reaction_board/repository.py

@@ -410,6 +410,75 @@ class Repository:
             return [dict(row) for row in rows]
         return self.db.read(run)
 
+    def refresh_video_metadata(
+        self,
+        *,
+        aid: int,
+        bvid: str | None,
+        title: str,
+        up_mid: int | None,
+        up_name: str | None,
+        pubdate: int,
+        duration: int | None = None,
+        pages: int | None = None,
+        cover: str | None = None,
+        up_face: str | None = None,
+        now: int | None = None,
+    ) -> bool:
+        """用接口返回字段覆盖更新已有视频元数据；不改 status，空值不冲掉旧值。
+
+        入库与快照成功路径共用。视频不存在时返回 False。
+        """
+        now = self._now() if now is None else int(now)
+        aid = int(aid)
+
+        def run(conn: sqlite3.Connection) -> bool:
+            existing = conn.execute(
+                "SELECT aid FROM video WHERE aid = ?", (aid,)
+            ).fetchone()
+            if existing is None:
+                return False
+            conn.execute(
+                """
+                UPDATE video SET
+                  bvid = COALESCE(?, bvid),
+                  title = CASE
+                    WHEN ? != '' THEN ?
+                    ELSE title
+                  END,
+                  up_mid = COALESCE(?, up_mid),
+                  up_name = COALESCE(?, up_name),
+                  pubdate = CASE
+                    WHEN ? > 0 THEN ?
+                    ELSE pubdate
+                  END,
+                  duration = COALESCE(?, duration),
+                  pages = COALESCE(?, pages),
+                  cover = COALESCE(?, cover),
+                  up_face = COALESCE(?, up_face),
+                  last_sync = ?
+                WHERE aid = ?
+                """,
+                (
+                    bvid,
+                    title or "",
+                    title or "(无标题)",
+                    up_mid,
+                    up_name,
+                    int(pubdate or 0),
+                    int(pubdate or 0),
+                    None if duration is None else int(duration),
+                    None if pages is None else int(pages),
+                    cover,
+                    up_face,
+                    now,
+                    aid,
+                ),
+            )
+            return True
+
+        return self.db.write(run)
+
     def ingest_video(
         self,
         *,
@@ -430,6 +499,7 @@ class Repository:
 
         不写 B 站收藏夹、不设置 ``folder_id``；同时删除同 aid 的候选行。
         既有的 ``removed`` / ``dropped`` / ``unfaved`` 状态都会被重新激活。
+        已存在时仍以本次接口字段为准刷新元数据。
         """
         now = self._now() if now is None else int(now)
         aid = int(aid)

reaction_board/services/library.py

@@ -116,11 +116,37 @@ class LibraryService:
         if candidate is None:
             video = self.repo.get_video(aid)
             if video is not None:
+                # 已在库：仍以接口为准刷新元数据（不重复建档）
+                try:
+                    detail = self.read_provider.get().view_detail(aid=aid)
+                except BilibiliError as exc:
+                    self._log("warning", "ingest.detail_failed", aid=aid, error=str(exc))
+                    return IngestResult(
+                        ok=True, aid=aid, state="exists",
+                        title=str(video.get("title") or ""),
+                        up_name=str(video.get("up_name") or ""),
+                        message="已在库，无需重复入库",
+                    )
+                self.repo.refresh_video_metadata(
+                    aid=detail.aid,
+                    bvid=detail.bvid,
+                    title=detail.title,
+                    up_mid=detail.up_mid,
+                    up_name=detail.up_name,
+                    pubdate=detail.pubdate,
+                    duration=detail.duration,
+                    pages=detail.pages,
+                    cover=detail.cover,
+                    up_face=detail.up_face,
+                    now=self._now(),
+                )
+                self._write_snapshot(detail)
+                self.repo.bump_cache_version("videos")
                 return IngestResult(
                     ok=True, aid=aid, state="exists",
-                    title=str(video.get("title") or ""),
+                    title=detail.title,
-                    up_name=str(video.get("up_name") or ""),
+                    up_name=detail.up_name or "",
-                    message="已在库，无需重复入库",
+                    message="已在库，已刷新元数据",
                 )
             return IngestResult(
                 ok=False, aid=aid, state="missing", message="候选不存在或已处理"

