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