Refactor: split run_freshrss_pipeline into internal helpers

Extract three internal helpers to reduce the main function from ~330
lines to ~80 lines:
- _process_item(): handles per-item extract/summarize/filter/build
- _build_and_persist_delivery(): builds payload and persists keyword index
- _build_run_report(): assembles status counts and run report dict

No behavior changes. External signature and return shape unchanged.

CODE_REVIEW.md

@@ -8,7 +8,7 @@
 
 ## 重构建议（中等成本）
 
-- [ ] `run_freshrss_pipeline` 函数过长（约300行）— 拆分为 `_process_single_item()`、`_build_and_persist_delivery()` 等内部函数，主函数只做编排
+- [x] `run_freshrss_pipeline` 函数过长（约300行）— 拆分为 `_process_item()`、`_build_and_persist_delivery()`、`_build_run_report()` 三个内部函数，主函数只做编排（已完成）
 - [ ] `load_filter_rules` 每次 pipeline 调用都重新读文件 — 加模块级缓存，MCP 服务长期运行时避免重复 I/O
 
 ## 功能补全

refactor-plan.md

@@ -0,0 +1,54 @@
+# 重构计划：拆分 run_freshrss_pipeline 为内部辅助函数
+
+## 背景
+
+`src/summary_mcp/workflows/freshrss_pipeline.py` 中的 `run_freshrss_pipeline` 函数约 330 行，
+将 6 个阶段全部写在一个函数体内，阅读、测试和后续扩展（如并发、重试策略）都比较困难。
+目标是在不改变任何外部行为的前提下，提取 3 个内部辅助函数。
+
+当前无测试覆盖，验证方式为函数签名和返回值结构保持不变。
+
+## 涉及文件
+
+- `src/summary_mcp/workflows/freshrss_pipeline.py`（唯一修改文件）
+
+## 提取 3 个内部辅助函数
+
+### 1. `_process_item(...)` — 单条 item 处理（当前 160-258 行）
+
+提取 for 循环体（约 100 行）为独立函数。
+返回 `item_report` dict；当 status 为 `delivered` 时，额外携带 `_candidate` 和 `_external_id`
+两个临时键供调用方解包，写盘前剥离这两个键。
+用 `return item_report` 替代循环中的 `continue`。
+
+### 2. `_build_and_persist_delivery(...)` — 阶段 4+5（当前 260-279 行）
+
+提取 payload 构建 + 词元索引持久化。
+返回 `(delivery_payload, keyword_index_result)`。
+
+### 3. `_build_run_report(...)` — 报告组装（当前 288-308 行）
+
+提取 status_counts 统计 + report dict 构建。
+返回 report dict，主函数拿到后再调用 `_save_json` 写盘。
+
+## 重构后主函数结构（约 80 行）
+
+1. 阶段 1：初始化（不变）
+2. 阶段 2：拉取 FreshRSS + 加载规则/context（不变）
+3. 阶段 3：for 循环调用 `_process_item(...)`，从返回值解包 candidate
+4. 阶段 4+5：`delivery_payload, keyword_index_result = _build_and_persist_delivery(...)`
+5. 阶段 6：标记已读，`report = _build_run_report(...)`，写盘，返回
+
+## 约束
+
+- `run_freshrss_pipeline` 外部签名不变
+- 返回 dict 的键结构不变
+- 所有文件写入路径不变
+- 纯结构性重构，无行为变化
+- 无需新增 import
+
+## 验证
+
+重构完成后运行：
+
+    python -c "from summary_mcp.workflows.freshrss_pipeline import run_freshrss_pipeline; print('ok')"

src/summary_mcp/workflows/freshrss_pipeline.py

@@ -71,91 +71,24 @@ def _maybe_path(enabled: bool, path: Path) -> Path | None:
     return path if enabled else None
 
