_ Git - youtube-dl/blob - test/test_download.py

   1 #!/usr/bin/env python
   2
   3 # Allow direct execution
   4 import os
   5 import sys
   6 import unittest
   7 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
   8
   9 from test.helper import (
  10     assertGreaterEqual,
  11     get_params,
  12     gettestcases,
  13     expect_info_dict,
  14     try_rm,
  15     report_warning,
  16 )
  17
  18
  19 import hashlib
  20 import io
  21 import json
  22 import socket
  23
  24 import youtube_dl.YoutubeDL
  25 from youtube_dl.utils import (
  26     compat_http_client,
  27     compat_urllib_error,
  28     compat_HTTPError,
  29     DownloadError,
  30     ExtractorError,
  31     UnavailableVideoError,
  32 )
  33 from youtube_dl.extractor import get_info_extractor
  34
  35 RETRIES = 3
  36
  37 class YoutubeDL(youtube_dl.YoutubeDL):
  38     def __init__(self, *args, **kwargs):
  39         self.to_stderr = self.to_screen
  40         self.processed_info_dicts = []
  41         super(YoutubeDL, self).__init__(*args, **kwargs)
  42     def report_warning(self, message):
  43         # Don't accept warnings during tests
  44         raise ExtractorError(message)
  45     def process_info(self, info_dict):
  46         self.processed_info_dicts.append(info_dict)
  47         return super(YoutubeDL, self).process_info(info_dict)
  48
  49 def _file_md5(fn):
  50     with open(fn, 'rb') as f:
  51         return hashlib.md5(f.read()).hexdigest()
  52
  53 defs = gettestcases()
  54
  55
  56 class TestDownload(unittest.TestCase):
  57     maxDiff = None
  58     def setUp(self):
  59         self.defs = defs
  60
  61 ### Dynamically generate tests
  62 def generator(test_case):
  63
  64     def test_template(self):
  65         ie = youtube_dl.extractor.get_info_extractor(test_case['name'])
  66         other_ies = [get_info_extractor(ie_key) for ie_key in test_case.get('add_ie', [])]
  67         is_playlist = any(k.startswith('playlist') for k in test_case)
  68         test_cases = test_case.get(
  69             'playlist', [] if is_playlist else [test_case])
  70
  71         def print_skipping(reason):
  72             print('Skipping %s: %s' % (test_case['name'], reason))
  73         if not ie.working():
  74             print_skipping('IE marked as not _WORKING')
  75             return
  76
  77         for tc in test_cases:
  78             info_dict = tc.get('info_dict', {})
  79             if not tc.get('file') and not (info_dict.get('id') and info_dict.get('ext')):
  80                 raise Exception('Test definition incorrect. The output file cannot be known. Are both \'id\' and \'ext\' keys present?')
  81
  82         if 'skip' in test_case:
  83             print_skipping(test_case['skip'])
  84             return
  85         for other_ie in other_ies:
  86             if not other_ie.working():
  87                 print_skipping(u'test depends on %sIE, marked as not WORKING' % other_ie.ie_key())
  88                 return
  89
  90         params = get_params(test_case.get('params', {}))
  91         if is_playlist and 'playlist' not in test_case:
  92             params.setdefault('extract_flat', True)
  93             params.setdefault('skip_download', True)
  94
  95         ydl = YoutubeDL(params)
  96         ydl.add_default_info_extractors()
  97         finished_hook_called = set()
  98         def _hook(status):
  99             if status['status'] == 'finished':
 100                 finished_hook_called.add(status['filename'])
 101         ydl.add_progress_hook(_hook)
 102
 103         def get_tc_filename(tc):
 104             return tc.get('file') or ydl.prepare_filename(tc.get('info_dict', {}))
 105
 106         res_dict = None
 107         def try_rm_tcs_files(tcs=None):
 108             if tcs is None:
 109                 tcs = test_cases
 110             for tc in tcs:
 111                 tc_filename = get_tc_filename(tc)
 112                 try_rm(tc_filename)
 113                 try_rm(tc_filename + '.part')
 114                 try_rm(os.path.splitext(tc_filename)[0] + '.info.json')
 115         try_rm_tcs_files()
 116         try:
 117             try_num = 1
 118             while True:
 119                 try:
 120                     # We're not using .download here sine that is just a shim
 121                     # for outside error handling, and returns the exit code
 122                     # instead of the result dict.
