]> Raphaël G. Git Repositories - youtubedl/blobdiff - youtube_dl/utils.py
debian/rules: Ignore tests temporarily.
[youtubedl] / youtube_dl / utils.py
index d39f313a41802f38f91418e12cc89dcd798b0286..562031fe110a9a6962c6e5efed2bb311f3059979 100644 (file)
@@ -4,6 +4,7 @@
 from __future__ import unicode_literals
 
 import base64
+import binascii
 import calendar
 import codecs
 import contextlib
@@ -13,8 +14,8 @@ import email.utils
 import errno
 import functools
 import gzip
-import itertools
 import io
+import itertools
 import json
 import locale
 import math
@@ -23,9 +24,8 @@ import os
 import pipes
 import platform
 import re
-import ssl
 import socket
-import struct
+import ssl
 import subprocess
 import sys
 import tempfile
@@ -34,29 +34,49 @@ import xml.etree.ElementTree
 import zlib
 
 from .compat import (
+    compat_HTMLParser,
     compat_basestring,
     compat_chr,
     compat_etree_fromstring,
     compat_html_entities,
+    compat_html_entities_html5,
     compat_http_client,
     compat_kwargs,
     compat_parse_qs,
+    compat_shlex_quote,
     compat_socket_create_connection,
     compat_str,
+    compat_struct_pack,
     compat_urllib_error,
     compat_urllib_parse,
+    compat_urllib_parse_urlencode,
     compat_urllib_parse_urlparse,
+    compat_urllib_parse_unquote_plus,
     compat_urllib_request,
     compat_urlparse,
-    shlex_quote,
+    compat_xpath,
+)
+
+from .socks import (
+    ProxyType,
+    sockssocket,
 )
 
 
+def register_socks_protocols():
+    # "Register" SOCKS protocols
+    # In Python < 2.6.5, urlsplit() suffers from bug https://bugs.python.org/issue7904
+    # URLs with protocols not in urlparse.uses_netloc are not handled correctly
+    for scheme in ('socks', 'socks4', 'socks4a', 'socks5'):
+        if scheme not in compat_urlparse.uses_netloc:
+            compat_urlparse.uses_netloc.append(scheme)
+
+
 # This is not clearly defined otherwise
 compiled_regex_type = type(re.compile(''))
 
 std_headers = {
-    'User-Agent': 'Mozilla/5.0 (X11; Linux x86_64; rv:10.0) Gecko/20150101 Firefox/20.0 (Chrome)',
+    'User-Agent': 'Mozilla/5.0 (X11; Linux x86_64; rv:10.0) Gecko/20150101 Firefox/47.0 (Chrome)',
     'Accept-Charset': 'ISO-8859-1,utf-8;q=0.7,*;q=0.7',
     'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',
     'Accept-Encoding': 'gzip, deflate',
@@ -70,6 +90,26 @@ ENGLISH_MONTH_NAMES = [
     'January', 'February', 'March', 'April', 'May', 'June',
     'July', 'August', 'September', 'October', 'November', 'December']
 
+KNOWN_EXTENSIONS = (
+    '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')
+
+# needed for sanitizing filenames in restricted mode
+ACCENT_CHARS = dict(zip('ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ',
+                        itertools.chain('AAAAAA', ['AE'], 'CEEEEIIIIDNOOOOOOO', ['OE'], 'UUUUUYP', ['ss'],
+                                        'aaaaaa', ['ae'], 'ceeeeiiiionooooooo', ['oe'], 'uuuuuypy')))
+
 
 def preferredencoding():
     """Get preferred encoding.
@@ -144,18 +184,11 @@ if sys.version_info >= (2, 7):
     def find_xpath_attr(node, xpath, key, val=None):
         """ Find the xpath xpath[@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=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):
+        for f in node.findall(compat_xpath(xpath)):
             if key not in f.attrib:
                 continue
             if val is None or f.attrib.get(key) == val:
@@ -180,9 +213,7 @@ def xpath_with_ns(path, ns_map):
 
 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)
+        return node.find(compat_xpath(xpath))
 
     if isinstance(xpath, (str, compat_str)):
         n = _find_xpath(xpath)
@@ -233,7 +264,7 @@ def xpath_attr(node, xpath, key, name=None, fatal=False, default=NO_DEFAULT):
 
 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)
+    return get_element_by_attribute('id', id, html)
 
 
 def get_element_by_attribute(attribute, value, html):
@@ -241,9 +272,9 @@ def get_element_by_attribute(attribute, value, html):
 
     m = re.search(r'''(?xs)
         <([a-zA-Z0-9:._-]+)
-         (?:\s+[a-zA-Z0-9:._-]+(?:=[a-zA-Z0-9:._-]+|="[^"]+"|='[^']+'))*?
+         (?:\s+[a-zA-Z0-9:._-]+(?:=[a-zA-Z0-9:._-]*|="[^"]*"|='[^']*'))*?
          \s+%s=['"]?%s['"]?
-         (?:\s+[a-zA-Z0-9:._-]+(?:=[a-zA-Z0-9:._-]+|="[^"]+"|='[^']+'))*?
+         (?:\s+[a-zA-Z0-9:._-]+(?:=[a-zA-Z0-9:._-]*|="[^"]*"|='[^']*'))*?
         \s*>
         (?P<content>.*?)
         </\1>
@@ -259,6 +290,38 @@ def get_element_by_attribute(attribute, value, html):
     return unescapeHTML(res)
 
