from __future__ import unicode_literals
+import base64
import calendar
import codecs
import contextlib
from .compat import (
compat_basestring,
compat_chr,
+ compat_etree_fromstring,
compat_html_entities,
compat_http_client,
compat_kwargs,
if sys.version_info >= (2, 7):
- def find_xpath_attr(node, xpath, key, val):
+ def find_xpath_attr(node, xpath, key, val=None):
""" Find the xpath xpath[@key=val] """
- assert re.match(r'^[a-zA-Z-]+$', key)
- assert re.match(r'^[a-zA-Z0-9@\s:._-]*$', val)
- expr = xpath + "[@%s='%s']" % (key, val)
+ assert re.match(r'^[a-zA-Z_-]+$', key)
+ if val:
+ assert re.match(r'^[a-zA-Z0-9@\s:._-]*$', val)
+ expr = xpath + ('[@%s]' % key if val is None else "[@%s='%s']" % (key, val))
return node.find(expr)
else:
- def find_xpath_attr(node, xpath, key, val):
+ def find_xpath_attr(node, xpath, key, val=None):
# Here comes the crazy part: In 2.6, if the xpath is a unicode,
# .//node does not match if a node is a direct child of . !
if isinstance(xpath, compat_str):
xpath = xpath.encode('ascii')
for f in node.findall(xpath):
- if f.attrib.get(key) == val:
+ if key not in f.attrib:
+ continue
+ if val is None or f.attrib.get(key) == val:
return f
return None
return '/'.join(replaced)
-def xpath_text(node, xpath, name=None, fatal=False, default=NO_DEFAULT):
- if sys.version_info < (2, 7): # Crazy 2.6
- xpath = xpath.encode('ascii')
+def xpath_element(node, xpath, name=None, fatal=False, default=NO_DEFAULT):
+ def _find_xpath(xpath):
+ if sys.version_info < (2, 7): # Crazy 2.6
+ xpath = xpath.encode('ascii')
+ return node.find(xpath)
- n = node.find(xpath)
- if n is None or n.text is None:
+ if isinstance(xpath, (str, compat_str)):
+ n = _find_xpath(xpath)
+ else:
+ for xp in xpath:
+ n = _find_xpath(xp)
+ if n is not None:
+ break
+
+ if n is None:
if default is not NO_DEFAULT:
return default
elif fatal:
raise ExtractorError('Could not find XML element %s' % name)
else:
return None
+ return n
+
+
+def xpath_text(node, xpath, name=None, fatal=False, default=NO_DEFAULT):
+ n = xpath_element(node, xpath, name, fatal=fatal, default=default)
+ if n is None or n == default:
+ return n
+ if n.text is None:
+ if default is not NO_DEFAULT:
+ return default
+ elif fatal:
+ name = xpath if name is None else name
+ raise ExtractorError('Could not find XML element\'s text %s' % name)
+ else:
+ return None
return n.text
+def xpath_attr(node, xpath, key, name=None, fatal=False, default=NO_DEFAULT):
+ n = find_xpath_attr(node, xpath, key)
+ if n is None:
+ if default is not NO_DEFAULT:
+ return default
+ elif fatal:
+ name = '%s[@%s]' % (xpath, key) if name is None else name
+ raise ExtractorError('Could not find XML attribute %s' % name)
+ else:
+ return None
+ return n.attrib[key]
+
+
def get_element_by_id(id, html):
"""Return the content of the tag with the specified ID in the passed HTML document"""
return get_element_by_attribute("id", id, html)
if drive_or_unc:
norm_path.pop(0)
sanitized_path = [
- path_part if path_part in ['.', '..'] else re.sub('(?:[/<>:"\\|\\\\?\\*]|\.$)', '#', path_part)
+ path_part if path_part in ['.', '..'] else re.sub('(?:[/<>:"\\|\\\\?\\*]|[\s.]$)', '#', path_part)
for path_part in norm_path]
if drive_or_unc:
sanitized_path.insert(0, drive_or_unc + os.path.sep)
return os.path.join(*sanitized_path)
+# Prepend protocol-less URLs with `http:` scheme in order to mitigate the number of
+# unwanted failures due to missing protocol
+def sanitized_Request(url, *args, **kwargs):
+ return compat_urllib_request.Request(
+ 'http:%s' % url if url.startswith('//') else url, *args, **kwargs)
+
+
def orderedSet(iterable):
""" Remove all duplicates from the input iterable """
res = []
numstr = '0%s' % numstr
else:
base = 10
- return compat_chr(int(numstr, base))
+ # See https://github.com/rg3/youtube-dl/issues/7518
+ try:
+ return compat_chr(int(numstr, base))
+ except ValueError:
+ pass
# Unknown entity in name, return its literal representation
- return ('&%s;' % entity)
+ return '&%s;' % entity
def unescapeHTML(s):
download is too small for what the server announced first, indicating
the connection was probably interrupted.
