You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

5571 lines
179 KiB

import atexit
import base64
import binascii
import calendar
import codecs
import collections
import contextlib
import ctypes
import datetime
import email.header
import email.utils
import errno
import gzip
import hashlib
import hmac
import html.entities
import html.parser
import http.client
import http.cookiejar
import importlib.util
import io
import itertools
import json
import locale
import math
import mimetypes
import operator
import os
import platform
import random
import re
import shlex
import socket
import ssl
import struct
import subprocess
import sys
import tempfile
import time
import traceback
import types
import urllib.error
import urllib.parse
import urllib.request
import xml.etree.ElementTree
import zlib
from .compat import asyncio, functools # isort: split
from .compat import (
from .dependencies import brotli, certifi, websockets, xattr
from .socks import ProxyType, sockssocket
def register_socks_protocols():
# "Register" SOCKS protocols
# In Python < 2.6.5, urlsplit() suffers from bug
# URLs with protocols not in urlparse.uses_netloc are not handled correctly
for scheme in ('socks', 'socks4', 'socks4a', 'socks5'):
if scheme not in urllib.parse.uses_netloc:
# This is not clearly defined otherwise
compiled_regex_type = type(re.compile(''))
def random_user_agent():
_USER_AGENT_TPL = 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/%s Safari/537.36'
return _USER_AGENT_TPL % random.choice(_CHROME_VERSIONS)
'gzip', 'deflate'
if brotli:
std_headers = {
'User-Agent': random_user_agent(),
'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',
'Accept-Language': 'en-us,en;q=0.5',
'Sec-Fetch-Mode': 'navigate',
'Safari': 'Mozilla/5.0 (X11; Linux x86_64; rv:10.0) AppleWebKit/533.20.25 (KHTML, like Gecko) Version/5.0.4 Safari/533.20.27',
NO_DEFAULT = object()
IDENTITY = lambda x: x
'January', 'February', 'March', 'April', 'May', 'June',
'July', 'August', 'September', 'October', 'November', 'December']
'fr': [
'janvier', 'février', 'mars', 'avril', 'mai', 'juin',
'juillet', 'août', 'septembre', 'octobre', 'novembre', 'décembre'],
'mp4', 'm4a', 'm4p', 'm4b', 'm4r', 'm4v', 'aac',
'flv', 'f4v', 'f4a', 'f4b',
'webm', 'ogg', 'ogv', 'oga', 'ogx', 'spx', 'opus',
'mkv', 'mka', 'mk3d',
'avi', 'divx',
'asf', 'wmv', 'wma',
'3gp', '3g2',
'f4f', 'f4m', 'm3u8', 'smil')
# needed for sanitizing filenames in restricted mode
ACCENT_CHARS = dict(zip('ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ',
itertools.chain('AAAAAA', ['AE'], 'CEEEEIIIIDNOOOOOOO', ['OE'], 'UUUUUY', ['TH', 'ss'],
'aaaaaa', ['ae'], 'ceeeeiiiionooooooo', ['oe'], 'uuuuuy', ['th'], 'y')))
'%d %B %Y',
'%d %b %Y',
'%B %d %Y',
'%B %dst %Y',
'%B %dnd %Y',
'%B %drd %Y',
'%B %dth %Y',
'%b %d %Y',
'%b %dst %Y',
'%b %dnd %Y',
'%b %drd %Y',
'%b %dth %Y',
'%b %dst %Y %I:%M',
'%b %dnd %Y %I:%M',
'%b %drd %Y %I:%M',
'%b %dth %Y %I:%M',
'%Y %m %d',
'%Y/%m/%d %H:%M',
'%Y/%m/%d %H:%M:%S',
'%Y-%m-%d %H:%M',
'%Y-%m-%d %H:%M:%S',
'%Y-%m-%d %H:%M:%S.%f',
'%Y-%m-%d %H:%M:%S:%f',
'%d.%m.%Y %H:%M',
'%d.%m.%Y %H.%M',
'%b %d %Y at %H:%M',
'%b %d %Y at %H:%M:%S',
'%B %d %Y at %H:%M',
'%B %d %Y at %H:%M:%S',
'%H:%M %d-%b-%Y',
'%d/%m/%Y %H:%M:%S',
'%m/%d/%Y %H:%M:%S',
PACKED_CODES_RE = r"}\('(.+)',(\d+),(\d+),'([^']+)'\.split\('\|'\)"
JSON_LD_RE = r'(?is)<script[^>]+type=(["\']?)application/ld\+json\1[^>]*>(?P<json_ld>.+?)</script>'
NUMBER_RE = r'\d+(?:\.\d+)?'
def preferredencoding():
"""Get preferred encoding.
Returns the best encoding scheme for the system, based on
locale.getpreferredencoding() and some further tweaks.
pref = locale.getpreferredencoding()
except Exception:
pref = 'UTF-8'
return pref
def write_json_file(obj, fn):
""" Encode obj as JSON and write it to fn, atomically if possible """
tf = tempfile.NamedTemporaryFile(
prefix=f'{os.path.basename(fn)}.', dir=os.path.dirname(fn),
suffix='.tmp', delete=False, mode='w', encoding='utf-8')
with tf:
json.dump(obj, tf, ensure_ascii=False)
if sys.platform == 'win32':
# Need to remove existing file on Windows, else os.rename raises
# WindowsError or FileExistsError.
with contextlib.suppress(OSError):
with contextlib.suppress(OSError):
mask = os.umask(0)
os.chmod(, 0o666 & ~mask)
os.rename(, fn)
except Exception:
with contextlib.suppress(OSError):
def find_xpath_attr(node, xpath, key, val=None):
""" Find the xpath xpath[@key=val] """
assert re.match(r'^[a-zA-Z_-]+$', key)
expr = xpath + ('[@%s]' % key if val is None else f"[@{key}='{val}']")
return node.find(expr)
# On python2.6 the xml.etree.ElementTree.Element methods don't support
# the namespace parameter
def xpath_with_ns(path, ns_map):
components = [c.split(':') for c in path.split('/')]
replaced = []
for c in components:
if len(c) == 1:
ns, tag = c
replaced.append('{%s}%s' % (ns_map[ns], tag))
return '/'.join(replaced)
def xpath_element(node, xpath, name=None, fatal=False, default=NO_DEFAULT):
def _find_xpath(xpath):
return node.find(xpath)
if isinstance(xpath, str):
n = _find_xpath(xpath)
for xp in xpath:
n = _find_xpath(xp)
if n is not None:
if n 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' % name)
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)
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 = f'{xpath}[@{key}]' if name is None else name
raise ExtractorError('Could not find XML attribute %s' % name)
return None
return n.attrib[key]
def get_element_by_id(id, html, **kwargs):
"""Return the content of the tag with the specified ID in the passed HTML document"""
return get_element_by_attribute('id', id, html, **kwargs)
def get_element_html_by_id(id, html, **kwargs):
"""Return the html of the tag with the specified ID in the passed HTML document"""
return get_element_html_by_attribute('id', id, html, **kwargs)
def get_element_by_class(class_name, html):
"""Return the content of the first tag with the specified class in the passed HTML document"""
retval = get_elements_by_class(class_name, html)
return retval[0] if retval else None
def get_element_html_by_class(class_name, html):
"""Return the html of the first tag with the specified class in the passed HTML document"""
retval = get_elements_html_by_class(class_name, html)
return retval[0] if retval else None
def get_element_by_attribute(attribute, value, html, **kwargs):
retval = get_elements_by_attribute(attribute, value, html, **kwargs)
return retval[0] if retval else None
def get_element_html_by_attribute(attribute, value, html, **kargs):
retval = get_elements_html_by_attribute(attribute, value, html, **kargs)
return retval[0] if retval else None
def get_elements_by_class(class_name, html, **kargs):
"""Return the content of all tags with the specified class in the passed HTML document as a list"""
return get_elements_by_attribute(
'class', r'[^\'"]*(?<=[\'"\s])%s(?=[\'"\s])[^\'"]*' % re.escape(class_name),
html, escape_value=False)
def get_elements_html_by_class(class_name, html):
"""Return the html of all tags with the specified class in the passed HTML document as a list"""
return get_elements_html_by_attribute(
'class', r'[^\'"]*(?<=[\'"\s])%s(?=[\'"\s])[^\'"]*' % re.escape(class_name),
html, escape_value=False)
def get_elements_by_attribute(*args, **kwargs):
"""Return the content of the tag with the specified attribute in the passed HTML document"""
return [content for content, _ in get_elements_text_and_html_by_attribute(*args, **kwargs)]
def get_elements_html_by_attribute(*args, **kwargs):
"""Return the html of the tag with the specified attribute in the passed HTML document"""
return [whole for _, whole in get_elements_text_and_html_by_attribute(*args, **kwargs)]
def get_elements_text_and_html_by_attribute(attribute, value, html, escape_value=True):
Return the text (content) and the html (whole) of the tag with the specified
attribute in the passed HTML document
quote = '' if re.match(r'''[\s"'`=<>]''', value) else '?'