 
+class HTMLAttributeParser(compat_HTMLParser):
+    """Trivial HTML parser to gather the attributes for a single element"""
+    def __init__(self):
+        self.attrs = {}
+        compat_HTMLParser.__init__(self)
+
+    def handle_starttag(self, tag, attrs):
+        self.attrs = dict(attrs)
+
+
+def extract_attributes(html_element):
+    """Given a string for an HTML element such as
+    <el
+         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': '\''
+    }.
+    NB HTMLParser is stricter in Python 2.6 & 3.2 than in later versions,
+    but the cases in the unit test will work for all of 2.6, 2.7, 3.2-3.5.
+    """
+    parser = HTMLAttributeParser()
+    parser.feed(html_element)
+    parser.close()
+    return parser.attrs
+
+
 def clean_html(html):
     """Clean an HTML snippet into a readable string"""
 
@@ -323,6 +386,8 @@ def sanitize_filename(s, restricted=False, is_id=False):
     Set is_id if this is not an arbitrary string, but an ID that should be kept if possible
     """
     def replace_insane(char):
+        if restricted and char in ACCENT_CHARS:
+            return ACCENT_CHARS[char]
         if char == '?' or ord(char) < 32 or ord(char) == 127:
             return ''
         elif char == '"':
@@ -373,6 +438,16 @@ def sanitize_path(s):
     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 sanitize_url(url):
+    return 'http:%s' % url if url.startswith('//') else url
+
+
+def sanitized_Request(url, *args, **kwargs):
+    return compat_urllib_request.Request(sanitize_url(url), *args, **kwargs)
+
+
 def orderedSet(iterable):
     """ Remove all duplicates from the input iterable """
     res = []
@@ -382,12 +457,19 @@ def orderedSet(iterable):
     return res
 
 
-def _htmlentity_transform(entity):
+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 compat_html_entities.name2codepoint:
         return compat_chr(compat_html_entities.name2codepoint[entity])
 
+    # TODO: HTML5 allows entities without a semicolon. For example,
+    # '&Eacuteric' should be decoded as 'Éric'.
+    if entity_with_semicolon in compat_html_entities_html5:
+        return compat_html_entities_html5[entity_with_semicolon]
+
     mobj = re.match(r'#(x[0-9a-fA-F]+|[0-9]+)', entity)
     if mobj is not None:
         numstr = mobj.group(1)
@@ -396,10 +478,14 @@ def _htmlentity_transform(entity):
             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):
@@ -408,7 +494,7 @@ def unescapeHTML(s):
     assert type(s) == compat_str
 
     return re.sub(
-        r'&([^;]+);', lambda m: _htmlentity_transform(m.group(1)), s)
+        r'&([^;]+;)', lambda m: _htmlentity_transform(m.group(1)), s)
 
 
 def get_subprocess_encoding():
@@ -440,6 +526,10 @@ def encodeFilename(s, for_subprocess=False):
     if not for_subprocess and sys.platform == 'win32' and sys.getwindowsversion()[0] >= 5:
         return s
 
+    # Jython assumes filenames are Unicode strings though reported as Python 2.x compatible
+    if sys.platform.startswith('java'):
+        return s
+
     return s.encode(get_subprocess_encoding(), 'ignore')
 
 
@@ -652,6 +742,16 @@ def _create_http_connection(ydl_handler, http_class, is_https, *args, **kwargs):
     return hc
 
 
+def handle_youtubedl_headers(headers):
+    filtered_headers = headers
+
+    if 'Youtubedl-no-compression' in filtered_headers:
+        filtered_headers = dict((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(compat_urllib_request.HTTPHandler):
     """Handler for HTTP requests and responses.
 
@@ -659,7 +759,7 @@ class YoutubeDLHandler(compat_urllib_request.HTTPHandler):
     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
+    to include the HTTP header "Youtubedl-no-compression", which will be
     removed before making the real request.
 
