#!/usr/bin/env python from __future__ import unicode_literals # Allow direct execution import os import sys import unittest sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__)))) from test.helper import FakeYDL, expect_dict from youtube_dl.extractor.common import InfoExtractor from youtube_dl.extractor import YoutubeIE, get_info_extractor from youtube_dl.utils import encode_data_uri, strip_jsonp, ExtractorError, RegexNotFoundError class TestIE(InfoExtractor): pass class TestInfoExtractor(unittest.TestCase): def setUp(self): self.ie = TestIE(FakeYDL()) def test_ie_key(self): self.assertEqual(get_info_extractor(YoutubeIE.ie_key()), YoutubeIE) def test_html_search_regex(self): html = '

Watch this video

' search = lambda re, *args: self.ie._html_search_regex(re, html, *args) self.assertEqual(search(r'

(.+?)

', 'foo'), 'Watch this video') def test_opengraph(self): ie = self.ie html = ''' ''' self.assertEqual(ie._og_search_title(html), 'Foo') self.assertEqual(ie._og_search_description(html), 'Some video\'s description ') self.assertEqual(ie._og_search_thumbnail(html), 'http://domain.com/pic.jpg?key1=val1&key2=val2') self.assertEqual(ie._og_search_video_url(html, default=None), None) self.assertEqual(ie._og_search_property('foobar', html), 'Foo') self.assertEqual(ie._og_search_property('test1', html), 'foo > < bar') self.assertEqual(ie._og_search_property('test2', html), 'foo >//< bar') self.assertEqual(ie._og_search_property(('test0', 'test1'), html), 'foo > < bar') self.assertRaises(RegexNotFoundError, ie._og_search_property, 'test0', html, None, fatal=True) self.assertRaises(RegexNotFoundError, ie._og_search_property, ('test0', 'test00'), html, None, fatal=True) def test_html_search_meta(self): ie = self.ie html = ''' ''' self.assertEqual(ie._html_search_meta('a', html), '1') self.assertEqual(ie._html_search_meta('b', html), '2') self.assertEqual(ie._html_search_meta('c', html), '3') self.assertEqual(ie._html_search_meta('d', html), '4') self.assertEqual(ie._html_search_meta('e', html), '5') self.assertEqual(ie._html_search_meta('f', html), '6') self.assertEqual(ie._html_search_meta(('a', 'b', 'c'), html), '1') self.assertEqual(ie._html_search_meta(('c', 'b', 'a'), html), '3') self.assertEqual(ie._html_search_meta(('z', 'x', 'c'), html), '3') self.assertRaises(RegexNotFoundError, ie._html_search_meta, 'z', html, None, fatal=True) self.assertRaises(RegexNotFoundError, ie._html_search_meta, ('z', 'x'), html, None, fatal=True) def test_download_json(self): uri = encode_data_uri(b'{"foo": "blah"}', 'application/json') self.assertEqual(self.ie._download_json(uri, None), {'foo': 'blah'}) uri = encode_data_uri(b'callback({"foo": "blah"})', 'application/javascript') self.assertEqual(self.ie._download_json(uri, None, transform_source=strip_jsonp), {'foo': 'blah'}) uri = encode_data_uri(b'{"foo": invalid}', 'application/json') self.assertRaises(ExtractorError, self.ie._download_json, uri, None) self.assertEqual(self.ie._download_json(uri, None, fatal=False), None) def test_extract_jwplayer_data_realworld(self): # from http://www.suffolk.edu/sjc/ expect_dict( self, self.ie._extract_jwplayer_data(r''' ''', None, require_title=False), { 'id': 'XEgvuql4', 'formats': [{ 'url': 'rtmp://192.138.214.154/live/sjclive', 'ext': 'flv' }] }) # from https://www.pornoxo.com/videos/7564/striptease-from-sexy-secretary/ expect_dict( self, self.ie._extract_jwplayer_data(r''' ''', 'dummy', require_title=False), { 'thumbnail': 'https://t03.vipstreamservice.com/thumbs/pxo-full/2009-12/14/a4b2157147afe5efa93ce1978e0265289c193874e02597.flv-full-13.jpg', 'formats': [{ 'url': 'https://cdn.pornoxo.com/key=MF+oEbaxqTKb50P-w9G3nA,end=1489689259,ip=104.199.146.27/ip=104.199.146.27/speed=6573765/buffer=3.0/2009-12/4b2157147afe5efa93ce1978e0265289c193874e02597.flv', 'ext': 'flv' }] }) # from http://www.indiedb.com/games/king-machine/videos expect_dict( self, self.ie._extract_jwplayer_data(r''' ''', 'dummy'), { 'title': 'king machine trailer 1', 'thumbnail': 'http://media.indiedb.com/cache/images/games/1/50/49678/thumb_620x2000/king-machine-trailer.mp4.jpg', 'formats': [{ 'url': 'http://cdn.dbolical.com/cache/videos/games/1/50/49678/encode_mp4/king-machine-trailer.mp4', 'height': 360, 'ext': 'mp4' }, { 'url': 'http://cdn.dbolical.com/cache/videos/games/1/50/49678/encode720p_mp4/king-machine-trailer.mp4', 'height': 720, 'ext': 'mp4' }] }) if __name__ == '__main__': unittest.main()