@vibe-cafe/vibe-usage 0.9.16 → 0.10.1

package/src/sync.js

@@ -34,6 +34,7 @@ export async function runSync({ throws = false, quiet = false } = {}) {
   const allBuckets = [];
   const allSessions = [];
   const parserResults = [];
+  const parserProgress = [];
   // Sources whose parser ran to completion this sync. pruneState() below is
   // scoped to these so a transient parser failure doesn't evict that tool's
   // state and force a full re-upload next run.
@@ -47,6 +48,14 @@ export async function runSync({ throws = false, quiet = false } = {}) {
       if (!Array.isArray(buckets) || !Array.isArray(sessions)) {
         throw new TypeError('Parser returned an invalid result');
       }
+      if (result?.indexing) {
+        parserProgress.push({ source, ...result.indexing });
+      }
+      if (Array.isArray(result?.warnings)) {
+        for (const message of result.warnings) {
+          process.stderr.write(`${dim(`  ${message}`)}\n`);
+        }
+      }
       // A parser may deliberately suppress a transient error (Cursor network
       // timeout) to keep daemon logs quiet. Its empty result is not proof that
       // its prior data disappeared, so it must not be pruned this run.
@@ -71,7 +80,13 @@ export async function runSync({ throws = false, quiet = false } = {}) {
     pruneState(state, new Set(), new Set(), okSources);
     const pruned = before - (Object.keys(state.buckets).length + Object.keys(state.sessions).length);
     if (pruned > 0) saveState(state);
-    if (!quiet) console.log(dim('暂无新数据。'));
+    if (!quiet && parserProgress.length > 0) {
+      for (const p of parserProgress) {
+        console.log(dim(`  ${p.source}: 正在建立本地索引 ${p.completed}/${p.total}（下次同步继续）`));
+      }
+    } else if (!quiet) {
+      console.log(dim('暂无新数据。'));
+    }
     return 0;
   }
 
@@ -83,6 +98,11 @@ export async function runSync({ throws = false, quiet = false } = {}) {
       console.log(`  ${dim(p.source.padEnd(14))}${parts.join(' · ')}`);
     }
   }
+  if (!quiet && parserProgress.length > 0) {
+    for (const p of parserProgress) {
+      console.log(dim(`  ${p.source}: 正在建立本地索引 ${p.completed}/${p.total}（下次同步继续）`));
+    }
+  }
 
   let host = config.hostname;
   if (!host) {
@@ -112,10 +132,11 @@ export async function runSync({ throws = false, quiet = false } = {}) {
     for (const s of allSessions) s.project = 'unknown';
   }
 
-  // Incremental diff: parsers above always read the full local history (cheap,
-  // local-only). Here we drop anything whose content matches what we already
-  // uploaded, so only new/changed items go over the network. A quiet machine
-  // sends zero bytes; an active one sends just the current 30-min bucket.
+  // Incremental upload diff: parsers above emit a complete view of live local
+  // data (Codex may assemble that view from its disposable parser cache). Here
+  // we drop anything whose content matches what we already uploaded, so only
+  // new/changed items go over the network. A quiet machine sends zero bytes;
+  // an active one sends just the current 30-min bucket.
   // Missing/corrupt state.json => empty maps => one-time full upload, then
   // incremental forever after.
   const state = loadState();
@@ -169,6 +190,8 @@ export async function runSync({ throws = false, quiet = false } = {}) {
   let totalIngested = 0;
   let totalSessionsSynced = 0;
   let totalDroppedBuckets = 0;
+  let totalDroppedUnknownModels = 0;
+  let totalDroppedImplausible = 0;
   const droppedSources = new Set();
   const bucketBatches = Math.ceil(allBucketsToSend.length / BATCH_SIZE);
   const sessionBatches = Math.ceil(allSessionsToSend.length / SESSION_BATCH_SIZE);
@@ -189,8 +212,11 @@ export async function runSync({ throws = false, quiet = false } = {}) {
       }, batchSessions.length > 0 ? batchSessions : undefined);
       totalIngested += result.ingested ?? batch.length;
       totalSessionsSynced += result.sessions ?? 0;
+      const batchUnknownSources = new Set(result.dropped?.unknownSources || []);
       if (result.dropped) {
         totalDroppedBuckets += Number(result.dropped.buckets) || 0;
+        totalDroppedUnknownModels += Number(result.dropped.unknownModels) || 0;
+        totalDroppedImplausible += Number(result.dropped.implausible) || 0;
         for (const s of result.dropped.unknownSources || []) droppedSources.add(s);
       }
 
