feat: 新增企微已推送新闻 link 去重缓存

Co-authored-by: Cursor <cursoragent@cursor.com>

daily/news/fetch.py

@@ -46,7 +46,7 @@ def _cn_enabled() -> bool:
 
 
 def _hours_window() -> int:
-    return max(1, env_int("DAILY_AI_NEWS_HOURS", 72))
+    return max(1, env_int("DAILY_AI_NEWS_HOURS", 24))
 
 
 def _per_feed_limit() -> int:
@@ -497,7 +497,7 @@ def _format_news_section(
     return lines
 
 
-def prepare_wecom_news_items(news: dict[str, Any]) -> list[dict[str, Any]]:
+def prepare_wecom_news_items(news: dict[str, Any], *, date_str: str | None = None) -> list[dict[str, Any]]:
     if not news.get("enabled"):
         return []
     limit = _wecom_limit()
@@ -528,10 +528,14 @@ def prepare_wecom_news_items(news: dict[str, Any]) -> list[dict[str, Any]]:
                 "desc_short": _clean_text(item.get("summary", ""), 36),
             }
         )
+    if date_str:
+        from daily.news.pushed_links import filter_unpushed_items
+
+        items = filter_unpushed_items(items, date_str=date_str)
     return items
 
 
-def prepare_wecom_cn_news_items(news: dict[str, Any]) -> list[dict[str, Any]]:
+def prepare_wecom_cn_news_items(news: dict[str, Any], *, date_str: str | None = None) -> list[dict[str, Any]]:
     if not news.get("enabled"):
         return []
     limit = _wecom_cn_limit()
@@ -577,4 +581,8 @@ def prepare_wecom_cn_news_items(news: dict[str, Any]) -> list[dict[str, Any]]:
                 "desc_short": _clean_text(item.get("summary", ""), 36),
             }
         )
+    if date_str:
+        from daily.news.pushed_links import filter_unpushed_items
+
+        items = filter_unpushed_items(items, date_str=date_str)
     return items

daily/news/pushed_links.py

@@ -0,0 +1,87 @@
+"""已推送企微早报的新闻 link 去重缓存。"""
+
+from __future__ import annotations
+
+import json
+from datetime import datetime, timedelta
+from pathlib import Path
+from typing import Any
+
+from daily.config import CACHE_DIR, news_dedup_days
+from daily.news.fetch import _normalize_link
+
+
+def _cache_path() -> Path:
+    return CACHE_DIR / "pushed-news-links.json"
+
+
+def _load_raw() -> dict[str, Any]:
+    path = _cache_path()
+    if not path.exists():
+        return {"dates": {}}
+    try:
+        data = json.loads(path.read_text(encoding="utf-8"))
+    except (OSError, ValueError):
+        return {"dates": {}}
+    if not isinstance(data.get("dates"), dict):
+        return {"dates": {}}
+    return data
+
+
+def _save_raw(data: dict[str, Any]) -> None:
+    path = _cache_path()
+    path.parent.mkdir(parents=True, exist_ok=True)
+    path.write_text(json.dumps(data, ensure_ascii=False, indent=2), encoding="utf-8")
+
+
+def _prune(data: dict[str, Any], *, keep_days: int) -> None:
+    dates: dict[str, list[str]] = data.setdefault("dates", {})
+    try:
+        anchor = max(datetime.strptime(d, "%Y-%m-%d") for d in dates)
+    except ValueError:
+        return
+    cutoff = anchor - timedelta(days=keep_days)
+    for key in list(dates.keys()):
+        try:
+            if datetime.strptime(key, "%Y-%m-%d") < cutoff:
+                dates.pop(key, None)
+        except ValueError:
+            dates.pop(key, None)
+
+
+def load_pushed_link_set() -> set[str]:
+    data = _load_raw()
+    out: set[str] = set()
+    for links in (data.get("dates") or {}).values():
+        if isinstance(links, list):
+            out.update(str(x) for x in links if x)
+    return out
+
+
+def filter_unpushed_items(
+    items: list[dict[str, Any]],
+    *,
+    date_str: str,
+) -> list[dict[str, Any]]:
+    del date_str  # reserved for per-day scoping if needed later
+    seen = load_pushed_link_set()
+    out: list[dict[str, Any]] = []
+    for item in items:
+        link = _normalize_link(str(item.get("link") or ""))
+        if not link or link in seen:
+            continue
+        out.append(item)
+    return out
+
+
+def record_pushed_links(date_str: str, links: list[str]) -> None:
+    data = _load_raw()
+    dates: dict[str, list[str]] = data.setdefault("dates", {})
+    normalized: list[str] = []
+    for link in links:
+        clean = _normalize_link(link)
+        if clean:
+            normalized.append(clean)
+    dates[date_str] = sorted(set(normalized))
+    _prune(data, keep_days=news_dedup_days())
+    _save_raw(data)