     Part of this code was copied from:
@@ -675,8 +775,15 @@ class YoutubeDLHandler(compat_urllib_request.HTTPHandler):
         self._params = params
 
     def http_open(self, req):
+        conn_class = compat_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, compat_http_client.HTTPConnection, False),
+            _create_http_connection, self, conn_class, False),
             req)
 
     @staticmethod
@@ -708,22 +815,15 @@ class YoutubeDLHandler(compat_urllib_request.HTTPHandler):
 
         # 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
+            req = update_Request(req, url=url_escaped)
 
         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
             if h.capitalize() not in req.headers:
                 req.add_header(h, v)
-        if 'Youtubedl-no-compression' in req.headers:
-            if 'Accept-encoding' in req.headers:
-                del req.headers['Accept-encoding']
-            del req.headers['Youtubedl-no-compression']
+
+        req.headers = handle_youtubedl_headers(req.headers)
 
         if sys.version_info < (2, 7) and '#' in req.get_full_url():
             # Python 2.6 is brain-dead when it comes to fragments
@@ -754,11 +854,13 @@ class YoutubeDLHandler(compat_urllib_request.HTTPHandler):
                     raise original_ioerror
             resp = self.addinfourl_wrapper(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.read()))
             resp = self.addinfourl_wrapper(gz, 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
         # https://github.com/rg3/youtube-dl/issues/6457).
         if 300 <= resp.code < 400:
@@ -767,9 +869,13 @@ class YoutubeDLHandler(compat_urllib_request.HTTPHandler):
                 # 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')
+                else:
+                    location = location.decode('utf-8')
                 location_escaped = escape_url(location)
                 if location != location_escaped:
                     del resp.headers['Location']
+                    if sys.version_info < (3, 0):
+                        location_escaped = location_escaped.encode('utf-8')
                     resp.headers['Location'] = location_escaped
         return resp
 
@@ -777,6 +883,49 @@ class YoutubeDLHandler(compat_urllib_request.HTTPHandler):
     https_response = http_response
 
 
+def make_socks_conn_class(base_class, socks_proxy):
+    assert issubclass(base_class, (
+        compat_http_client.HTTPConnection, compat_http_client.HTTPSConnection))
+
+    url_components = compat_urlparse.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 compat_urllib_parse_unquote_plus(s)
+
+    proxy_args = (
+        socks_type,
+        url_components.hostname, url_components.port or 1080,
+        True,  # Remote DNS
+        unquote_if_non_empty(url_components.username),
+        unquote_if_non_empty(url_components.password),
+    )
+
+    class SocksConnection(base_class):
+        def connect(self):
+            self.sock = sockssocket()
+            self.sock.setproxy(*proxy_args)
+            if type(self.timeout) in (int, float):
+                self.sock.settimeout(self.timeout)
+            self.sock.connect((self.host, self.port))
+
+            if isinstance(self, compat_http_client.HTTPSConnection):
+                if hasattr(self, '_context'):  # Python > 2.6
+                    self.sock = self._context.wrap_socket(
+                        self.sock, server_hostname=self.host)
+                else:
+                    self.sock = ssl.wrap_socket(self.sock)
+
+    return SocksConnection
+
+
 class YoutubeDLHTTPSHandler(compat_urllib_request.HTTPSHandler):
     def __init__(self, params, https_conn_class=None, *args, **kwargs):
         compat_urllib_request.HTTPSHandler.__init__(self, *args, **kwargs)
@@ -785,12 +934,20 @@ class YoutubeDLHTTPSHandler(compat_urllib_request.HTTPSHandler):
 
     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, self._https_conn_class, True),
+            _create_http_connection, self, conn_class, True),
             req, **kwargs)
 
 
@@ -868,9 +1025,9 @@ def unified_strdate(date_str, day_first=True):
         '%d %b %Y',
         '%B %d %Y',
         '%b %d %Y',
-        '%b %dst %Y %I:%M%p',
-        '%b %dnd %Y %I:%M%p',
-        '%b %dth %Y %I:%M%p',
+        '%b %dst %Y %I:%M',
+        '%b %dnd %Y %I:%M',
+        '%b %dth %Y %I:%M',
         '%Y %m %d',
         '%Y-%m-%d',
         '%Y/%m/%d',
@@ -890,6 +1047,7 @@ def unified_strdate(date_str, day_first=True):
         format_expressions.extend([
             '%d-%m-%Y',
             '%d.%m.%Y',
+            '%d.%m.%y',
             '%d/%m/%Y',
             '%d/%m/%y',
             '%d/%m/%Y %H:%M:%S',
@@ -910,7 +1068,10 @@ def unified_strdate(date_str, day_first=True):
     if upload_date is None:
         timetuple = email.utils.parsedate_tz(date_str)
         if timetuple:
-            upload_date = datetime.datetime(*timetuple[:6]).strftime('%Y%m%d')
+            try:
+                upload_date = datetime.datetime(*timetuple[:6]).strftime('%Y%m%d')
+            except ValueError:
+                pass
     if upload_date is not None:
         return compat_str(upload_date)
 
@@ -921,6 +1082,9 @@ 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
+    # Try extract ext from URLs like http://example.com/foo/bar.mp4/?download
+    elif guess.rstrip('/') in KNOWN_EXTENSIONS:
+        return guess.rstrip('/')
     else:
         return default_ext
 
@@ -945,7 +1109,7 @@ def date_from_str(date_str):
         if sign == '-':
             time = -time
         unit = match.group('unit')
-        # A bad aproximation?
+        # A bad approximation?
         if unit == 'month':
             unit = 'day'
             time *= 30
@@ -955,7 +1119,7 @@ def date_from_str(date_str):
         unit += 's'
         delta = datetime.timedelta(**{unit: time})
         return today + delta
-    return datetime.datetime.strptime(date_str, "%Y%m%d").date()
+    return datetime.datetime.strptime(date_str, '%Y%m%d').date()
 
 
 def hyphenate_date(date_str):
@@ -1035,22 +1199,22 @@ def _windows_write_string(s, out):
 
     GetStdHandle = ctypes.WINFUNCTYPE(
         ctypes.wintypes.HANDLE, ctypes.wintypes.DWORD)(
-        (b"GetStdHandle", ctypes.windll.kernel32))
+        (b'GetStdHandle', ctypes.windll.kernel32))
     h = GetStdHandle(WIN_OUTPUT_IDS[fileno])
 
     WriteConsoleW = ctypes.WINFUNCTYPE(
         ctypes.wintypes.BOOL, ctypes.wintypes.HANDLE, ctypes.wintypes.LPWSTR,
         ctypes.wintypes.DWORD, ctypes.POINTER(ctypes.wintypes.DWORD),
-        ctypes.wintypes.LPVOID)((b"WriteConsoleW", ctypes.windll.kernel32))
+        ctypes.wintypes.LPVOID)((b'WriteConsoleW', ctypes.windll.kernel32))
     written = ctypes.wintypes.DWORD(0)
 
-    GetFileType = ctypes.WINFUNCTYPE(ctypes.wintypes.DWORD, ctypes.wintypes.DWORD)((b"GetFileType", ctypes.windll.kernel32))
+    GetFileType = ctypes.WINFUNCTYPE(ctypes.wintypes.DWORD, ctypes.wintypes.DWORD)((b'GetFileType', ctypes.windll.kernel32))
     FILE_TYPE_CHAR = 0x0002
     FILE_TYPE_REMOTE = 0x8000
     GetConsoleMode = ctypes.WINFUNCTYPE(
         ctypes.wintypes.BOOL, ctypes.wintypes.HANDLE,
         ctypes.POINTER(ctypes.wintypes.DWORD))(
-        (b"GetConsoleMode", ctypes.windll.kernel32))
+        (b'GetConsoleMode', ctypes.windll.kernel32))
     INVALID_HANDLE_VALUE = ctypes.wintypes.DWORD(-1).value
 
     def not_a_console(handle):
@@ -1118,7 +1282,7 @@ def bytes_to_intlist(bs):
 def intlist_to_bytes(xs):
     if not xs:
         return b''
-    return struct_pack('%dB' % len(xs), *xs)
+    return compat_struct_pack('%dB' % len(xs), *xs)
 
 
 # Cross-platform file locking
@@ -1177,13 +1341,23 @@ if sys.platform == 'win32':
             raise OSError('Unlocking file failed: %r' % ctypes.FormatError())
 
 else:
-    import fcntl
+    # Some platforms, such as Jython, is missing fcntl
+    try:
+        import fcntl
 
-    def _lock_file(f, exclusive):
-        fcntl.flock(f, fcntl.LOCK_EX if exclusive else fcntl.LOCK_SH)
+        def _lock_file(f, exclusive):
+            fcntl.flock(f, fcntl.LOCK_EX if exclusive else fcntl.LOCK_SH)
 
-    def _unlock_file(f):
-        fcntl.flock(f, fcntl.LOCK_UN)
+        def _unlock_file(f):
+            fcntl.flock(f, fcntl.LOCK_UN)
+    except ImportError:
+        UNSUPPORTED_MSG = 'file locking is not supported on this platform'
+
+        def _lock_file(f, exclusive):
+            raise IOError(UNSUPPORTED_MSG)
+
+        def _unlock_file(f):
+            raise IOError(UNSUPPORTED_MSG)
 
 
 class locked_file(object):
@@ -1236,7 +1410,7 @@ def shell_quote(args):
 def smuggle_url(url, data):
     """ Pass additional data in a URL for internal use. """
 
-    sdata = compat_urllib_parse.urlencode(
+    sdata = compat_urllib_parse_urlencode(
         {'__youtubedl_smuggle': json.dumps(data)})
     return url + '#' + sdata
 
@@ -1264,11 +1438,22 @@ def format_bytes(bytes):
     return '%.2f%s' % (converted, suffix)
 
 
+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 = m.group('num').replace(',', '.')
+    mult = unit_table[m.group('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 inofficial,
+    # The lower-case forms are of course incorrect and unofficial,
     # but we support those too
     _UNIT_TABLE = {
         'B': 1,
@@ -1307,15 +1492,28 @@ def parse_filesize(s):
         'Yb': 1000 ** 8,
     }
 
-    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)' % units_re, s)
-    if not m:
+    return lookup_unit_table(_UNIT_TABLE, s)
+
+
+def parse_count(s):
+    if s is None:
         return None
 