value = re.escape(value) if escape_value else value
partial_element_re = rf'''(?x)
for m in re.finditer(partial_element_re, html):
content, whole = get_element_text_and_html_by_tag('tag'), html[m.start():])
yield (
unescapeHTML(re.sub(r'^(?P<q>["\'])(?P<content>.*)(?P=q)$', r'\g<content>', content, flags=re.DOTALL)),
class HTMLBreakOnClosingTagParser(html.parser.HTMLParser):
HTML parser which raises HTMLBreakOnClosingTagException upon reaching the
closing tag for the first opening tag it has encountered, and can be used
as a context manager
class HTMLBreakOnClosingTagException(Exception):
def __init__(self):
self.tagstack = collections.deque()
def __enter__(self):
return self
def __exit__(self, *_):
def close(self):
# handle_endtag does not return upon raising HTMLBreakOnClosingTagException,
# so data remains buffered; we no longer have any interest in it, thus
# override this method to discard it
def handle_starttag(self, tag, _):
def handle_endtag(self, tag):
if not self.tagstack:
raise compat_HTMLParseError('no tags in the stack')
while self.tagstack:
inner_tag = self.tagstack.pop()
if inner_tag == tag:
raise compat_HTMLParseError(f'matching opening tag for closing {tag} tag not found')
if not self.tagstack:
raise self.HTMLBreakOnClosingTagException()
def get_element_text_and_html_by_tag(tag, html):
For the first element with the specified tag in the passed HTML document
return its' content (text) and the whole element (html)
def find_or_raise(haystack, needle, exc):
return haystack.index(needle)
except ValueError:
raise exc
closing_tag = f'</{tag}>'
whole_start = find_or_raise(
html, f'<{tag}', compat_HTMLParseError(f'opening {tag} tag not found'))
content_start = find_or_raise(
html[whole_start:], '>', compat_HTMLParseError(f'malformed opening {tag} tag'))
content_start += whole_start + 1
with HTMLBreakOnClosingTagParser() as parser:
if not parser.tagstack or parser.tagstack[0] != tag:
raise compat_HTMLParseError(f'parser did not match opening {tag} tag')
offset = content_start
while offset < len(html):
next_closing_tag_start = find_or_raise(
html[offset:], closing_tag,
compat_HTMLParseError(f'closing {tag} tag not found'))
next_closing_tag_end = next_closing_tag_start + len(closing_tag)
parser.feed(html[offset:offset + next_closing_tag_end])
offset += next_closing_tag_end
except HTMLBreakOnClosingTagParser.HTMLBreakOnClosingTagException:
return html[content_start:offset + next_closing_tag_start], \
html[whole_start:offset + next_closing_tag_end]
raise compat_HTMLParseError('unexpected end of html')
class HTMLAttributeParser(html.parser.HTMLParser):
"""Trivial HTML parser to gather the attributes for a single element"""
def __init__(self):
self.attrs = {}
def handle_starttag(self, tag, attrs):
self.attrs = dict(attrs)
class HTMLListAttrsParser(html.parser.HTMLParser):
"""HTML parser to gather the attributes for the elements of a list"""
def __init__(self):
self.items = []
self._level = 0
def handle_starttag(self, tag, attrs):
if tag == 'li' and self._level == 0:
self._level += 1
def handle_endtag(self, tag):
self._level -= 1
def extract_attributes(html_element):
"""Given a string for an HTML element such as
a="foo" B="bar" c="&98;az" d=boz
empty= noval entity="&amp;"
sq='"' dq="'"
Decode and return a dictionary of attributes.
'a': 'foo', 'b': 'bar', c: 'baz', d: 'boz',
'empty': '', 'noval': None, 'entity': '&',
'sq': '"', 'dq': '\''
parser = HTMLAttributeParser()
with contextlib.suppress(compat_HTMLParseError):
return parser.attrs
def parse_list(webpage):
"""Given a string for an series of HTML <li> elements,
return a dictionary of their attributes"""
parser = HTMLListAttrsParser()
return parser.items
def clean_html(html):
"""Clean an HTML snippet into a readable string"""
if html is None: # Convenience for sanitizing descriptions etc.
return html
html = re.sub(r'\s+', ' ', html)
html = re.sub(r'(?u)\s?<\s?br\s?/?\s?>\s?', '\n', html)
html = re.sub(r'(?u)<\s?/\s?p\s?>\s?<\s?p[^>]*>', '\n', html)
# Strip html tags
html = re.sub('<.*?>', '', html)
# Replace html entities
html = unescapeHTML(html)
return html.strip()
class LenientJSONDecoder(json.JSONDecoder):
def __init__(self, *args, transform_source=None, ignore_extra=False, **kwargs):
self.transform_source, self.ignore_extra = transform_source, ignore_extra
super().__init__(*args, **kwargs)
def decode(self, s):
if self.transform_source:
s = self.transform_source(s)
if self.ignore_extra:
return self.raw_decode(s.lstrip())[0]
return super().decode(s)
def sanitize_open(filename, open_mode):
"""Try to open the given filename, and slightly tweak it if this fails.
Attempts to open the given filename. If this fails, it tries to change
the filename slightly, step by step, until it's either able to open it
or it fails and raises a final exception, like the standard open()
It returns the tuple (stream, definitive_file_name).
if filename == '-':
if sys.platform == 'win32':
import msvcrt
msvcrt.setmode(sys.stdout.fileno(), os.O_BINARY)
return (sys.stdout.buffer if hasattr(sys.stdout, 'buffer') else sys.stdout, filename)
for attempt in range(2):
if sys.platform == 'win32':
# FIXME: An exclusive lock also locks the file from being read.
# Since windows locks are mandatory, don't lock the file on windows (for now).
# Ref:
raise LockingUnsupportedError()
stream = locked_file(filename, open_mode, block=False).__enter__()
except OSError:
stream = open(filename, open_mode)
return stream, filename
except OSError as err:
if attempt or err.errno in (errno.EACCES,):
old_filename, filename = filename, sanitize_path(filename)
if old_filename == filename:
def timeconvert(timestr):
"""Convert RFC 2822 defined time string into system timestamp"""
timestamp = None
timetuple = email.utils.parsedate_tz(timestr)
if timetuple is not None:
timestamp = email.utils.mktime_tz(timetuple)
return timestamp
def sanitize_filename(s, restricted=False, is_id=NO_DEFAULT):
"""Sanitizes a string so it could be used as part of a filename.
@param restricted Use a stricter subset of allowed characters
@param is_id Whether this is an ID that should be kept unchanged if possible.
If unset, yt-dlp's new sanitization rules are in effect
if s == '':
return ''
def replace_insane(char):
if restricted and char in ACCENT_CHARS:
return ACCENT_CHARS[char]
elif not restricted and char == '\n':
return '\0 '
elif char == '?' or ord(char) < 32 or ord(char) == 127:
return ''
elif char == '"':
return '' if restricted else '\''
elif char == ':':
return '\0_\0-' if restricted else '\0 \0-'
elif char in '\\/|*<>':
return '\0_'
if restricted and (char in '!&\'()[]{}$;`^,#' or char.isspace() or ord(char) > 127):
return '\0_'
return char
s = re.sub(r'[0-9]+(?::[0-9]+)+', lambda m:':', '_'), s) # Handle timestamps
result = ''.join(map(replace_insane, s))
if is_id is NO_DEFAULT:
result = re.sub('(\0.)(?:(?=\\1)..)+', r'\1', result) # Remove repeated substitute chars
STRIP_RE = '(?:\0.|[ _-])*'
result = re.sub(f'^\0.{STRIP_RE}|{STRIP_RE}\0.$', '', result) # Remove substitute chars from start/end
result = result.replace('\0', '') or '_'
if not is_id:
while '__' in result:
result = result.replace('__', '_')
result = result.strip('_')
# Common case of "Foreign band name - English song title"
if restricted and result.startswith('-_'):
result = result[2:]
if result.startswith('-'):
result = '_' + result[len('-'):]
result = result.lstrip('.')
