From 47c4e84fdd7b1766f1a80da0ae8fd8b5949d37c5 Mon Sep 17 00:00:00 2001
From: jxxghp <jxxghp@gmail.com>
Date: Wed, 20 May 2026 16:54:01 +0800
Subject: [PATCH] fix: handle tmdb gzip json responses

---
 app/modules/themoviedb/tmdbv3api/tmdb.py | 43 ++++++++++++++++++++----
 tests/test_tmdb_response_cache.py        | 36 ++++++++++++++++++++
 2 files changed, 73 insertions(+), 6 deletions(-)

diff --git a/app/modules/themoviedb/tmdbv3api/tmdb.py b/app/modules/themoviedb/tmdbv3api/tmdb.py
index 3e93de4f..52bea513 100644
--- a/app/modules/themoviedb/tmdbv3api/tmdb.py
+++ b/app/modules/themoviedb/tmdbv3api/tmdb.py
@@ -22,6 +22,7 @@ logger = logging.getLogger(__name__)
 class TMDb(object):
     _RESPONSE_SNAPSHOT_MARKER = "__mp_tmdb_response_snapshot__"
     _JSON_DECODE_FAILED = object()
+    _MAX_GZIP_DECODE_DEPTH = 3
 
     def __init__(self, session=None, language=None):
         self._api_key = settings.TMDB_API_KEY
@@ -37,14 +38,27 @@ class TMDb(object):
 
         if not self._session:
             self._session = requests.Session()
-        self._req = RequestUtils(ua=settings.NORMAL_USER_AGENT, session=self._session, proxies=self.proxies)
+        request_headers = self._build_request_headers()
+        self._req = RequestUtils(headers=request_headers, session=self._session, proxies=self.proxies)
 
-        self._async_req = AsyncRequestUtils(ua=settings.NORMAL_USER_AGENT, proxies=self.proxies)
+        self._async_req = AsyncRequestUtils(headers=request_headers, proxies=self.proxies)
 
         self._remaining = 40
         self._reset = None
         self._timeout = 15
 
+    @staticmethod
+    def _build_request_headers():
+        """
+        构造TMDB JSON请求头，避免小体积JSON被代理重复压缩。
+        """
+        return {
+            "User-Agent": settings.NORMAL_USER_AGENT,
+            "Content-Type": "application/x-www-form-urlencoded; charset=UTF-8",
+            "Accept": "application/json",
+            "Accept-Encoding": "identity",
+        }
+
     @property
     def page(self):
         return self._page
@@ -197,10 +211,27 @@ class TMDb(object):
         if "gzip" not in encodings and not content_bytes.startswith(b"\x1f\x8b"):
             return cls._JSON_DECODE_FAILED
 
-        try:
-            return jsonlib.loads(gzip.decompress(content_bytes))
-        except (OSError, EOFError, ValueError, UnicodeDecodeError):
-            return cls._JSON_DECODE_FAILED
+        for json_payload in cls._iter_gzip_decoded_payloads(content_bytes):
+            try:
+                return jsonlib.loads(json_payload)
+            except (ValueError, UnicodeDecodeError):
+                continue
+        return cls._JSON_DECODE_FAILED
+
+    @classmethod
+    def _iter_gzip_decoded_payloads(cls, content_bytes: bytes):
+        """
+        逐层解开gzip响应体，兼容客户端或代理只解压了部分层级的情况。
+        """
+        current_payload = content_bytes
+        for _ in range(cls._MAX_GZIP_DECODE_DEPTH):
+            if not current_payload.startswith(b"\x1f\x8b"):
+                return
+            try:
+                current_payload = gzip.decompress(current_payload)
+            except (OSError, EOFError):
+                return
+            yield current_payload
 
     @staticmethod
     def _get_header_value(headers, name):
diff --git a/tests/test_tmdb_response_cache.py b/tests/test_tmdb_response_cache.py
index 3421be74..ce573dd8 100644
--- a/tests/test_tmdb_response_cache.py
+++ b/tests/test_tmdb_response_cache.py
@@ -174,7 +174,29 @@ class _GzipJsonResponse(_UnicodeDecodeErrorResponse):
         super().__init__(gzip.compress(json.dumps(payload).encode("utf-8")))
 
 
+class _DoubleGzipJsonResponse(_UnicodeDecodeErrorResponse):
+    """
+    模拟代理或上游重复gzip压缩后的JSON响应。
+    """
+
+    def __init__(self, payload):
+        """
+        将JSON载荷压缩两次，复现客户端只自动解开外层gzip的情况。
+        """
+        inner_payload = gzip.compress(json.dumps(payload).encode("utf-8"))
+        super().__init__(gzip.compress(inner_payload))
+
+
 class TmdbResponseCacheTest(TestCase):
+    def test_build_request_headers_disables_response_compression(self):
+        """
+        TMDB请求应避免主动接受压缩JSON，减少代理保留gzip响应头的兼容问题。
+        """
+        headers = TMDb._build_request_headers()
+
+        self.assertEqual(headers["Accept"], "application/json")
+        self.assertEqual(headers["Accept-Encoding"], "identity")
+
     def test_request_returns_pickleable_snapshot(self):
         tmdb = TMDb()
         response = _FakeResponse(
@@ -252,6 +274,20 @@ class TmdbResponseCacheTest(TestCase):
         self.assertTrue(result[TMDb._RESPONSE_SNAPSHOT_MARKER])
         self.assertEqual(result["json"]["results"], [{"id": 100}])
 
+    def test_request_decodes_nested_gzip_json_response(self):
+        """
+        响应体仍是gzip字节时，应逐层解压直到得到可解析的JSON。
+        """
+        tmdb = TMDb()
+        tmdb._req.get_res = lambda *args, **kwargs: _DoubleGzipJsonResponse(
+            {"page": 1, "results": [{"id": 101}]}
+        )
+
+        result = TMDb.request.__wrapped__(tmdb, "GET", "https://example.com", None, None)
+
+        self.assertTrue(result[TMDb._RESPONSE_SNAPSHOT_MARKER])
+        self.assertEqual(result["json"]["results"], [{"id": 101}])
+
     def test_get_response_json_rejects_invalid_live_response(self):
         """
         未缓存的实时响应解析失败时也应输出统一诊断信息。