X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/ced7488f6d3a519b2c1b1cbd31048743fb8285bd..af014acd27e0b471d5903630847eabb26437b46c:/test/test_utils.py?ds=sidebyside diff --git a/test/test_utils.py b/test/test_utils.py index 74fcf91..efa73d0 100644 --- a/test/test_utils.py +++ b/test/test_utils.py @@ -34,11 +34,17 @@ from youtube_dl.utils import ( find_xpath_attr, fix_xml_ampersands, get_element_by_class, + get_element_by_attribute, + get_elements_by_class, + get_elements_by_attribute, InAdvancePagedList, intlist_to_bytes, is_html, js_to_json, limit_length, + mimetype2ext, + month_by_name, + multipart_encode, ohdave_rsa_encrypt, OnDemandPagedList, orderedSet, @@ -47,9 +53,11 @@ from youtube_dl.utils import ( parse_filesize, parse_count, parse_iso8601, + pkcs1pad, read_batch_urls, sanitize_filename, sanitize_path, + expand_path, prepend_extension, replace_extension, remove_start, @@ -67,6 +75,8 @@ from youtube_dl.utils import ( uppercase_escape, lowercase_escape, url_basename, + base_url, + urljoin, urlencode_postdata, urshift, update_url_query, @@ -87,6 +97,9 @@ from youtube_dl.utils import ( from youtube_dl.compat import ( compat_chr, compat_etree_fromstring, + compat_getenv, + compat_os_name, + compat_setenv, compat_urlparse, compat_parse_qs, ) @@ -206,6 +219,18 @@ class TestUtil(unittest.TestCase): self.assertEqual(sanitize_path('./abc'), 'abc') self.assertEqual(sanitize_path('./../abc'), '..\\abc') + def test_expand_path(self): + def env(var): + return '%{0}%'.format(var) if sys.platform == 'win32' else '${0}'.format(var) + + compat_setenv('YOUTUBE_DL_EXPATH_PATH', 'expanded') + self.assertEqual(expand_path(env('YOUTUBE_DL_EXPATH_PATH')), 'expanded') + self.assertEqual(expand_path(env('HOME')), compat_getenv('HOME')) + self.assertEqual(expand_path('~'), compat_getenv('HOME')) + self.assertEqual( + expand_path('~/%s' % env('YOUTUBE_DL_EXPATH_PATH')), + '%s/expanded' % compat_getenv('HOME')) + def test_prepend_extension(self): self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext') self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext') @@ -254,6 +279,7 @@ class TestUtil(unittest.TestCase): self.assertEqual(unescapeHTML('/'), '/') self.assertEqual(unescapeHTML('é'), 'é') self.assertEqual(unescapeHTML(''), '') + self.assertEqual(unescapeHTML('&a"'), '&a"') # HTML5 entities self.assertEqual(unescapeHTML('.''), '.\'') @@ -290,6 +316,10 @@ class TestUtil(unittest.TestCase): self.assertEqual(unified_strdate('25-09-2014'), '20140925') self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227') self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None) + self.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207') + self.assertEqual(unified_strdate('July 15th, 2013'), '20130715') + self.assertEqual(unified_strdate('September 1st, 2013'), '20130901') + self.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902') def test_unified_timestamps(self): self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600) @@ -310,6 +340,9 @@ class TestUtil(unittest.TestCase): self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200) self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None) self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500) + self.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100) + self.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361) + self.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540) def test_determine_ext(self): self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4') @@ -417,7 +450,9 @@ class TestUtil(unittest.TestCase): def test_shell_quote(self): args = ['ffmpeg', '-i', encodeFilename('ñâ¬Ã\'.mp4')] - self.assertEqual(shell_quote(args), """ffmpeg -i 'ñâ¬Ã'"'"'.mp4'""") + self.assertEqual( + shell_quote(args), + """ffmpeg -i 'ñâ¬Ã'"'"'.mp4'""" if compat_os_name != 'nt' else '''ffmpeg -i "ñâ¬Ã'.mp4"''') def test_str_to_int(self): self.assertEqual(str_to_int('123,456'), 123456) @@ -433,6 +468,33 @@ class TestUtil(unittest.TestCase): url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'), 'trailer.mp4') + def test_base_url(self): + self.assertEqual(base_url('http://foo.de/'), 'http://foo.de/') + self.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/') + self.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/') + self.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/') + self.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/') + + def test_urljoin(self): + self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt') + self.assertEqual(urljoin(b'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt') + self.assertEqual(urljoin('http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt') + self.assertEqual(urljoin(b'http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt') + self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt') + self.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt') + self.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt') + self.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt') + self.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt') + self.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt') + self.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt') + self.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt') + self.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt') + self.