Update Standards-Version to 3.9.8 (no changes required).

[youtubedl] / youtube_dl / downloader / http.py

diff --git a/youtube_dl/downloader/http.py b/youtube_dl/downloader/http.py
index 49170cf9d47634602efe7832b235e4a751e25817..f8b69d186ac5ee93c8402f85bc66e7ed59570118 100644 (file)
--- a/youtube_dl/downloader/http.py
+++ b/youtube_dl/downloader/http.py
@@ -1,21 +1,18 @@
 from __future__ import unicode_literals
 
 from __future__ import unicode_literals
 

+import errno

 import os
 import os

+import socket

 import time
 import time

-
-from socket import error as SocketError
-import errno
+import re

 
 from .common import FileDownloader
 
 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,
 from ..utils import (
     ContentTooShortError,
     encodeFilename,
     sanitize_open,

-    format_bytes,
+    sanitized_Request,

 )
 
 
 )
 
 


@@ -30,13 +27,8 @@ class HttpFD(FileDownloader):
         add_headers = info_dict.get('http_headers')
         if add_headers:
             headers.update(add_headers)
         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)
 
 
         is_test = self.params.get('test', False)
 


@@ -51,7 +43,7 @@ class HttpFD(FileDownloader):
 
         open_mode = 'wb'
         if resume_len != 0:
 
         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'
                 self.report_resuming_byte(resume_len)
                 request.add_header('Range', 'bytes=%d-' % resume_len)
                 open_mode = 'ab'


@@ -64,6 +56,24 @@ 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)
+                        # 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:
                 break
             except (compat_urllib_error.HTTPError, ) as err:
                 if (err.code < 500 or err.code >= 600) and err.code != 416:


@@ -94,6 +104,8 @@ class HttpFD(FileDownloader):
                             self._hook_progress({
                                 'filename': filename,
                                 'status': 'finished',
                             self._hook_progress({
                                 'filename': filename,
                                 'status': 'finished',

+                                'downloaded_bytes': resume_len,
+                                'total_bytes': resume_len,

                             })
                             return True
                         else:
                             })
                             return True
                         else:


@@ -102,7 +114,7 @@ class HttpFD(FileDownloader):
                             resume_len = 0
                             open_mode = 'wb'
                             break
                             resume_len = 0
                             open_mode = 'wb'
                             break

-            except SocketError as e:
+            except socket.error as e:

                 if e.errno != errno.ECONNRESET:
                     # Connection reset is no problem, just retry
                     raise
                 if e.errno != errno.ECONNRESET:
                     # Connection reset is no problem, just retry
                     raise


@@ -128,8 +140,8 @@ class HttpFD(FileDownloader):
 
         if data_len is not None:
             data_len = int(data_len) + resume_len
 
         if data_len is not None:
             data_len = int(data_len) + resume_len

-            min_data_len = self.params.get("min_filesize", None)
-            max_data_len = self.params.get("max_filesize", None)
+            min_data_len = self.params.get('min_filesize')
+            max_data_len = self.params.get('max_filesize')

             if min_data_len is not None and data_len < min_data_len:
                 self.to_screen('\r[download] File is smaller than min-filesize (%s bytes < %s bytes). Aborting.' % (data_len, min_data_len))
                 return False
             if min_data_len is not None and data_len < min_data_len:
                 self.to_screen('\r[download] File is smaller than min-filesize (%s bytes < %s bytes). Aborting.' % (data_len, min_data_len))
                 return False


@@ -137,7 +149,6 @@ class HttpFD(FileDownloader):
                 self.to_screen('\r[download] File is larger than max-filesize (%s bytes > %s bytes). Aborting.' % (data_len, max_data_len))
                 return False
 
                 self.to_screen('\r[download] File is larger than max-filesize (%s bytes > %s bytes). Aborting.' % (data_len, max_data_len))
                 return False
 

-        data_len_str = format_bytes(data_len)

         byte_counter = 0 + resume_len
         block_size = self.params.get('buffersize', 1024)
         start = time.time()
         byte_counter = 0 + resume_len
         block_size = self.params.get('buffersize', 1024)
         start = time.time()


@@ -196,20 +207,19 @@ class HttpFD(FileDownloader):
             # Progress message
             speed = self.calc_speed(start, now, byte_counter - resume_len)
             if data_len is None:
             # Progress message
             speed = self.calc_speed(start, now, byte_counter - resume_len)
             if data_len is None:

-                eta = percent = None
+                eta = None

             else:
             else:

-                percent = self.calc_percent(byte_counter, data_len)

                 eta = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len)
                 eta = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len)

-            self.report_progress(percent, data_len_str, speed, eta)

 
             self._hook_progress({
 
             self._hook_progress({

+                'status': 'downloading',

                 'downloaded_bytes': byte_counter,
                 'total_bytes': data_len,
                 'tmpfilename': tmpfilename,
                 'filename': filename,
                 'downloaded_bytes': byte_counter,
                 'total_bytes': data_len,
                 'tmpfilename': tmpfilename,
                 'filename': filename,

-                'status': 'downloading',

                 'eta': eta,
                 'speed': speed,
                 'eta': eta,
                 'speed': speed,

+                'elapsed': now - start,

             })
 
             if is_test and byte_counter == data_len:
             })
 
             if is_test and byte_counter == data_len:


@@ -221,7 +231,7 @@ class HttpFD(FileDownloader):
             return False
         if tmpfilename != '-':
             stream.close()
             return False
         if tmpfilename != '-':
             stream.close()

-        self.report_finish(data_len_str, (time.time() - start))
+

         if data_len is not None and byte_counter != data_len:
             raise ContentTooShortError(byte_counter, int(data_len))
         self.try_rename(tmpfilename, filename)
         if data_len is not None and byte_counter != data_len:
             raise ContentTooShortError(byte_counter, int(data_len))
         self.try_rename(tmpfilename, filename)


@@ -235,6 +245,7 @@ class HttpFD(FileDownloader):
             'total_bytes': byte_counter,
             'filename': filename,
             'status': 'finished',
             'total_bytes': byte_counter,
             'filename': filename,
             'status': 'finished',

+            'elapsed': time.time() - start,

         })
 
         return True
         })
 
         return True