 
-def run_freshrss_pipeline(
+def _process_item(
     *,
-    api_base_url: str | None = None,
+    index: int,
-    username: str | None = None,
+    item: Any,
-    api_password: str | None = None,
+    resolved_output_dir: Path,
-    stream_id: str = "user/-/state/com.google/reading-list",
+    resolved_prompt_path: Path,
-    limit: int = 5,
+    resolved_run_id: str,
-    continuation: str | None = None,
+    debug_artifacts: bool,
-    include_read: bool = False,
+    loaded_rules: list,
-    mark_read: bool = False,
+    filter_context: Any,
-    debug_artifacts: bool = False,
+    max_retries: int,
-    prompt: Path | None = None,
+    timeout_seconds: float,
-    rules: Path | None = None,
+    resolved_llm_api_key: str,
-    context: dict[str, Any] | None = None,
+    resolved_llm_model: str,
-    context_path: Path | None = None,
+    resolved_llm_api_url: str,
-    max_retries: int = 2,
-    timeout_seconds: float = 60.0,
-    llm_api_key: str | None = None,
-    llm_model: str | None = None,
-    llm_api_url: str | None = None,
-    run_id: str | None = None,
-    delivery_date: date | None = None,
-    output_dir: Path | None = None,
 ) -> dict[str, Any]:
-    started_at = datetime.now(tz=UTC)
+    # 处理单条 item：提取 -> LLM 摘要 -> 规则过滤 -> 候选构建
-    resolved_output_dir = output_dir or default_output_dir()
+    # 返回 item_report dict；delivered 时额外携带 _candidate/_external_id 供调用方解包
-    run_stamp = started_at.strftime("%Y%m%d-%H%M%S")
-    resolved_run_id = run_id or f"freshrss-pipeline-{run_stamp}"
-    resolved_delivery_date = delivery_date or datetime.now(tz=UTC).date()
-
-    resolved_api_base_url = _load_required_env("FRESHRSS_API_BASE_URL", api_base_url)
-    resolved_username = _load_required_env("FRESHRSS_USERNAME", username)
-    resolved_api_password = _load_required_env("FRESHRSS_API_PASSWORD", api_password)
-    resolved_llm_api_key, resolved_llm_model, resolved_llm_api_url = resolve_llm_settings(
-        api_key=llm_api_key,
-        model=llm_model,
-        api_url=llm_api_url,
-    )
-
-    resolved_prompt_path = prompt or DEFAULT_PROMPT_PATH
-    resolved_rules_path = rules or DEFAULT_RULES_PATH
-    raw_output = resolved_output_dir / "raw" / "freshrss.raw.json"
-    delivery_output = resolved_output_dir / "candidates" / "openclaw-delivery-payload.json"
-    report_output = resolved_output_dir / "run-report.json"
-    items_list_output = _maybe_path(debug_artifacts, resolved_output_dir / "items" / "freshrss.items.json")
-
-    client = FreshRSSClient(
-        api_base_url=resolved_api_base_url,
-        username=resolved_username,
-        api_password=resolved_api_password,
-        timeout_seconds=timeout_seconds,
-    )
-    auth_token = client.client_login()
-    payload = client.fetch_stream_contents(
-        auth_token=auth_token,
-        stream_id=stream_id,
-        limit=limit,
-        continuation=continuation,
-        exclude_targets=[] if include_read else [READ_TAG],
-    )
-    entries = payload.get("items")
-    if not isinstance(entries, list):
-        raise RuntimeError("FreshRSS stream response does not contain an items array.")
-
-    _save_json(raw_output, payload)
-
-    items = [map_entry_to_item(entry) for entry in entries]
-    if items_list_output is not None:
-        _save_json(items_list_output, [item.model_dump(mode="json") for item in items])
-
-    loaded_rules = load_filter_rules(resolved_rules_path)
-    # context 优先使用直接传入的 dict，其次读取 context_path 文件，两者均缺失则使用空 context
-    if context is not None:
-        filter_context = FilterContext.model_validate(context)
-    elif context_path is not None:
-        filter_context = FilterContext.model_validate(_load_json(context_path))
-    else:
-        filter_context = FilterContext()
-
-    delivered_candidates: list[OpenClawCandidateInput] = []
-    delivered_item_ids: list[str] = []
-    item_reports: list[dict[str, Any]] = []
-
-    for index, item in enumerate(items, start=1):
-        # 逐条处理：提取 -> 摘要 -> 过滤 -> 候选构建；任一步骤失败则记录状态后 continue
     item_key = f"item-{index:02d}"
     item_path = _maybe_path(debug_artifacts, resolved_output_dir / "items" / f"{item_key}.item.json")
     extracted_path = resolved_output_dir / "extracted" / f"{item_key}.extracted.json"
@@ -192,8 +125,7 @@ def run_freshrss_pipeline(
     if not extraction.success or extraction.article is None:
         item_report["status"] = "extract_failed"
         item_report["error"] = extraction.error.model_dump(mode="json") if extraction.error else None
-            item_reports.append(item_report)
+        return item_report
-            continue
 
     item_report["status"] = "extracted"
     summary_exit_code, summary_payload, summary_report = run_loop_payload(
@@ -210,8 +142,7 @@ def run_freshrss_pipeline(
         item_report["status"] = "summary_failed"
         if summary_report is not None:
             item_report["summary_errors"] = summary_report.errors
-            item_reports.append(item_report)
+        return item_report
-            continue
 
     summary = LlmSummaryResult.model_validate(summary_payload)
     decision = evaluate_filter_rules(
@@ -245,15 +176,22 @@ def run_freshrss_pipeline(
     if openclaw_path is not None:
         _save_json(openclaw_path, openclaw_input.model_dump(mode="json"))
 