-    num_str = m.group('num').replace(',', '.')
-    mult = _UNIT_TABLE[m.group('unit')]
-    return int(float(num_str) * mult)
+    s = s.strip()
+
+    if re.match(r'^[\d,.]+$', s):
+        return str_to_int(s)
+
+    _UNIT_TABLE = {
+        'k': 1000,
+        'K': 1000,
+        'm': 1000 ** 2,
+        'M': 1000 ** 2,
+        'kk': 1000 ** 2,
+        'KK': 1000 ** 2,
+    }
+
+    return lookup_unit_table(_UNIT_TABLE, s)
 
 
 def month_by_name(name):
@@ -1347,8 +1545,14 @@ def fix_xml_ampersands(xml_str):
 
 def setproctitle(title):
     assert isinstance(title, compat_str)
+
+    # ctypes in Jython is not complete
+    # http://bugs.jython.org/issue2148
+    if sys.platform.startswith('java'):
+        return
+
     try:
-        libc = ctypes.cdll.LoadLibrary("libc.so.6")
+        libc = ctypes.cdll.LoadLibrary('libc.so.6')
     except OSError:
         return
     title_bytes = title.encode('utf-8')
@@ -1361,14 +1565,19 @@ def setproctitle(title):
 
 
 def remove_start(s, start):
-    if s.startswith(start):
-        return s[len(start):]
-    return s
+    return s[len(start):] if s is not None and s.startswith(start) else s
 
 
 def remove_end(s, end):
-    if s.endswith(end):
-        return s[:-len(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
 
 
@@ -1379,7 +1588,7 @@ def url_basename(url):
 
 class HEADRequest(compat_urllib_request.Request):
     def get_method(self):
-        return "HEAD"
+        return 'HEAD'
 
 
 def int_or_none(v, scale=1, default=None, get_attr=None, invscale=1):
@@ -1423,44 +1632,46 @@ def parse_duration(s):
 
     s = s.strip()
 
-    m = re.match(
-        r'''(?ix)(?:P?T)?
-        (?:
-            (?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*|
-            (?:
+    days, hours, mins, secs, ms = [None] * 5
+    m = re.match(r'(?:(?:(?:(?P<days>[0-9]+):)?(?P<hours>[0-9]+):)?(?P<mins>[0-9]+):)?(?P<secs>[0-9]+)(?P<ms>\.[0-9]+)?$', s)
+    if m:
+        days, hours, mins, secs, ms = m.groups()
+    else:
+        m = re.match(
+            r'''(?ix)(?:P?T)?
                 (?:
-                    (?:(?P<days>[0-9]+)\s*(?:[:d]|days?)\s*)?
-                    (?P<hours>[0-9]+)\s*(?:[:h]|hours?)\s*
+                    (?P<days>[0-9]+)\s*d(?:ays?)?\s*
                 )?
-                (?P<mins>[0-9]+)\s*(?:[:m]|mins?|minutes?)\s*
-            )?
-            (?P<secs>[0-9]+)(?P<ms>\.[0-9]+)?\s*(?:s|secs?|seconds?)?
-        )$''', s)
-    if not m:
-        return None
-    res = 0
-    if m.group('only_mins'):
-        return float_or_none(m.group('only_mins'), invscale=60)
-    if m.group('only_hours'):
-        return float_or_none(m.group('only_hours'), invscale=60 * 60)
-    if m.group('secs'):
-        res += int(m.group('secs'))
-    if m.group('mins_reversed'):
-        res += int(m.group('mins_reversed')) * 60
-    if m.group('mins'):
-        res += int(m.group('mins')) * 60
-    if m.group('hours'):
-        res += int(m.group('hours')) * 60 * 60
-    if m.group('hours_reversed'):
-        res += int(m.group('hours_reversed')) * 60 * 60
-    if m.group('days'):
-        res += int(m.group('days')) * 24 * 60 * 60
-    if m.group('ms'):
-        res += float(m.group('ms'))
-    return res
+                (?:
+                    (?P<hours>[0-9]+)\s*h(?:ours?)?\s*
+                )?
+                (?:
+                    (?P<mins>[0-9]+)\s*m(?:in(?:ute)?s?)?\s*
+                )?
+                (?:
+                    (?P<secs>[0-9]+)(?P<ms>\.[0-9]+)?\s*s(?:ec(?:ond)?s?)?\s*
+                )?$''', s)
+        if m:
+            days, hours, mins, secs, ms = m.groups()
+        else:
+            m = re.match(r'(?i)(?:(?P<hours>[0-9.]+)\s*(?:hours?)|(?P<mins>[0-9.]+)\s*(?:mins?\.?|minutes?)\s*)$', s)
+            if m:
+                hours, mins = m.groups()
+            else:
+                return None
+
+    duration = 0
+    if secs:
+        duration += float(secs)
+    if mins:
+        duration += float(mins) * 60
+    if hours:
+        duration += float(hours) * 60 * 60
+    if days:
+        duration += float(days) * 24 * 60 * 60
+    if ms:
+        duration += float(ms)
+    return duration
 
 
 def prepend_extension(filename, ext, expected_real_ext=None):
@@ -1521,9 +1732,12 @@ class PagedList(object):
 
 
 class OnDemandPagedList(PagedList):
-    def __init__(self, pagefunc, pagesize):
+    def __init__(self, pagefunc, pagesize, use_cache=False):
         self._pagefunc = pagefunc
         self._pagesize = pagesize
+        self._use_cache = use_cache
+        if use_cache:
+            self._cache = {}
 
     def getslice(self, start=0, end=None):
         res = []
@@ -1533,7 +1747,13 @@ class OnDemandPagedList(PagedList):
             if start >= nextfirstid:
                 continue
 
