X-Git-Url: http://git.bitcoin.ninja/index.cgi?p=youtube-dl;a=blobdiff_plain;f=youtube_dl%2Fdownloader%2Fhttp.py;h=56840e026cc710a7c709da13f637e62842ccfd84;hp=2e3dac8251dbaf5d8b3e1a90bc459f362d14f72e;hb=06e4874c997fb523bd9d8e675d1ec0e69ab501ed;hpb=09ab40b7d1d1838aaf0ad2086e691a0d68e165f1 diff --git a/youtube_dl/downloader/http.py b/youtube_dl/downloader/http.py index 2e3dac825..56840e026 100644 --- a/youtube_dl/downloader/http.py +++ b/youtube_dl/downloader/http.py @@ -4,16 +4,15 @@ import errno import os import socket import time +import re from .common import FileDownloader -from ..compat import ( - compat_urllib_request, - compat_urllib_error, -) +from ..compat import compat_urllib_error from ..utils import ( ContentTooShortError, encodeFilename, sanitize_open, + sanitized_Request, ) @@ -28,13 +27,8 @@ class HttpFD(FileDownloader): add_headers = info_dict.get('http_headers') if add_headers: headers.update(add_headers) - data = info_dict.get('http_post_data') - http_method = info_dict.get('http_method') - basic_request = compat_urllib_request.Request(url, data, headers) - request = compat_urllib_request.Request(url, data, headers) - if http_method is not None: - basic_request.get_method = lambda: http_method - request.get_method = lambda: http_method + basic_request = sanitized_Request(url, None, headers) + request = sanitized_Request(url, None, headers) is_test = self.params.get('test', False) @@ -49,7 +43,7 @@ class HttpFD(FileDownloader): open_mode = 'wb' if resume_len != 0: - if self.params.get('continuedl', False): + if self.params.get('continuedl', True): self.report_resuming_byte(resume_len) request.add_header('Range', 'bytes=%d-' % resume_len) open_mode = 'ab' @@ -62,6 +56,24 @@ class HttpFD(FileDownloader): # 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) + # 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' break except (compat_urllib_error.HTTPError, ) as err: if (err.code < 500 or err.code >= 600) and err.code != 416: @@ -92,6 +104,8 @@ class HttpFD(FileDownloader): self._hook_progress({ 'filename': filename, 'status': 'finished', + 'downloaded_bytes': resume_len, + 'total_bytes': resume_len, }) return True else: @@ -218,12 +232,6 @@ class HttpFD(FileDownloader): if tmpfilename != '-': stream.close() - self._hook_progress({ - 'downloaded_bytes': byte_counter, - 'total_bytes': data_len, - 'tmpfilename': tmpfilename, - 'status': 'error', - }) if data_len is not None and byte_counter != data_len: raise ContentTooShortError(byte_counter, int(data_len)) self.try_rename(tmpfilename, filename)