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

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()