]>
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 |
b91a7a4e | 7 | import random |
8d5b8b47 | 8 | import re |
3bc2ddcc JMF |
9 | |
10 | from .common import FileDownloader | |
ba515388 S |
11 | from ..compat import ( |
12 | compat_str, | |
13 | compat_urllib_error, | |
14 | ) | |
1cc79574 | 15 | from ..utils import ( |
3bc2ddcc | 16 | ContentTooShortError, |
3bc2ddcc | 17 | encodeFilename, |
ba515388 | 18 | int_or_none, |
3bc2ddcc | 19 | sanitize_open, |
5c2266df | 20 | sanitized_Request, |
d7753d19 YCH |
21 | write_xattr, |
22 | XAttrMetadataError, | |
23 | XAttrUnavailableError, | |
3bc2ddcc JMF |
24 | ) |
25 | ||
26 | ||
27 | class HttpFD(FileDownloader): | |
d3f0687c | 28 | def real_download(self, filename, info_dict): |
3bc2ddcc | 29 | url = info_dict['url'] |
a3c3a1e1 S |
30 | |
31 | class DownloadContext(dict): | |
32 | __getattr__ = dict.get | |
33 | __setattr__ = dict.__setitem__ | |
34 | __delattr__ = dict.__delitem__ | |
35 | ||
36 | ctx = DownloadContext() | |
37 | ctx.filename = filename | |
38 | ctx.tmpfilename = self.temp_name(filename) | |
39 | ctx.stream = None | |
3bc2ddcc JMF |
40 | |
41 | # Do not include the Accept-Encoding header | |
42 | headers = {'Youtubedl-no-compression': 'True'} | |
d769be6c PH |
43 | add_headers = info_dict.get('http_headers') |
44 | if add_headers: | |
45 | headers.update(add_headers) | |
3bc2ddcc | 46 | |
df297c87 | 47 | is_test = self.params.get('test', False) |
ba515388 | 48 | chunk_size = self._TEST_FILE_SIZE if is_test else ( |
3089bc74 S |
49 | info_dict.get('downloader_options', {}).get('http_chunk_size') |
50 | or self.params.get('http_chunk_size') or 0) | |
3bc2ddcc | 51 | |
a3c3a1e1 S |
52 | ctx.open_mode = 'wb' |
53 | ctx.resume_len = 0 | |
ba515388 S |
54 | ctx.data_len = None |
55 | ctx.block_size = self.params.get('buffersize', 1024) | |
56 | ctx.start_time = time.time() | |
b91a7a4e | 57 | ctx.chunk_size = None |
a3c3a1e1 S |
58 | |
59 | if self.params.get('continuedl', True): | |
60 | # Establish possible resume length | |
61 | if os.path.isfile(encodeFilename(ctx.tmpfilename)): | |
ba515388 S |
62 | ctx.resume_len = os.path.getsize( |
63 | encodeFilename(ctx.tmpfilename)) | |
64 | ||
65 | ctx.is_resume = ctx.resume_len > 0 | |
3bc2ddcc JMF |
66 | |
67 | count = 0 | |
68 | retries = self.params.get('retries', 0) | |
a3c3a1e1 S |
69 | |
70 | class SucceedDownload(Exception): | |
71 | pass | |
72 | ||
73 | class RetryDownload(Exception): | |
74 | def __init__(self, source_error): | |
75 | self.source_error = source_error | |
76 | ||
ba515388 S |
77 | class NextFragment(Exception): |
78 | pass | |
79 | ||
80 | def set_range(req, start, end): | |
81 | range_header = 'bytes=%d-' % start | |
82 | if end: | |
83 | range_header += compat_str(end) | |
84 | req.add_header('Range', range_header) | |
85 | ||
a3c3a1e1 | 86 | def establish_connection(): |
b91a7a4e S |
87 | ctx.chunk_size = (random.randint(int(chunk_size * 0.95), chunk_size) |
88 | if not is_test and chunk_size else chunk_size) | |
ba515388 S |
89 | if ctx.resume_len > 0: |
90 | range_start = ctx.resume_len | |
91 | if ctx.is_resume: | |
92 | self.report_resuming_byte(ctx.resume_len) | |
a3c3a1e1 | 93 | ctx.open_mode = 'ab' |
b91a7a4e | 94 | elif ctx.chunk_size > 0: |
ba515388 S |
95 | range_start = 0 |
96 | else: | |
97 | range_start = None | |
98 | ctx.is_resume = False | |
b91a7a4e | 99 | range_end = range_start + ctx.chunk_size - 1 if ctx.chunk_size else None |
ba515388 S |
100 | if range_end and ctx.data_len is not None and range_end >= ctx.data_len: |
101 | range_end = ctx.data_len - 1 | |
102 | has_range = range_start is not None | |
103 | ctx.has_range = has_range | |
cf7259bc | 104 | request = sanitized_Request(url, None, headers) |
ba515388 S |
105 | if has_range: |