if not result:
result = '_'
return result
def sanitize_path(s, force=False):
"""Sanitizes and normalizes path on Windows"""
if sys.platform == 'win32':
force = False
drive_or_unc, _ = os.path.splitdrive(s)
elif force:
drive_or_unc = ''
return s
norm_path = os.path.normpath(remove_start(s, drive_or_unc)).split(os.path.sep)
if drive_or_unc:
sanitized_path = [
path_part if path_part in ['.', '..'] else re.sub(r'(?:[/<>:"\|\\?\*]|[\s.]$)', '#', path_part)
for path_part in norm_path]
if drive_or_unc:
sanitized_path.insert(0, drive_or_unc + os.path.sep)
elif force and s and s[0] == os.path.sep:
sanitized_path.insert(0, os.path.sep)
return os.path.join(*sanitized_path)
def sanitize_url(url):
# Prepend protocol-less URLs with `http:` scheme in order to mitigate
# the number of unwanted failures due to missing protocol
if url is None:
elif url.startswith('//'):
return 'http:%s' % url
# Fix some common typos seen so far
(r'^httpss://', r'https://'),
(r'^rmtp([es]?)://', r'rtmp\1://'),
for mistake, fixup in COMMON_TYPOS:
if re.match(mistake, url):
return re.sub(mistake, fixup, url)
return url
def extract_basic_auth(url):
parts = urllib.parse.urlsplit(url)
if parts.username is None:
return url, None
url = urllib.parse.urlunsplit(parts._replace(netloc=(
parts.hostname if parts.port is None
else '%s:%d' % (parts.hostname, parts.port))))
auth_payload = base64.b64encode(
('%s:%s' % (parts.username, parts.password or '')).encode())
return url, f'Basic {auth_payload.decode()}'
def sanitized_Request(url, *args, **kwargs):
url, auth_header = extract_basic_auth(escape_url(sanitize_url(url)))
if auth_header is not None:
headers = args[1] if len(args) >= 2 else kwargs.setdefault('headers', {})
headers['Authorization'] = auth_header
return urllib.request.Request(url, *args, **kwargs)
def expand_path(s):
"""Expand shell variables and ~"""
return os.path.expandvars(compat_expanduser(s))
def orderedSet(iterable, *, lazy=False):
"""Remove all duplicates from the input iterable"""
def _iter():
seen = [] # Do not use set since the items can be unhashable
for x in iterable:
if x not in seen:
yield x
return _iter() if lazy else list(_iter())
def _htmlentity_transform(entity_with_semicolon):
"""Transforms an HTML entity to a character."""
entity = entity_with_semicolon[:-1]
# Known non-numeric HTML entity
if entity in html.entities.name2codepoint:
return chr(html.entities.name2codepoint[entity])
# TODO: HTML5 allows entities without a semicolon. For example,
# '&Eacuteric' should be decoded as 'Éric'.
if entity_with_semicolon in html.entities.html5:
return html.entities.html5[entity_with_semicolon]
mobj = re.match(r'#(x[0-9a-fA-F]+|[0-9]+)', entity)
if mobj is not None:
numstr =
if numstr.startswith('x'):
base = 16
numstr = '0%s' % numstr
base = 10
# See
with contextlib.suppress(ValueError):
return chr(int(numstr, base))
# Unknown entity in name, return its literal representation
return '&%s;' % entity
def unescapeHTML(s):
if s is None:
return None
assert isinstance(s, str)
return re.sub(
r'&([^&;]+;)', lambda m: _htmlentity_transform(, s)
def escapeHTML(text):
return (
.replace('&', '&amp;')
.replace('<', '&lt;')
.replace('>', '&gt;')
.replace('"', '&quot;')
.replace("'", '&#39;')
def process_communicate_or_kill(p, *args, **kwargs):
write_string('DeprecationWarning: yt_dlp.utils.process_communicate_or_kill is deprecated '
'and may be removed in a future version. Use yt_dlp.utils.Popen.communicate_or_kill instead')
return Popen.communicate_or_kill(p, *args, **kwargs)
class Popen(subprocess.Popen):
if sys.platform == 'win32':
_startupinfo = subprocess.STARTUPINFO()
_startupinfo.dwFlags |= subprocess.STARTF_USESHOWWINDOW
_startupinfo = None
def __init__(self, *args, text=False, **kwargs):
if text is True:
kwargs['universal_newlines'] = True # For 3.6 compatibility
kwargs.setdefault('encoding', 'utf-8')
kwargs.setdefault('errors', 'replace')
super().__init__(*args, **kwargs, startupinfo=self._startupinfo)
def communicate_or_kill(self, *args, **kwargs):
return self.communicate(*args, **kwargs)
except BaseException: # Including KeyboardInterrupt
def kill(self, *, timeout=0):
if timeout != 0:
def run(cls, *args, **kwargs):
with cls(*args, **kwargs) as proc:
stdout, stderr = proc.communicate_or_kill()
return stdout or '', stderr or '', proc.returncode
def get_subprocess_encoding():
if sys.platform == 'win32' and sys.getwindowsversion()[0] >= 5:
# For subprocess calls, encode with locale encoding
# Refer to
encoding = preferredencoding()
encoding = sys.getfilesystemencoding()
if encoding is None:
encoding = 'utf-8'
return encoding
def encodeFilename(s, for_subprocess=False):
assert isinstance(s, str)
return s
def decodeFilename(b, for_subprocess=False):
return b
def encodeArgument(s):
# Legacy code that uses byte strings
# Uncomment the following line after fixing all post processors
# assert isinstance(s, str), 'Internal error: %r should be of type %r, is %r' % (s, str, type(s))
return s if isinstance(s, str) else s.decode('ascii')
def decodeArgument(b):
return b
def decodeOption(optval):
if optval is None:
return optval
if isinstance(optval, bytes):
optval = optval.decode(preferredencoding())
assert isinstance(optval, str)
return optval
_timetuple = collections.namedtuple('Time', ('hours', 'minutes', 'seconds', 'milliseconds'))
def timetuple_from_msec(msec):
secs, msec = divmod(msec, 1000)
mins, secs = divmod(secs, 60)
hrs, mins = divmod(mins, 60)
return _timetuple(hrs, mins, secs, msec)
def formatSeconds(secs, delim=':', msec=False):
time = timetuple_from_msec(secs * 1000)
if time.hours:
ret = '%d%s%02d%s%02d' % (time.hours, delim, time.minutes, delim, time.seconds)
elif time.minutes:
ret = '%d%s%02d' % (time.minutes, delim, time.seconds)
ret = '%d' % time.seconds
return '%s.%03d' % (ret, time.milliseconds) if msec else ret
def _ssl_load_windows_store_certs(ssl_context, storename):
# Code adapted from _load_windows_store_certs in
certs = [cert for cert, encoding, trust in ssl.enum_certificates(storename)
if encoding == 'x509_asn' and (
trust is True or ssl.Purpose.SERVER_AUTH.oid in trust)]
except PermissionError:
for cert in certs:
with contextlib.suppress(ssl.SSLError):
def make_HTTPS_handler(params, **kwargs):
opts_check_certificate = not params.get('nocheckcertificate')
context = ssl.SSLContext(ssl.PROTOCOL_TLS_CLIENT)
context.check_hostname = opts_check_certificate
if params.get('legacyserverconnect'):
context.options |= 4 # SSL_OP_LEGACY_SERVER_CONNECT
# Allow use of weaker ciphers in Python 3.10+. See
context.verify_mode = ssl.CERT_REQUIRED if opts_check_certificate else ssl.CERT_NONE
if opts_check_certificate:
if has_certifi and 'no-certifi' not in params.get('compat_opts', []):
# Work around the issue in load_default_certs when there are bad certificates. See:
except ssl.SSLError:
# enum_certificates is not present in mingw python. See
if sys.platform == 'win32' and hasattr(ssl, 'enum_certificates'):
for storename in ('CA', 'ROOT'):
_ssl_load_windows_store_certs(context, storename)
client_certfile = params.get('client_certificate')
if client_certfile:
client_certfile, keyfile=params.get('client_certificate_key'),
except ssl.SSLError:
raise YoutubeDLError('Unable to load client certificate')
# Some servers may reject requests if ALPN extension is not sent. See:
with contextlib.suppress(NotImplementedError):
return YoutubeDLHTTPSHandler(params, context=context, **kwargs)
def bug_reports_message(before=';'):
from .update import REPOSITORY
msg = (f'please report this issue on{REPOSITORY}/issues?q= , '
'filling out the appropriate issue template. Confirm you are on the latest version using yt-dlp -U')
before = before.rstrip()
if not before or before.endswith(('.', '!', '?')):
msg = msg[0].title() + msg[1:]
return (before + ' ' if before else '') + msg
class YoutubeDLError(Exception):
"""Base exception for YoutubeDL errors."""
msg = None
def __init__(self, msg=None):
if msg is not None:
self.msg = msg
elif self.msg is None:
self.msg = type(self).__name__
network_exceptions = [urllib.error.URLError, http.client.HTTPException, socket.error]
if hasattr(ssl, 'CertificateError'):
network_exceptions = tuple(network_exceptions)
class ExtractorError(YoutubeDLError):
"""Error during info extraction."""
def __init__(self, msg, tb=None, expected=False, cause=None, video_id=None, ie=None):
""" tb, if given, is the original traceback (so that it can be printed out).
