]>
Commit | Line | Data |
---|---|---|
4d2d638d PH |
1 | from __future__ import unicode_literals |
2 | ||
16e7711e | 3 | import errno |
3bc2ddcc | 4 | import os |
16e7711e | 5 | import socket |
3bc2ddcc | 6 | import time |
8d5b8b47 | 7 | import re |
3bc2ddcc JMF |
8 | |
9 | from .common import FileDownloader | |
1cc79574 | 10 | from ..compat import ( |
3bc2ddcc JMF |
11 | compat_urllib_request, |
12 | compat_urllib_error, | |
1cc79574 PH |
13 | ) |
14 | from ..utils import ( | |
3bc2ddcc | 15 | ContentTooShortError, |
3bc2ddcc JMF |
16 | encodeFilename, |
17 | sanitize_open, | |
3bc2ddcc JMF |
18 | ) |
19 | ||
20 | ||
21 | class HttpFD(FileDownloader): | |
22 | def real_download(self, filename, info_dict): | |
23 | url = info_dict['url'] | |
24 | tmpfilename = self.temp_name(filename) | |
25 | stream = None | |
26 | ||
27 | # Do not include the Accept-Encoding header | |
28 | headers = {'Youtubedl-no-compression': 'True'} | |
d769be6c PH |
29 | add_headers = info_dict.get('http_headers') |
30 | if add_headers: | |
31 | headers.update(add_headers) | |
2ddcd881 JMF |
32 | basic_request = compat_urllib_request.Request(url, None, headers) |
33 | request = compat_urllib_request.Request(url, None, headers) | |
3bc2ddcc | 34 | |
df297c87 S |
35 | is_test = self.params.get('test', False) |
36 | ||
37 | if is_test: | |
38 | request.add_header('Range', 'bytes=0-%s' % str(self._TEST_FILE_SIZE - 1)) | |
3bc2ddcc JMF |
39 | |
40 | # Establish possible resume length | |
41 | if os.path.isfile(encodeFilename(tmpfilename)): | |
42 | resume_len = os.path.getsize(encodeFilename(tmpfilename)) | |
43 | else: | |
44 | resume_len = 0 | |
45 | ||
46 | open_mode = 'wb' | |
47 | if resume_len != 0: | |
c59e701e | 48 | if self.params.get('continuedl', True): |
3bc2ddcc | 49 | self.report_resuming_byte(resume_len) |
f89197d7 | 50 | request.add_header('Range', 'bytes=%d-' % resume_len) |
3bc2ddcc JMF |
51 | open_mode = 'ab' |
52 | else: | |
53 | resume_len = 0 | |
54 | ||
55 | count = 0 | |
56 | retries = self.params.get('retries', 0) | |
57 | while count <= retries: | |
58 | # Establish connection | |
59 | try: | |
22ff5d21 | 60 | data = self.ydl.urlopen(request) |
84bc4dcb S |
61 | # When trying to resume, Content-Range HTTP header of response has to be checked |
62 | # to match the value of requested Range HTTP header. This is due to a webservers | |
63 | # that don't support resuming and serve a whole file with no Content-Range | |
64 | # set in response despite of requested Range (see | |
65 | # https://github.com/rg3/youtube-dl/issues/6057#issuecomment-126129799) | |
d7d2a9a3 | 66 | if resume_len > 0: |
67 | content_range = data.headers.get('Content-Range') | |
68 | if content_range: | |
69 | content_range_m = re.search(r'bytes (\d+)-', content_range) | |
84bc4dcb S |
70 | # Content-Range is present and matches requested Range, resume is possible |
71 | if content_range_m and resume_len == int(content_range_m.group(1)): | |
72 | break | |
73 | # Content-Range is either not present or invalid. Assuming remote webserver is | |
74 | # trying to send the whole file, resume is not possible, so wiping the local file | |
75 | # and performing entire redownload | |
10eaa8ef | 76 | self.report_unable_to_resume() |
d7d2a9a3 | 77 | resume_len = 0 |
78 | open_mode = 'wb' | |
3bc2ddcc JMF |
79 | break |
80 | except (compat_urllib_error.HTTPError, ) as err: | |
81 | if (err.code < 500 or err.code >= 600) and err.code != 416: | |
82 | # Unexpected HTTP error | |
83 | raise | |
84 | elif err.code == 416: | |
85 | # Unable to resume (requested range not satisfiable) | |
86 | try: | |
87 | # Open the connection again without the range header | |
