    //
+    //   (1) cc subprocess sits idle in epoll_wait for the full 240s budget
+    //       after the report is fully written, then gets destroyForcibly'd
+    //       by Kotlin LocalCcRunner → false "timeout after 240000ms" UI
+    //       error even though vault writes committed. Root cause: a
+    //       lingering libuv handle in getHub() wiring (suspect: aichat-
+    //       health setInterval or sidecar supervisor health ping) keeps
+    //       the event loop alive forever. Long-run fix is to close those
+    //       cleanly; until then we force-exit here.
+    //
+    //   (2) First-pass force-exit landed `process.exit(0)` after a normal
+    //       `printJson(report) = console.log(...)`. But console.log →
+    //       process.stdout.write is ASYNC for pipes (only sync for TTY),
+    //       so process.exit fires before the JSON buffer drains → Kotlin
+    //       parent reads truncated/empty stdout → "parse-failed" reason
+    //       in CcResult.Failed → UI "写入本地数据库失败".
+    //
+    // Fix: explicit drain-then-exit (jsonAndExit helper at top of file).
     if (options.json) {
-      printJson(report);
-    } else {
-      logger.log(
-        `ingested=${report.ingested} kgTriples=${report.kgTriples} ragDocs=${report.ragDocs} durationMs=${report.durationMs}`,
-      );
+      jsonAndExit(report);
+      return;
     }
+    logger.log(
+      `ingested=${report.ingested} kgTriples=${report.kgTriples} ragDocs=${report.ragDocs} durationMs=${report.durationMs}`,
+    );
+    process.exit(0);
   } catch (err) {
     fail(spinner, err, options.json);
   }
@@ -239,6 +299,46 @@ async function cmdSyncAll(options) {
   }
 }
 
+// ─── rederive ────────────────────────────────────────────────────────
+
+async function cmdRederive(options) {
+  const spinner = options.json
+    ? null
+    : ora(
+        options.adapter
+          ? `re-deriving ${options.adapter} from raw_events...`
+          : "re-deriving all adapters from raw_events...",
+      ).start();
+  try {
+    const hub = await getHub();
+    const opts = {};
+    if (options.adapter) opts.adapter = String(options.adapter);
+    if (options.batchSize) opts.batchSize = Number(options.batchSize);
+    const report = await hub.registry.rederive(opts);
+    if (spinner) {
+      spinner.succeed(
+        `re-derived ${report.rawSeen} raw → events:${report.entityCounts.events} ` +
+          `persons:${report.entityCounts.persons} items:${report.entityCounts.items} ` +
+          `(invalid:${report.invalidCount} missing:${report.adapterMissing} ` +
+          `${report.durationMs}ms)`,
+      );
+    }
+    if (options.json) {
+      jsonAndExit(report);
+      return;
+    }
+    if (report.errors.length > 0) {
+      logger.log(chalk.red(`errors: ${report.errors.length}`));
+      for (const e of report.errors) {
+        logger.log(`  ${chalk.red(e.adapter)}: ${e.error}`);
+      }
+    }
+    process.exit(0);
+  } catch (err) {
+    fail(spinner, err, options.json);
+  }
+}
+
 // ─── query-events / recent-audit ─────────────────────────────────────
 
