[utils] Don't use Content-length with encoding (#6176)

Authored by: felixonmars
Closes #3772, #6178

yt_dlp/downloader/http.py

@@ -211,7 +211,12 @@ class HttpFD(FileDownloader):
                 ctx.stream = None
 
         def download():
-            data_len = ctx.data.info().get('Content-length', None)
+            data_len = ctx.data.info().get('Content-length')
+
+            if ctx.data.info().get('Content-encoding'):
+                # Content-encoding is present, Content-length is not reliable anymore as we are
+                # doing auto decompression. (See: https://github.com/yt-dlp/yt-dlp/pull/6176)
+                data_len = None
 
             # Range HTTP header may be ignored/unsupported by a webserver
             # (e.g. extractor/scivee.py, extractor/bambuser.py).

yt_dlp/utils.py

@@ -1438,19 +1438,16 @@ class YoutubeDLHandler(urllib.request.HTTPHandler):
                     raise original_ioerror
             resp = urllib.request.addinfourl(uncompressed, old_resp.headers, old_resp.url, old_resp.code)
             resp.msg = old_resp.msg
-            del resp.headers['Content-encoding']
         # deflate
         if resp.headers.get('Content-encoding', '') == 'deflate':
             gz = io.BytesIO(self.deflate(resp.read()))
             resp = urllib.request.addinfourl(gz, old_resp.headers, old_resp.url, old_resp.code)
             resp.msg = old_resp.msg
-            del resp.headers['Content-encoding']
         # brotli
         if resp.headers.get('Content-encoding', '') == 'br':
             resp = urllib.request.addinfourl(
                 io.BytesIO(self.brotli(resp.read())), old_resp.headers, old_resp.url, old_resp.code)
             resp.msg = old_resp.msg
-            del resp.headers['Content-encoding']
         # Percent-encode redirect URL of Location HTTP header to satisfy RFC 3986 (see
         # https://github.com/ytdl-org/youtube-dl/issues/6457).
         if 300 <= resp.code < 400: