[ie] Detect invalid m3u8 playlist data (#13563)

Authored by: Grub4K

test/test_InfoExtractor.py

@@ -36,6 +36,18 @@ class InfoExtractorTestRequestHandler(http.server.BaseHTTPRequestHandler):
             self.send_header('Content-Type', 'text/html; charset=utf-8')
             self.end_headers()
             self.wfile.write(TEAPOT_RESPONSE_BODY.encode())
+        elif self.path == '/fake.m3u8':
+            self.send_response(200)
+            self.send_header('Content-Length', '1024')
+            self.end_headers()
+            self.wfile.write(1024 * b'\x00')
+        elif self.path == '/bipbop.m3u8':
+            with open('test/testdata/m3u8/bipbop_16x9.m3u8', 'rb') as f:
+                data = f.read()
+            self.send_response(200)
+            self.send_header('Content-Length', str(len(data)))
+            self.end_headers()
+            self.wfile.write(data)
         else:
             assert False
 
@@ -2079,5 +2091,45 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/
                 self.ie._search_nuxt_json(HTML_TMPL.format(data), None, default=DEFAULT), DEFAULT)
 
 
+class TestInfoExtractorNetwork(unittest.TestCase):
+    def setUp(self, /):
+        self.httpd = http.server.HTTPServer(
+            ('127.0.0.1', 0), InfoExtractorTestRequestHandler)
+        self.port = http_server_port(self.httpd)
+
+        self.server_thread = threading.Thread(target=self.httpd.serve_forever)
+        self.server_thread.daemon = True
+        self.server_thread.start()
+
+        self.called = False
+
+        def require_warning(*args, **kwargs):
+            self.called = True
+
+        self.ydl = FakeYDL()
+        self.ydl.report_warning = require_warning
+        self.ie = DummyIE(self.ydl)
+
+    def tearDown(self, /):
+        self.ydl.close()
+        self.httpd.shutdown()
+        self.httpd.server_close()
+        self.server_thread.join(1)
+
+    def test_extract_m3u8_formats(self):
+        formats, subtitles = self.ie._extract_m3u8_formats_and_subtitles(
+            f'http://127.0.0.1:{self.port}/bipbop.m3u8', None, fatal=False)
+        self.assertFalse(self.called)
+        self.assertTrue(formats)
+        self.assertTrue(subtitles)
+
+    def test_extract_m3u8_formats_warning(self):
+        formats, subtitles = self.ie._extract_m3u8_formats_and_subtitles(
+            f'http://127.0.0.1:{self.port}/fake.m3u8', None, fatal=False)
+        self.assertTrue(self.called, 'Warning was not issued for binary m3u8 file')
+        self.assertFalse(formats)
+        self.assertFalse(subtitles)
+
+
 if __name__ == '__main__':
     unittest.main()

yt_dlp/extractor/common.py

@@ -1,5 +1,6 @@
 import base64
 import collections
+import contextlib
 import functools
 import getpass
 import http.client
@@ -2129,21 +2130,33 @@ class InfoExtractor:
                     raise ExtractorError(errnote, video_id=video_id)
                 self.report_warning(f'{errnote}{bug_reports_message()}')
             return [], {}
-
-        res = self._download_webpage_handle(
-            m3u8_url, video_id,
-            note='Downloading m3u8 information' if note is None else note,
-            errnote='Failed to download m3u8 information' if errnote is None else errnote,
+        if note is None:
+            note = 'Downloading m3u8 information'
+        if errnote is None:
+            errnote = 'Failed to download m3u8 information'
+        response = self._request_webpage(
+            m3u8_url, video_id, note=note, errnote=errnote,
             fatal=fatal, data=data, headers=headers, query=query)
+        if response is False:
+            return [], {}
 
-        if res is False:
+        with contextlib.closing(response):
+            prefix = response.read(512)
+            if not prefix.startswith(b'#EXTM3U'):
+                msg = 'Response data has no m3u header'
+                if fatal:
+                    raise ExtractorError(msg, video_id=video_id)
+                self.report_warning(f'{msg}{bug_reports_message()}', video_id=video_id)
                 return [], {}
 
-        m3u8_doc, urlh = res
-        m3u8_url = urlh.url
+            content = self._webpage_read_content(
+                response, m3u8_url, video_id, note=note, errnote=errnote,
+                fatal=fatal, prefix=prefix, data=data)
+        if content is False:
+            return [], {}
 
         return self._parse_m3u8_formats_and_subtitles(
-            m3u8_doc, m3u8_url, ext=ext, entry_protocol=entry_protocol,
+            content, response.url, ext=ext, entry_protocol=entry_protocol,
             preference=preference, quality=quality, m3u8_id=m3u8_id,
             note=note, errnote=errnote, fatal=fatal, live=live, data=data,
             headers=headers, query=query, video_id=video_id)