106 | set_range(request, range_start, range_end) | |
3bc2ddcc JMF |
107 | # Establish connection |
108 | try: | |
a3c3a1e1 | 109 | ctx.data = self.ydl.urlopen(request) |
84bc4dcb S |
110 | # When trying to resume, Content-Range HTTP header of response has to be checked |
111 | # to match the value of requested Range HTTP header. This is due to a webservers | |
112 | # that don't support resuming and serve a whole file with no Content-Range | |
113 | # set in response despite of requested Range (see | |
067aa17e | 114 | # https://github.com/ytdl-org/youtube-dl/issues/6057#issuecomment-126129799) |
ba515388 | 115 | if has_range: |
a3c3a1e1 | 116 | content_range = ctx.data.headers.get('Content-Range') |
d7d2a9a3 | 117 | if content_range: |
ba515388 | 118 | content_range_m = re.search(r'bytes (\d+)-(\d+)?(?:/(\d+))?', content_range) |
84bc4dcb | 119 | # Content-Range is present and matches requested Range, resume is possible |
ba515388 S |
120 | if content_range_m: |
121 | if range_start == int(content_range_m.group(1)): | |
122 | content_range_end = int_or_none(content_range_m.group(2)) | |
123 | content_len = int_or_none(content_range_m.group(3)) | |
124 | accept_content_len = ( | |
125 | # Non-chunked download | |
3089bc74 | 126 | not ctx.chunk_size |
ba515388 S |
127 | # Chunked download and requested piece or |
128 | # its part is promised to be served | |
3089bc74 S |
129 | or content_range_end == range_end |
130 | or content_len < range_end) | |
ba515388 S |
131 | if accept_content_len: |
132 | ctx.data_len = content_len | |
133 | return | |
84bc4dcb S |
134 | # Content-Range is either not present or invalid. Assuming remote webserver is |
135 | # trying to send the whole file, resume is not possible, so wiping the local file | |
136 | # and performing entire redownload | |
10eaa8ef | 137 | self.report_unable_to_resume() |
a3c3a1e1 S |
138 | ctx.resume_len = 0 |
139 | ctx.open_mode = 'wb' | |
ba515388 | 140 | ctx.data_len = int_or_none(ctx.data.info().get('Content-length', None)) |
a3c3a1e1 | 141 | return |
3bc2ddcc | 142 | except (compat_urllib_error.HTTPError, ) as err: |
ba515388 | 143 | if err.code == 416: |
3bc2ddcc JMF |
144 | # Unable to resume (requested range not satisfiable) |
145 | try: | |
146 | # Open the connection again without the range header | |
cf7259bc S |
147 | ctx.data = self.ydl.urlopen( |
148 | sanitized_Request(url, None, headers)) | |
a3c3a1e1 | 149 | content_length = ctx.data.info()['Content-Length'] |
3bc2ddcc JMF |
150 | except (compat_urllib_error.HTTPError, ) as err: |
151 | if err.code < 500 or err.code >= 600: | |
152 | raise | |
153 | else: | |
154 | # Examine the reported length | |
3089bc74 S |
155 | if (content_length is not None |
156 | and (ctx.resume_len - 100 < int(content_length) < ctx.resume_len + 100)): | |
3bc2ddcc JMF |
157 | # The file had already been fully downloaded. |
158 | # Explanation to the above condition: in issue #175 it was revealed that | |
159 | # YouTube sometimes adds or removes a few bytes from the end of the file, | |
160 | # changing the file size slightly and causing problems for some users. So | |
161 | # I decided to implement a suggested change and consider the file | |
162 | # completely downloaded if the file size differs less than 100 bytes from | |
163 | # the one in the hard drive. | |
a3c3a1e1 S |
164 | self.report_file_already_downloaded(ctx.filename) |
165 | self.try_rename(ctx.tmpfilename, ctx.filename) | |
3bc2ddcc | 166 | self._hook_progress({ |
a3c3a1e1 | 167 | 'filename': ctx.filename, |
3bc2ddcc | 168 | 'status': 'finished', |
a3c3a1e1 S |
169 | 'downloaded_bytes': ctx.resume_len, |
170 | 'total_bytes': ctx.resume_len, | |
3bc2ddcc | 171 | }) |
a3c3a1e1 | 172 | raise SucceedDownload() |
3bc2ddcc JMF |