"""
- # Both in bytes
- downloaded = None
- expected = None
def __init__(self, downloaded, expected):
+ # Both in bytes
self.downloaded = downloaded
self.expected = expected
def _create_http_connection(ydl_handler, http_class, is_https, *args, **kwargs):
+ # Working around python 2 bug (see http://bugs.python.org/issue17849) by limiting
+ # expected HTTP responses to meet HTTP/1.0 or later (see also
+ # https://github.com/rg3/youtube-dl/issues/6727)
+ if sys.version_info < (3, 0):
+ kwargs[b'strict'] = True
hc = http_class(*args, **kwargs)
source_address = ydl_handler._params.get('source_address')
if source_address is not None:
return ret
def http_request(self, req):
+ # According to RFC 3986, URLs can not contain non-ASCII characters, however this is not
+ # always respected by websites, some tend to give out URLs with non percent-encoded
+ # non-ASCII characters (see telemb.py, ard.py [#3412])
+ # urllib chokes on URLs with non-ASCII characters (see http://bugs.python.org/issue3991)
+ # To work around aforementioned issue we will replace request's original URL with
+ # percent-encoded one
+ # Since redirects are also affected (e.g. http://www.southpark.de/alle-episoden/s18e09)
+ # the code of this workaround has been moved here from YoutubeDL.urlopen()
+ url = req.get_full_url()
+ url_escaped = escape_url(url)
+
+ # Substitute URL if any change after escaping
+ if url != url_escaped:
+ req_type = HEADRequest if req.get_method() == 'HEAD' else compat_urllib_request.Request
+ new_req = req_type(
+ url_escaped, data=req.data, headers=req.headers,
+ origin_req_host=req.origin_req_host, unverifiable=req.unverifiable)
+ new_req.timeout = req.timeout
+ req = new_req
+
for h, v in std_headers.items():
# Capitalize is needed because of Python bug 2275: http://bugs.python.org/issue2275
# The dict keys are capitalized because of this bug by urllib
gz = io.BytesIO(self.deflate(resp.read()))
resp = self.addinfourl_wrapper(gz, old_resp.headers, old_resp.url, old_resp.code)
resp.msg = old_resp.msg
+ # Percent-encode redirect URL of Location HTTP header to satisfy RFC 3986 (see
+ # https://github.com/rg3/youtube-dl/issues/6457).
+ if 300 <= resp.code < 400:
+ location = resp.headers.get('Location')
+ if location:
+ # As of RFC 2616 default charset is iso-8859-1 that is respected by python 3
+ if sys.version_info >= (3, 0):
+ location = location.encode('iso-8859-1').decode('utf-8')
+ location_escaped = escape_url(location)
+ if location != location_escaped:
+ del resp.headers['Location']
+ resp.headers['Location'] = location_escaped
return resp
https_request = http_request
req, **kwargs)
+class YoutubeDLCookieProcessor(compat_urllib_request.HTTPCookieProcessor):
+ def __init__(self, cookiejar=None):
+ compat_urllib_request.HTTPCookieProcessor.__init__(self, cookiejar)
+
+ def http_response(self, request, response):
+ # Python 2 will choke on next HTTP request in row if there are non-ASCII
+ # characters in Set-Cookie HTTP header of last response (see
+ # https://github.com/rg3/youtube-dl/issues/6769).