@@ -199,6 +225,10 @@ export async function runSync({ throws = false, quiet = false } = {}) {
       // state, so the next sync re-sends exactly those items — no data loss,
       // no silent gaps.
       for (const b of batch) {
+        // A source unknown to an older backend may become valid after deploy.
+        // Leave those hashes uncommitted so the next sync retries them instead
+        // of turning a temporary release-order mismatch into permanent loss.
+        if (batchUnknownSources.has(b.source)) continue;
         const key = bucketKey(b);
         const entry = pendingBucketState.get(key);
         if (entry) state.buckets[key] = entry;
@@ -219,11 +249,14 @@ export async function runSync({ throws = false, quiet = false } = {}) {
     console.log(success(`已同步 ${syncParts.join(' · ')}`));
 
     if (totalDroppedBuckets > 0) {
-      // Server doesn't (yet) recognize these source IDs — usually means the
-      // CLI is newer than the deployed vibe-cafe. Surface so the user knows
-      // the data wasn't lost on their end, just not stored upstream.
-      const sourcesList = Array.from(droppedSources).sort().join(', ');
-      console.log(dim(`  ${totalDroppedBuckets} buckets dropped (服务端未收录的 source: ${sourcesList})`));
+      const reasons = [];
+      if (droppedSources.size > 0) {
+        reasons.push(`服务端未收录的 source: ${Array.from(droppedSources).sort().join(', ')}`);
+      }
+      if (totalDroppedUnknownModels > 0) reasons.push(`模型未知: ${totalDroppedUnknownModels}`);
+      if (totalDroppedImplausible > 0) reasons.push(`超出合理范围: ${totalDroppedImplausible}`);
+      if (reasons.length === 0) reasons.push('服务端拒绝');
+      console.log(dim(`  ${totalDroppedBuckets} buckets dropped (${reasons.join('；')})`));
     }
 
     if (!quiet && totalSessionsSynced > 0) {

package/src/tools.js

@@ -1,6 +1,7 @@
 import { existsSync, readdirSync, statSync } from 'node:fs';
 import { join } from 'node:path';
 import { homedir } from 'node:os';
+import { findClaudeCodeDataDirs } from './claude-roots.js';
 
 function getCursorStateDbPath() {
   const rel = join('User', 'globalStorage', 'state.vscdb');
@@ -80,20 +81,6 @@ function findOpenclawDataDirs() {
   return dirs;
 }
 
-// Claude Code lives in ~/.claude/projects, but $CLAUDE_CONFIG_DIR relocates its
-// whole tree. Detect either so a user who only set CLAUDE_CONFIG_DIR is still
-// recognized (the parser scans both roots; see parsers/claude-code.js).
-function findClaudeCodeDataDirs() {
-  const dirs = [join(homedir(), '.claude', 'projects')];
-  const cfg = process.env.CLAUDE_CONFIG_DIR?.trim();
-  if (cfg) {
-    let custom = cfg.startsWith('~') ? join(homedir(), cfg.slice(1)) : cfg;
-    custom = custom.replace(/[/\\]+$/, '') || custom;
-    dirs.push(join(custom, 'projects'));
-  }
-  return dirs.filter(existsSync);
-}
-
 // Codex keeps live sessions in ~/.codex/sessions and moves completed ones to
 // ~/.codex/archived_sessions. Detect Codex if either dir exists, so a user
 // whose sessions have all been archived is still recognized.