 async function cmdQueryEvents(options) {
@@ -253,16 +353,93 @@ async function cmdQueryEvents(options) {
     if (options.limit) q.limit = Number(options.limit);
     const events = hub.vault.queryEvents(q);
     if (options.json) {
-      printJson(events);
-    } else {
-      logger.log(`${events.length} events:`);
-      for (const ev of events) {
-        const at = new Date(ev.at).toISOString();
-        logger.log(
-          `  ${chalk.gray(at)} ${chalk.cyan(ev.subtype)} ${ev.summary || ev.id}`,
-        );
-      }
+      jsonAndExit(events);
+      return;
+    }
+    logger.log(`${events.length} events:`);
+    for (const ev of events) {
+      const at = new Date(ev.at).toISOString();
+      logger.log(
+        `  ${chalk.gray(at)} ${chalk.cyan(ev.subtype)} ${ev.summary || ev.id}`,
+      );
+    }
+    process.exit(0);
+  } catch (err) {
+    fail(null, err, options.json);
+  }
+}
+
+// ─── search / facet-counts (Phase 16 Vault Browser) ──────────────────
+//
+// Surfaces vault.searchEvents + vault.facetCounts for headless use
+// (Android LocalCcRunner.searchEvents calls this; CI smoke too).
+async function cmdSearchEvents(options) {
+  try {
+    const hub = await getHub();
+    const q = {};
+    if (options.q) q.q = String(options.q);
+    if (options.adapter) q.adapter = options.adapter;
+    if (options.category) q.category = options.category;
+    if (options.subtype) q.subtype = options.subtype;
+    if (options.since) q.since = Number(options.since);
+    if (options.until) q.until = Number(options.until);
+    if (options.limit) q.limit = Number(options.limit);
+    // cursor passed as `--cursor "<occurredAt>:<id>"` (cli-friendly)
+    if (options.cursor) {
+      const m = String(options.cursor).match(/^(\d+):(.+)$/);
+      if (m) q.cursor = { occurredAt: Number(m[1]), id: m[2] };
     }
+    const result = hub.vault.searchEvents(q);
+    if (options.json) {
+      jsonAndExit(result);
+      return;
+    }
+    logger.log(
+      `${result.rows.length} events (mode=${result.mode}${result.shortQuery ? " shortQuery!" : ""})` +
+        (result.nextCursor
+          ? `  nextCursor=${result.nextCursor.occurredAt}:${result.nextCursor.id}`
+          : ""),
+    );
+    for (const ev of result.rows) {
+      const at = new Date(ev.occurredAt).toISOString();
+      const summary =
+        (ev.content &&
+          (ev.content.text || ev.content.title || ev.content.subject)) ||
+        ev.id;
+      logger.log(
+        `  ${chalk.gray(at)} ${chalk.cyan(ev.source.adapter)} ${chalk.dim(ev.subtype)} ${summary}`,
+      );
+    }
+    process.exit(0);
+  } catch (err) {
+    fail(null, err, options.json);
+  }
+}
+
+async function cmdFacetCounts(options) {
+  try {
+    const hub = await getHub();
+    const q = {};
+    if (options.q) q.q = String(options.q);
+    if (options.since) q.since = Number(options.since);
+    if (options.until) q.until = Number(options.until);
+    const result = hub.vault.facetCounts(q);
+    if (options.json) {
+      jsonAndExit(result);
+      return;
+    }
+    logger.log(
+      `total=${result.total} mode=${result.mode}${result.shortQuery ? " shortQuery!" : ""}`,
+    );
+    logger.log(chalk.bold("by category:"));
+    for (const [k, n] of Object.entries(result.byCategory)) {
+      logger.log(`  ${chalk.cyan(k.padEnd(10))} ${n}`);
+    }
+    logger.log(chalk.bold("by adapter:"));
+    for (const [k, n] of Object.entries(result.byAdapter)) {
+      logger.log(`  ${chalk.cyan(k.padEnd(24))} ${n}`);
+    }
+    process.exit(0);
   } catch (err) {
     fail(null, err, options.json);
   }
@@ -277,16 +454,17 @@ async function cmdRecentAudit(options) {
     if (options.limit) q.limit = Number(options.limit);
     const rows = hub.vault.queryAudit(q);
     if (options.json) {
-      printJson(rows);
-    } else {
-      logger.log(`${rows.length} audit rows:`);
-      for (const r of rows) {
-        const at = new Date(r.at).toISOString();
-        logger.log(
-          `  ${chalk.gray(at)} ${chalk.cyan(r.action)} ${r.adapter || ""} ${r.eventId || ""}`,
-        );
-      }
+      jsonAndExit(rows);
+      return;
     }
+    logger.log(`${rows.length} audit rows:`);
+    for (const r of rows) {
+      const at = new Date(r.at).toISOString();
+      logger.log(
+        `  ${chalk.gray(at)} ${chalk.cyan(r.action)} ${r.adapter || ""} ${r.eventId || ""}`,
+      );
+    }
+    process.exit(0);
   } catch (err) {
     fail(null, err, options.json);
   }
@@ -1044,6 +1222,14 @@ export function registerHubCommand(program) {
       "--accept-non-local",
       "Allow non-local LLM (sends data off device — required when provider is not Ollama/vLLM)",
     )
+    .option(
+      "--max-facts <n>",
+      "Cap facts in prompt (default 80; on-device small models e.g. Qwen2.5-1.5B should pass 20)",
+    )
+    .option(
+      "--max-query-limit <n>",
+      "Cap vault queryEvents limit (default 200; small-model callers should pass 50)",
+    )
     .option("--json", "Output JSON")
     .action(cmdAsk);
 
