Merge branch 'sohu_fix' of https://github.com/yan12125/youtube-dl into yan12125-sohu_fix

[youtube-dl] / test / test_utils.py

diff --git a/test/test_utils.py b/test/test_utils.py
index 80c765bc496f0ce51e3deda0b39689f8136b597a..4f0ffd4824506f2f171e8b0e374d034d6ae8ccea 100644 (file)
--- a/test/test_utils.py
+++ b/test/test_utils.py
@@ -38,6 +38,7 @@ from youtube_dl.utils import (
     parse_iso8601,
     read_batch_urls,
     sanitize_filename,
+    sanitize_path,
     shell_quote,
     smuggle_url,
     str_to_int,
@@ -53,6 +54,8 @@ from youtube_dl.utils import (
     version_tuple,
     xpath_with_ns,
     render_table,
+    match_str,
+    url_sanitize_consecutive_slashes,
 )
 
 
@@ -85,6 +88,11 @@ class TestUtil(unittest.TestCase):
             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:
@@ -125,6 +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([]), [])
@@ -243,6 +287,7 @@ class TestUtil(unittest.TestCase):
         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(
@@ -369,6 +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"})
@@ -459,6 +508,52 @@ ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
             '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 <? 50 & description',
+            {'like_count': 90, 'description': 'foo'}))
+        self.assertTrue(match_str(
+            'like_count > 100 & dislike_count <? 50 & description',
+            {'like_count': 190, 'description': 'foo'}))
+        self.assertFalse(match_str(
+            'like_count > 100 & dislike_count <? 50 & description',
+            {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
+        self.assertFalse(match_str(
+            'like_count > 100 & dislike_count <? 50 & description',
+            {'like_count': 190, 'dislike_count': 10}))
+
+    def test_url_sanitize_consecutive_slashes(self):
+        self.assertEqual(url_sanitize_consecutive_slashes(
+            'http://hostname/foo//bar/filename.html'),
+            'http://hostname/foo/bar/filename.html')
+        self.assertEqual(url_sanitize_consecutive_slashes(
+            'http://hostname//foo/bar/filename.html'),
+            'http://hostname/foo/bar/filename.html')
+        self.assertEqual(url_sanitize_consecutive_slashes(
+            'http://hostname//'), 'http://hostname/')
+        self.assertEqual(url_sanitize_consecutive_slashes(
+            'http://hostname/foo/bar/filename.html'),
+            'http://hostname/foo/bar/filename.html')
+        self.assertEqual(url_sanitize_consecutive_slashes(
+            'http://hostname/'), 'http://hostname/')
+
 
 if __name__ == '__main__':
     unittest.main()