22ff5d21 | 88 | data = self.ydl.urlopen(basic_request) |
3bc2ddcc JMF |
89 | content_length = data.info()['Content-Length'] |
90 | except (compat_urllib_error.HTTPError, ) as err: | |
91 | if err.code < 500 or err.code >= 600: | |
92 | raise | |
93 | else: | |
94 | # Examine the reported length | |
95 | if (content_length is not None and | |
96 | (resume_len - 100 < int(content_length) < resume_len + 100)): | |
97 | # The file had already been fully downloaded. | |
98 | # Explanation to the above condition: in issue #175 it was revealed that | |
99 | # YouTube sometimes adds or removes a few bytes from the end of the file, | |
100 | # changing the file size slightly and causing problems for some users. So | |
101 | # I decided to implement a suggested change and consider the file | |
102 | # completely downloaded if the file size differs less than 100 bytes from | |
103 | # the one in the hard drive. | |
104 | self.report_file_already_downloaded(filename) | |
105 | self.try_rename(tmpfilename, filename) | |
106 | self._hook_progress({ | |
107 | 'filename': filename, | |
108 | 'status': 'finished', | |
f28fe669 JMF |
109 | 'downloaded_bytes': resume_len, |
110 | 'total_bytes': resume_len, | |
3bc2ddcc JMF |
111 | }) |
112 | return True | |
113 | else: | |
114 | # The length does not match, we start the download over | |
115 | self.report_unable_to_resume() | |
60daf7f0 | 116 | resume_len = 0 |
3bc2ddcc JMF |
117 | open_mode = 'wb' |
118 | break | |
16e7711e | 119 | except socket.error as e: |
e77d2975 | 120 | if e.errno != errno.ECONNRESET: |
12832049 PH |
121 | # Connection reset is no problem, just retry |
122 | raise | |
123 | ||
3bc2ddcc JMF |
124 | # Retry |
125 | count += 1 | |
126 | if count <= retries: | |
127 | self.report_retry(count, retries) | |
128 | ||
129 | if count > retries: | |
4d2d638d | 130 | self.report_error('giving up after %s retries' % retries) |
3bc2ddcc JMF |
131 | return False |
132 | ||
133 | data_len = data.info().get('Content-length', None) | |
df297c87 S |
134 | |
135 | # Range HTTP header may be ignored/unsupported by a webserver | |
136 | # (e.g. extractor/scivee.py, extractor/bambuser.py). | |
137 | # However, for a test we still would like to download just a piece of a file. | |
138 | # To achieve this we limit data_len to _TEST_FILE_SIZE and manually control | |
139 | # block size when downloading a file. | |
32fd27ec | 140 | if is_test and (data_len is None or int(data_len) > self._TEST_FILE_SIZE): |
df297c87 S |
141 | data_len = self._TEST_FILE_SIZE |
142 | ||
3bc2ddcc JMF |
143 | if data_len is not None: |
144 | data_len = int(data_len) + resume_len | |
145 | min_data_len = self.params.get("min_filesize", None) | |
f89197d7 | 146 | max_data_len = self.params.get("max_filesize", None) |
3bc2ddcc | 147 | if min_data_len is not None and data_len < min_data_len: |
4d2d638d | 148 | self.to_screen('\r[download] File is smaller than min-filesize (%s bytes < %s bytes). Aborting.' % (data_len, min_data_len)) |
3bc2ddcc JMF |
149 | return False |
150 | if max_data_len is not None and data_len > max_data_len: | |
4d2d638d | 151 | self.to_screen('\r[download] File is larger than max-filesize (%s bytes > %s bytes). Aborting.' % (data_len, max_data_len)) |
3bc2ddcc JMF |
152 | return False |
153 | ||
3bc2ddcc JMF |
154 | byte_counter = 0 + resume_len |
155 | block_size = self.params.get('buffersize', 1024) | |
156 | start = time.time() | |
c7667c2d S |
157 | |
158 | # measure time over whole while-loop, so slow_down() and best_block_size() work together properly | |
159 | now = None # needed for slow_down() in the first loop run | |
160 | before = start # start measuring | |