If expected is set, this is a normal error message and most likely not a bug in yt-dlp.
if sys.exc_info()[0] in network_exceptions:
expected = True
self.orig_msg = str(msg)
self.traceback = tb
self.expected = expected
self.cause = cause
self.video_id = video_id = ie
self.exc_info = sys.exc_info() # preserve original exception
if isinstance(self.exc_info[1], ExtractorError):
self.exc_info = self.exc_info[1].exc_info
format_field(ie, None, '[%s] '),
format_field(video_id, None, '%s: '),
format_field(cause, None, ' (caused by %r)'),
'' if expected else bug_reports_message())))
def format_traceback(self):
return join_nonempty(
self.traceback and ''.join(traceback.format_tb(self.traceback)),
self.cause and ''.join(traceback.format_exception(None, self.cause, self.cause.__traceback__)[1:]),
delim='\n') or None
class UnsupportedError(ExtractorError):
def __init__(self, url):
'Unsupported URL: %s' % url, expected=True)
self.url = url
class RegexNotFoundError(ExtractorError):
"""Error when a regex didn't match"""
class GeoRestrictedError(ExtractorError):
"""Geographic restriction Error exception.
This exception may be thrown when a video is not available from your
geographic location due to geographic restrictions imposed by a website.
def __init__(self, msg, countries=None, **kwargs):
kwargs['expected'] = True
super().__init__(msg, **kwargs)
self.countries = countries
class DownloadError(YoutubeDLError):
"""Download Error exception.
This exception may be thrown by FileDownloader objects if they are not
configured to continue on errors. They will contain the appropriate
error message.
def __init__(self, msg, exc_info=None):
""" exc_info, if given, is the original exception that caused the trouble (as returned by sys.exc_info()). """
self.exc_info = exc_info
class EntryNotInPlaylist(YoutubeDLError):
"""Entry not in playlist exception.
This exception will be thrown by YoutubeDL when a requested entry
is not found in the playlist info_dict
msg = 'Entry not found in info'
class SameFileError(YoutubeDLError):
"""Same File exception.
This exception will be thrown by FileDownloader objects if they detect
multiple files would have to be downloaded to the same file on disk.
msg = 'Fixed output name but more than one file to download'
def __init__(self, filename=None):
if filename is not None:
self.msg += f': {filename}'
class PostProcessingError(YoutubeDLError):
"""Post Processing exception.
This exception may be raised by PostProcessor's .run() method to
indicate an error in the postprocessing task.
class DownloadCancelled(YoutubeDLError):
""" Exception raised when the download queue should be interrupted """
msg = 'The download was cancelled'
class ExistingVideoReached(DownloadCancelled):
""" --break-on-existing triggered """
msg = 'Encountered a video that is already in the archive, stopping due to --break-on-existing'
class RejectedVideoReached(DownloadCancelled):
""" --break-on-reject triggered """
msg = 'Encountered a video that did not match filter, stopping due to --break-on-reject'
class MaxDownloadsReached(DownloadCancelled):
""" --max-downloads limit has been reached. """
msg = 'Maximum number of downloads reached, stopping due to --max-downloads'
class ReExtractInfo(YoutubeDLError):
""" Video info needs to be re-extracted. """
def __init__(self, msg, expected=False):
self.expected = expected
class ThrottledDownload(ReExtractInfo):
""" Download speed below --throttled-rate. """
msg = 'The download speed is below throttle limit'
def __init__(self):
super().__init__(self.msg, expected=False)
class UnavailableVideoError(YoutubeDLError):
"""Unavailable Format exception.
This exception will be thrown when a video is requested
in a format that is not available for that video.
msg = 'Unable to download video'
def __init__(self, err=None):
if err is not None:
self.msg += f': {err}'
class ContentTooShortError(YoutubeDLError):
"""Content Too Short exception.
This exception may be raised by FileDownloader objects when a file they
download is too small for what the server announced first, indicating
the connection was probably interrupted.
def __init__(self, downloaded, expected):
super().__init__(f'Downloaded {downloaded} bytes, expected {expected} bytes')
# Both in bytes
self.downloaded = downloaded
self.expected = expected
class XAttrMetadataError(YoutubeDLError):
def __init__(self, code=None, msg='Unknown error'):
self.code = code
self.msg = msg
# Parsing code and msg
if (self.code in (errno.ENOSPC, errno.EDQUOT)
or 'No space left' in self.msg or 'Disk quota exceeded' in self.msg):
self.reason = 'NO_SPACE'
elif self.code == errno.E2BIG or 'Argument list too long' in self.msg:
self.reason = 'VALUE_TOO_LONG'
self.reason = 'NOT_SUPPORTED'
class XAttrUnavailableError(YoutubeDLError):
def _create_http_connection(ydl_handler, http_class, is_https, *args, **kwargs):
hc = http_class(*args, **kwargs)
source_address = ydl_handler._params.get('source_address')
if source_address is not None:
# This is to workaround _create_connection() from socket where it will try all
# address data from getaddrinfo() including IPv6. This filters the result from
# getaddrinfo() based on the source_address value.
# This is based on the cpython socket.create_connection() function.
def _create_connection(address, timeout=socket._GLOBAL_DEFAULT_TIMEOUT, source_address=None):
host, port = address
err = None
addrs = socket.getaddrinfo(host, port, 0, socket.SOCK_STREAM)
af = socket.AF_INET if '.' in source_address[0] else socket.AF_INET6
ip_addrs = [addr for addr in addrs if addr[0] == af]
if addrs and not ip_addrs:
ip_version = 'v4' if af == socket.AF_INET else 'v6'
raise OSError(
"No remote IP%s addresses available for connect, can't use '%s' as source address"
% (ip_version, source_address[0]))
for res in ip_addrs:
af, socktype, proto, canonname, sa = res
sock = None
sock = socket.socket(af, socktype, proto)
if timeout is not socket._GLOBAL_DEFAULT_TIMEOUT:
err = None # Explicitly break reference cycle
return sock
except OSError as _:
err = _
if sock is not None:
if err is not None:
raise err
raise OSError('getaddrinfo returns an empty list')
if hasattr(hc, '_create_connection'):
hc._create_connection = _create_connection
hc.source_address = (source_address, 0)
return hc
def handle_youtubedl_headers(headers):
filtered_headers = headers
if 'Youtubedl-no-compression' in filtered_headers:
filtered_headers = {k: v for k, v in filtered_headers.items() if k.lower() != 'accept-encoding'}
del filtered_headers['Youtubedl-no-compression']
return filtered_headers
class YoutubeDLHandler(urllib.request.HTTPHandler):
"""Handler for HTTP requests and responses.
This class, when installed with an OpenerDirector, automatically adds
the standard headers to every HTTP request and handles gzipped and
deflated responses from web servers. If compression is to be avoided in
a particular request, the original request in the program code only has
to include the HTTP header "Youtubedl-no-compression", which will be
removed before making the real request.
