[downloader/http] Clarify rationale for Content-Range check (#6426)

diff --git a/youtube_dl/downloader/http.py b/youtube_dl/downloader/http.py
index 2f8490f023cf681c0c6767ab50462295cdc21645..a29f5cf31fddf3f3b15f47311fafeef9ab4f3d7a 100644 (file)
--- a/youtube_dl/downloader/http.py
+++ b/youtube_dl/downloader/http.py
@@ -58,14 +58,21 @@ class HttpFD(FileDownloader):
             # Establish connection
             try:
                 data = self.ydl.urlopen(request)
             # Establish connection
             try:
                 data = self.ydl.urlopen(request)

+                # When trying to resume, Content-Range HTTP header of response has to be checked
+                # to match the value of requested Range HTTP header. This is due to a webservers
+                # that don't support resuming and serve a whole file with no Content-Range
+                # set in response despite of requested Range (see
+                # https://github.com/rg3/youtube-dl/issues/6057#issuecomment-126129799)

                 if resume_len > 0:
                     content_range = data.headers.get('Content-Range')
                     if content_range:
                         content_range_m = re.search(r'bytes (\d+)-', content_range)
                 if resume_len > 0:
                     content_range = data.headers.get('Content-Range')
                     if content_range:
                         content_range_m = re.search(r'bytes (\d+)-', content_range)

-                        # Content-Range is correct - go on
-                    if content_range_m and resume_len == int(content_range_m.group(1)):
-                        break
-                    # Content-Range is invalid - wipe the file and do entire redownload
+                        # Content-Range is present and matches requested Range, resume is possible
+                        if content_range_m and resume_len == int(content_range_m.group(1)):
+                            break
+                    # Content-Range is either not present or invalid. Assuming remote webserver is
+                    # trying to send the whole file, resume is not possible, so wiping the local file
+                    # and performing entire redownload

                     self.report_unable_to_resume()
                     resume_len = 0
                     open_mode = 'wb'
                     self.report_unable_to_resume()
                     resume_len = 0
                     open_mode = 'wb'