+ # In order to at least prevent crashing we will percent encode Set-Cookie
+ # header before HTTPCookieProcessor starts processing it.
+ # if sys.version_info < (3, 0) and response.headers:
+ # for set_cookie_header in ('Set-Cookie', 'Set-Cookie2'):
+ # set_cookie = response.headers.get(set_cookie_header)
+ # if set_cookie:
+ # set_cookie_escaped = compat_urllib_parse.quote(set_cookie, b"%/;:@&=+$,!~*'()?#[] ")
+ # if set_cookie != set_cookie_escaped:
+ # del response.headers[set_cookie_header]
+ # response.headers[set_cookie_header] = set_cookie_escaped
+ return compat_urllib_request.HTTPCookieProcessor.http_response(self, request, response)
+
+ https_request = compat_urllib_request.HTTPCookieProcessor.http_request
+ https_response = http_response
+
+
def parse_iso8601(date_str, delimiter='T', timezone=None):
""" Return a UNIX timestamp from the given date """
if date_str is None:
return None
+ date_str = re.sub(r'\.[0-9]+', '', date_str)
+
if timezone is None:
m = re.search(
- r'(\.[0-9]+)?(?:Z$| ?(?P<sign>\+|-)(?P<hours>[0-9]{2}):?(?P<minutes>[0-9]{2})$)',
+ r'(?:Z$| ?(?P<sign>\+|-)(?P<hours>[0-9]{2}):?(?P<minutes>[0-9]{2})$)',
date_str)
if not m:
timezone = datetime.timedelta()
timezone = datetime.timedelta(
hours=sign * int(m.group('hours')),
minutes=sign * int(m.group('minutes')))
- date_format = '%Y-%m-%d{0}%H:%M:%S'.format(delimiter)
- dt = datetime.datetime.strptime(date_str, date_format) - timezone
- return calendar.timegm(dt.timetuple())
+ try:
+ date_format = '%Y-%m-%d{0}%H:%M:%S'.format(delimiter)
+ dt = datetime.datetime.strptime(date_str, date_format) - timezone
+ return calendar.timegm(dt.timetuple())
+ except ValueError:
+ pass
def unified_strdate(date_str, day_first=True):
timetuple = email.utils.parsedate_tz(date_str)
if timetuple:
upload_date = datetime.datetime(*timetuple[:6]).strftime('%Y%m%d')
- return upload_date
+ if upload_date is not None:
+ return compat_str(upload_date)
def determine_ext(url, default_ext='unknown_video'):
guess = url.partition('?')[0].rpartition('.')[2]
if re.match(r'^[A-Za-z0-9]+$', guess):
return guess
+ elif guess.rstrip('/') in (
+ 'mp4', 'm4a', 'm4p', 'm4b', 'm4r', 'm4v', 'aac',
+ 'flv', 'f4v', 'f4a', 'f4b',
+ 'webm', 'ogg', 'ogv', 'oga', 'ogx', 'spx', 'opus',
+ 'mkv', 'mka', 'mk3d',
+ 'avi', 'divx',
+ 'mov',
+ 'asf', 'wmv', 'wma',
+ '3gp', '3g2',
+ 'mp3',
+ 'flac',
+ 'ape',
+ 'wav',
+ 'f4f', 'f4m', 'm3u8', 'smil'):
+ return guess.rstrip('/')
else:
return default_ext
v = getattr(v, get_attr, None)
if v == '':
v = None
- return default if v is None else (int(v) * invscale // scale)
+ if v is None:
+ return default
+ try:
+ return int(v) * invscale // scale
+ except ValueError:
+ return default
def str_or_none(v, default=None):
def float_or_none(v, scale=1, invscale=1, default=None):
- return default if v is None else (float(v) * invscale / scale)
+ if v is None:
+ return default
+ try:
+ return float(v) * invscale / scale
+ except ValueError:
+ return default
def parse_duration(s):
m = re.match(
r'''(?ix)(?:P?T)?
