X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=test%2Ftest_utils.py;h=ed61e4c2751c82c062db6339b4245cb4cae6c9bf;hb=e3a6747d8f19ad0ba8aee7c3214cdb64903beba0;hp=918a7a9ef886611753a790b38da81dc59b318ce6;hpb=d01949dc89feb2441f251e42e8a6bfa4711b9715;p=youtube-dl
diff --git a/test/test_utils.py b/test/test_utils.py
index 918a7a9ef..ed61e4c27 100644
--- a/test/test_utils.py
+++ b/test/test_utils.py
@@ -18,12 +18,18 @@ import xml.etree.ElementTree
from youtube_dl.utils import (
age_restricted,
args_to_str,
+ encode_base_n,
clean_html,
+ date_from_str,
DateRange,
detect_exe_version,
+ determine_ext,
+ dict_get,
+ encode_compat_str,
encodeFilename,
escape_rfc3986,
escape_url,
+ extract_attributes,
ExtractorError,
find_xpath_attr,
fix_xml_ampersands,
@@ -32,29 +38,36 @@ from youtube_dl.utils import (
is_html,
js_to_json,
limit_length,
+ ohdave_rsa_encrypt,
OnDemandPagedList,
orderedSet,
parse_duration,
parse_filesize,
+ parse_count,
parse_iso8601,
read_batch_urls,
sanitize_filename,
sanitize_path,
prepend_extension,
replace_extension,
+ remove_start,
+ remove_end,
+ remove_quotes,
shell_quote,
smuggle_url,
str_to_int,
strip_jsonp,
- struct_unpack,
timeconvert,
unescapeHTML,
unified_strdate,
+ unified_timestamp,
unsmuggle_url,
uppercase_escape,
lowercase_escape,
url_basename,
urlencode_postdata,
+ urshift,
+ update_url_query,
version_tuple,
xpath_with_ns,
xpath_element,
@@ -68,6 +81,12 @@ from youtube_dl.utils import (
cli_valueless_option,
cli_bool_option,
)
+from youtube_dl.compat import (
+ compat_chr,
+ compat_etree_fromstring,
+ compat_urlparse,
+ compat_parse_qs,
+)
class TestUtil(unittest.TestCase):
@@ -123,8 +142,8 @@ 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 = 'a\xe4b\u4e2d\u56fd\u7684c'
- self.assertEqual(sanitize_filename(tests, restricted=True), 'a_b_c')
+ tests = 'aäb\u4e2d\u56fd\u7684c'
+ self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
self.assertTrue(sanitize_filename('\xf6', restricted=True) != '') # No empty filename
forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
@@ -139,6 +158,10 @@ class TestUtil(unittest.TestCase):
self.assertTrue(sanitize_filename('-', restricted=True) != '')
self.assertTrue(sanitize_filename(':', restricted=True) != '')
+ self.assertEqual(sanitize_filename(
+ 'ÃÃÃÃÃÃ
ÃÃÃÃÃÃÃÃÃÃÃÃÃÃÃÃÃÅÃÅÃÃÃÃÅ°ÃÃÃà áâãäåæçèéêëìÃîïðñòóôõöÅøÅùúûüűýþÿ', restricted=True),
+ 'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYPssaaaaaaaeceeeeiiiionooooooooeuuuuuypy')
+
def test_sanitize_ids(self):
self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
@@ -196,6 +219,25 @@ class TestUtil(unittest.TestCase):
self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
+ def test_remove_start(self):
+ self.assertEqual(remove_start(None, 'A - '), None)
+ self.assertEqual(remove_start('A - B', 'A - '), 'B')
+ self.assertEqual(remove_start('B - A', 'A - '), 'B - A')
+
+ def test_remove_end(self):
+ self.assertEqual(remove_end(None, ' - B'), None)
+ self.assertEqual(remove_end('A - B', ' - B'), 'A')
+ self.assertEqual(remove_end('B - A', ' - B'), 'B - A')
+
+ def test_remove_quotes(self):
+ self.assertEqual(remove_quotes(None), None)
+ self.assertEqual(remove_quotes('"'), '"')
+ self.assertEqual(remove_quotes("'"), "'")
+ self.assertEqual(remove_quotes(';'), ';')
+ self.assertEqual(remove_quotes('";'), '";')
+ self.assertEqual(remove_quotes('""'), '')
+ self.assertEqual(remove_quotes('";"'), ';')
+
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([]), [])
@@ -207,8 +249,17 @@ class TestUtil(unittest.TestCase):
self.assertEqual(unescapeHTML('%20;'), '%20;')
self.assertEqual(unescapeHTML('/'), '/')
self.assertEqual(unescapeHTML('/'), '/')
- self.assertEqual(
- unescapeHTML('é'), 'é')
+ self.assertEqual(unescapeHTML('é'), 'é')
+ self.assertEqual(unescapeHTML(''), '')
+ # HTML5 entities
+ self.assertEqual(unescapeHTML('.''), '.\'')
+
+ def test_date_from_str(self):
+ self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
+ self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