@@ -1087,6 +1273,24 @@ export function registerHubCommand(program) {
     .option("--json", "Output JSON")
     .action(cmdSyncAll);
 
+  hub
+    .command("rederive")
+    .description(
+      "Re-derive canonical events from raw_events without re-fetching " +
+        "from source (recovers orphan raws from a past sync where putBatch " +
+        "silently failed, e.g. trap #25 partial-index drift).",
+    )
+    .option(
+      "--adapter <name>",
+      "Filter to one adapter; default = all registered",
+    )
+    .option(
+      "--batch-size <n>",
+      "Raws per partitionBatch+putBatch tx (default 100)",
+    )
+    .option("--json", "Output JSON")
+    .action(cmdRederive);
+
   hub
     .command("query-events")
     .description("Query vault events with filters")
@@ -1099,6 +1303,34 @@ export function registerHubCommand(program) {
     .option("--json", "Output JSON")
     .action(cmdQueryEvents);
 
+  hub
+    .command("search")
+    .description("Vault Browser search — events_fts (FTS5) + faceted filters")
+    .option("--q <text>", "Keyword (FTS5 phrase match if mode=fts5)")
+    .option("--adapter <name>", "Filter by exact adapter")
+    .option(
+      "--category <cat>",
+      "Filter by category (chat/social/email/shopping/travel/system/ai-chat/other)",
+    )
+    .option("--subtype <t>", "Filter by event subtype")
+    .option("--since <ms>", "Start of time window (unix-ms)")
+    .option("--until <ms>", "End of time window (unix-ms)")
+    .option("--cursor <occurredAt:id>", "Page cursor from previous response")
+    .option("--limit <n>", "Page size (max 500)", "50")
+    .option("--json", "Output JSON")
+    .action(cmdSearchEvents);
+
+  hub
+    .command("facet-counts")
+    .description(
+      "Sidebar / chip counts — events grouped by category / adapter / subtype",
+    )
+    .option("--q <text>", "Keyword filter (same semantics as `cc hub search`)")
+    .option("--since <ms>", "Start of time window (unix-ms)")
+    .option("--until <ms>", "End of time window (unix-ms)")
+    .option("--json", "Output JSON")
+    .action(cmdFacetCounts);
+
   hub
     .command("recent-audit")
     .description("Recent audit log entries")
@@ -1270,6 +1502,8 @@ export function registerHubCommand(program) {
 // `_wizard` / `_factoryDeps` / `_knownVendors` injected, bypassing the
 // real `getHub()` call. The commander wiring above is the runtime path.
 export const _internal = {
+  cmdAsk,
+  parsePositiveInt,
   cmdAIChatList,
   cmdAIChatLogin,
   cmdAIChatProbe,