(?:
- (?P<only_mins>[0-9.]+)\s*(?:mins?|minutes?)\s*|
+ (?P<only_mins>[0-9.]+)\s*(?:mins?\.?|minutes?)\s*|
(?P<only_hours>[0-9.]+)\s*(?:hours?)|
- \s*(?P<hours_reversed>[0-9]+)\s*(?:[:h]|hours?)\s*(?P<mins_reversed>[0-9]+)\s*(?:[:m]|mins?|minutes?)\s*|
+ \s*(?P<hours_reversed>[0-9]+)\s*(?:[:h]|hours?)\s*(?P<mins_reversed>[0-9]+)\s*(?:[:m]|mins?\.?|minutes?)\s*|
(?:
(?:
(?:(?P<days>[0-9]+)\s*(?:[:d]|days?)\s*)?
return compat_urllib_parse.urlencode(*args, **kargs).encode('ascii')
-try:
- etree_iter = xml.etree.ElementTree.Element.iter
-except AttributeError: # Python <=2.6
- etree_iter = lambda n: n.findall('.//*')
-
-
-def parse_xml(s):
- class TreeBuilder(xml.etree.ElementTree.TreeBuilder):
- def doctype(self, name, pubid, system):
- pass # Ignore doctypes
-
- parser = xml.etree.ElementTree.XMLParser(target=TreeBuilder())
- kwargs = {'parser': parser} if sys.version_info >= (2, 7) else {}
- tree = xml.etree.ElementTree.XML(s.encode('utf-8'), **kwargs)
- # Fix up XML parser in Python 2.x
- if sys.version_info < (3, 0):
- for n in etree_iter(tree):
- if n.text is not None:
- if not isinstance(n.text, compat_str):
- n.text = n.text.decode('utf-8')
- return tree
+def encode_dict(d, encoding='utf-8'):
+ def encode(v):
+ return v.encode(encoding) if isinstance(v, compat_basestring) else v
+ return dict((encode(k), encode(v)) for k, v in d.items())
US_RATINGS = {
if v in ('true', 'false', 'null'):
return v
if v.startswith('"'):
- return v
- if v.startswith("'"):
+ v = re.sub(r"\\'", "'", v[1:-1])
+ elif v.startswith("'"):
v = v[1:-1]
v = re.sub(r"\\\\|\\'|\"", lambda m: {
'\\\\': '\\\\',
return mimetype2ext(getheader('Content-Type'))
+def encode_data_uri(data, mime_type):
+ return 'data:%s;base64,%s' % (mime_type, base64.b64encode(data).decode('ascii'))
+
+
def age_restricted(content_limit, age_limit):
""" Returns True iff the content should be blocked """
return out
- dfxp = xml.etree.ElementTree.fromstring(dfxp_data.encode('utf-8'))
+ dfxp = compat_etree_fromstring(dfxp_data.encode('utf-8'))
out = []
paras = dfxp.findall(_x('.//ttml:p')) or dfxp.findall(_x('.//ttaf1:p')) or dfxp.findall('.//p')
return ''.join(out)
+def cli_option(params, command_option, param):
+ param = params.get(param)
+ return [command_option, param] if param is not None else []
+
+
+def cli_bool_option(params, command_option, param, true_value='true', false_value='false', separator=None):
+ param = params.get(param)
+ assert isinstance(param, bool)
+ if separator:
+ return [command_option + separator + (true_value if param else false_value)]
+ return [command_option, true_value if param else false_value]
+
+
+def cli_valueless_option(params, command_option, param, expected_value=True):
+ param = params.get(param)
+ return [command_option] if param == expected_value else []
+
+
+def cli_configuration_args(params, param, default=[]):
+ ex_args = params.get(param)
+ if ex_args is None:
+ return default
+ assert isinstance(ex_args, list)
+ return ex_args
+
+
class ISO639Utils(object):
# See http://www.loc.gov/standards/iso639-2/ISO-639-2_utf-8.txt
_lang_map = {
Rapsys Git / youtubedl / blobdiff