Part of this code was copied from:
Andrew Rowls, the author of that code, agreed to release it to the
public domain.
def __init__(self, params, *args, **kwargs):
urllib.request.HTTPHandler.__init__(self, *args, **kwargs)
self._params = params
def http_open(self, req):
conn_class = http.client.HTTPConnection
socks_proxy = req.headers.get('Ytdl-socks-proxy')
if socks_proxy:
conn_class = make_socks_conn_class(conn_class, socks_proxy)
del req.headers['Ytdl-socks-proxy']
return self.do_open(functools.partial(
_create_http_connection, self, conn_class, False),
def deflate(data):
if not data:
return data
return zlib.decompress(data, -zlib.MAX_WBITS)
except zlib.error:
return zlib.decompress(data)
def brotli(data):
if not data:
return data
return brotli.decompress(data)
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, [#3412])
# urllib chokes on URLs with non-ASCII characters (see
# To work around aforementioned issue we will replace request's original URL with
# percent-encoded one
# Since redirects are also affected (e.g.
# 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 = update_Request(req, url=url_escaped)
for h, v in self._params.get('http_headers', std_headers).items():
# Capitalize is needed because of Python bug 2275:
# The dict keys are capitalized because of this bug by urllib
if h.capitalize() not in req.headers:
req.add_header(h, v)
if 'Accept-encoding' not in req.headers:
req.add_header('Accept-encoding', ', '.join(SUPPORTED_ENCODINGS))
req.headers = handle_youtubedl_headers(req.headers)
return super().do_request_(req)
def http_response(self, req, resp):
old_resp = resp
# gzip
if resp.headers.get('Content-encoding', '') == 'gzip':
content =
gz = gzip.GzipFile(fileobj=io.BytesIO(content), mode='rb')
uncompressed = io.BytesIO(
except OSError as original_ioerror:
# There may be junk add the end of the file
# See for details
for i in range(1, 1024):
gz = gzip.GzipFile(fileobj=io.BytesIO(content[:-i]), mode='rb')
uncompressed = io.BytesIO(
except OSError:
raise original_ioerror
resp = urllib.request.addinfourl(uncompressed, old_resp.headers, old_resp.url, old_resp.code)
resp.msg = old_resp.msg
del resp.headers['Content-encoding']
# deflate
if resp.headers.get('Content-encoding', '') == 'deflate':
gz = io.BytesIO(self.deflate(
resp = urllib.request.addinfourl(gz, old_resp.headers, old_resp.url, old_resp.code)
resp.msg = old_resp.msg
del resp.headers['Content-encoding']
# brotli
if resp.headers.get('Content-encoding', '') == 'br':
resp = urllib.request.addinfourl(
io.BytesIO(self.brotli(, old_resp.headers, old_resp.url, old_resp.code)
resp.msg = old_resp.msg
del resp.headers['Content-encoding']
# Percent-encode redirect URL of Location HTTP header to satisfy RFC 3986 (see
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
location = location.encode('iso-8859-1').decode()
location_escaped = escape_url(location)
if location != location_escaped:
del resp.headers['Location']
resp.headers['Location'] = location_escaped
return resp
https_request = http_request
https_response = http_response
def make_socks_conn_class(base_class, socks_proxy):
assert issubclass(base_class, (
http.client.HTTPConnection, http.client.HTTPSConnection))
url_components = urllib.parse.urlparse(socks_proxy)
if url_components.scheme.lower() == 'socks5':
socks_type = ProxyType.SOCKS5
elif url_components.scheme.lower() in ('socks', 'socks4'):
socks_type = ProxyType.SOCKS4
elif url_components.scheme.lower() == 'socks4a':
socks_type = ProxyType.SOCKS4A
def unquote_if_non_empty(s):
if not s:
return s
return urllib.parse.unquote_plus(s)
proxy_args = (
url_components.hostname, url_components.port or 1080,
True, # Remote DNS
class SocksConnection(base_class):
def connect(self):
self.sock = sockssocket()
if isinstance(self.timeout, (int, float)):
self.sock.connect((, self.port))
if isinstance(self, http.client.HTTPSConnection):
if hasattr(self, '_context'): # Python > 2.6
self.sock = self._context.wrap_socket(
self.sock = ssl.wrap_socket(self.sock)
return SocksConnection
class YoutubeDLHTTPSHandler(urllib.request.HTTPSHandler):
def __init__(self, params, https_conn_class=None, *args, **kwargs):
urllib.request.HTTPSHandler.__init__(self, *args, **kwargs)
self._https_conn_class = https_conn_class or http.client.HTTPSConnection
self._params = params
def https_open(self, req):
kwargs = {}
conn_class = self._https_conn_class
if hasattr(self, '_context'): # python > 2.6
kwargs['context'] = self._context
if hasattr(self, '_check_hostname'): # python 3.x
kwargs['check_hostname'] = self._check_hostname
socks_proxy = req.headers.get('Ytdl-socks-proxy')
if socks_proxy:
conn_class = make_socks_conn_class(conn_class, socks_proxy)
del req.headers['Ytdl-socks-proxy']
return self.do_open(
functools.partial(_create_http_connection, self, conn_class, True), req, **kwargs)
except urllib.error.URLError as e:
if (isinstance(e.reason, ssl.SSLError)
and getattr(e.reason, 'reason', None) == 'SSLV3_ALERT_HANDSHAKE_FAILURE'):
raise YoutubeDLError('SSLV3_ALERT_HANDSHAKE_FAILURE: Try using --legacy-server-connect')
class YoutubeDLCookieJar(http.cookiejar.MozillaCookieJar):
See [1] for cookie file format.
_HEADER = '''# Netscape HTTP Cookie File
# This file is generated by yt-dlp. Do not edit.
_CookieFileEntry = collections.namedtuple(
('domain_name', 'include_subdomains', 'path', 'https_only', 'expires_at', 'name', 'value'))
def __init__(self, filename=None, *args, **kwargs):
super().__init__(None, *args, **kwargs)
if self.is_path(filename):
filename = os.fspath(filename)
self.filename = filename
def _true_or_false(cndn):
return 'TRUE' if cndn else 'FALSE'
def is_path(file):
return isinstance(file, (str, bytes, os.PathLike))
def open(self, file, *, write=False):
if self.is_path(file):
with open(file, 'w' if write else 'r', encoding='utf-8') as f:
yield f
if write:
yield file
def _really_save(self, f, ignore_discard=False, ignore_expires=False):
now = time.time()
for cookie in self:
if (not ignore_discard and cookie.discard
or not ignore_expires and cookie.is_expired(now)):
name, value =, cookie.value
if value is None:
# cookies.txt regards 'Set-Cookie: foo' as a cookie
# with no name, whereas http.cookiejar regards it as a
# cookie with no value.
name, value = '', name
f.write('%s\n' % '\t'.join((
str_or_none(cookie.expires, default=''),
name, value
def save(self, filename=None, *args, **kwargs):
Save cookies to a file.
Code is taken from CPython 3.6 """
if filename is None:
if self.filename is not None:
filename = self.filename
raise ValueError(http.cookiejar.MISSING_FILENAME_TEXT)
# Store session cookies with `expires` set to 0 instead of an empty string
for cookie in self:
if cookie.expires is None:
cookie.expires = 0
with, write=True) as f:
self._really_save(f, *args, **kwargs)
def load(self, filename=None, ignore_discard=False, ignore_expires=False):
"""Load cookies from a file."""
if filename is None:
if self.filename is not None:
filename = self.filename
raise ValueError(http.cookiejar.MISSING_FILENAME_TEXT)
def prepare_line(line):
if line.startswith(self._HTTPONLY_PREFIX):
line = line[len(self._HTTPONLY_PREFIX):]
# comments and empty lines are fine
if line.startswith('#') or not line.strip():
return line
cookie_list = line.split('\t')
if len(cookie_list) != self._ENTRY_LEN:
raise http.cookiejar.LoadError('invalid length %d' % len(cookie_list))
cookie = self._CookieFileEntry(*cookie_list)
if cookie.expires_at and not cookie.expires_at.isdigit():
raise http.cookiejar.LoadError('invalid expires at %s' % cookie.expires_at)
return line
cf = io.StringIO()
with as f:
for line in f:
except http.cookiejar.LoadError as e:
if f'{line.strip()} '[0] in '[{"':
raise http.cookiejar.LoadError(
'Cookies file must be Netscape formatted, not JSON. See '
write_string(f'WARNING: skipping cookie file entry due to {e}: {line!r}\n')
self._really_load(cf, filename, ignore_discard, ignore_expires)
# Session cookies are denoted by either `expires` field set to
# an empty string or 0. MozillaCookieJar only recognizes the former
# (see [1]). So we need force the latter to be recognized as session
# cookies on our own.
# Session cookies may be important for cookies-based authentication,
# e.g. usually, when user does not check 'Remember me' check box while
# logging in on a site, some important cookies are stored as session
# cookies so that not recognizing them will result in failed login.
# 1.
for cookie in self:
# Treat `expires=0` cookies as session cookies
if cookie.expires == 0:
cookie.expires = None
cookie.discard = True
class YoutubeDLCookieProcessor(urllib.request.HTTPCookieProcessor):
def __init__(self, cookiejar=None):
urllib.request.HTTPCookieProcessor.__init__(self, cookiejar)
def http_response(self, request, response):
return urllib.request.HTTPCookieProcessor.http_response(self, request, response)
https_request = urllib.request.HTTPCookieProcessor.http_request
https_response = http_response
class YoutubeDLRedirectHandler(urllib.request.HTTPRedirectHandler):
"""YoutubeDL redirect handler
The code is based on HTTPRedirectHandler implementation from CPython [1].
