X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=test%2Ftest_utils.py;h=a3a23fbb4a5a71da86488c951b151c5c42e17456;hb=18ea0cefc370807e1d31fd7b462691f29f48c8fe;hp=614ef1b03afb429f8c62f192a3c227b207e38032;hpb=9993976ae436c62ef86c2c45f0f001e5f7e471bd;p=youtube-dl

diff --git a/test/test_utils.py b/test/test_utils.py
index 614ef1b03..a3a23fbb4 100644
--- a/test/test_utils.py
+++ b/test/test_utils.py
@@ -2,22 +2,78 @@
 
 # Various small unit tests
 
+import unittest
+
+#from youtube_dl.utils import htmlentity_transform
+from youtube_dl.utils import timeconvert
 from youtube_dl.utils import sanitize_filename
+from youtube_dl.utils import unescapeHTML
+from youtube_dl.utils import orderedSet
+
+
+class TestUtil(unittest.TestCase):
+	def test_timeconvert(self):
+		self.assertTrue(timeconvert('') is None)
+		self.assertTrue(timeconvert('bougrg') is None)
+
+	def test_sanitize_filename(self):
+		self.assertEqual(sanitize_filename(u'abc'), u'abc')
+		self.assertEqual(sanitize_filename(u'abc_d-e'), u'abc_d-e')
+
+		self.assertEqual(sanitize_filename(u'123'), u'123')
+
+		self.assertEqual(u'abc_de', sanitize_filename(u'abc/de'))
+		self.assertFalse(u'/' in sanitize_filename(u'abc/de///'))
+
+		self.assertEqual(u'abc_de', sanitize_filename(u'abc/<>\\*|de'))
+		self.assertEqual(u'xxx', sanitize_filename(u'xxx/<>\\*|'))
+		self.assertEqual(u'yes no', sanitize_filename(u'yes? no'))
+		self.assertEqual(u'this - that', sanitize_filename(u'this: that'))
+
+		self.assertEqual(sanitize_filename(u'AT&T'), u'AT&T')
+		self.assertEqual(sanitize_filename(u'ä'), u'ä')
+		self.assertEqual(sanitize_filename(u'кириллица'), u'кириллица')
+
+		forbidden = u'"\0\\/'
+		for fc in forbidden:
+			for fbc in forbidden:
+				self.assertTrue(fbc not in sanitize_filename(fc))
+
+	def test_sanitize_filename_restricted(self):
+		self.assertEqual(sanitize_filename(u'abc', restricted=True), u'abc')
+		self.assertEqual(sanitize_filename(u'abc_d-e', restricted=True), u'abc_d-e')
+
+		self.assertEqual(sanitize_filename(u'123', restricted=True), u'123')
+
+		self.assertEqual(u'abc_de', sanitize_filename(u'abc/de', restricted=True))
+		self.assertFalse(u'/' in sanitize_filename(u'abc/de///', restricted=True))
+
+		self.assertEqual(u'abc_de', sanitize_filename(u'abc/<>\\*|de', restricted=True))
+		self.assertEqual(u'xxx', sanitize_filename(u'xxx/<>\\*|', restricted=True))
+		self.assertEqual(u'yes_no', sanitize_filename(u'yes? no', restricted=True))
+		self.assertEqual(u'this_-_that', sanitize_filename(u'this: that', restricted=True))
 
-def test_sanitize_filename():
-	assert sanitize_filename(u'abc') == u'abc'
-	assert sanitize_filename(u'abc_d-e') == u'abc_d-e'
+		self.assertEqual(sanitize_filename(u'aäb中国的c', restricted=True), u'a_b_c')
+		self.assertTrue(sanitize_filename(u'ö', restricted=True) != u'') # No empty filename
 
-	assert sanitize_filename(u'123') == u'123'
+		forbidden = u'"\0\\/&!: \'\t\n'
+		for fc in forbidden:
+			for fbc in forbidden:
+				self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
 
-	assert u'/' not in sanitize_filename(u'abc/de')
-	assert u'abc' in sanitize_filename(u'abc/de')
-	assert u'de' in sanitize_filename(u'abc/de')
-	assert u'/' not in sanitize_filename(u'abc/de///')
+		# Handle a common case more neatly
+		self.assertEqual(sanitize_filename(u'大声带 - Song', restricted=True), u'Song')
+		self.assertEqual(sanitize_filename(u'总统: Speech', restricted=True), u'Speech')
+		# .. but make sure the file name is never empty
+		self.assertTrue(sanitize_filename(u'-', restricted=True) != u'')
+		self.assertTrue(sanitize_filename(u':', restricted=True) != u'')
 
-	assert u'\\' not in sanitize_filename(u'abc\\de')
-	assert u'abc' in sanitize_filename(u'abc\\de')
-	assert u'de' in sanitize_filename(u'abc\\de')
+	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([]), [])
+		self.assertEqual(orderedSet([1]), [1])
+		#keep the list ordered
+		self.assertEqual(orderedSet([135,1,1,1]), [135,1])
 
-	assert sanitize_filename(u'ä') == u'ä'
-	assert sanitize_filename(u'кириллица') == u'кириллица'
+	def test_unescape_html(self):
+		self.assertEqual(unescapeHTML(u"%20;"), u"%20;")