package/src/constants.js

@@ -145,6 +145,13 @@ export const DEFAULT_CONFIG = {
     apiKey: null,
     baseUrl: "https://ark.cn-beijing.volces.com/api/v3",
     model: "doubao-seed-1-6-251015",
+    // When true, `cc ask` (provider=ollama only) overrides baseUrl to point at
+    // the Android LocalLlmServer (127.0.0.1:18484). User opt-in toggle:
+    //   cc config set llm.preferAndroidLocal true
+    // Resolution priority in ask.js:
+    //   --base-url > CC_HUB_OLLAMA_URL env > --prefer-android-local flag >
+    //   this config > config.llm.baseUrl > hardcoded localhost:11434
+    preferAndroidLocal: false,
   },
   enterprise: {
     serverUrl: null,

package/src/gateways/ws/personal-data-hub-protocol.js

@@ -22,12 +22,103 @@ import {
   close as closeHub,
 } from "../../lib/personal-data-hub-wiring.js";
 import { getAIChatWizard } from "../../lib/personal-data-hub-aichat-wizard.js";
-import { existsSync, unlinkSync, readdirSync } from "node:fs";
+import {
+  existsSync,
+  unlinkSync,
+  readdirSync,
+  mkdirSync,
+  writeFileSync,
+} from "node:fs";
 import { join } from "node:path";
 import pdhPkg from "@chainlesschain/personal-data-hub";
 
 const { ingestSystemDataAndroidSnapshot } = pdhPkg;
 
+/**
+ * If the caller didn't pass `inputPath`, try to pull a snapshot for
+ * this adapter off the attached Android phone via `adb shell run-as
+ * <pkg.debug> cat files/.chainlesschain/staging/<name>.json`. On
+ * success, write the JSON to `hub.hubDir/staging/` and return an
+ * `options` object with `inputPath` set so the adapter's _syncViaSnapshot
+ * picks it up.
+ *
+ * Why this is needed: every social adapter (bilibili, weibo, douyin,
+ * xiaohongshu, toutiao, kuaishou, qq, wechat, baidu-map, tencent-map,
+ * jd, meituan, pinduoduo) is snapshot-only and the Android in-app
+ * collectors write their JSON into the app's filesDir/.chainlesschain/
+ * staging/. Without an auto-pull, the desktop 同步 button can never
+ * reach a working state for these adapters — it just throws "needs
+ * opts.inputPath".
+ *
+ * Best-effort: if anything fails (no device, package not debuggable,
+ * snapshot file missing for that adapter, bridge module unavailable),
+ * return the original options unchanged so the adapter's normal error
+ * path fires and the UI banner shows a meaningful message.
+ */
+async function _tryAdbAutoPullInputPath(hub, name, options) {
+  if (
+    options &&
+    typeof options.inputPath === "string" &&
+    options.inputPath.length > 0
+  ) {
+    return options; // caller already supplied a path
+  }
+  // Skip auto-pull for adapters that have a live bridge mode — pulling
+  // a stale snapshot file would short-circuit the bridge path which
+  // yields fresher / richer data (e.g. system-data-android's bridge
+  // mode pulls contacts + apps + sms + call_log live via ADB, while
+  // an Android-collected snapshot only contains contacts + apps).
+  // Keep this list in sync with adapters whose _syncViaBridge is
+  // strictly richer than their snapshot output.
+  // browser-history-* and vscode read desktop-local files (browser History
+  // SQLite + Bookmarks JSON, or VS Code workspaceStorage + state.vscdb);
+  // ADB snapshot is meaningless for them.
+  const BRIDGE_PREFERRED = new Set([
+    "system-data-android",
+    "browser-history-chrome",
+    "browser-history-edge",
+    "vscode",
+    "win-recent",
+    "git-activity",
+    "shell-history",
+    "local-files",
+  ]);
+  if (BRIDGE_PREFERRED.has(name)) {
+    return options || {};
+  }
+  try {
+    const { createHostAdbBridge } =
+      await import("../../lib/host-adb-bridge.js");
+    const bridge = createHostAdbBridge();
+    const content = await bridge.invoke("snapshot.read", {
+      fileName: `${name}.json`,
+    });
+    // Sanity: must be parseable JSON. If not, the adapter would fail
+    // with a confusing parse error — surface a clearer one here.
+    try {
+      JSON.parse(content);
+    } catch (_e) {
+      return options; // leave to adapter — it'll report parse error
+    }
+    const stagingDir = join(hub.hubDir, "staging");
+    mkdirSync(stagingDir, { recursive: true });
+    const stagingPath = join(
+      stagingDir,
+      `${name}-${Date.now()}-${Math.random().toString(36).slice(2, 8)}.json`,
+    );
+    writeFileSync(stagingPath, content, "utf-8");
+    return {
+      ...(options || {}),
+      inputPath: stagingPath,
+      _autoPulledViaAdb: true,
+    };
+  } catch (_e) {
+    // Bridge unavailable / device not attached / file missing — let the
+    // adapter throw its normal error so the banner explains.
+    return options || {};
+  }
+}
+
 async function withHub(fn) {
   try {
     const hub = await getHub();
@@ -83,10 +174,26 @@ export const PERSONAL_DATA_HUB_HANDLERS = {
     withHub((hub) => hub.registry.list()),
 
