Merge pull request #421 from pumpkinperson996/fix/normalize-bilibili-url

fix: 规范化含 BV 号的任意 B 站链接（稍后再看/收藏夹/追踪参数）

backend/app/routers/note.py

@@ -7,7 +7,7 @@ from typing import Optional
 from urllib.parse import urlparse
 
 from fastapi import APIRouter, HTTPException, BackgroundTasks, UploadFile, File
-from pydantic import BaseModel, validator, field_validator
+from pydantic import BaseModel, validator, field_validator, model_validator
 from dataclasses import asdict
 
 from app.db.video_task_dao import get_task_by_video
@@ -17,7 +17,7 @@ from app.exceptions.note import NoteError
 from app.services.note import NoteGenerator, logger
 from app.services.task_serial_executor import task_serial_executor
 from app.utils.response import ResponseWrapper as R
-from app.utils.url_parser import extract_video_id
+from app.utils.url_parser import extract_video_id, normalize_video_url
 from app.validators.video_url_validator import is_supported_video_url
 from fastapi import APIRouter, Request, HTTPException
 from fastapi.responses import StreamingResponse
@@ -55,6 +55,15 @@ class VideoRequest(BaseModel):
     #   {"language": "zh", "full_text": "...", "segments": [{"start","end","text"}, ...]}
     prefetched_transcript: Optional[dict] = None
 
+    @model_validator(mode="before")
+    @classmethod
+    def normalize_url(cls, data):
+        # 稍后再看/收藏夹/带追踪参数的 B 站链接先规范化成标准 /video/BVxxx 形式，
+        # 后续校验和 yt-dlp 下载拿到的都是干净链接
+        if isinstance(data, dict) and data.get("platform") == "bilibili" and data.get("video_url"):
+            data["video_url"] = normalize_video_url(str(data["video_url"]))
+        return data
+
     @field_validator("video_url")
     def validate_supported_url(cls, v):
         url = str(v)

backend/app/utils/url_parser.py

@@ -35,6 +35,29 @@ def extract_video_id(url: str, platform: str) -> Optional[str]:
     return None
 
 
+def normalize_video_url(url: str) -> str:
+    """
+    将任意包含 BV 号的 B 站链接规范化为标准视频链接。
+
+    支持稍后再看（/list/watchlater/?bvid=BV...）、收藏夹播放页（/list/mlXXX?bvid=BV...）、
+    带追踪参数的分享链接等。保留分 P 参数，丢弃其余查询参数。
+
+    b23.tv 短链与无 BV 号的链接原样返回（后者交由校验器拒绝）。
+    """
+    if "b23.tv" in url:
+        return url
+
+    match = re.search(r"BV([0-9A-Za-z]+)", url)
+    if not match:
+        return url
+
+    normalized = f"https://www.bilibili.com/video/BV{match.group(1)}"
+    p = extract_bilibili_p_number(url)
+    if p:
+        normalized += f"?p={p}"
+    return normalized
+
+
 def resolve_bilibili_short_url(short_url: str) -> Optional[str]:
     """
     解析哔哩哔哩短链接以获取真实视频链接

backend/app/validators/video_url_validator.py

@@ -1,7 +1,9 @@
-from pydantic import AnyUrl, validator, BaseModel, field_validator
+from pydantic import AnyUrl, validator, BaseModel, field_validator, model_validator
 import re
 from urllib.parse import urlparse
 
+from app.utils.url_parser import normalize_video_url
+
 SUPPORTED_PLATFORMS = {
     "bilibili": r"(https?://)?(www\.)?bilibili\.com/video/[a-zA-Z0-9]+",
     "youtube": r"(https?://)?(www\.)?(youtube\.com/(watch\?v=|shorts/)|youtu\.be/)[\w\-]+",
@@ -31,6 +33,13 @@ class VideoRequest(BaseModel):
     url: AnyUrl
     platform: str
 
+    @model_validator(mode="before")
+    @classmethod
+    def normalize_url(cls, data):
+        if isinstance(data, dict) and data.get("platform") == "bilibili" and data.get("url"):
+            data["url"] = normalize_video_url(str(data["url"]))
+        return data
+
     @field_validator("url")
     def validate_video_url(cls, v):
         if not is_supported_video_url(str(v)):

backend/tests/test_url_normalize.py

@@ -0,0 +1,67 @@
+import pytest
+
+from app.utils.url_parser import normalize_video_url
+from app.validators.video_url_validator import VideoRequest
+
+
+def test_watchlater_url():
+    url = ("https://www.bilibili.com/list/watchlater/?bvid=BV1CPXpBYEui"
+           "&oid=116294762371214&spm_id_from=333.881.0.0&vd_source=abc")
+    assert normalize_video_url(url) == "https://www.bilibili.com/video/BV1CPXpBYEui"
+
+
+def test_favlist_url():
+    url = "https://www.bilibili.com/list/ml123456?bvid=BV1xx411c7mD&oid=987"
+    assert normalize_video_url(url) == "https://www.bilibili.com/video/BV1xx411c7mD"
+
+
+def test_tracking_params_stripped():
+    url = "https://www.bilibili.com/video/BV1xx411c7mD/?spm_id_from=333.881&vd_source=abc"
+    assert normalize_video_url(url) == "https://www.bilibili.com/video/BV1xx411c7mD"
+
+
+def test_p_number_preserved():
+    url = "https://www.bilibili.com/video/BV1xx411c7mD?p=36&spm_id_from=333.881"
+    assert normalize_video_url(url) == "https://www.bilibili.com/video/BV1xx411c7mD?p=36"
+
+
+def test_no_bv_returned_unchanged():
+    url = "https://www.bilibili.com/anime/timeline"
+    assert normalize_video_url(url) == url
+
+
+def test_b23_short_url_unchanged():
+    url = "https://b23.tv/abc123"
+    assert normalize_video_url(url) == url
+
+
+def test_video_request_accepts_watchlater():
+    req = VideoRequest(
+        url="https://www.bilibili.com/list/watchlater/?bvid=BV1CPXpBYEui&oid=116294762371214",
+        platform="bilibili",
+    )
+    assert str(req.url) == "https://www.bilibili.com/video/BV1CPXpBYEui"
+
+
+def test_video_request_rejects_no_bv():
+    with pytest.raises(ValueError):
+        VideoRequest(url="https://www.bilibili.com/anime/timeline", platform="bilibili")
+
+
+def test_note_router_request_accepts_watchlater():
+    # note.py 里的 VideoRequest 才是 /generate_note 实际使用的请求模型
+    from app.routers.note import VideoRequest as NoteVideoRequest
+
+    req = NoteVideoRequest(
+        video_url="https://www.bilibili.com/list/watchlater/?bvid=BV1CPXpBYEui&oid=116294762371214&spm_id_from=333.881.0.0",
+        platform="bilibili",
+        quality="fast",
+        model_name="test-model",
+        provider_id="test-provider",
+    )
+    assert req.video_url == "https://www.bilibili.com/video/BV1CPXpBYEui"
+
+
+def test_video_request_youtube_unaffected():
+    req = VideoRequest(url="https://www.youtube.com/watch?v=dQw4w9WgXcQ", platform="youtube")
+    assert str(req.url) == "https://www.youtube.com/watch?v=dQw4w9WgXcQ"