 123                     res_dict = ydl.extract_info(test_case['url'])
 124                 except (DownloadError, ExtractorError) as err:
 125                     # Check if the exception is not a network related one
 126                     if not err.exc_info[0] in (compat_urllib_error.URLError, socket.timeout, UnavailableVideoError, compat_http_client.BadStatusLine) or (err.exc_info[0] == compat_HTTPError and err.exc_info[1].code == 503):
 127                         raise
 128
 129                     if try_num == RETRIES:
 130                         report_warning(u'Failed due to network errors, skipping...')
 131                         return
 132
 133                     print('Retrying: {0} failed tries\n\n##########\n\n'.format(try_num))
 134
 135                     try_num += 1
 136                 else:
 137                     break
 138
 139             if is_playlist:
 140                 self.assertEqual(res_dict['_type'], 'playlist')
 141                 expect_info_dict(self, test_case.get('info_dict', {}), res_dict)
 142             if 'playlist_mincount' in test_case:
 143                 assertGreaterEqual(
 144                     self,
 145                     len(res_dict['entries']),
 146                     test_case['playlist_mincount'],
 147                     'Expected at least %d in playlist %s, but got only %d' % (
 148                         test_case['playlist_mincount'], test_case['url'],
 149                         len(res_dict['entries'])))
 150             if 'playlist_count' in test_case:
 151                 self.assertEqual(
 152                     len(res_dict['entries']),
 153                     test_case['playlist_count'],
 154                     'Expected %d entries in playlist %s, but got %d.' % (
 155                         len(res_dict['entries']),
 156                         test_case['url'],
 157                         test_case['playlist_count']))
 158             if 'playlist_duration_sum' in test_case:
 159                 got_duration = sum(e['duration'] for e in res_dict['entries'])
 160                 self.assertEqual(
 161                     test_case['playlist_duration_sum'], got_duration)
 162
 163             for tc in test_cases:
 164                 tc_filename = get_tc_filename(tc)
 165                 if not test_case.get('params', {}).get('skip_download', False):
 166                     self.assertTrue(os.path.exists(tc_filename), msg='Missing file ' + tc_filename)
 167                     self.assertTrue(tc_filename in finished_hook_called)
 168                 info_json_fn = os.path.splitext(tc_filename)[0] + '.info.json'
 169                 self.assertTrue(os.path.exists(info_json_fn))
 170                 if 'md5' in tc:
 171                     md5_for_file = _file_md5(tc_filename)
 172                     self.assertEqual(md5_for_file, tc['md5'])
 173                 with io.open(info_json_fn, encoding='utf-8') as infof:
 174                     info_dict = json.load(infof)
 175
 176                 expect_info_dict(self, tc.get('info_dict', {}), info_dict)
 177         finally:
 178             try_rm_tcs_files()
 179             if is_playlist and res_dict is not None:
 180                 # Remove all other files that may have been extracted if the
 181                 # extractor returns full results even with extract_flat
 182                 res_tcs = [{'info_dict': e} for e in res_dict['entries']]
 183                 try_rm_tcs_files(res_tcs)
 184
 185     return test_template
 186
 187 ### And add them to TestDownload
 188 for n, test_case in enumerate(defs):
 189     test_method = generator(test_case)
 190     tname = 'test_' + str(test_case['name'])
 191     i = 1
 192     while hasattr(TestDownload, tname):
 193         tname = 'test_'  + str(test_case['name']) + '_' + str(i)
 194         i += 1
 195     test_method.__name__ = tname
 196     setattr(TestDownload, test_method.__name__, test_method)
 197     del test_method
 198
 199
 200 if __name__ == '__main__':
 201     unittest.main()