   "personal-data-hub.sync-adapter": async (msg) =>
-    withHub(
-      async (hub) =>
-        await hub.registry.syncAdapter(msg.name, msg.options || {}),
-    ),
+    withHub(async (hub) => {
+      const options = await _tryAdbAutoPullInputPath(
+        hub,
+        msg.name,
+        msg.options,
+      );
+      try {
+        return await hub.registry.syncAdapter(msg.name, options);
+      } finally {
+        // Best-effort cleanup of staging file we just wrote (don't shadow
+        // a real adapter error with cleanup noise).
+        if (options && options._autoPulledViaAdb && options.inputPath) {
+          try {
+            if (existsSync(options.inputPath)) unlinkSync(options.inputPath);
+          } catch (_e) {
+            /* ignore */
+          }
+        }
+      }
+    }),
 
   "personal-data-hub.sync-all": async (msg) =>
     withHub(async (hub) => await hub.registry.syncAll(msg.options || {})),
@@ -116,6 +223,36 @@ export const PERSONAL_DATA_HUB_HANDLERS = {
       }),
     ),
 
+  // Phase 16 Vault Browser — full-text + faceted search over events.
+  // See packages/personal-data-hub/lib/vault.js#searchEvents for the
+  // query/result shape. The desktop browser view + Android "我的数据" tab
+  // both consume this topic.
+  "personal-data-hub.search-events": async (msg) =>
+    withHub((hub) =>
+      hub.vault.searchEvents({
+        q: msg.q,
+        adapter: msg.adapter,
+        category: msg.category,
+        subtype: msg.subtype,
+        since: msg.since,
+        until: msg.until,
+        cursor: msg.cursor,
+        limit: msg.limit,
+      }),
+    ),
+
+  // Phase 16 Vault Browser — counts grouped by category/adapter/subtype,
+  // honoring the same q + since/until filters as search-events. Powers
+  // the sidebar badges + adapter chip counts in the browser UI.
+  "personal-data-hub.facet-counts": async (msg) =>
+    withHub((hub) =>
+      hub.vault.facetCounts({
+        q: msg.q,
+        since: msg.since,
+        until: msg.until,
+      }),
+    ),
+
   "personal-data-hub.recent-audit": async (msg) =>
     withHub((hub) =>
       hub.vault.queryAudit({
@@ -379,14 +516,19 @@ export const PERSONAL_DATA_HUB_STREAMING_HANDLERS = {
         } catch (_e) {}
       }
     };
+    const options = await _tryAdbAutoPullInputPath(hub, msg.name, msg.options);
     try {
-      const report = await hub.registry.syncAdapter(
-        msg.name,
-        msg.options || {},
-      );
+      const report = await hub.registry.syncAdapter(msg.name, options);
       return { result: report };
     } finally {
       hub.registry.onSyncEvent = original;
+      if (options && options._autoPulledViaAdb && options.inputPath) {
+        try {
+          if (existsSync(options.inputPath)) unlinkSync(options.inputPath);
+        } catch (_e) {
+          /* ignore */
+        }
+      }
     }
   },