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt') + self.assertEqual(urljoin('http://foo.de/', None), None) + self.assertEqual(urljoin('http://foo.de/', ''), None) + self.assertEqual(urljoin('http://foo.de/', ['foobar']), None) + self.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt') + def test_parse_age_limit(self): self.assertEqual(parse_age_limit(None), None) self.assertEqual(parse_age_limit(False), None) @@ -477,6 +539,7 @@ class TestUtil(unittest.TestCase): self.assertEqual(parse_duration('1 hour 3 minutes'), 3780) self.assertEqual(parse_duration('87 Min.'), 5220) self.assertEqual(parse_duration('PT1H0.040S'), 3600.04) + self.assertEqual(parse_duration('PT00H03M30SZ'), 210) def test_fix_xml_ampersands(self): self.assertEqual( @@ -563,6 +626,16 @@ class TestUtil(unittest.TestCase): 'http://example.com/path', {'test': '第äºè¡ÑеÑÑ'})), query_dict('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82')) + def test_multipart_encode(self): + self.assertEqual( + multipart_encode({b'field': b'value'}, boundary='AAAAAA')[0], + b'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n') + self.assertEqual( + multipart_encode({'æ¬ä½'.encode('utf-8'): 'å¼'.encode('utf-8')}, boundary='AAAAAA')[0], + b'--AAAAAA\r\nContent-Disposition: form-data; name="\xe6\xac\x84\xe4\xbd\x8d"\r\n\r\n\xe5\x80\xbc\r\n--AAAAAA--\r\n') + self.assertRaises( + ValueError, multipart_encode, {b'field': b'value'}, boundary='value') + def test_dict_get(self): FALSE_VALUES = { 'none': None, @@ -610,6 +683,14 @@ class TestUtil(unittest.TestCase): d = json.loads(stripped) self.assertEqual(d, {'status': 'success'}) + stripped = strip_jsonp('window.cb && window.cb({"status": "success"});') + d = json.loads(stripped) + self.assertEqual(d, {'status': 'success'}) + + stripped = strip_jsonp('window.cb && cb({"status": "success"});') + d = json.loads(stripped) + self.assertEqual(d, {'status': 'success'}) + def test_uppercase_escape(self): self.assertEqual(uppercase_escape('aä'), 'aä') self.assertEqual(uppercase_escape('\\U0001d550'), 'ð') @@ -625,6 +706,22 @@ class TestUtil(unittest.TestCase): limit_length('foo bar baz asd', 12).startswith('foo bar')) self.assertTrue('...' in limit_length('foo bar baz asd', 12)) + def test_mimetype2ext(self): + self.assertEqual(mimetype2ext(None), None) + self.assertEqual(mimetype2ext('video/x-flv'), 'flv') + self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8') + self.assertEqual(mimetype2ext('text/vtt'), 'vtt') + self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt') + self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html') + + def test_month_by_name(self): + self.assertEqual(month_by_name(None), None) + self.assertEqual(month_by_name('December', 'en'), 12) + self.assertEqual(month_by_name('décembre', 'fr'), 12) + self.assertEqual(month_by_name('December'), 12) + self.assertEqual(month_by_name('décembre'), None) + self.assertEqual(month_by_name('Unknown', 'unknown'), None) + def test_parse_codecs(self): self.assertEqual(parse_codecs(''), {}) self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), { @@ -712,6 +809,9 @@ class TestUtil(unittest.TestCase): inp = '''{"foo":101}''' self.assertEqual(js_to_json(inp), '''{"foo":101}''') + inp = '''{"duration": "00:01:07"}''' + self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''') + def test_js_to_json_edgecases(self): on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}") self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"}) @@ -732,12 +832,27 @@ class TestUtil(unittest.TestCase): on = js_to_json('["abc", "def",]') self.assertEqual(json.loads(on), ['abc', 'def']) + on = js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]') + self.assertEqual(json.loads(on), ['abc', 'def']) + + on = js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]') + self.assertEqual(json.loads(on), ['abc', 'def']) + on = js_to_json('{"abc": "def",}') self.assertEqual(json.loads(on), {'abc': 'def'}) + on = js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}') + self.assertEqual(json.loads(on), {'abc': 'def'}) + on = js_to_json('{ 0: /* " \n */ ",]" , }') self.assertEqual(json.loads(on), {'0': ',]'}) + on = js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }') + self.assertEqual(json.loads(on), {'0': ',]'}) + + on = js_to_json('{ 0: // comment\n1 }') + self.assertEqual(json.loads(on), {'0': 1}) + on = js_to_json(r'["
x<\/p>"]') self.assertEqual(json.loads(on), ['
x
']) @@ -747,15 +862,27 @@ class TestUtil(unittest.TestCase): on = js_to_json("['a\\\nb']") self.assertEqual(json.loads(on), ['ab']) + on = js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/") + self.assertEqual(json.loads(on), ['ab']) + on = js_to_json('{0xff:0xff}') self.assertEqual(json.loads(on), {'255': 255}) + on = js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}') + self.assertEqual(json.loads(on), {'255': 255}) + on = js_to_json('{077:077}') self.assertEqual(json.loads(on), {'63': 63}) + on = js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}') + self.assertEqual(json.loads(on), {'63': 63}) + on = js_to_json('{42:42}') self.assertEqual(json.loads(on), {'42': 42}) + on = js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}') + self.assertEqual(json.loads(on), {'42': 42}) + def test_extract_attributes(self): self.assertEqual(extract_attributes('Ignored, three