X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=test%2Ftest_utils.py;h=8d89979775c75db9f54c041716f5bfc92a77150a;hb=cc7fec5818254f4679896823c7de9d17f50201ca;hp=5f4fdb771377cb30c8c1e4bd67c57d419e7cc645;hpb=29eb5174031cfc0b5de556da3da7761ac377de4e;p=youtube-dl diff --git a/test/test_utils.py b/test/test_utils.py index 5f4fdb771..8d8997977 100644 --- a/test/test_utils.py +++ b/test/test_utils.py @@ -1,6 +1,8 @@ #!/usr/bin/env python # coding: utf-8 +from __future__ import unicode_literals + # Allow direct execution import os import sys @@ -9,32 +11,37 @@ sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__)))) # Various small unit tests +import io +import json import xml.etree.ElementTree -#from youtube_dl.utils import htmlentity_transform from youtube_dl.utils import ( DateRange, encodeFilename, find_xpath_attr, + fix_xml_ampersands, get_meta_content, orderedSet, + PagedList, + parse_duration, + read_batch_urls, sanitize_filename, shell_quote, smuggle_url, str_to_int, + struct_unpack, timeconvert, unescapeHTML, unified_strdate, unsmuggle_url, url_basename, + urlencode_postdata, xpath_with_ns, + parse_iso8601, + strip_jsonp, + uppercase_escape, ) -if sys.version_info < (3, 0): - _compat_str = lambda b: b.decode('unicode-escape') -else: - _compat_str = lambda s: s - class TestUtil(unittest.TestCase): def test_timeconvert(self): @@ -56,9 +63,9 @@ class TestUtil(unittest.TestCase): self.assertEqual('this - that', sanitize_filename('this: that')) self.assertEqual(sanitize_filename('AT&T'), 'AT&T') - aumlaut = _compat_str('\xe4') + aumlaut = 'ä' self.assertEqual(sanitize_filename(aumlaut), aumlaut) - tests = _compat_str('\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430') + tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430' self.assertEqual(sanitize_filename(tests), tests) forbidden = '"\0\\/' @@ -80,9 +87,9 @@ class TestUtil(unittest.TestCase): self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True)) self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True)) - tests = _compat_str('a\xe4b\u4e2d\u56fd\u7684c') + tests = 'a\xe4b\u4e2d\u56fd\u7684c' self.assertEqual(sanitize_filename(tests, restricted=True), 'a_b_c') - self.assertTrue(sanitize_filename(_compat_str('\xf6'), restricted=True) != '') # No empty filename + self.assertTrue(sanitize_filename('\xf6', restricted=True) != '') # No empty filename forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#' for fc in forbidden: @@ -90,8 +97,8 @@ class TestUtil(unittest.TestCase): self.assertTrue(fbc not in sanitize_filename(fc, restricted=True)) # Handle a common case more neatly - self.assertEqual(sanitize_filename(_compat_str('\u5927\u58f0\u5e26 - Song'), restricted=True), 'Song') - self.assertEqual(sanitize_filename(_compat_str('\u603b\u7edf: Speech'), restricted=True), 'Speech') + self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song') + self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech') # .. but make sure the file name is never empty self.assertTrue(sanitize_filename('-', restricted=True) != '') self.assertTrue(sanitize_filename(':', restricted=True) != '') @@ -109,7 +116,9 @@ class TestUtil(unittest.TestCase): self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1]) def test_unescape_html(self): - self.assertEqual(unescapeHTML(_compat_str('%20;')), _compat_str('%20;')) + self.assertEqual(unescapeHTML('%20;'), '%20;') + self.assertEqual( + unescapeHTML('é'), 'é') def test_daterange(self): _20century = DateRange("19000101","20000101") @@ -124,9 +133,10 @@ 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') def test_find_xpath_attr(self): - testxml = u''' + testxml = ''' @@ -139,18 +149,18 @@ class TestUtil(unittest.TestCase): self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2]) def test_meta_parser(self): - testhtml = u''' + testhtml = ''' ''' get_meta = lambda name: get_meta_content(name, testhtml) - self.assertEqual(get_meta('description'), u'foo & bar') + self.assertEqual(get_meta('description'), 'foo & bar') self.assertEqual(get_meta('author'), 'Plato') def test_xpath_with_ns(self): - testxml = u''' + testxml = ''' The Author http://server.com/download.mp3 @@ -159,8 +169,8 @@ class TestUtil(unittest.TestCase): doc = xml.etree.ElementTree.fromstring(testxml) find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'})) self.assertTrue(find('media:song') is not None) - self.assertEqual(find('media:song/media:author').text, u'The