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 get_params, get_testcases, global_setup, try_rm, md5
  10 global_setup()
  11
  12
  13 import hashlib
  14 import io
  15 import json
  16 import socket
  17
  18 import youtube_dl.YoutubeDL
  19 from youtube_dl.utils import (
  20     compat_str,
  21     compat_urllib_error,
  22     DownloadError,
  23     ExtractorError,
  24     UnavailableVideoError,
  25 )
  26
  27 RETRIES = 3
  28
  29 class YoutubeDL(youtube_dl.YoutubeDL):
  30     def __init__(self, *args, **kwargs):
  31         self.to_stderr = self.to_screen
  32         self.processed_info_dicts = []
  33         super(YoutubeDL, self).__init__(*args, **kwargs)
  34     def report_warning(self, message):
  35         # Don't accept warnings during tests
  36         raise ExtractorError(message)
  37     def process_info(self, info_dict):
  38         self.processed_info_dicts.append(info_dict)
  39         return super(YoutubeDL, self).process_info(info_dict)
  40
  41 def _file_md5(fn):
  42     with open(fn, 'rb') as f:
  43         return hashlib.md5(f.read()).hexdigest()
  44
  45 defs = get_testcases()
  46
  47
  48 class TestDownload(unittest.TestCase):
  49     maxDiff = None
  50     def setUp(self):
  51         self.defs = defs
  52
  53 ### Dynamically generate tests
  54 def generator(test_case):
  55
  56     def test_template(self):
  57         ie = youtube_dl.extractor.get_info_extractor(test_case['name'])
  58         def print_skipping(reason):
  59             print('Skipping %s: %s' % (test_case['name'], reason))
  60         if not ie._WORKING:
  61             print_skipping('IE marked as not _WORKING')
  62             return
  63         if 'playlist' not in test_case:
  64             info_dict = test_case.get('info_dict', {})
  65             if not test_case.get('file') and not (info_dict.get('id') and info_dict.get('ext')):
  66                 print_skipping('The output file cannot be know, the "file" '
  67                     'key is missing or the info_dict is incomplete')
  68                 return
  69         if 'skip' in test_case:
  70             print_skipping(test_case['skip'])
  71             return
  72
  73         params = get_params(test_case.get('params', {}))
  74
  75         ydl = YoutubeDL(params)
  76         ydl.add_default_info_extractors()
  77         finished_hook_called = set()
  78         def _hook(status):
  79             if status['status'] == 'finished':
  80                 finished_hook_called.add(status['filename'])
  81         ydl.fd.add_progress_hook(_hook)
  82
  83         def get_tc_filename(tc):
  84             return tc.get('file') or ydl.prepare_filename(tc.get('info_dict', {}))
  85
  86         test_cases = test_case.get('playlist', [test_case])
  87         def try_rm_tcs_files():
  88             for tc in test_cases:
  89                 tc_filename = get_tc_filename(tc)
  90                 try_rm(tc_filename)
  91                 try_rm(tc_filename + '.part')
  92                 try_rm(tc_filename + '.info.json')
  93         try_rm_tcs_files()
  94         try:
  95             for retry in range(1, RETRIES + 1):
  96                 try:
  97                     ydl.download([test_case['url']])
  98                 except (DownloadError, ExtractorError) as err:
  99                     if retry == RETRIES: raise
 100
 101                     # Check if the exception is not a network related one
 102                     if not err.exc_info[0] in (compat_urllib_error.URLError, socket.timeout, UnavailableVideoError):
 103                         raise
 104
 105                     print('Retrying: {0} failed tries\n\n##########\n\n'.format(retry))
 106                 else:
 107                     break
 108
 109             for tc in test_cases:
 110                 tc_filename = get_tc_filename(tc)
 111                 if not test_case.get('params', {}).get('skip_download', False):
 112                     self.assertTrue(os.path.exists(tc_filename), msg='Missing file ' + tc_filename)
 113                     self.assertTrue(tc_filename in finished_hook_called)
 114                 self.assertTrue(os.path.exists(tc_filename + '.info.json'))
 115                 if 'md5' in tc:
 116                     md5_for_file = _file_md5(tc_filename)
 117                     self.assertEqual(md5_for_file, tc['md5'])
 118                 with io.open(tc_filename + '.info.json', encoding='utf-8') as infof:
 119                     info_dict = json.load(infof)
 120                 for (info_field, expected) in tc.get('info_dict', {}).items():
 121                     if isinstance(expected, compat_str) and expected.startswith('md5:'):
 122                         got = 'md5:' + md5(info_dict.get(info_field))
 123                     else:
 124                         got = info_dict.get(info_field)
 125                     self.assertEqual(expected, got,
 126                         u'invalid value for field %s, expected %r, got %r' % (info_field, expected, got))
 127
 128                 # If checkable fields are missing from the test case, print the info_dict
 129                 test_info_dict = dict((key, value if not isinstance(value, compat_str) or len(value) < 250 else 'md5:' + md5(value))
 130                     for key, value in info_dict.items()
 131                     if value and key in ('title', 'description', 'uploader', 'upload_date', 'uploader_id', 'location'))
 132                 if not all(key in tc.get('info_dict', {}).keys() for key in test_info_dict.keys()):
 133                     sys.stderr.write(u'\n"info_dict": ' + json.dumps(test_info_dict, ensure_ascii=False, indent=2) + u'\n')
 134
 135                 # Check for the presence of mandatory fields
 136                 for key in ('id', 'url', 'title', 'ext'):
 137                     self.assertTrue(key in info_dict.keys() and info_dict[key])
 138         finally:
 139             try_rm_tcs_files()
 140
 141     return test_template
 142
 143 ### And add them to TestDownload
 144 for n, test_case in enumerate(defs):
 145     test_method = generator(test_case)
 146     tname = 'test_' + str(test_case['name'])
 147     i = 1
 148     while hasattr(TestDownload, tname):
 149         tname = 'test_'  + str(test_case['name']) + '_' + str(i)
 150         i += 1
 151     test_method.__name__ = tname
 152     setattr(TestDownload, test_method.__name__, test_method)
 153     del test_method
 154
 155
 156 if __name__ == '__main__':
 157     unittest.main()