reaction_board/services/snapshot.py

@@ -3,12 +3,13 @@
 
 规则（DEVELOPMENT.md §7.4）：
 
-* 对 ``active`` / ``unfaved`` 视频逐条请求 ``view``；
+* 对 ``active`` / ``unfaved`` 视频逐条请求 ``view_detail``（完整元数据 + 七项）；
-* 成功写入七项指标；``ok=1`` 的同日快照直接跳过（断点续跑；含小时任务 1 小时内刚刷新的）；
+* 成功时以接口字段覆盖更新 ``video`` 元数据（title/duration/pages/cover/UP 等），并写入七项快照；
+* ``ok=1`` 的同日快照直接跳过（断点续跑；含小时任务 1 小时内刚刷新的）；
 * 午夜 ``seed_at`` 用上一完成日 ok=1 行克隆出当日壳（``snapshot:seeded`` / ``fans:seeded``），不推进完成日；
 * 晚间 ``snapshot_at`` 全量必须 force 打穿 seed，成功后才写完成日并清除 seeded；
 * ``run_recent_videos`` 只刷新近 N 天新片的当日行，不改完成日 / 备份；
-* 视频不可访问时标记 ``removed``；
+* 视频不可访问时标记 ``removed``（保留元数据与历史快照，站内继续展示，后续采集跳过）；
 * 其他错误最多重试 3 次，仍失败写 ``ok=0`` 占位，不写假零值；
 * 对 ``track_fans`` 逐个请求 ``relation/stat``（独立任务，不影响视频完成日）；
 * 视频采集全部成功后写入当日完成标记并生成日备份。
@@ -224,7 +225,7 @@ class SnapshotService:
         force: bool = False,
         keep_scraped: bool = False,
     ) -> SnapshotSummary:
-        """采集视频七项指标并在全部成功后标记完成日 / 备份。
+        """采集视频元数据 + 七项指标，并在全部成功后标记完成日 / 备份。
 
         ``force``：忽略完成标记，重跑。
         ``keep_scraped``：force 时仍跳过非 seed 的 ok=1（seal 断点续跑）；
@@ -264,7 +265,7 @@ class SnapshotService:
                     continue
                 # force 且非 keep_scraped：CLI 全量重采，不 skip
 
-            stat, error, unavailable = self._fetch_view(client, aid)
+            detail, error, unavailable = self._fetch_view_detail(client, aid)
             if unavailable:
                 summary.removed += 1
                 self.repo.set_video_status(aid, "removed", now=now)
@@ -272,17 +273,16 @@ class SnapshotService:
                     "warning", "snapshot.removed",
                     stat_date=stat_date, aid=aid, error=error,
                 )
-            elif stat is not None:
+            elif detail is not None:
+                self._apply_detail(detail, now=self._now())
                 self.repo.upsert_snapshot(
                     stat_date,
                     aid,
-                    metrics=stat.metrics(),
+                    metrics=detail.metrics(),
                     ok=True,
                     err=None,
                     captured_at=self._now(),
                 )
-                if getattr(stat, "up_face", None):
-                    self.repo.apply_video_up_face(aid, stat.up_face)
                 summary.success += 1
             else:
                 summary.failed += 1
@@ -369,7 +369,7 @@ class SnapshotService:
         videos = self.repo.video_aids_recent_for_snapshot(since_pubdate=since_pubdate)
         summary.total = len(videos)
         for index, aid in enumerate(videos, 1):
-            stat, error, unavailable = self._fetch_view(client, aid)
+            detail, error, unavailable = self._fetch_view_detail(client, aid)
             if unavailable:
                 summary.removed += 1
                 self.repo.set_video_status(aid, "removed", now=now)
@@ -377,17 +377,16 @@ class SnapshotService:
                     "warning", "snapshot.recent_removed",
                     stat_date=stat_date, aid=aid, error=error,
                 )
-            elif stat is not None:
+            elif detail is not None:
+                self._apply_detail(detail, now=self._now())
                 self.repo.upsert_snapshot(
                     stat_date,
                     aid,
-                    metrics=stat.metrics(),
+                    metrics=detail.metrics(),
                     ok=True,
                     err=None,
                     captured_at=self._now(),
                 )
-                if getattr(stat, "up_face", None):
-                    self.repo.apply_video_up_face(aid, stat.up_face)
                 summary.success += 1
             else:
                 summary.failed += 1
@@ -693,11 +692,27 @@ class SnapshotService:
             )
         return state.is_login, state.message
 