Author') - self.assertEqual(find('media:song/url').text, u'http://server.com/download.mp3') + self.assertEqual(find('media:song/media:author').text, 'The Author') + self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3') def test_smuggle_url(self): data = {u"ö": u"ö", u"abc": [3]} @@ -175,19 +185,106 @@ class TestUtil(unittest.TestCase): self.assertEqual(res_data, None) def test_shell_quote(self): - args = ['ffmpeg', '-i', encodeFilename(u'ñ€ß\'.mp4')] - self.assertEqual(shell_quote(args), u"""ffmpeg -i 'ñ€ß'"'"'.mp4'""") + args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')] + self.assertEqual(shell_quote(args), """ffmpeg -i 'ñ€ß'"'"'.mp4'""") def test_str_to_int(self): self.assertEqual(str_to_int('123,456'), 123456) self.assertEqual(str_to_int('123.456'), 123456) def test_url_basename(self): - self.assertEqual(url_basename(u'http://foo.de/'), u'') - self.assertEqual(url_basename(u'http://foo.de/bar/baz'), u'baz') - self.assertEqual(url_basename(u'http://foo.de/bar/baz?x=y'), u'baz') - self.assertEqual(url_basename(u'http://foo.de/bar/baz#x=y'), u'baz') - self.assertEqual(url_basename(u'http://foo.de/bar/baz/'), u'baz') + self.assertEqual(url_basename('http://foo.de/'), '') + self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz') + self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz') + self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz') + self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz') + self.assertEqual( + url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'), + 'trailer.mp4') + + def test_parse_duration(self): + self.assertEqual(parse_duration(None), None) + self.assertEqual(parse_duration('1'), 1) + self.assertEqual(parse_duration('1337:12'), 80232) + self.assertEqual(parse_duration('9:12:43'), 33163) + self.assertEqual(parse_duration('12:00'), 720) + self.assertEqual(parse_duration('00:01:01'), 61) + self.assertEqual(parse_duration('x:y'), None) + self.assertEqual(parse_duration('3h11m53s'), 11513) + self.assertEqual(parse_duration('3h 11m 53s'), 11513) + self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513) + self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513) + self.assertEqual(parse_duration('62m45s'), 3765) + self.assertEqual(parse_duration('6m59s'), 419) + self.assertEqual(parse_duration('49s'), 49) + self.assertEqual(parse_duration('0h0m0s'), 0) + self.assertEqual(parse_duration('0m0s'), 0) + self.assertEqual(parse_duration('0s'), 0) + self.assertEqual(parse_duration('01:02:03.05'), 3723.05) + + def test_fix_xml_ampersands(self): + self.assertEqual( + fix_xml_ampersands('"&x=y&z=a'), '"&x=y&z=a') + self.assertEqual( + fix_xml_ampersands('"&x=y&wrong;&z=a'), + '"&x=y&wrong;&z=a') + self.assertEqual( + fix_xml_ampersands('&'><"'), + '&'><"') + self.assertEqual( + fix_xml_ampersands('Ӓ᪼'), 'Ӓ᪼') + self.assertEqual(fix_xml_ampersands('&#&#'), '&#&#') + + def test_paged_list(self): + def testPL(size, pagesize, sliceargs, expected): + def get_page(pagenum): + firstid = pagenum * pagesize + upto = min(size, pagenum * pagesize + pagesize) + for i in range(firstid, upto): + yield i + + pl = PagedList(get_page, pagesize) + got = pl.getslice(*sliceargs) + self.assertEqual(got, expected) + + testPL(5, 2, (), [0, 1, 2, 3, 4]) + testPL(5, 2, (1,), [1, 2, 3, 4]) + testPL(5, 2, (2,), [2, 3, 4]) + testPL(5, 2, (4,), [4]) + testPL(5, 2, (0, 3), [0, 1, 2]) + testPL(5, 2, (1, 4), [1, 2, 3]) + testPL(5, 2, (2, 99), [2, 3, 4]) + testPL(5, 2, (20, 99), []) + + def test_struct_unpack(self): + self.assertEqual(struct_unpack('!B', b'\x00'), (0,)) + + def test_read_batch_urls(self): + f = io.StringIO('''\xef\xbb\xbf foo + bar\r + baz + # More after this line\r + ; or after this + bam''') + self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam']) + + def test_urlencode_postdata(self): + data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'}) + self.assertTrue(isinstance(data, bytes)) + + def test_parse_iso8601(self): + self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266) + self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266) + self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266) + + def test_strip_jsonp(self): + stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);') + d = json.loads(stripped) + self.assertEqual(d, [{"id": "532cb", "x": 3}]) + + def test_uppercase_escape(self): + self.assertEqual(uppercase_escape('aä'), 'aä') + self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐') if __name__ == '__main__': unittest.main()