-        delivered_candidates.append(openclaw_input)
-        if item.external_id:
-            delivered_item_ids.append(item.external_id)
-
     item_report["status"] = "delivered"
     item_report["selection_decision"] = decision.decision
     item_report["candidate_id"] = openclaw_input.candidate_id
-        item_reports.append(item_report)
+    item_report["_candidate"] = openclaw_input
+    item_report["_external_id"] = item.external_id
+    return item_report
 
+
+def _build_and_persist_delivery(
+    *,
+    delivered_candidates: list[OpenClawCandidateInput],
+    resolved_run_id: str,
+    resolved_delivery_date: Any,
+    delivery_output: Path,
+) -> tuple[OpenClawDeliveryPayload, dict[str, Any]]:
+    # 按 digest_rank 排序，构建 delivery payload，持久化词元索引
     delivered_candidates.sort(key=lambda candidate: candidate.digest_rank, reverse=True)
     delivery_payload = build_openclaw_delivery_payload(
         delivered_candidates,
@@ -272,19 +210,31 @@ def run_freshrss_pipeline(
         aliases_path=DEFAULT_TERM_ALIASES_PATH,
         stopwords_path=DEFAULT_TERM_STOPWORDS_PATH,
     )
+    return delivery_payload, keyword_index_result
 
-    marked_count = 0
-    if mark_read and delivered_item_ids:
-        # 仅标记成功投递（delivered）的 item；drop/review 的 item 保持未读状态
-        client.mark_items_as_read(auth_token=auth_token, item_ids=delivered_item_ids)
-        marked_count = len({item_id for item_id in delivered_item_ids if item_id})
 
+def _build_run_report(
+    *,
+    resolved_run_id: str,
+    started_at: datetime,
+    limit: int,
+    items: list,
+    delivered_candidates: list,
+    marked_count: int,
+    mark_read: bool,
+    debug_artifacts: bool,
+    raw_output: Path,
+    delivery_output: Path,
+    keyword_index_result: dict[str, Any],
+    item_reports: list[dict[str, Any]],
+) -> dict[str, Any]:
+    # 统计各状态计数，组装 run report dict
     status_counts: dict[str, int] = {}
     for item_report in item_reports:
         status = str(item_report["status"])
         status_counts[status] = status_counts.get(status, 0) + 1
 