+ self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
+ self.assertEqual(date_from_str('now+365day'), date_from_str('now+1year'))
+ self.assertEqual(date_from_str('now+30day'), date_from_str('now+1month'))
def test_daterange(self):
_20century = DateRange("19000101", "20000101")
@@ -232,7 +283,36 @@ class TestUtil(unittest.TestCase):
self.assertEqual(
unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
'20150202')
+ self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
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)
+
+ def test_unified_timestamps(self):
+ self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
+ self.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
+ self.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
+ self.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
+ self.assertEqual(unified_timestamp('1968 12 10'), -33436800)
+ self.assertEqual(unified_timestamp('1968-12-10'), -33436800)
+ self.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
+ self.assertEqual(
+ unified_timestamp('11/26/2014 11:30:00 AM PST', day_first=False),
+ 1417001400)
+ self.assertEqual(
+ unified_timestamp('2/2/2015 6:47:40 PM', day_first=False),
+ 1422902860)
+ self.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
+ self.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
+ self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
+ self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
+
+ def test_determine_ext(self):
+ self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
+ self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
+ self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
+ self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
+ self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
def test_find_xpath_attr(self):
testxml = ''' Foo
Foo
''' - doc = xml.etree.ElementTree.fromstring(testxml) + doc = compat_etree_fromstring(testxml) self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a') self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None) self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None) @@ -365,6 +452,7 @@ class TestUtil(unittest.TestCase): self.assertEqual(parse_duration('01:02:03:04'), 93784) self.assertEqual(parse_duration('1 hour 3 minutes'), 3780) self.assertEqual(parse_duration('87 Min.'), 5220) + self.assertEqual(parse_duration('PT1H0.040S'), 3600.04) def test_fix_xml_ampersands(self): self.assertEqual( @@ -404,9 +492,6 @@ class TestUtil(unittest.TestCase): 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 @@ -420,11 +505,73 @@ class TestUtil(unittest.TestCase): data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'}) self.assertTrue(isinstance(data, bytes)) + def test_update_url_query(self): + def query_dict(url): + return compat_parse_qs(compat_urlparse.urlparse(url).query) + self.assertEqual(query_dict(update_url_query( + 'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})), + query_dict('http://example.com/path?quality=HD&format=mp4')) + self.assertEqual(query_dict(update_url_query( + 'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})), + query_dict('http://example.com/path?system=LINUX&system=WINDOWS')) + self.assertEqual(query_dict(update_url_query( + 'http://example.com/path', {'fields': 'id,formats,subtitles'})), + query_dict('http://example.com/path?fields=id,formats,subtitles')) + self.assertEqual(query_dict(update_url_query( + 'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})), + query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails')) + self.assertEqual(query_dict(update_url_query( + 'http://example.com/path?manifest=f4m', {'manifest': []})), + query_dict('http://example.com/path')) + self.assertEqual(query_dict(update_url_query( + 'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})), + query_dict('http://example.com/path?system=LINUX')) + self.assertEqual(query_dict(update_url_query( + 'http://example.com/path', {'fields': b'id,formats,subtitles'})), + query_dict('http://example.com/path?fields=id,formats,subtitles')) + self.assertEqual(query_dict(update_url_query( + 