173 | else: |
174 | # The length does not match, we start the download over | |
175 | self.report_unable_to_resume() | |
a3c3a1e1 S |
176 | ctx.resume_len = 0 |
177 | ctx.open_mode = 'wb' | |
178 | return | |
ba515388 S |
179 | elif err.code < 500 or err.code >= 600: |
180 | # Unexpected HTTP error | |
181 | raise | |
a3c3a1e1 S |
182 | raise RetryDownload(err) |
183 | except socket.error as err: | |
184 | if err.errno != errno.ECONNRESET: | |
12832049 PH |
185 | # Connection reset is no problem, just retry |
186 | raise | |
a3c3a1e1 S |
187 | raise RetryDownload(err) |
188 | ||
189 | def download(): | |
190 | data_len = ctx.data.info().get('Content-length', None) | |
191 | ||
192 | # Range HTTP header may be ignored/unsupported by a webserver | |
193 | # (e.g. extractor/scivee.py, extractor/bambuser.py). | |
194 | # However, for a test we still would like to download just a piece of a file. | |
195 | # To achieve this we limit data_len to _TEST_FILE_SIZE and manually control | |
196 | # block size when downloading a file. | |
197 | if is_test and (data_len is None or int(data_len) > self._TEST_FILE_SIZE): | |
198 | data_len = self._TEST_FILE_SIZE | |
199 | ||
200 | if data_len is not None: | |
201 | data_len = int(data_len) + ctx.resume_len | |
202 | min_data_len = self.params.get('min_filesize') | |
203 | max_data_len = self.params.get('max_filesize') | |
204 | if min_data_len is not None and data_len < min_data_len: | |
205 | self.to_screen('\r[download] File is smaller than min-filesize (%s bytes < %s bytes). Aborting.' % (data_len, min_data_len)) | |
206 | return False | |
207 | if max_data_len is not None and data_len > max_data_len: | |
208 | self.to_screen('\r[download] File is larger than max-filesize (%s bytes > %s bytes). Aborting.' % (data_len, max_data_len)) | |
209 | return False | |
12832049 | 210 | |
a3c3a1e1 | 211 | byte_counter = 0 + ctx.resume_len |
ba515388 | 212 | block_size = ctx.block_size |
a3c3a1e1 | 213 | start = time.time() |
c7667c2d | 214 | |
a3c3a1e1 S |
215 | # measure time over whole while-loop, so slow_down() and best_block_size() work together properly |
216 | now = None # needed for slow_down() in the first loop run | |
217 | before = start # start measuring | |
c7667c2d | 218 | |
a3c3a1e1 | 219 | def retry(e): |
5d6c81b6 S |
220 | to_stdout = ctx.tmpfilename == '-' |
221 | if not to_stdout: | |
a3c3a1e1 S |
222 | ctx.stream.close() |
223 | ctx.stream = None | |
5d6c81b6 | 224 | ctx.resume_len = byte_counter if to_stdout else os.path.getsize(encodeFilename(ctx.tmpfilename)) |
a3c3a1e1 | 225 | raise RetryDownload(e) |
c7667c2d | 226 | |
a3c3a1e1 S |
227 | while True: |
228 | try: | |
229 | # Download and write | |
230 | data_block = ctx.data.read(block_size if not is_test else min(block_size, data_len - byte_counter)) | |
231 | # socket.timeout is a subclass of socket.error but may not have | |
232 | # errno set | |
233 | except socket.timeout as e: | |
234 | retry(e) | |
235 | except socket.error as e: | |
236 | if e.errno not in (errno.ECONNRESET, errno.ETIMEDOUT): | |
237 | raise | |
238 | retry(e) | |
239 | ||
240 | byte_counter += len(data_block) | |
241 | ||
242 | # exit loop when download is finished | |
243 | if len(data_block) == 0: | |
244 | break | |
245 | ||
246 | # Open destination file just in time | |
247 | if ctx.stream is None: | |
248 | try: | |
249 | ctx.stream, ctx.tmpfilename = sanitize_open( | |
250 | ctx.tmpfilename, ctx.open_mode) | |
251 | assert ctx.stream is not None | |
252 | ctx.filename = self.undo_temp_name(ctx.tmpfilename) | |
253 | self.report_destination(ctx.filename) | |
254 | except (OSError, IOError) as err: | |
255 | self.report_error('unable to open for writing: %s' % str(err)) | |
256 | return False | |
257 | ||
