X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=test%2Ftest_utils.py;h=4f0ffd4824506f2f171e8b0e374d034d6ae8ccea;hb=dc03a42537cba83597ca8acb2bbe03f686f2136c;hp=16e1a1ddfdef5be7c0ba941d8793b3c8b88faf08;hpb=51897bb77c504ad206abbef5ae7504fcd082b5b0;p=youtube-dl diff --git a/test/test_utils.py b/test/test_utils.py index 16e1a1ddf..4f0ffd482 100644 --- a/test/test_utils.py +++ b/test/test_utils.py @@ -28,6 +28,7 @@ from youtube_dl.utils import ( fix_xml_ampersands, InAdvancePagedList, intlist_to_bytes, + is_html, js_to_json, limit_length, OnDemandPagedList, @@ -37,6 +38,7 @@ from youtube_dl.utils import ( parse_iso8601, read_batch_urls, sanitize_filename, + sanitize_path, shell_quote, smuggle_url, str_to_int, @@ -51,6 +53,9 @@ from youtube_dl.utils import ( urlencode_postdata, version_tuple, xpath_with_ns, + render_table, + match_str, + url_sanitize_consecutive_slashes, ) @@ -79,6 +84,15 @@ class TestUtil(unittest.TestCase): tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430' self.assertEqual(sanitize_filename(tests), tests) + self.assertEqual( + sanitize_filename('New World record at 0:12:34'), + 'New World record at 0_12_34') + + self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf') + self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf') + self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf') + self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf') + forbidden = '"\0\\/' for fc in forbidden: for fbc in forbidden: @@ -119,6 +133,42 @@ class TestUtil(unittest.TestCase): self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw') self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI') + def test_sanitize_path(self): + if sys.platform != 'win32': + return + + self.assertEqual(sanitize_path('abc'), 'abc') + self.assertEqual(sanitize_path('abc/def'), 'abc\\def') + self.assertEqual(sanitize_path('abc\\def'), 'abc\\def') + self.assertEqual(sanitize_path('abc|def'), 'abc#def') + self.assertEqual(sanitize_path('<>:"|?*'), '#######') + self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def') + self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def') + + self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc') + self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc') + + self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc') + self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc') + self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f') + self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc') + + self.assertEqual( + sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'), + 'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s') + + self.assertEqual( + sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'), + 'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part') + self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#') + self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def') + self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#') + + self.assertEqual(sanitize_path('../abc'), '..\\abc') + self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc') + self.assertEqual(sanitize_path('./abc'), 'abc') + self.assertEqual(sanitize_path('./../abc'), '..\\abc') + def test_ordered_set(self): self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7]) self.assertEqual(orderedSet([]), []) @@ -144,11 +194,15 @@ class TestUtil(unittest.TestCase): self.assertEqual(unified_strdate('8/7/2009'), '20090708') self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214') self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011') + self.assertEqual(unified_strdate('1968 12 10'), '19681210') self.assertEqual(unified_strdate('1968-12-10'), '19681210') self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128') self.assertEqual( unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False), '20141126') + self.assertEqual( + unified_strdate('2/2/2015 6:47:40 PM', day_first=False), + '20150202') def test_find_xpath_attr(self): testxml = ''' @@ -208,6 +262,8 @@ class TestUtil(unittest.TestCase): def test_parse_duration(self): self.assertEqual(parse_duration(None), None) + self.assertEqual(parse_duration(False), None) + self.assertEqual(parse_duration('invalid'), None) self.assertEqual(parse_duration('1'), 1) self.assertEqual(parse_duration('1337:12'), 80232) self.assertEqual(parse_duration('9:12:43'), 33163) @@ -229,6 +285,9 @@ class TestUtil(unittest.TestCase): self.assertEqual(parse_duration('5 s'), 5) self.assertEqual(parse_duration('3 min'), 180) self.assertEqual(parse_duration('2.5 hours'), 9000) + self.assertEqual(parse_duration('02:03:04'), 7384) + self.assertEqual(parse_duration('01:02:03:04'), 93784) + self.assertEqual(parse_duration('1 hour 3 minutes'), 3780) def test_fix_xml_ampersands(self): self.assertEqual( @@ -355,6 +414,10 @@ class TestUtil(unittest.TestCase): "playlist":[{"controls":{"all":null}}] }''') + inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"' + json_code = js_to_json(inp) + self.assertEqual(json.loads(json_code), json.loads(inp)) + 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"}) @@ -362,6 +425,16 @@ class TestUtil(unittest.TestCase): on = js_to_json('{"abc": true}') self.assertEqual(json.loads(on), {'abc': True}) + # Ignore JavaScript code as well + on = js_to_json('''{ + "x": 1, + y: "a", + z: some.code + }''') + d = json.loads(on) + self.assertEqual(d['x'], 1) + self.assertEqual(d['y'], 'a') + def test_clean_html(self): self.assertEqual(clean_html('a:\nb'), 'a: b') self.assertEqual(clean_html('a:\n "b"'), 'a: "b"') @@ -410,5 +483,77 @@ ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4') self.assertTrue(age_restricted(18, 14)) self.assertFalse(age_restricted(18, 18)) + def test_is_html(self): + self.assertFalse(is_html(b'\x49\x44\x43\xaaa')) + self.assertTrue(is_html( # UTF-8 with BOM + b'\xef\xbb\xbf\xaaa')) + self.assertTrue(is_html( # UTF-16-LE + b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00' + )) + self.assertTrue(is_html( # UTF-16-BE + b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4' + )) + self.assertTrue(is_html( # UTF-32-BE + b'\x00\x00\xFE\xFF\x00\x00\x00<\x00\x00\x00h\x00\x00\x00t\x00\x00\x00m\x00\x00\x00l\x00\x00\x00>\x00\x00\x00\xe4')) + self.assertTrue(is_html( # UTF-32-LE + b'\xFF\xFE\x00\x00<\x00\x00\x00h\x00\x00\x00t\x00\x00\x00m\x00\x00\x00l\x00\x00\x00>\x00\x00\x00\xe4\x00\x00\x00')) + + def test_render_table(self): + self.assertEqual( + render_table( + ['a', 'bcd'], + [[123, 4], [9999, 51]]), + 'a bcd\n' + '123 4\n' + '9999 51') + + def test_match_str(self): + self.assertRaises(ValueError, match_str, 'xy>foobar', {}) + self.assertFalse(match_str('xy', {'x': 1200})) + self.assertTrue(match_str('!xy', {'x': 1200})) + self.assertTrue(match_str('x', {'x': 1200})) + self.assertFalse(match_str('!x', {'x': 1200})) + self.assertTrue(match_str('x', {'x': 0})) + self.assertFalse(match_str('x>0', {'x': 0})) + self.assertFalse(match_str('x>0', {})) + self.assertTrue(match_str('x>?0', {})) + self.assertTrue(match_str('x>1K', {'x': 1200})) + self.assertFalse(match_str('x>2K', {'x': 1200})) + self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200})) + self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200})) + self.assertFalse(match_str('y=a212', {'y': 'foobar42'})) + self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'})) + self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'})) + self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'})) + self.assertFalse(match_str( + 'like_count > 100 & dislike_count 100 & dislike_count 100 & dislike_count 100 & dislike_count