This redirect handler solves two issues:
- ensures redirect URL is always unicode under python 2
- introduces support for experimental HTTP response status code
308 Permanent Redirect [2] used by some sites [3]
http_error_301 = http_error_303 = http_error_307 = http_error_308 = urllib.request.HTTPRedirectHandler.http_error_302
def redirect_request(self, req, fp, code, msg, headers, newurl):
"""Return a Request or None in response to a redirect.
This is called by the http_error_30x methods when a
redirection response is received. If a redirection should
take place, return a new Request to allow http_error_30x to
perform the redirect. Otherwise, raise HTTPError if no-one
else should try to handle this url. Return None if you can't
but another Handler might.
m = req.get_method()
if (not (code in (301, 302, 303, 307, 308) and m in ("GET", "HEAD")
or code in (301, 302, 303) and m == "POST")):
raise urllib.error.HTTPError(req.full_url, code, msg, headers, fp)
# Strictly (according to RFC 2616), 301 or 302 in response to
# a POST MUST NOT cause a redirection without confirmation
# from the user (of urllib.request, in this case). In practice,
# essentially all clients do redirect in this case, so we do
# the same.
# Be conciliant with URIs containing a space. This is mainly
# redundant with the more complete encoding done in http_error_302(),
# but it is kept for compatibility with other callers.
newurl = newurl.replace(' ', '%20')
CONTENT_HEADERS = ("content-length", "content-type")
# NB: don't use dict comprehension for python 2.6 compatibility
newheaders = {k: v for k, v in req.headers.items() if k.lower() not in CONTENT_HEADERS}
# A 303 must either use GET or HEAD for subsequent request
if code == 303 and m != 'HEAD':
m = 'GET'
# 301 and 302 redirects are commonly turned into a GET from a POST
# for subsequent requests by browsers, so we'll do the same.
if code in (301, 302) and m == 'POST':
m = 'GET'
return urllib.request.Request(
newurl, headers=newheaders, origin_req_host=req.origin_req_host,
unverifiable=True, method=m)
def extract_timezone(date_str):
m =
^.{8,}? # >=8 char non-TZ prefix, if present
(?P<tz>Z| # just the UTC Z, or
(?:(?<=.\b\d{4}|\b\d{2}:\d\d)| # preceded by 4 digits or hh:mm or
(?<!.\b[a-zA-Z]{3}|[a-zA-Z]{4}|..\b\d\d)) # not preceded by 3 alpha word or >= 4 alpha or 2 digits
[ ]? # optional space
(?P<sign>\+|-) # +/-
(?P<hours>[0-9]{2}):?(?P<minutes>[0-9]{2}) # hh[:]mm
''', date_str)
if not m:
timezone = datetime.timedelta()
date_str = date_str[:-len('tz'))]
if not'sign'):
timezone = datetime.timedelta()
sign = 1 if'sign') == '+' else -1
timezone = datetime.timedelta(
hours=sign * int('hours')),
minutes=sign * int('minutes')))
return timezone, date_str
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:
timezone, date_str = extract_timezone(date_str)
with contextlib.suppress(ValueError):
date_format = f'%Y-%m-%d{delimiter}%H:%M:%S'
dt = datetime.datetime.strptime(date_str, date_format) - timezone
return calendar.timegm(dt.timetuple())
def date_formats(day_first=True):
def unified_strdate(date_str, day_first=True):
"""Return a string with the date in the format YYYYMMDD"""
if date_str is None:
return None
upload_date = None
# Replace commas
date_str = date_str.replace(',', ' ')
# Remove AM/PM + timezone
date_str = re.sub(r'(?i)\s*(?:AM|PM)(?:\s+[A-Z]+)?', '', date_str)
_, date_str = extract_timezone(date_str)
for expression in date_formats(day_first):
with contextlib.suppress(ValueError):
upload_date = datetime.datetime.strptime(date_str, expression).strftime('%Y%m%d')
if upload_date is None:
timetuple = email.utils.parsedate_tz(date_str)
if timetuple:
with contextlib.suppress(ValueError):
upload_date = datetime.datetime(*timetuple[:6]).strftime('%Y%m%d')
if upload_date is not None:
return str(upload_date)
def unified_timestamp(date_str, day_first=True):
if date_str is None:
return None
date_str = re.sub(r'[,|]', '', date_str)
pm_delta = 12 if'(?i)PM', date_str) else 0
timezone, date_str = extract_timezone(date_str)
# Remove AM/PM + timezone
date_str = re.sub(r'(?i)\s*(?:AM|PM)(?:\s+[A-Z]+)?', '', date_str)
# Remove unrecognized timezones from ISO 8601 alike timestamps
m ='\d{1,2}:\d{1,2}(?:\.\d+)?(?P<tz>\s*[A-Z]+)$', date_str)
if m:
date_str = date_str[:-len('tz'))]
# Python only supports microseconds, so remove nanoseconds
m ='^([0-9]{4,}-[0-9]{1,2}-[0-9]{1,2}T[0-9]{1,2}:[0-9]{1,2}:[0-9]{1,2}\.[0-9]{6})[0-9]+$', date_str)
if m:
date_str =
for expression in date_formats(day_first):
with contextlib.suppress(ValueError):
dt = datetime.datetime.strptime(date_str, expression) - timezone + datetime.timedelta(hours=pm_delta)
return calendar.timegm(dt.timetuple())
timetuple = email.utils.parsedate_tz(date_str)
if timetuple:
return calendar.timegm(timetuple) + pm_delta * 3600
def determine_ext(url, default_ext='unknown_video'):
if url is None or '.' not in url:
return default_ext
guess = url.partition('?')[0].rpartition('.')[2]
if re.match(r'^[A-Za-z0-9]+$', guess):
return guess
# Try extract ext from URLs like
elif guess.rstrip('/') in KNOWN_EXTENSIONS:
return guess.rstrip('/')
return default_ext
def subtitles_filename(filename, sub_lang, sub_format, expected_real_ext=None):
return replace_extension(filename, sub_lang + '.' + sub_format, expected_real_ext)
def datetime_from_str(date_str, precision='auto', format='%Y%m%d'):
Return a datetime object from a string.
Supported format:
@param format strftime format of DATE
@param precision Round the datetime object: auto|microsecond|second|minute|hour|day
auto: round to the unit provided in date_str (if applicable).
auto_precision = False
if precision == 'auto':
auto_precision = True
precision = 'microsecond'
today = datetime_round(datetime.datetime.utcnow(), precision)
if date_str in ('now', 'today'):
return today
if date_str == 'yesterday':
return today - datetime.timedelta(days=1)
match = re.match(
if match is not None:
start_time = datetime_from_str('start'), precision, format)
time = int('time')) * (-1 if'sign') == '-' else 1)
unit ='unit')
if unit == 'month' or unit == 'year':
new_date = datetime_add_months(start_time, time * 12 if unit == 'year' else time)
unit = 'day'
if unit == 'week':
unit = 'day'
time *= 7
delta = datetime.timedelta(**{unit + 's': time})
new_date = start_time + delta
if auto_precision:
return datetime_round(new_date, unit)
return new_date
return datetime_round(datetime.datetime.strptime(date_str, format), precision)
def date_from_str(date_str, format='%Y%m%d', strict=False):
Return a date object from a string using datetime_from_str
@param strict Restrict allowed patterns to "YYYYMMDD" and
if strict and not re.fullmatch(r'\d{8}|(now|today|yesterday)(-\d+(day|week|month|year)s?)?', date_str):
raise ValueError(f'Invalid date format "{date_str}"')
return datetime_from_str(date_str, precision='microsecond', format=format).date()
def datetime_add_months(dt, months):
"""Increment/Decrement a datetime object by months."""