258 | if self.params.get('xattr_set_filesize', False) and data_len is not None: | |
259 | try: | |
260 | write_xattr(ctx.tmpfilename, 'user.ytdl.filesize', str(data_len).encode('utf-8')) | |
261 | except (XAttrUnavailableError, XAttrMetadataError) as err: | |
262 | self.report_error('unable to set filesize xattr: %s' % str(err)) | |
3bc2ddcc | 263 | |
3bc2ddcc | 264 | try: |
a3c3a1e1 S |
265 | ctx.stream.write(data_block) |
266 | except (IOError, OSError) as err: | |
267 | self.to_stderr('\n') | |
268 | self.report_error('unable to write data: %s' % str(err)) | |
3bc2ddcc | 269 | return False |
881e6a1f | 270 | |
a3c3a1e1 S |
271 | # Apply rate limit |
272 | self.slow_down(start, now, byte_counter - ctx.resume_len) | |
273 | ||
274 | # end measuring of one loop run | |
275 | now = time.time() | |
276 | after = now | |
277 | ||
278 | # Adjust block size | |
279 | if not self.params.get('noresizebuffer', False): | |
280 | block_size = self.best_block_size(after - before, len(data_block)) | |
281 | ||
282 | before = after | |
283 | ||
284 | # Progress message | |
285 | speed = self.calc_speed(start, now, byte_counter - ctx.resume_len) | |
ba515388 | 286 | if ctx.data_len is None: |
a3c3a1e1 S |
287 | eta = None |
288 | else: | |
ba515388 | 289 | eta = self.calc_eta(start, time.time(), ctx.data_len - ctx.resume_len, byte_counter - ctx.resume_len) |
a3c3a1e1 S |
290 | |
291 | self._hook_progress({ | |
292 | 'status': 'downloading', | |
293 | 'downloaded_bytes': byte_counter, | |
ba515388 | 294 | 'total_bytes': ctx.data_len, |
a3c3a1e1 S |
295 | 'tmpfilename': ctx.tmpfilename, |
296 | 'filename': ctx.filename, | |
297 | 'eta': eta, | |
298 | 'speed': speed, | |
ba515388 | 299 | 'elapsed': now - ctx.start_time, |
a3c3a1e1 S |
300 | }) |
301 | ||
302 | if is_test and byte_counter == data_len: | |
303 | break | |
304 | ||
b91a7a4e | 305 | if not is_test and ctx.chunk_size and ctx.data_len is not None and byte_counter < ctx.data_len: |
ba515388 S |
306 | ctx.resume_len = byte_counter |
307 | # ctx.block_size = block_size | |
308 | raise NextFragment() | |
309 | ||
a3c3a1e1 | 310 | if ctx.stream is None: |
4d2d638d | 311 | self.to_stderr('\n') |
a3c3a1e1 | 312 | self.report_error('Did not get any data blocks') |
3bc2ddcc | 313 | return False |
a3c3a1e1 S |
314 | if ctx.tmpfilename != '-': |
315 | ctx.stream.close() | |
c7667c2d | 316 | |
a3c3a1e1 S |
317 | if data_len is not None and byte_counter != data_len: |
318 | err = ContentTooShortError(byte_counter, int(data_len)) | |
319 | if count <= retries: | |
320 | retry(err) | |
321 | raise err | |
c7667c2d | 322 | |
a3c3a1e1 | 323 | self.try_rename(ctx.tmpfilename, ctx.filename) |
c7667c2d | 324 | |
a3c3a1e1 S |
325 | # Update file modification time |
326 | if self.params.get('updatetime', True): | |
327 | info_dict['filetime'] = self.try_utime(ctx.filename, ctx.data.info().get('last-modified', None)) | |
3bc2ddcc JMF |
328 | |
329 | self._hook_progress({ | |
330 | 'downloaded_bytes': byte_counter, | |
a3c3a1e1 S |
331 | 'total_bytes': byte_counter, |
332 | 'filename': ctx.filename, | |
333 | 'status': 'finished', | |
ba515388 | 334 | 'elapsed': time.time() - ctx.start_time, |
3bc2ddcc JMF |
335 | }) |
336 | ||
a3c3a1e1 S |
337 | return True |
338 | ||
339 | while count <= retries: | |
340 | try: | |
341 | establish_connection() | |
bec49996 | 342 | return download() |
a3c3a1e1 S |
343 | except RetryDownload as e: |
344 | count += 1 | |
345 | if count <= retries: | |
346 | self.report_retry(e.source_error, count, retries) | |
347 | continue | |
ba515388 S |
348 | except NextFragment: |
349 | continue | |
a3c3a1e1 S |
350 | except SucceedDownload: |
351 | return True | |
352 | ||
353 | self.report_error('giving up after %s retries' % retries) | |
354 | return False |