-    report = {
+    return {
         "run_id": resolved_run_id,
         "started_at": started_at.isoformat(),
         "completed_at": datetime.now(tz=UTC).isoformat(),
@@ -300,6 +250,146 @@ def run_freshrss_pipeline(
         "status_counts": status_counts,
         "items": item_reports,
     }
+
+
+def run_freshrss_pipeline(
+    *,
+    api_base_url: str | None = None,
+    username: str | None = None,
+    api_password: str | None = None,
+    stream_id: str = "user/-/state/com.google/reading-list",
+    limit: int = 5,
+    continuation: str | None = None,
+    include_read: bool = False,
+    mark_read: bool = False,
+    debug_artifacts: bool = False,
+    prompt: Path | None = None,
+    rules: Path | None = None,
+    context: dict[str, Any] | None = None,
+    context_path: Path | None = None,
+    max_retries: int = 2,
+    timeout_seconds: float = 60.0,
+    llm_api_key: str | None = None,
+    llm_model: str | None = None,
+    llm_api_url: str | None = None,
+    run_id: str | None = None,
+    delivery_date: date | None = None,
+    output_dir: Path | None = None,
+) -> dict[str, Any]:
+    # --- 阶段 1：初始化 run_id、输出路径、凭证 ---
+    started_at = datetime.now(tz=UTC)
+    resolved_output_dir = output_dir or default_output_dir()
+    run_stamp = started_at.strftime("%Y%m%d-%H%M%S")
+    resolved_run_id = run_id or f"freshrss-pipeline-{run_stamp}"
+    resolved_delivery_date = delivery_date or datetime.now(tz=UTC).date()
+
+    resolved_api_base_url = _load_required_env("FRESHRSS_API_BASE_URL", api_base_url)
+    resolved_username = _load_required_env("FRESHRSS_USERNAME", username)
+    resolved_api_password = _load_required_env("FRESHRSS_API_PASSWORD", api_password)
+    resolved_llm_api_key, resolved_llm_model, resolved_llm_api_url = resolve_llm_settings(
+        api_key=llm_api_key,
+        model=llm_model,
+        api_url=llm_api_url,
+    )
+
+    resolved_prompt_path = prompt or DEFAULT_PROMPT_PATH
+    resolved_rules_path = rules or DEFAULT_RULES_PATH
+    raw_output = resolved_output_dir / "raw" / "freshrss.raw.json"
+    delivery_output = resolved_output_dir / "candidates" / "openclaw-delivery-payload.json"
+    report_output = resolved_output_dir / "run-report.json"
+    items_list_output = _maybe_path(debug_artifacts, resolved_output_dir / "items" / "freshrss.items.json")
+
+    # --- 阶段 2：登录 FreshRSS，拉取未读条目，写原始 payload ---
+    client = FreshRSSClient(
+        api_base_url=resolved_api_base_url,
+        username=resolved_username,
+        api_password=resolved_api_password,
+        timeout_seconds=timeout_seconds,
+    )
+    auth_token = client.client_login()
+    payload = client.fetch_stream_contents(
+        auth_token=auth_token,
+        stream_id=stream_id,
+        limit=limit,
+        continuation=continuation,
+        exclude_targets=[] if include_read else [READ_TAG],
+    )
+    entries = payload.get("items")
+    if not isinstance(entries, list):
+        raise RuntimeError("FreshRSS stream response does not contain an items array.")
+
+    _save_json(raw_output, payload)
+
+    items = [map_entry_to_item(entry) for entry in entries]
+    if items_list_output is not None:
+        _save_json(items_list_output, [item.model_dump(mode="json") for item in items])
+
+    loaded_rules = load_filter_rules(resolved_rules_path)
+    # context 优先使用直接传入的 dict，其次读取 context_path 文件，两者均缺失则使用空 context
+    if context is not None:
+        filter_context = FilterContext.model_validate(context)
+    elif context_path is not None:
+        filter_context = FilterContext.model_validate(_load_json(context_path))
+    else:
+        filter_context = FilterContext()
+
+    # --- 阶段 3：逐条处理（提取 -> LLM 摘要 -> 规则过滤 -> 候选构建） ---
+    delivered_candidates: list[OpenClawCandidateInput] = []
+    delivered_item_ids: list[str] = []
+    item_reports: list[dict[str, Any]] = []
+
+    for index, item in enumerate(items, start=1):
+        result = _process_item(
+            index=index,
+            item=item,
+            resolved_output_dir=resolved_output_dir,
+            resolved_prompt_path=resolved_prompt_path,
+            resolved_run_id=resolved_run_id,
+            debug_artifacts=debug_artifacts,
+            loaded_rules=loaded_rules,
+            filter_context=filter_context,
+            max_retries=max_retries,
+            timeout_seconds=timeout_seconds,
+            resolved_llm_api_key=resolved_llm_api_key,
+            resolved_llm_model=resolved_llm_model,
+            resolved_llm_api_url=resolved_llm_api_url,
+        )
+        if result.get("status") == "delivered":
+            delivered_candidates.append(result.pop("_candidate"))
+            external_id = result.pop("_external_id", None)
+            if external_id:
+                delivered_item_ids.append(external_id)
+        item_reports.append(result)
+
+    # --- 阶段 4+5：构建 delivery payload 并持久化词元索引 ---
+    delivery_payload, keyword_index_result = _build_and_persist_delivery(
+        delivered_candidates=delivered_candidates,
+        resolved_run_id=resolved_run_id,
+        resolved_delivery_date=resolved_delivery_date,
+        delivery_output=delivery_output,
+    )
+
+    # --- 阶段 6：标记已读，汇总报告，返回结果 ---
+    marked_count = 0
+    if mark_read and delivered_item_ids:
+        # 仅标记成功投递（delivered）的 item；drop/review 的 item 保持未读状态
+        client.mark_items_as_read(auth_token=auth_token, item_ids=delivered_item_ids)
+        marked_count = len({item_id for item_id in delivered_item_ids if item_id})
+
+    report = _build_run_report(
+        resolved_run_id=resolved_run_id,
+        started_at=started_at,
+        limit=limit,
+        items=items,
+        delivered_candidates=delivered_candidates,
+        marked_count=marked_count,
+        mark_read=mark_read,
+        debug_artifacts=debug_artifacts,
+        raw_output=raw_output,
+        delivery_output=delivery_output,
+        keyword_index_result=keyword_index_result,
+        item_reports=item_reports,
+    )
     _save_json(report_output, report)
 
     return {
@@ -312,7 +402,7 @@ def run_freshrss_pipeline(
         "pulled_count": len(items),
         "delivered_count": len(delivered_candidates),
         "marked_read_count": marked_count,
-        "status_counts": status_counts,
+        "status_counts": report["status_counts"],
         "debug_artifacts": debug_artifacts,
         "delivery_payload": delivery_payload.model_dump(mode="json"),
         "items": item_reports,