month = dt.month + months - 1
year = dt.year + month // 12
month = month % 12 + 1
day = min(, calendar.monthrange(year, month)[1])
return dt.replace(year, month, day)
def datetime_round(dt, precision='day'):
Round a datetime object's time to a specific precision
if precision == 'microsecond':
return dt
unit_seconds = {
'day': 86400,
'hour': 3600,
'minute': 60,
'second': 1,
roundto = lambda x, n: ((x + n / 2) // n) * n
timestamp = calendar.timegm(dt.timetuple())
return datetime.datetime.utcfromtimestamp(roundto(timestamp, unit_seconds[precision]))
def hyphenate_date(date_str):
Convert a date in 'YYYYMMDD' format to 'YYYY-MM-DD' format"""
match = re.match(r'^(\d\d\d\d)(\d\d)(\d\d)$', date_str)
if match is not None:
return '-'.join(match.groups())
return date_str
class DateRange:
"""Represents a time interval between two dates"""
def __init__(self, start=None, end=None):
"""start and end must be strings in the format accepted by date"""
if start is not None:
self.start = date_from_str(start, strict=True)
self.start =
if end is not None:
self.end = date_from_str(end, strict=True)
self.end =
if self.start > self.end:
raise ValueError('Date range: "%s" , the start date must be before the end date' % self)
def day(cls, day):
"""Returns a range that only contains the given day"""
return cls(day, day)
def __contains__(self, date):
"""Check if the date is in the range"""
if not isinstance(date,
date = date_from_str(date)
return self.start <= date <= self.end
def __str__(self):
return f'{self.start.isoformat()} - {self.end.isoformat()}'
def platform_name():
""" Returns the platform name as a str """
res = platform.platform()
if isinstance(res, bytes):
res = res.decode(preferredencoding())
assert isinstance(res, str)
return res
def get_windows_version():
''' Get Windows version. returns () if it's not running on Windows '''
if compat_os_name == 'nt':
return version_tuple(platform.win32_ver()[1])
return ()
def write_string(s, out=None, encoding=None):
assert isinstance(s, str)
out = out or sys.stderr
if compat_os_name == 'nt' and supports_terminal_sequences(out):
s = re.sub(r'([\r\n]+)', r' \1', s)
enc, buffer = None, out
if 'b' in getattr(out, 'mode', ''):
enc = encoding or preferredencoding()
elif hasattr(out, 'buffer'):
buffer = out.buffer
enc = encoding or getattr(out, 'encoding', None) or preferredencoding()
buffer.write(s.encode(enc, 'ignore') if enc else s)
def bytes_to_intlist(bs):
if not bs:
return []
if isinstance(bs[0], int): # Python 3
return list(bs)
return [ord(c) for c in bs]
def intlist_to_bytes(xs):
if not xs:
return b''
return struct.pack('%dB' % len(xs), *xs)
class LockingUnsupportedError(OSError):
msg = 'File locking is not supported'
def __init__(self):
# Cross-platform file locking
if sys.platform == 'win32':
import ctypes.wintypes
import msvcrt
class OVERLAPPED(ctypes.Structure):
_fields_ = [
('Internal', ctypes.wintypes.LPVOID),
('InternalHigh', ctypes.wintypes.LPVOID),
('Offset', ctypes.wintypes.DWORD),
('OffsetHigh', ctypes.wintypes.DWORD),
('hEvent', ctypes.wintypes.HANDLE),
kernel32 = ctypes.windll.kernel32
LockFileEx = kernel32.LockFileEx
LockFileEx.argtypes = [
ctypes.wintypes.HANDLE, # hFile
ctypes.wintypes.DWORD, # dwFlags
ctypes.wintypes.DWORD, # dwReserved
ctypes.wintypes.DWORD, # nNumberOfBytesToLockLow
ctypes.wintypes.DWORD, # nNumberOfBytesToLockHigh
ctypes.POINTER(OVERLAPPED) # Overlapped
LockFileEx.restype = ctypes.wintypes.BOOL
UnlockFileEx = kernel32.UnlockFileEx
UnlockFileEx.argtypes = [
ctypes.wintypes.HANDLE, # hFile
ctypes.wintypes.DWORD, # dwReserved
ctypes.wintypes.DWORD, # nNumberOfBytesToLockLow
ctypes.wintypes.DWORD, # nNumberOfBytesToLockHigh
ctypes.POINTER(OVERLAPPED) # Overlapped
UnlockFileEx.restype = ctypes.wintypes.BOOL
whole_low = 0xffffffff
whole_high = 0x7fffffff
def _lock_file(f, exclusive, block):
overlapped = OVERLAPPED()
overlapped.Offset = 0
overlapped.OffsetHigh = 0
overlapped.hEvent = 0
f._lock_file_overlapped_p = ctypes.pointer(overlapped)
if not LockFileEx(msvcrt.get_osfhandle(f.fileno()),
(0x2 if exclusive else 0x0) | (0x0 if block else 0x1),
0, whole_low, whole_high, f._lock_file_overlapped_p):
# NB: No argument form of "ctypes.FormatError" does not work on PyPy
raise BlockingIOError(f'Locking file failed: {ctypes.FormatError(ctypes.GetLastError())!r}')
def _unlock_file(f):
assert f._lock_file_overlapped_p
handle = msvcrt.get_osfhandle(f.fileno())
if not UnlockFileEx(handle, 0, whole_low, whole_high, f._lock_file_overlapped_p):
raise OSError('Unlocking file failed: %r' % ctypes.FormatError())
import fcntl
def _lock_file(f, exclusive, block):
flags = fcntl.LOCK_EX if exclusive else fcntl.LOCK_SH
if not block:
flags |= fcntl.LOCK_NB
fcntl.flock(f, flags)
except BlockingIOError:
except OSError: # AOSP does not have flock()
fcntl.lockf(f, flags)
def _unlock_file(f):
fcntl.flock(f, fcntl.LOCK_UN)
except OSError:
fcntl.lockf(f, fcntl.LOCK_UN)
except ImportError:
def _lock_file(f, exclusive, block):
raise LockingUnsupportedError()
def _unlock_file(f):
raise LockingUnsupportedError()
class locked_file:
locked = False
def __init__(self, filename, mode, block=True, encoding=None):
if mode not in {'r', 'rb', 'a', 'ab', 'w', 'wb'}:
raise NotImplementedError(mode)
self.mode, self.block = mode, block
writable = any(f in mode for f in 'wax+')
readable = any(f in mode for f in 'r+')
flags = functools.reduce(operator.ior, (
getattr(os, 'O_CLOEXEC', 0), # UNIX only
getattr(os, 'O_BINARY', 0), # Windows only
getattr(os, 'O_NOINHERIT', 0), # Windows only
os.O_CREAT if writable else 0, # O_TRUNC only after locking
os.O_APPEND if 'a' in mode else 0,
os.O_EXCL if 'x' in mode else 0,
os.O_RDONLY if not writable else os.O_RDWR if readable else os.O_WRONLY,
self.f = os.fdopen(, flags, 0o666), mode, encoding=encoding)
def __enter__(self):
exclusive = 'r' not in self.mode
_lock_file(self.f, exclusive, self.block)
self.locked = True
except OSError:
if 'w' in self.mode:
except OSError as e:
if e.errno not in (
errno.ESPIPE, # Illegal seek - expected for FIFO
errno.EINVAL, # Invalid argument - expected for /dev/null
return self
def unlock(self):
if not self.locked:
self.locked = False
def __exit__(self, *_):
open = __enter__
close = __exit__
def __getattr__(self, attr):
return getattr(self.f, attr)
def __iter__(self):
return iter(self.f)
def get_filesystem_encoding():
encoding = sys.getfilesystemencoding()
return encoding if encoding is not None else 'utf-8'
def shell_quote(args):
quoted_args = []
encoding = get_filesystem_encoding()
for a in args:
if isinstance(a, bytes):
# We may get a filename encoded with 'encodeFilename'
a = a.decode(encoding)
return ' '.join(quoted_args)
def smuggle_url(url, data):
""" Pass additional data in a URL for internal use. """
url, idata = unsmuggle_url(url, {})
sdata = urllib.parse.urlencode(
{'__youtubedl_smuggle': json.dumps(data)})
return url + '#' + sdata
def unsmuggle_url(smug_url, default=None):
if '#__youtubedl_smuggle' not in smug_url:
return smug_url, default
url, _, sdata = smug_url.rpartition('#')
jsond = urllib.parse.parse_qs(sdata)['__youtubedl_smuggle'][0]
data = json.loads(jsond)
return url, data
def format_decimal_suffix(num, fmt='%d%s', *, factor=1000):
""" Formats numbers with decimal sufixes like K, M, etc """
num, factor = float_or_none(num), float(factor)
if num is None or num < 0:
return None
exponent = 0 if num == 0 else min(int(math.log(num, factor)), len(POSSIBLE_SUFFIXES))
suffix = ['', *POSSIBLE_SUFFIXES][exponent]
if factor == 1024:
suffix = {'k': 'Ki', '': ''}.get(suffix, f'{suffix}i')
converted = num / (factor ** exponent)
return fmt % (converted, suffix)
def format_bytes(bytes):
return format_decimal_suffix(bytes, '%.2f%sB', factor=1024) or 'N/A'
def lookup_unit_table(unit_table, s):
units_re = '|'.join(re.escape(u) for u in unit_table)
m = re.match(
r'(?P<num>[0-9]+(?:[,.][0-9]*)?)\s*(?P<unit>%s)\b' % units_re, s)
if not m:
return None
num_str ='num').replace(',', '.')