3bc2ddcc | 161 | while True: |
c7667c2d | 162 | |
3bc2ddcc | 163 | # Download and write |
df297c87 | 164 | data_block = data.read(block_size if not is_test else min(block_size, data_len - byte_counter)) |
c7667c2d S |
165 | byte_counter += len(data_block) |
166 | ||
167 | # exit loop when download is finished | |
3bc2ddcc JMF |
168 | if len(data_block) == 0: |
169 | break | |
3bc2ddcc | 170 | |
c7667c2d | 171 | # Open destination file just in time |
3bc2ddcc JMF |
172 | if stream is None: |
173 | try: | |
174 | (stream, tmpfilename) = sanitize_open(tmpfilename, open_mode) | |
175 | assert stream is not None | |
176 | filename = self.undo_temp_name(tmpfilename) | |
177 | self.report_destination(filename) | |
178 | except (OSError, IOError) as err: | |
4d2d638d | 179 | self.report_error('unable to open for writing: %s' % str(err)) |
3bc2ddcc | 180 | return False |
881e6a1f PH |
181 | |
182 | if self.params.get('xattr_set_filesize', False) and data_len is not None: | |
183 | try: | |
184 | import xattr | |
185 | xattr.setxattr(tmpfilename, 'user.ytdl.filesize', str(data_len)) | |
186 | except(OSError, IOError, ImportError) as err: | |
187 | self.report_error('unable to set filesize xattr: %s' % str(err)) | |
188 | ||
3bc2ddcc JMF |
189 | try: |
190 | stream.write(data_block) | |
7976fcac | 191 | except (IOError, OSError) as err: |
4d2d638d PH |
192 | self.to_stderr('\n') |
193 | self.report_error('unable to write data: %s' % str(err)) | |
3bc2ddcc | 194 | return False |
c7667c2d S |
195 | |
196 | # Apply rate limit | |
197 | self.slow_down(start, now, byte_counter - resume_len) | |
198 | ||
199 | # end measuring of one loop run | |
200 | now = time.time() | |
201 | after = now | |
202 | ||
203 | # Adjust block size | |
3bc2ddcc JMF |
204 | if not self.params.get('noresizebuffer', False): |
205 | block_size = self.best_block_size(after - before, len(data_block)) | |
206 | ||
c7667c2d S |
207 | before = after |
208 | ||
3bc2ddcc | 209 | # Progress message |
c7667c2d | 210 | speed = self.calc_speed(start, now, byte_counter - resume_len) |
3bc2ddcc | 211 | if data_len is None: |
a91a2c1a | 212 | eta = None |
3bc2ddcc | 213 | else: |
3bc2ddcc | 214 | eta = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len) |
3bc2ddcc JMF |
215 | |
216 | self._hook_progress({ | |
5cda4eda | 217 | 'status': 'downloading', |
3bc2ddcc JMF |
218 | 'downloaded_bytes': byte_counter, |
219 | 'total_bytes': data_len, | |
220 | 'tmpfilename': tmpfilename, | |
221 | 'filename': filename, | |
3bc2ddcc JMF |
222 | 'eta': eta, |
223 | 'speed': speed, | |
5cda4eda | 224 | 'elapsed': now - start, |
3bc2ddcc JMF |
225 | }) |
226 | ||
df297c87 S |
227 | if is_test and byte_counter == data_len: |
228 | break | |
229 | ||
3bc2ddcc | 230 | if stream is None: |
4d2d638d PH |
231 | self.to_stderr('\n') |
232 | self.report_error('Did not get any data blocks') | |
3bc2ddcc | 233 | return False |
4d2d638d | 234 | if tmpfilename != '-': |
88a23aef | 235 | stream.close() |
5cda4eda | 236 | |
3bc2ddcc JMF |
237 | if data_len is not None and byte_counter != data_len: |
238 | raise ContentTooShortError(byte_counter, int(data_len)) | |
239 | self.try_rename(tmpfilename, filename) | |
240 | ||
241 | # Update file modification time | |
242 | if self.params.get('updatetime', True): | |
243 | info_dict['filetime'] = self.try_utime(filename, data.info().get('last-modified', None)) | |
244 | ||
245 | self._hook_progress({ | |
246 | 'downloaded_bytes': byte_counter, | |
247 | 'total_bytes': byte_counter, | |
248 | 'filename': filename, | |
249 | 'status': 'finished', | |
5cda4eda | 250 | 'elapsed': time.time() - start, |
3bc2ddcc JMF |
251 | }) |
252 | ||
253 | return True |