-    def _fetch_view(self, client, aid: int):
+    def _apply_detail(self, detail, *, now: int) -> None:
+        """快照成功：以 view_detail 字段覆盖 video 元数据（不改 status）。"""
+        self.repo.refresh_video_metadata(
+            aid=detail.aid,
+            bvid=detail.bvid,
+            title=detail.title,
+            up_mid=detail.up_mid,
+            up_name=detail.up_name,
+            pubdate=detail.pubdate,
+            duration=detail.duration,
+            pages=detail.pages,
+            cover=detail.cover,
+            up_face=detail.up_face,
+            now=now,
+        )
+
+    def _fetch_view_detail(self, client, aid: int):
         last_error = ""
         for attempt in range(1, self.retries + 1):
             try:
-                return client.view(aid), None, False
+                return client.view_detail(aid=aid), None, False
             except BilibiliAPIError as exc:
                 last_error = str(exc)
                 if exc.code in VIDEO_UNAVAILABLE_CODES:

reaction_board/web/queries.py

@@ -45,6 +45,8 @@ from ..timeutil import (
 from .cache import DASHBOARD_CACHE, TTL_FANS, TTL_VIDEOS, build_cache_key
 
 TRACKED_STATUSES = ("active", "unfaved")
+# 站内列表可见：含原站失效 removed（继续展示、badge 可标 status-removed）；不含人工 dropped
+VISIBLE_STATUSES = ("active", "unfaved", "removed")
 VIDEO_SORTS = {
     "heat", "view", "like", "coin", "favorite", "share", "reply", "danmaku",
     "interaction", "growth", "growth_rate", "added", "published",
@@ -955,6 +957,7 @@ class DashboardQueries:
                 within_start = _date_minus(latest, int(within_days) - 1)
             videos = self._videos(
                 conn,
+                statuses=VISIBLE_STATUSES,
                 folder_id=folder_id,
                 q=q,
                 within_field=within_field,
@@ -1157,7 +1160,7 @@ class DashboardQueries:
         aid = int(aid)
 
         def run(conn: sqlite3.Connection) -> dict[str, Any] | None:
-            video_rows = self._videos(conn)
+            video_rows = self._videos(conn, statuses=VISIBLE_STATUSES)
             video = next((item for item in video_rows if int(item["aid"]) == aid), None)
             if video is None:
                 row = conn.execute(

