X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Futils.py;h=f77ab865093604f217ab6b2da824f340327df652;hb=daa0df9e8beac1325e5fb55d828e7a3a38e74bf6;hp=9fd0ec8d5856cbee27534c0f4a02cc90b05f8389;hpb=49dea4913bea3b8e5c7d65dd932aa68ada526088;p=youtube-dl diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py index 9fd0ec8d5..f77ab8650 100644 --- a/youtube_dl/utils.py +++ b/youtube_dl/utils.py @@ -14,8 +14,8 @@ import email.utils import errno import functools import gzip -import itertools import io +import itertools import json import locale import math @@ -24,9 +24,8 @@ import os import pipes import platform import re -import ssl import socket -import struct +import ssl import subprocess import sys import tempfile @@ -35,23 +34,43 @@ 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('')) @@ -86,6 +105,11 @@ KNOWN_EXTENSIONS = ( '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. @@ -164,12 +188,7 @@ if sys.version_info >= (2, 7): 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: @@ -194,9 +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) @@ -255,9 +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.*?) @@ -273,6 +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 + + 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""" @@ -337,6 +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 == '"': @@ -389,9 +440,12 @@ def sanitize_path(s): # 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( - 'http:%s' % url if url.startswith('//') else url, *args, **kwargs) + return compat_urllib_request.Request(sanitize_url(url), *args, **kwargs) def orderedSet(iterable): @@ -403,12 +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, + # 'Éric' 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) @@ -433,7 +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(): @@ -714,8 +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 @@ -747,12 +815,7 @@ 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 @@ -806,9 +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 @@ -816,6 +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) @@ -824,12 +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) @@ -929,6 +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', @@ -949,7 +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) @@ -1160,7 +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 @@ -1288,7 +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 @@ -1319,7 +1441,7 @@ def format_bytes(bytes): def lookup_unit_table(unit_table, s): units_re = '|'.join(re.escape(u) for u in unit_table) m = re.match( - r'(?P[0-9]+(?:[,.][0-9]*)?)\s*(?P%s)' % units_re, s) + r'(?P[0-9]+(?:[,.][0-9]*)?)\s*(?P%s)\b' % units_re, s) if not m: return None num_str = m.group('num').replace(',', '.') @@ -1443,15 +1565,11 @@ 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 + return s[:-len(end)] if s is not None and s.endswith(end) else s def remove_quotes(s): @@ -1514,44 +1632,46 @@ def parse_duration(s): s = s.strip() - m = re.match( - r'''(?ix)(?:P?T)? - (?: - (?P[0-9.]+)\s*(?:mins?\.?|minutes?)\s*| - (?P[0-9.]+)\s*(?:hours?)| - - \s*(?P[0-9]+)\s*(?:[:h]|hours?)\s*(?P[0-9]+)\s*(?:[:m]|mins?\.?|minutes?)\s*| - (?: + days, hours, mins, secs, ms = [None] * 5 + m = re.match(r'(?:(?:(?:(?P[0-9]+):)?(?P[0-9]+):)?(?P[0-9]+):)?(?P[0-9]+)(?P\.[0-9]+)?$', s) + if m: + days, hours, mins, secs, ms = m.groups() + else: + m = re.match( + r'''(?ix)(?:P?T)? (?: - (?:(?P[0-9]+)\s*(?:[:d]|days?)\s*)? - (?P[0-9]+)\s*(?:[:h]|hours?)\s* + (?P[0-9]+)\s*d(?:ays?)?\s* )? - (?P[0-9]+)\s*(?:[:m]|mins?|minutes?)\s* - )? - (?P[0-9]+)(?P\.[0-9]+)?\s*(?:s|secs?|seconds?)? - )$''', 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[0-9]+)\s*h(?:ours?)?\s* + )? + (?: + (?P[0-9]+)\s*m(?:in(?:ute)?s?)?\s* + )? + (?: + (?P[0-9]+)(?P\.[0-9]+)?\s*s(?:ec(?:ond)?s?)?\s* + )?$''', s) + if m: + days, hours, mins, secs, ms = m.groups() + else: + m = re.match(r'(?i)(?:(?P[0-9.]+)\s*(?:hours?)|(?P[0-9.]+)\s*(?:mins?\.?|minutes?)\s*)$', 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): @@ -1719,29 +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): @@ -1760,22 +1864,31 @@ 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 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) - qs = encode_dict(qs) return compat_urlparse.urlunparse(parsed_url._replace( - query=compat_urllib_parse.urlencode(qs, True))) + query=compat_urllib_parse_urlencode(qs, True))) -def encode_dict(d, encoding='utf-8'): - def encode(v): - return v.encode(encoding) if isinstance(v, compat_basestring) else v - return dict((encode(k), encode(v)) for k, v in d.items()) +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): @@ -1810,7 +1923,7 @@ def parse_age_limit(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): @@ -1818,24 +1931,38 @@ 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]*| + (?: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): @@ -1883,7 +2010,7 @@ 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): @@ -1896,8 +2023,14 @@ def error_to_compat_str(err): 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 @@ -1918,11 +2051,7 @@ def mimetype2ext(mt): 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: @@ -2096,6 +2225,7 @@ 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', }) class TTMLPElementParser(object): @@ -2122,7 +2252,7 @@ def dfxp2srt(dfxp_data): 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') @@ -2651,6 +2781,10 @@ 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)