-            page_results = list(self._pagefunc(pagenum))
+            page_results = None
+            if self._use_cache:
+                page_results = self._cache.get(pagenum)
+            if page_results is None:
+                page_results = list(self._pagefunc(pagenum))
+            if self._use_cache:
+                self._cache[pagenum] = page_results
 
             startv = (
                 start % self._pagesize
@@ -1619,29 +1839,13 @@ def escape_url(url):
     """Escape URL as suggested by RFC 3986"""
     url_parsed = compat_urllib_parse_urlparse(url)
     return url_parsed._replace(
+        netloc=url_parsed.netloc.encode('idna').decode('ascii'),
         path=escape_rfc3986(url_parsed.path),
         params=escape_rfc3986(url_parsed.params),
         query=escape_rfc3986(url_parsed.query),
         fragment=escape_rfc3986(url_parsed.fragment)
     ).geturl()
 
-try:
-    struct.pack('!I', 0)
-except TypeError:
-    # In Python 2.6 (and some 2.7 versions), struct requires a bytes argument
-    def struct_pack(spec, *args):
-        if isinstance(spec, compat_str):
-            spec = spec.encode('ascii')
-        return struct.pack(spec, *args)
-
-    def struct_unpack(spec, *args):
-        if isinstance(spec, compat_str):
-            spec = spec.encode('ascii')
-        return struct.unpack(spec, *args)
-else:
-    struct_pack = struct.pack
-    struct_unpack = struct.unpack
-
 
 def read_batch_urls(batch_fd):
     def fixup(url):
@@ -1660,11 +1864,55 @@ def read_batch_urls(batch_fd):
 
 
 def urlencode_postdata(*args, **kargs):
-    return compat_urllib_parse.urlencode(*args, **kargs).encode('ascii')
+    return compat_urllib_parse_urlencode(*args, **kargs).encode('ascii')
 
 
-def encode_dict(d, encoding='utf-8'):
-    return dict((k.encode(encoding), v.encode(encoding)) for k, v in d.items())
+def update_url_query(url, query):
+    if not query:
+        return url
+    parsed_url = compat_urlparse.urlparse(url)
+    qs = compat_parse_qs(parsed_url.query)
+    qs.update(query)
+    return compat_urlparse.urlunparse(parsed_url._replace(
+        query=compat_urllib_parse_urlencode(qs, True)))
+
+
+def update_Request(req, url=None, data=None, headers={}, query={}):
+    req_headers = req.headers.copy()
+    req_headers.update(headers)
+    req_data = data or req.data
+    req_url = update_url_query(url or req.get_full_url(), query)
+    req_type = HEADRequest if req.get_method() == 'HEAD' else compat_urllib_request.Request
+    new_req = req_type(
+        req_url, data=req_data, headers=req_headers,
+        origin_req_host=req.origin_req_host, unverifiable=req.unverifiable)
+    if hasattr(req, 'timeout'):
+        new_req.timeout = req.timeout
+    return new_req
+
+
+def dict_get(d, key_or_keys, default=None, skip_false_values=True):
+    if isinstance(key_or_keys, (list, tuple)):
+        for key in key_or_keys:
+            if key not in d or d[key] is None or skip_false_values and not d[key]:
+                continue
+            return d[key]
+        return default
+    return d.get(key_or_keys, default)
+
+
+def try_get(src, getter, expected_type=None):
+    try:
+        v = getter(src)
+    except (AttributeError, KeyError, TypeError, IndexError):
+        pass
+    else:
+        if expected_type is None or isinstance(v, expected_type):
+            return v
+
+
+def encode_compat_str(string, encoding=preferredencoding(), errors='strict'):
+    return string if isinstance(string, compat_str) else compat_str(string, encoding, errors)
 
 
 US_RATINGS = {
@@ -1680,12 +1928,12 @@ def parse_age_limit(s):
     if s is None:
         return None
     m = re.match(r'^(?P<age>\d{1,2})\+?$', s)
-    return int(m.group('age')) if m else US_RATINGS.get(s, None)
+    return int(m.group('age')) if m else US_RATINGS.get(s)
 
 
 def strip_jsonp(code):
     return re.sub(
-        r'(?s)^[a-zA-Z0-9_]+\s*\(\s*(.*)\);?\s*?(?://[^\n]*)*$', r'\1', code)
+        r'(?s)^[a-zA-Z0-9_.$]+\s*\(\s*(.*)\);?\s*?(?://[^\n]*)*$', r'\1', code)
 