tests/fixtures.py

@@ -256,11 +256,41 @@ def view_for(
     reply: int = 16,
     danmaku: int = 7,
     bvid: str | None = None,
+    title: str | None = None,
+    duration: int | None = None,
+    pages: int | None = None,
+    cover: str | None = None,
+    up_mid: int | None = None,
+    up_name: str | None = None,
+    up_face: str | None = None,
+    pubdate: int | None = None,
 ) -> dict[str, Any]:
     item = copy.deepcopy(VIEW_PAYLOAD)
     data = item["data"]
     data["aid"] = aid
     data["bvid"] = bvid or f"BV{aid}"
+    if title is not None:
+        data["title"] = title
+    if duration is not None:
+        data["duration"] = duration
+    if cover is not None:
+        data["pic"] = cover
+    if pubdate is not None:
+        data["pubdate"] = pubdate
+        data["ctime"] = pubdate
+    if pages is not None:
+        data["videos"] = pages
+        data["pages"] = [
+            {"cid": 1000 + i, "page": i + 1, "from": "vupload", "part": f"P{i + 1}"}
+            for i in range(max(1, int(pages)))
+        ]
+    owner = data.setdefault("owner", {})
+    if up_mid is not None:
+        owner["mid"] = up_mid
+    if up_name is not None:
+        owner["name"] = up_name
+    if up_face is not None:
+        owner["face"] = up_face
     data["stat"].update({
         "aid": aid,
         "view": view,

tests/test_repository.py

@@ -408,6 +408,72 @@ def test_up_follower_snapshot_and_backfill():
             db.close()
 
 
+def test_refresh_video_metadata_overwrites_api_fields_keeps_nulls():
+    with tempfile.TemporaryDirectory() as tmp:
+        db, repo = setup(tmp)
+        try:
+            repo.upsert_from_fav(resource(9, pages=1), F1.media_id, was_added=True)
+            ok = repo.refresh_video_metadata(
+                aid=9,
+                bvid="BVnew9",
+                title="刷新后",
+                up_mid=200,
+                up_name="新名",
+                pubdate=1_710_000_000,
+                duration=999,
+                pages=5,
+                cover="http://example/new.png",
+                up_face="http://example/face.png",
+                now=1_800_000_100,
+            )
+            assert ok is True
+            video = repo.get_video(9)
+            assert video["title"] == "刷新后"
+            assert video["duration"] == 999
+            assert video["pages"] == 5
+            assert video["bvid"] == "BVnew9"
+            assert video["up_mid"] == 200
+            assert video["cover"] == "http://example/new.png"
+            assert video["status"] == "active"
+            # 空值不冲掉
+            ok = repo.refresh_video_metadata(
+                aid=9,
+                bvid=None,
+                title="",
+                up_mid=None,
+                up_name=None,
+                pubdate=0,
+                duration=None,
+                pages=None,
+                cover=None,
+                up_face=None,
+                now=1_800_000_200,
+            )
+            assert ok is True
+            video = repo.get_video(9)
+            assert video["title"] == "刷新后"
+            assert video["duration"] == 999
+            assert video["pages"] == 5
+            assert video["bvid"] == "BVnew9"
+            # 不存在的 aid
+            assert repo.refresh_video_metadata(
+                aid=404,
+                bvid=None, title="x", up_mid=None, up_name=None, pubdate=1,
+            ) is False
+            # removed 后 refresh 也不改 status
+            repo.set_video_status(9, "removed")
+            repo.refresh_video_metadata(
+                aid=9, bvid="BVkeep", title="仍在", up_mid=200, up_name="新名",
+                pubdate=1_710_000_000, duration=1000, pages=6,
+            )
+            video = repo.get_video(9)
+            assert video["status"] == "removed"
+            assert video["title"] == "仍在"
+            assert video["pages"] == 6
+        finally:
+            db.close()
+
+
 def main():
     test_upsert_and_folder_resource()
     test_invalid_attr_and_missing()
@@ -422,8 +488,10 @@ def main():
     test_pending_snapshot_aids()
     test_upsert_snapshot_preserves_view_open()
     test_up_follower_snapshot_and_backfill()
+    test_refresh_video_metadata_overwrites_api_fields_keeps_nulls()
     print("OK: 收藏夹/视频状态机与黑名单测试全部通过")
 