'http://example.com/path', {'width': 1080, 'height': 720})), + query_dict('http://example.com/path?width=1080&height=720')) + self.assertEqual(query_dict(update_url_query( + 'http://example.com/path', {'bitrate': 5020.43})), + query_dict('http://example.com/path?bitrate=5020.43')) + self.assertEqual(query_dict(update_url_query( + '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_dict_get(self): + FALSE_VALUES = { + 'none': None, + 'false': False, + 'zero': 0, + 'empty_string': '', + 'empty_list': [], + } + d = FALSE_VALUES.copy() + d['a'] = 42 + self.assertEqual(dict_get(d, 'a'), 42) + self.assertEqual(dict_get(d, 'b'), None) + self.assertEqual(dict_get(d, 'b', 42), 42) + self.assertEqual(dict_get(d, ('a', )), 42) + self.assertEqual(dict_get(d, ('b', 'a', )), 42) + self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42) + self.assertEqual(dict_get(d, ('b', 'c', )), None) + self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42) + for key, false_value in FALSE_VALUES.items(): + self.assertEqual(dict_get(d, ('b', 'c', key, )), None) + self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value) + + def test_encode_compat_str(self): + self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'ÑеÑÑ') + self.assertEqual(encode_compat_str('ÑеÑÑ', 'utf-8'), 'ÑеÑÑ') + 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) self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266) + self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251) + self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None) def test_strip_jsonp(self): stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);') @@ -435,6 +582,10 @@ class TestUtil(unittest.TestCase): d = json.loads(stripped) self.assertEqual(d, {'STATUS': 'OK'}) + stripped = strip_jsonp('ps.embedHandler({"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') @@ -471,11 +622,11 @@ class TestUtil(unittest.TestCase): ) self.assertEqual( escape_url('http://ÑеÑÑ.ÑÑ/ÑÑагменÑ'), - 'http://ÑеÑÑ.ÑÑ/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82' + 'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82' ) self.assertEqual( escape_url('http://ÑеÑÑ.ÑÑ/абв?абв=абв#абв'), - 'http://ÑеÑÑ.ÑÑ/%D0%B0%D0%B1%D0%B2?%D0%B0%D0%B1%D0%B2=%D0%B0%D0%B1%D0%B2#%D0%B0%D0%B1%D0%B2' + 'http://xn--e1aybc.xn--p1ai/%D0%B0%D0%B1%D0%B2?%D0%B0%D0%B1%D0%B2=%D0%B0%D0%B1%D0%B2#%D0%B0%D0%B1%D0%B2' ) self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0') @@ -502,6 +653,18 @@ class TestUtil(unittest.TestCase): json_code = js_to_json(inp) self.assertEqual(json.loads(json_code), json.loads(inp)) + inp = '''{ + 0:{src:'skipped', type: 'application/dash+xml'}, + 1:{src:'skipped', type: 'application/vnd.apple.mpegURL'}, + }''' + self.assertEqual(js_to_json(inp), '''{ + "0":{"src":"skipped", "type": "application/dash+xml"}, + "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"} + }''') + + inp = '''{"foo":101}''' + self.assertEqual(js_to_json(inp), '''{"foo":101}''') + 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"}) @@ -525,6 +688,65 @@ class TestUtil(unittest.TestCase): on = js_to_json('{"abc": "def",}') self.assertEqual(json.loads(on), {'abc': 'def'}) + on = js_to_json('{ 0: /* " \n */ ",]" , }') + self.assertEqual(json.loads(on), {'0': ',]'}) + + on = js_to_json(r'["x<\/p>"]') + self.assertEqual(json.loads(on), ['
x
']) + + on = js_to_json(r'["\xaa"]') + self.assertEqual(json.loads(on), ['\u00aa']) + + on = js_to_json("['a\\\nb']") + self.assertEqual(json.loads(on), ['ab']) + + on = js_to_json('{0xff:0xff}') + self.assertEqual(json.loads(on), {'255': 255}) + + on = js_to_json('{077:077}') + self.assertEqual(json.loads(on), {'63': 63}) + + on = js_to_json('{42:42}') + self.assertEqual(json.loads(on), {'42': 42}) + + def test_extract_attributes(self): + self.assertEqual(extract_attributes('The following line contains Chinese characters and special symbols
第äºè¡
âªâª
Third
Line
Lines with invalid timestamps are ignored
+Ignore, two
+Ignored, three