parsehub 2.3.4__tar.gz → 2.3.5__tar.gz

{parsehub-2.3.4/src/parsehub.egg-info → parsehub-2.3.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: parsehub
-Version: 2.3.4
+Version: 2.3.5
 Summary: 轻量、异步、开箱即用的社交媒体聚合解析库
 Author-email: 梓澪 <zilingmio@gmail.com>
 License: MIT

{parsehub-2.3.4 → parsehub-2.3.5}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "parsehub"
-version = "2.3.4"
+version = "2.3.5"
 description = "轻量、异步、开箱即用的社交媒体聚合解析库"
 readme = "README.md"
 requires-python = ">=3.12.0"

{parsehub-2.3.4 → parsehub-2.3.5}/src/parsehub/parsers/parser/bilibili.py

@@ -31,6 +31,8 @@ class BiliParse(BaseParser):
     __redirect_keywords__ = ["b23.tv", "bili2233.cn"]
 
     async def _do_parse(self, raw_url: str) -> YtVideoParseResult | BiliVideoParseResult | ImageParseResult:
+        if urlparse(raw_url).path.startswith("/bangumi/"):
+            raise ParseError("不支持解析番剧和影视")
         if await self.is_dynamic(raw_url):
             dynamic = await self.get_dynamic_info(raw_url)
             content = self.hashtag_handler(dynamic.content or "")

{parsehub-2.3.4 → parsehub-2.3.5}/src/parsehub/parsers/parser/kuaishou.py

@@ -1,5 +1,13 @@
 from ...provider_api.kuaishou import KuaiShouAPI, KuaishouParser
-from ...types import ImageParseResult, ImageRef, ParseError, Platform, VideoParseResult, VideoRef
+from ...types import (
+    ContentUnavailableError,
+    ImageParseResult,
+    ImageRef,
+    ParseError,
+    Platform,
+    VideoParseResult,
+    VideoRef,
+)
 from ...utils.helpers import SecretCookie
 from ..base.base import BaseParser, FallbackStage
 
@@ -26,13 +34,15 @@ COOKIE = SecretCookie(
 class KuaiShouParser(BaseParser):
     __platform__ = Platform.KUAISHOU
     __supported_type__ = ["视频", "图文"]
-    __match__ = r"^(http(s)?://)?(www|v|live|v\.m)\.(kuaishou|chenzhongtech).com/.+"
+    __match__ = r"^(http(s)?://)?(www|v|live|v\.m|[a-z0-9]+\.m)\.(kuaishou|chenzhongtech).com/.+"
     __redirect_keywords__ = ["v.kuaishou", "/f/"]
 
     async def _do_parse(self, raw_url: str) -> VideoParseResult | ImageParseResult:
         ksp = await KuaishouParser.create(
             raw_url, proxy=self.proxy, cookie=self.cookie.get_value() or COOKIE.get_value()
         )
+        if ksp.page_type == "UNKNOWN" and ksp.error_result in (223, 20113):
+            raise ContentUnavailableError(f"{ksp.error_result} {ksp.error_msg}")
         if (
             ksp.page_type == "VIDEO" or "/short-video/" in raw_url
         ):  # 地区风控导致获取到的 page_type 可能为 ATLAS 但实际为视频

{parsehub-2.3.4 → parsehub-2.3.5}/src/parsehub/provider_api/kuaishou.py

@@ -212,6 +212,8 @@ class KuaishouParser:
         self.page_type = "UNKNOWN"
         self.structured_data: dict = {}
         self.client: dict = {}
+        self.error_result: int | None = None
+        self.error_msg: str | None = None
 
     @classmethod
     async def create(cls, real_url, proxy: str | None = None, cookie: dict | None = None):
@@ -319,8 +321,10 @@ class KuaishouParser:
             return False
         if default_client.get("VisionVideoSetRepresentation:1"):
             return True
-        for key in default_client.keys():
-            if "visionVideoDetail" in key or "VisionVideoDetailPhoto:" in key:
+        for key, value in default_client.items():
+            if "visionVideoDetail" in key and isinstance(value, dict) and value.get("photo"):
+                return True
+            if "VisionVideoDetailPhoto:" in key:
                 return True
         return False
 
@@ -333,6 +337,10 @@ class KuaishouParser:
         page_type, structured_data = self._identify_and_parse_data()
         if page_type == "UNKNOWN" or not structured_data:
             return False
+        error_payload = self._find_nested_dict(structured_data, ("result", "error_msg"))
+        if error_payload and error_payload["result"] != 1 and error_payload["error_msg"]:
+            self.error_result = error_payload["result"]
+            self.error_msg = error_payload["error_msg"]
         if not self._is_valid_video_state(page_type, structured_data):
             logger.warning(f"Kuaishou route {candidate_url} returned incomplete video state, trying fallback")
             return False

{parsehub-2.3.4 → parsehub-2.3.5}/src/parsehub/provider_api/twitter.py

@@ -91,6 +91,13 @@ class Twitter:
         if not legacy:
             if result.get("__typename") == "TweetTombstone":
                 raise Exception("error -2: 该推文开启了限制, 匿名用户无法查看")
+            match result.get("reason"):
+                case "Suspended":
+                    raise ContentUnavailableError("error -5: 这个帖子来自一个被冻结的账号")
+                case "BounceDeleted":
+                    raise ContentUnavailableError("error -5: 这个帖子违反了 X 规则")
+                case "Protected":
+                    raise Exception("error -3: 你无法查看这个帖子，因为该账号所有者限制了可以查看其帖子的用户")
             raise Exception(f"error -3: {result.get('reason')}")
 
         if article := result.get("article", {}):

{parsehub-2.3.4 → parsehub-2.3.5/src/parsehub.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: parsehub
-Version: 2.3.4
+Version: 2.3.5
 Summary: 轻量、异步、开箱即用的社交媒体聚合解析库
 Author-email: 梓澪 <zilingmio@gmail.com>
 License: MIT