 
+
 if __name__ == "__main__":
     main()

tests/test_snapshot.py

@@ -216,14 +216,86 @@ def test_unavailable_marks_removed():
         try:
             add_video(repo, 1)
             add_video(repo, 2)
+            before = repo.get_video(1)
             result = service.run(date=DATE)
             assert result.removed == 2
             assert result.failed == 0
             assert result.completed is True
-            assert repo.get_video(1)["status"] == "removed"
+            video1 = repo.get_video(1)
-            assert repo.get_video(2)["status"] == "removed"
+            video2 = repo.get_video(2)
+            assert video1["status"] == "removed"
+            assert video2["status"] == "removed"
+            # 失效只改 status，不得抹元数据 / 不得写空快照
+            assert video1["title"] == before["title"]
+            assert video1["duration"] == before["duration"]
+            assert video1["bvid"] == before["bvid"]
             assert repo.get_snapshot(DATE, 1) is None
             assert repo.get_snapshot(DATE, 2) is None
+            # removed 不再进入后续快照队列
+            assert 1 not in repo.video_aids_for_snapshot()
+            assert 2 not in repo.video_aids_for_snapshot()
+        finally:
+            db.close()
+
+
+def test_snapshot_refreshes_video_metadata():
+    """快照成功路径以 view_detail 覆盖 pages/duration/title 等元数据。"""
+    with tempfile.TemporaryDirectory() as tmp:
+        transport = SnapshotTransport({
+            1: view_for(
+                1, view=111, title="新标题P3", duration=3600, pages=3,
+                cover="http://example/new.jpg", up_name="新UP",
+            ),
+        })
+        db, repo, service = make_service(tmp, transport, track_fans=())
+        try:
+            add_video(repo, 1)
+            old = repo.get_video(1)
+            assert old["title"] == "视频1"
+            assert old["duration"] == 600
+            assert old.get("pages") in (None, 0) or old["pages"] != 3
+            result = service.run_videos(date=DATE)
+            assert result.success == 1
+            video = repo.get_video(1)
+            assert video["title"] == "新标题P3"
+            assert video["duration"] == 3600
+            assert video["pages"] == 3
+            assert video["cover"] == "http://example/new.jpg"
+            assert video["up_name"] == "新UP"
+            assert video["status"] == "active"
+            snap = repo.get_snapshot(DATE, 1)
+            assert snap["view"] == 111 and snap["ok"] == 1
+        finally:
+            db.close()
+
+
+def test_recent_snapshot_refreshes_metadata():
+    with tempfile.TemporaryDirectory() as tmp:
+        now = 1_800_000_000
+        transport = SnapshotTransport({
+            7: view_for(7, view=77, title="近新片加P", duration=1200, pages=2,
+                        pubdate=now - 3600),
+        })
+        db, repo, service = make_service(tmp, transport, track_fans=())
+        try:
+            repo.upsert_from_fav(
+                FavResource(
+                    aid=7, bvid="BV7", title="旧标题", up_mid=100,
+                    up_name="UP", pubdate=now - 3600, duration=300,
+                    cover=None, attr=0, fav_time=None, pages=1,
+                ),
+                FOLDER.media_id,
+                was_added=True,
+            )
+            result = service.run_recent_videos(date=DATE, recent_days=2)
+            assert result.success == 1
+            video = repo.get_video(7)
+            assert video["title"] == "近新片加P"
+            assert video["duration"] == 1200
+            assert video["pages"] == 2
+            assert repo.get_snapshot(DATE, 7)["view"] == 77
+            # recent 不推进完成日
+            assert repo.kv_get("snapshot:last_completed") in (None, "")
         finally:
             db.close()
 