 
 def js_to_json(code):
@@ -1693,24 +1941,38 @@ def js_to_json(code):
         v = m.group(0)
         if v in ('true', 'false', 'null'):
             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: {
-                '\\\\': '\\\\',
-                "\\'": "'",
+        elif v.startswith('/*') or v == ',':
+            return ""
+
+        if v[0] in ("'", '"'):
+            v = re.sub(r'(?s)\\.|"', lambda m: {
                 '"': '\\"',
-            }[m.group(0)], v)
+                "\\'": "'",
+                '\\\n': '',
+                '\\x': '\\u00',
+            }.get(m.group(0), m.group(0)), v[1:-1])
+
+        INTEGER_TABLE = (
+            (r'^0[xX][0-9a-fA-F]+', 16),
+            (r'^0+[0-7]+', 8),
+        )
+
+        for regex, base in INTEGER_TABLE:
+            im = re.match(regex, v)
+            if im:
+                i = int(im.group(0), base)
+                return '"%d":' % i if v.endswith(':') else '%d' % i
+
         return '"%s"' % v
 
-    res = re.sub(r'''(?x)
-        "(?:[^"\\]*(?:\\\\|\\['"nu]))*[^"\\]*"|
-        '(?:[^'\\]*(?:\\\\|\\['"nu]))*[^'\\]*'|
-        [a-zA-Z_][.a-zA-Z_0-9]*
+    return re.sub(r'''(?sx)
+        "(?:[^"\\]*(?:\\\\|\\['"nurtbfx/\n]))*[^"\\]*"|
+        '(?:[^'\\]*(?:\\\\|\\['"nurtbfx/\n]))*[^'\\]*'|
+        /\*.*?\*/|,(?=\s*[\]}])|
+        [a-zA-Z_][.a-zA-Z_0-9]*|
+        \b(?:0[xX][0-9a-fA-F]+|0+[0-7]+)(?:\s*:)?|
+        [0-9]+(?=\s*:)
         ''', fix_kv, code)
-    res = re.sub(r',(\s*[\]}])', lambda m: m.group(1), res)
-    return res
 
 
 def qualities(quality_ids):
@@ -1758,25 +2020,48 @@ def ytdl_is_updateable():
 
 def args_to_str(args):
     # Get a short string representation for a subprocess command
-    return ' '.join(shlex_quote(a) for a in args)
+    return ' '.join(compat_shlex_quote(a) for a in args)
+
+
+def error_to_compat_str(err):
+    err_str = str(err)
+    # On python 2 error byte string must be decoded with proper
+    # encoding rather than ascii
+    if sys.version_info[0] < 3:
+        err_str = err_str.decode(preferredencoding())
+    return err_str
 
 
 def mimetype2ext(mt):
+    if mt is None:
+        return None
+
+    ext = {
+        'audio/mp4': 'm4a',
+        # Per RFC 3003, audio/mpeg can be .mp1, .mp2 or .mp3. Here use .mp3 as
+        # it's the most popular one
+        'audio/mpeg': 'mp3',
+    }.get(mt)
+    if ext is not None:
+        return ext
+
     _, _, res = mt.rpartition('/')
 
     return {
-        'x-ms-wmv': 'wmv',
-        'x-mp4-fragmented': 'mp4',
+        '3gpp': '3gp',
+        'smptett+xml': 'tt',
+        'srt': 'srt',
+        'ttaf+xml': 'dfxp',
         'ttml+xml': 'ttml',
+        'vtt': 'vtt',
+        'x-flv': 'flv',
+        'x-mp4-fragmented': 'mp4',
+        'x-ms-wmv': 'wmv',
     }.get(res, res)
 
 
 def urlhandle_detect_ext(url_handle):
-    try:
-        url_handle.headers
-        getheader = lambda h: url_handle.headers[h]
-    except AttributeError:  # Python < 3
-        getheader = url_handle.info().getheader
+    getheader = url_handle.headers.get
 
     cd = getheader('Content-Disposition')
     if cd:
@@ -1931,15 +2216,15 @@ def match_filter_func(filter_str):
 
 def parse_dfxp_time_expr(time_expr):
     if not time_expr:
-        return 0.0
+        return
 
     mobj = re.match(r'^(?P<time_offset>\d+(?:\.\d+)?)s?$', time_expr)
     if mobj:
         return float(mobj.group('time_offset'))
 
-    mobj = re.match(r'^(\d+):(\d\d):(\d\d(?:\.\d+)?)$', time_expr)
+    mobj = re.match(r'^(\d+):(\d\d):(\d\d(?:(?:\.|:)\d+)?)$', time_expr)
     if mobj:
-        return 3600 * int(mobj.group(1)) + 60 * int(mobj.group(2)) + float(mobj.group(3))
+        return 3600 * int(mobj.group(1)) + 60 * int(mobj.group(2)) + float(mobj.group(3).replace(':', '.'))
 