mult = unit_table['unit')]
return int(float(num_str) * mult)
def parse_filesize(s):
if s is None:
return None
# The lower-case forms are of course incorrect and unofficial,
# but we support those too
'B': 1,
'b': 1,
'bytes': 1,
'KiB': 1024,
'KB': 1000,
'kB': 1024,
'Kb': 1000,
'kb': 1000,
'kilobytes': 1000,
'kibibytes': 1024,
'MiB': 1024 ** 2,
'MB': 1000 ** 2,
'mB': 1024 ** 2,
'Mb': 1000 ** 2,
'mb': 1000 ** 2,
'megabytes': 1000 ** 2,
'mebibytes': 1024 ** 2,
'GiB': 1024 ** 3,
'GB': 1000 ** 3,
'gB': 1024 ** 3,
'Gb': 1000 ** 3,
'gb': 1000 ** 3,
'gigabytes': 1000 ** 3,
'gibibytes': 1024 ** 3,
'TiB': 1024 ** 4,
'TB': 1000 ** 4,
'tB': 1024 ** 4,
'Tb': 1000 ** 4,
'tb': 1000 ** 4,
'terabytes': 1000 ** 4,
'tebibytes': 1024 ** 4,
'PiB': 1024 ** 5,
'PB': 1000 ** 5,
'pB': 1024 ** 5,
'Pb': 1000 ** 5,
'pb': 1000 ** 5,
'petabytes': 1000 ** 5,
'pebibytes': 1024 ** 5,
'EiB': 1024 ** 6,
'EB': 1000 ** 6,
'eB': 1024 ** 6,
'Eb': 1000 ** 6,
'eb': 1000 ** 6,
'exabytes': 1000 ** 6,
'exbibytes': 1024 ** 6,
'ZiB': 1024 ** 7,
'ZB': 1000 ** 7,
'zB': 1024 ** 7,
'Zb': 1000 ** 7,
'zb': 1000 ** 7,
'zettabytes': 1000 ** 7,
'zebibytes': 1024 ** 7,
'YiB': 1024 ** 8,
'YB': 1000 ** 8,
'yB': 1024 ** 8,
'Yb': 1000 ** 8,
'yb': 1000 ** 8,
'yottabytes': 1000 ** 8,
'yobibytes': 1024 ** 8,
return lookup_unit_table(_UNIT_TABLE, s)
def parse_count(s):
if s is None:
return None
s = re.sub(r'^[^\d]+\s', '', s).strip()
if re.match(r'^[\d,.]+$', s):
return str_to_int(s)
'k': 1000,
'K': 1000,
'm': 1000 ** 2,
'M': 1000 ** 2,
'kk': 1000 ** 2,
'KK': 1000 ** 2,
'b': 1000 ** 3,
'B': 1000 ** 3,
ret = lookup_unit_table(_UNIT_TABLE, s)
if ret is not None:
return ret
mobj = re.match(r'([\d,.]+)(?:$|\s)', s)
if mobj:
return str_to_int(
def parse_resolution(s, *, lenient=False):
if s is None:
return {}
if lenient:
mobj ='(?P<w>\d+)\s*[xX×,]\s*(?P<h>\d+)', s)
mobj ='(?<![a-zA-Z0-9])(?P<w>\d+)\s*[xX×,]\s*(?P<h>\d+)(?![a-zA-Z0-9])', s)
if mobj:
return {
'width': int('w')),
'height': int('h')),
mobj ='(?<![a-zA-Z0-9])(\d+)[pPiI](?![a-zA-Z0-9])', s)
if mobj:
return {'height': int(}
mobj ='\b([48])[kK]\b', s)
if mobj:
return {'height': int( * 540}
return {}
def parse_bitrate(s):
if not isinstance(s, str):
mobj ='\b(\d+)\s*kbps', s)
if mobj:
return int(
def month_by_name(name, lang='en'):
""" Return the number of a month by (locale-independently) English name """
month_names = MONTH_NAMES.get(lang, MONTH_NAMES['en'])
return month_names.index(name) + 1
except ValueError:
return None
def month_by_abbreviation(abbrev):
""" Return the number of a month by (locale-independently) English
abbreviations """
return [s[:3] for s in ENGLISH_MONTH_NAMES].index(abbrev) + 1
except ValueError:
return None
def fix_xml_ampersands(xml_str):
"""Replace all the '&' by '&amp;' in XML"""
return re.sub(
def setproctitle(title):
assert isinstance(title, str)
# ctypes in Jython is not complete
if sys.platform.startswith('java'):
libc = ctypes.cdll.LoadLibrary('')
except OSError:
except TypeError:
# LoadLibrary in Windows Python 2.7.13 only expects
# a bytestring, but since unicode_literals turns
# every string into a unicode string, it fails.
title_bytes = title.encode()
buf = ctypes.create_string_buffer(len(title_bytes))
buf.value = title_bytes
libc.prctl(15, buf, 0, 0, 0)
except AttributeError:
return # Strange libc, just skip this
def remove_start(s, start):
return s[len(start):] if s is not None and s.startswith(start) else s
def remove_end(s, end):
return s[:-len(end)] if s is not None and s.endswith(end) else s
def remove_quotes(s):
if s is None or len(s) < 2:
return s
for quote in ('"', "'", ):
if s[0] == quote and s[-1] == quote:
return s[1:-1]
return s
def get_domain(url):
domain = re.match(r'(?:https?:\/\/)?(?:www\.)?(?P<domain>[^\n\/]+\.[^\n\/]+)(?:\/(.*))?', url)
return'domain') if domain else None
def url_basename(url):
path = urllib.parse.urlparse(url).path
return path.strip('/').split('/')[-1]
def base_url(url):
return re.match(r'https?://[^?#&]+/', url).group()
def urljoin(base, path):
if isinstance(path, bytes):
path = path.decode()
if not isinstance(path, str) or not path:
return None
if re.match(r'^(?:[a-zA-Z][a-zA-Z0-9+-.]*:)?//', path):
return path
if isinstance(base, bytes):
base = base.decode()
if not isinstance(base, str) or not re.match(
r'^(?:https?:)?//', base):
return None
return urllib.parse.urljoin(base, path)
class HEADRequest(urllib.request.Request):
def get_method(self):
return 'HEAD'
class PUTRequest(urllib.request.Request):
def get_method(self):
return 'PUT'
def int_or_none(v, scale=1, default=None, get_attr=None, invscale=1):
if get_attr and v is not None:
v = getattr(v, get_attr, None)
return int(v) * invscale // scale
except (ValueError, TypeError, OverflowError):
return default
def str_or_none(v, default=None):
return default if v is None else str(v)
def str_to_int(int_str):
""" A more relaxed version of int_or_none """
if isinstance(int_str, int):
return int_str
elif isinstance(int_str, str):
int_str = re.sub(r'[,\.\+]', '', int_str)
return int_or_none(int_str)
def float_or_none(v, scale=1, invscale=1, default=None):
if v is None:
return default
return float(v) * invscale / scale
except (ValueError, TypeError):
return default
def bool_or_none(v, default=None):
return v if isinstance(v, bool) else default
def strip_or_none(v, default=None):
return v.strip() if isinstance(v, str) else