@@ -440,11 +512,12 @@ def test_run_recent_videos_refreshes_without_completion_markers():
     """小时任务只刷新近窗新片，不写完成日 / 备份。"""
     clock = 1_800_000_000
     with tempfile.TemporaryDirectory() as tmp:
+        # view_for 须带与库内一致的 pubdate，否则快照回写会把近窗片挤出窗口
         transport = SnapshotTransport({
-            1: view_for(1, view=111),
+            1: view_for(1, view=111, pubdate=clock - 3600),
-            2: view_for(2, view=222),
+            2: view_for(2, view=222, pubdate=clock - 86400),
-            3: view_for(3, view=333),
+            3: view_for(3, view=333, pubdate=clock - 10 * 86400),
-            4: view_for(4, view=444),
+            4: view_for(4, view=444, pubdate=clock - 10 * 86400),
         })
         db, repo, service = make_service(tmp, transport, track_fans=())
         try:
@@ -776,6 +849,8 @@ def main():
     test_snapshot_success_resume_and_force()
     test_resume_skips_ok_and_retries_failed()
     test_unavailable_marks_removed()
+    test_snapshot_refreshes_video_metadata()
+    test_recent_snapshot_refreshes_metadata()
     test_other_error_writes_failed_placeholder()
     test_fan_failure_placeholder()
     test_auth_invalid_is_warned_but_snapshot_continues()

tests/test_web_queries.py

@@ -68,13 +68,16 @@ def test_video_listing_filters_and_sorting():
 
             heat = q.list_videos(sort="heat")
             aids = [item["aid"] for item in heat["items"]]
-            assert 104 not in aids, "removed 视频不应进入榜单"
+            # 热度榜仍按门槛过滤；104 removed 无成功快照，不进热度
+            assert 104 not in aids
             # 综合热度榜应用最小播放门槛 2000：101（1200）与 103（120）被排除
             assert aids == [102], aids
             assert heat["items"][0]["heat_percentile"] == 100.0
 
             views = q.list_videos(sort="view")
-            assert [item["aid"] for item in views["items"]] == [102, 101, 103]
+            # removed(104) 站内继续展示；无指标排在末尾
+            assert [item["aid"] for item in views["items"]] == [102, 101, 103, 104]
+            assert views["items"][-1]["status"] == "removed"
 
             interactions = q.list_videos(sort="interaction")
             assert interactions["items"][0]["aid"] == 101
@@ -91,14 +94,19 @@ def test_video_listing_filters_and_sorting():
             assert [item["aid"] for item in searched["items"]] == [102]
 
             folder = q.list_videos(sort="view", folder_id=3700023351)
-            assert folder["total"] == 3
+            assert folder["total"] == 4  # 含 removed 104
             folder = q.list_videos(sort="view", folder_id=999)
             assert folder["total"] == 0
 
             page = q.list_videos(sort="view", page=2, page_size=1)
-            assert page["total"] == 3
+            assert page["total"] == 4
-            assert page["pages"] == 3
+            assert page["pages"] == 4
             assert [item["aid"] for item in page["items"]] == [101]
+
+            detail = q.video_detail(104)
+            assert detail is not None
+            assert detail["status"] == "removed"
+            assert detail["aid"] == 104
         finally:
             db.close()
 
@@ -457,7 +465,7 @@ def test_year_filter_and_published_sort():
         db, q = setup(tmp)
         try:
             by_year = q.list_videos(sort="view", year=2026)
-            assert [item["aid"] for item in by_year["items"]] == [102, 101, 103]
+            assert [item["aid"] for item in by_year["items"]] == [102, 101, 103, 104]
             assert by_year["year"] == 2026
 
             none = q.list_videos(sort="view", year=2025)
@@ -465,10 +473,10 @@ def test_year_filter_and_published_sort():
             assert none["total"] == 0
 
             published = q.list_videos(sort="published")
-            assert [item["aid"] for item in published["items"]] == [103, 101, 102]
+            assert [item["aid"] for item in published["items"]] == [103, 101, 102, 104]
 
             added = q.list_videos(sort="added")
-            assert added["total"] == 3
+            assert added["total"] == 4  # 含 removed
 
             ups = q.list_ups(sort="efficiency", year=2026)
             assert ups["year"] == 2026 and ups["total"] == 2