 
 def srt_subtitles_timecode(seconds):
@@ -1950,35 +2235,48 @@ def dfxp2srt(dfxp_data):
     _x = functools.partial(xpath_with_ns, ns_map={
         'ttml': 'http://www.w3.org/ns/ttml',
         'ttaf1': 'http://www.w3.org/2006/10/ttaf1',
+        'ttaf1_0604': 'http://www.w3.org/2006/04/ttaf1',
     })
 
-    def parse_node(node):
-        str_or_empty = functools.partial(str_or_none, default='')
+    class TTMLPElementParser(object):
+        out = ''
 
-        out = str_or_empty(node.text)
+        def start(self, tag, attrib):
+            if tag in (_x('ttml:br'), _x('ttaf1:br'), 'br'):
+                self.out += '\n'
 
-        for child in node:
-            if child.tag in (_x('ttml:br'), _x('ttaf1:br'), 'br'):
-                out += '\n' + str_or_empty(child.tail)
-            elif child.tag in (_x('ttml:span'), _x('ttaf1:span'), 'span'):
-                out += str_or_empty(parse_node(child))
-            else:
-                out += str_or_empty(xml.etree.ElementTree.tostring(child))
+        def end(self, tag):
+            pass
 
-        return out
+        def data(self, data):
+            self.out += data
+
+        def close(self):
+            return self.out.strip()
+
+    def parse_node(node):
+        target = TTMLPElementParser()
+        parser = xml.etree.ElementTree.XMLParser(target=target)
+        parser.feed(xml.etree.ElementTree.tostring(node))
+        return parser.close()
 
     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')
+    paras = dfxp.findall(_x('.//ttml:p')) or dfxp.findall(_x('.//ttaf1:p')) or dfxp.findall(_x('.//ttaf1_0604:p')) or dfxp.findall('.//p')
 
     if not paras:
         raise ValueError('Invalid dfxp/TTML subtitle')
 
     for para, index in zip(paras, itertools.count(1)):
-        begin_time = parse_dfxp_time_expr(para.attrib['begin'])
+        begin_time = parse_dfxp_time_expr(para.attrib.get('begin'))
         end_time = parse_dfxp_time_expr(para.attrib.get('end'))
+        dur = parse_dfxp_time_expr(para.attrib.get('dur'))
+        if begin_time is None:
+            continue
         if not end_time:
-            end_time = begin_time + parse_dfxp_time_expr(para.attrib['dur'])
+            if not dur:
+                continue
+            end_time = begin_time + dur
         out.append('%d\n%s --> %s\n%s\n\n' % (
             index,
             srt_subtitles_timecode(begin_time),
@@ -2493,5 +2791,73 @@ class PerRequestProxyHandler(compat_urllib_request.ProxyHandler):
 
         if proxy == '__noproxy__':
             return None  # No Proxy
+        if compat_urlparse.urlparse(proxy).scheme.lower() in ('socks', 'socks4', 'socks4a', 'socks5'):
+            req.add_header('Ytdl-socks-proxy', proxy)
+            # youtube-dl's http/https handlers do wrapping the socket with socks
+            return None
         return compat_urllib_request.ProxyHandler.proxy_open(
             self, req, proxy, type)
+
+
+def ohdave_rsa_encrypt(data, exponent, modulus):
+    '''
+    Implement OHDave's RSA algorithm. See http://www.ohdave.com/rsa/
+
+    Input:
+        data: data to encrypt, bytes-like object
+        exponent, modulus: parameter e and N of RSA algorithm, both integer
+    Output: hex string of encrypted data
+
+    Limitation: supports one block encryption only
+    '''
+
+    payload = int(binascii.hexlify(data[::-1]), 16)
+    encrypted = pow(payload, exponent, modulus)
+    return '%x' % encrypted
+
+
+def encode_base_n(num, n, table=None):
+    FULL_TABLE = '0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ'
+    if not table:
+        table = FULL_TABLE[:n]
+
+    if n > len(table):
+        raise ValueError('base %d exceeds table length %d' % (n, len(table)))
+
+    if num == 0:
+        return table[0]
+
+    ret = ''
+    while num:
+        ret = table[num % n] + ret
+        num = num // n
+    return ret
+
+
+def decode_packed_codes(code):
+    mobj = re.search(
+        r"}\('(.+)',(\d+),(\d+),'([^']+)'\.split\('\|'\)",
+        code)
+    obfucasted_code, base, count, symbols = mobj.groups()
+    base = int(base)
+    count = int(count)
+    symbols = symbols.split('|')
+    symbol_table = {}
+
+    while count:
+        count -= 1
+        base_n_count = encode_base_n(count, base)
+        symbol_table[base_n_count] = symbols[count] or base_n_count
+
+    return re.sub(
+        r'\b(\w+)\b', lambda mobj: symbol_table[mobj.group(0)],
+        obfucasted_code)
+
+
+def parse_m3u8_attributes(attrib):
+    info = {}
+    for (key, val) in re.findall(r'(?P<key>[A-Z0-9-]+)=(?P<val>"[^"]+"|[^",]+)(?:,|$)', attrib):
+        if val.startswith('"'):
+            val = val[1:-1]
+        info[key] = val
+    return info