ai-worklog 1.0.2 → 1.0.3

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ai-worklog",
-  "version": "1.0.2",
+  "version": "1.0.3",
   "description": "AI 对话工作日志自动收集工具 —— 从 Claude Code / Codex 对话记录生成每日工作日志并推送到 GitLab",
   "bin": {
     "ai-worklog": "./bin/index.js"

package/scripts/collect_work_log.py

@@ -278,21 +278,31 @@ def generate_summary(
     if total_sessions == 0:
         return _generate_empty_log(target_date)
 
-    # 构建给 AI 的原始数据
+    # 构建给 AI 的原始数据（去重 + 截断，控制 token 消耗）
+    def dedup_and_trim(msgs: list[str], max_chars: int = 200) -> list[str]:
+        seen: set[str] = set()
+        result = []
+        for m in msgs:
+            key = m[:80]  # 用前 80 字符去重
+            if key in seen:
+                continue
+            seen.add(key)
+            result.append(m if len(m) <= max_chars else m[:max_chars] + "…")
+        return result
+
     project_sections = []
     for proj_name, data in all_projects.items():
         msgs_parts = []
         if data["claude"]:
-            msgs_parts.append(f"[Claude Code 对话 {len(data['claude'])} 条]")
-            for i, m in enumerate(data["claude"], 1):
-                # 截断过长消息
-                msg = m if len(m) <= 500 else m[:500] + "..."
-                msgs_parts.append(f"  {i}. {msg}")
+            trimmed = dedup_and_trim(data["claude"])
+            msgs_parts.append(f"[Claude Code {len(data['claude'])} 条，去重后 {len(trimmed)} 条]")
+            for i, m in enumerate(trimmed, 1):
+                msgs_parts.append(f"  {i}. {m}")
         if data["codex"]:
-            msgs_parts.append(f"[Codex 对话 {len(data['codex'])} 条]")
-            for i, m in enumerate(data["codex"], 1):
-                msg = m if len(m) <= 500 else m[:500] + "..."
-                msgs_parts.append(f"  {i}. {msg}")
+            trimmed = dedup_and_trim(data["codex"])
+            msgs_parts.append(f"[Codex {len(data['codex'])} 条，去重后 {len(trimmed)} 条]")
+            for i, m in enumerate(trimmed, 1):
+                msgs_parts.append(f"  {i}. {m}")
         project_sections.append(f"项目: {proj_name}\n" + "\n".join(msgs_parts))
 
     raw_data = "\n\n".join(project_sections)
@@ -462,7 +472,9 @@ def git_commit_and_push(log_file: Path, target_date: str, push: bool = True) ->
     """全自动 git init → GitLab 项目创建 → commit → push"""
 
     def run(cmd: list[str], check_err: bool = True) -> subprocess.CompletedProcess:
-        r = subprocess.run(cmd, cwd=REPO_DIR, capture_output=True, text=True)
+        env = os.environ.copy()
+        env["LC_ALL"] = "C"  # 强制英文输出，保证字符串匹配不受本地化影响
+        r = subprocess.run(cmd, cwd=REPO_DIR, capture_output=True, text=True, env=env)
         if check_err and r.returncode != 0:
             print(f"命令失败: {' '.join(cmd)}\n{r.stderr.strip()}", file=sys.stderr)
         return r
@@ -496,7 +508,7 @@ def git_commit_and_push(log_file: Path, target_date: str, push: bool = True) ->
         run(["git", "add", "-A"])
         run(["git", "commit", "-m", "init: 初始化工作日志仓库"])
 
-    # ── 4. 提交日志 ────────────────────────────────────────────────────────────
+    # ── 4. 提交日志 ───────────────────────────────────────────────────────────
     rel_path = log_file.relative_to(REPO_DIR)
     print(f"Git: 添加 {rel_path}")
     if run(["git", "add", str(rel_path)]).returncode != 0: