feat(ingest): parse video refs and fetch detail by bvid

tests/fixtures.py

@@ -176,6 +176,12 @@ VIEW_PAYLOAD: dict[str, Any] = {
         "pubdate": 1793370433,
         "ctime": 1793370433,
         "duration": 9504,
+        "pic": "http://i1.hdslb.com/bfs/archive/e5d3d4f25777b08c1dfae8bc9dc3d56b21d548f1.jpg",
+        "videos": 2,
+        "pages": [
+            {"cid": 43184885058, "page": 1, "from": "vupload", "part": "第一部分"},
+            {"cid": 43184885059, "page": 2, "from": "vupload", "part": "第二部分"},
+        ],
         "owner": {"mid": 434442072, "name": "_-抱抱-_", "face": ""},
         "stat": {
             "aid": 117530286889638,

tests/test_bilibili_client.py

@@ -340,6 +340,100 @@ def test_view_unavailable_and_malformed():
         raise AssertionError("缺少 stat 字段应报协议错误")
 
 
+def test_view_detail_by_aid_and_bvid():
+    transport = QueueTransport([http_json(VIEW_PAYLOAD)])
+    client, _ = make_client(transport)
+    detail = client.view_detail(aid=117530286889638)
+    assert detail.aid == 117530286889638
+    assert detail.bvid == "BV1xvxi6LEZD"
+    assert detail.up_mid == 434442072
+    assert detail.up_name == "_-抱抱-_"
+    assert detail.pubdate == 1793370433
+    assert detail.duration == 9504
+    assert detail.pages == 2
+    assert detail.cover and detail.cover.startswith("http")
+    assert detail.metrics()["view"] == 819
+    assert "aid=117530286889638" in transport.calls[0]["url"]
+
+    transport = QueueTransport([http_json(VIEW_PAYLOAD)])
+    client, _ = make_client(transport)
+    detail = client.view_detail(bvid="BV1xvxi6LEZD")
+    assert detail.aid == 117530286889638
+    assert "bvid=BV1xvxi6LEZD" in transport.calls[0]["url"]
+
+
+def test_view_detail_fallbacks_and_required_arg():
+    payload = view_for(1)
+    for key in ("owner", "duration", "pages", "videos", "pic"):
+        payload["data"].pop(key, None)
+    transport = QueueTransport([http_json(payload)])
+    client, _ = make_client(transport)
+    detail = client.view_detail(aid=1)
+    assert detail.up_mid is None and detail.up_name is None
+    assert detail.duration == 0
+    assert detail.pages is None
+    assert detail.cover is None
+
+    client, _ = make_client(QueueTransport([]))
+    for kwargs in ({}, {"aid": 1, "bvid": "BV1xvxi6LEZD"}):
+        try:
+            client.view_detail(**kwargs)
+        except ValueError:
+            pass
+        else:
+            raise AssertionError("必须且只能提供 aid 或 bvid")
+
+
+def test_view_detail_unavailable():
+    transport = QueueTransport([
+        http_json(VIEW_UNAVAILABLE), http_json(VIEW_UNAVAILABLE),
+    ])
+    client, _ = make_client(transport)
+    try:
+        client.view_detail(bvid="BV1xvxi6LEZD")
+    except BilibiliAPIError as exc:
+        assert exc.code == 62012
+    else:
+        raise AssertionError("不可访问视频应抛业务错误")
+
+
+def test_resolve_short_link():
+    redirect = HTTPResponse(
+        302,
+        {"location": "https://www.bilibili.com/video/BV1x8x166EzY"},
+        b"",
+    )
+    transport = QueueTransport([redirect])
+    client, _ = make_client(transport)
+    final = client.resolve_short_link("https://b23.tv/abc")
+    assert final == "https://www.bilibili.com/video/BV1x8x166EzY"
+    assert transport.calls[0]["method"] == "GET"
+    assert "b23.tv" in transport.calls[0]["url"]
+
+    # 多跳重定向，大小写不同的 Location 头
+    transport = QueueTransport([
+        HTTPResponse(301, {"Location": "https://b23.tv/next"}, b""),
+        HTTPResponse(302, {"location": "https://www.bilibili.com/video/BV1x8x166EzY"}, b""),
+    ])
+    client, _ = make_client(transport)
+    assert client.resolve_short_link("https://b23.tv/start").endswith("BV1x8x166EzY")
+
+    # 非重定向直接返回最终 URL
+    transport = QueueTransport([HTTPResponse(200, {}, b"<html>")])
+    client, _ = make_client(transport)
+    assert client.resolve_short_link("https://b23.tv/x") == "https://b23.tv/x"
+
+    # HTTP 错误应抛出
+    transport = QueueTransport([HTTPResponse(404, {}, b"not found")])
+    client, _ = make_client(transport)
+    try:
+        client.resolve_short_link("https://b23.tv/x")
+    except BilibiliHTTPError as exc:
+        assert exc.status == 404
+    else:
+        raise AssertionError("短链 HTTP 404 应抛出")
+
+
 def test_relation_stat():
     transport = QueueTransport([http_json(RELATION_PAYLOAD)])
     client, _ = make_client(transport)
@@ -436,6 +530,10 @@ def main():
     test_timeout_and_method_forwarded()
     test_view_seven_metrics()
     test_view_unavailable_and_malformed()
+    test_view_detail_by_aid_and_bvid()
+    test_view_detail_fallbacks_and_required_arg()
+    test_view_detail_unavailable()
+    test_resolve_short_link()
     test_relation_stat()
     test_check_login()
     test_check_login_without_cookie_makes_no_request()

tests/test_video_ref.py

@@ -0,0 +1,93 @@
+# -*- coding: utf-8 -*-
+"""视频引用解析纯函数测试: uv run python tests/test_video_ref.py"""
+
+import os
+import sys
+
+sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+sys.stdout.reconfigure(encoding="utf-8")
+
+from reaction_board.video_ref import VideoRef, extract_video_refs  # noqa: E402
+
+BV = "BV1x8x166EzY"
+
+
+def test_bv_url_and_bare_bvid():
+    assert extract_video_refs(f"https://www.bilibili.com/video/{BV}") == [
+        VideoRef("bvid", BV)
+    ]
+    assert extract_video_refs(BV) == [VideoRef("bvid", BV)]
+    assert extract_video_refs(
+        f"https://m.bilibili.com/video/{BV}?p=2&spm_id_from=333.999"
+    ) == [VideoRef("bvid", BV)]
+    assert extract_video_refs(f"https://www.bilibili.com/video/{BV}/?share=1") == [
+        VideoRef("bvid", BV)
+    ]
+
+
+def test_arbitrary_text_containing_bvid():
+    assert extract_video_refs(f"看看这个 {BV} 很有意思") == [VideoRef("bvid", BV)]
+    assert extract_video_refs(f"前缀{BV}后缀") == [VideoRef("bvid", BV)]
+
+
+def test_av_id_forms():
+    assert extract_video_refs("av12345") == [VideoRef("aid", 12345)]
+    assert extract_video_refs("AV12345") == [VideoRef("aid", 12345)]
+    assert extract_video_refs("https://www.bilibili.com/video/av12345") == [
+        VideoRef("aid", 12345)
+    ]
+
+
+def test_short_link():
+    assert extract_video_refs("https://b23.tv/abcDEF12") == [
+        VideoRef("short", "abcDEF12")
+    ]
+    assert extract_video_refs("b23.tv/xyz") == [VideoRef("short", "xyz")]
+
+
+def test_multiple_and_order_and_dedup():
+    text = f"{BV} 还有 av100 再看 {BV} 以及 https://b23.tv/short1"
+    assert extract_video_refs(text) == [
+        VideoRef("bvid", BV),
+        VideoRef("aid", 100),
+        VideoRef("short", "short1"),
+    ]
+
+
+def test_limit():
+    text = " ".join(["av1", "av2", "av3"])
+    assert extract_video_refs(text, limit=2) == [
+        VideoRef("aid", 1),
+        VideoRef("aid", 2),
+    ]
+    assert len(extract_video_refs(text, limit=None)) == 3
+
+
+def test_invalid_and_boundaries():
+    assert extract_video_refs("") == []
+    assert extract_video_refs(None) == []
+    assert extract_video_refs(12345) == []
+    assert extract_video_refs("hello world") == []
+    # BV 号长度不足
+    assert extract_video_refs("BV1x8") == []
+    # 更长的字母数字串不应被截断成假 BV 号
+    assert extract_video_refs(f"{BV}EXTRA") == []
+    # "have123" 不应命中 av
+    assert extract_video_refs("have123") == []
+    # av 后紧跟字母不应命中
+    assert extract_video_refs("av12x") == []
+
+
+def main():
+    test_bv_url_and_bare_bvid()
+    test_arbitrary_text_containing_bvid()
+    test_av_id_forms()
+    test_short_link()
+    test_multiple_and_order_and_dedup()
+    test_limit()
+    test_invalid_and_boundaries()
+    print("OK: 视频引用（BV/av/URL/短链）解析测试全部通过")
+
+
+if __name__ == "__main__":
+    main()