X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/a316b1d93c357b5edf19d5e3100526a94191c029..a3df54c98511f5038b0a4c959f87a1220198e076:/youtube_dl/extractor/youtube.py diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index 9e2373b..a88cba2 100644 --- a/youtube_dl/extractor/youtube.py +++ b/youtube_dl/extractor/youtube.py @@ -1,26 +1,39 @@ # coding: utf-8 +import collections +import errno +import io +import itertools import json -import netrc +import os.path import re import socket -import itertools +import string +import struct +import traceback +import xml.etree.ElementTree +import zlib from .common import InfoExtractor, SearchInfoExtractor +from .subtitles import SubtitlesInfoExtractor from ..utils import ( + compat_chr, compat_http_client, compat_parse_qs, compat_urllib_error, compat_urllib_parse, compat_urllib_request, + compat_urlparse, compat_str, clean_html, + get_cachedir, get_element_by_id, ExtractorError, unescapeHTML, unified_strdate, orderedSet, + write_json_file, ) class YoutubeBaseInfoExtractor(InfoExtractor): @@ -130,13 +143,15 @@ class YoutubeBaseInfoExtractor(InfoExtractor): return self._confirm_age() -class YoutubeIE(YoutubeBaseInfoExtractor): + +class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor): IE_DESC = u'YouTube.com' _VALID_URL = r"""^ ( (?:https?://)? # http(s):// (optional) - (?:youtu\.be/|(?:\w+\.)?youtube(?:-nocookie)?\.com/| - tube\.majestyc\.net/) # the various hostnames, with wildcard subdomains + (?:(?:(?:(?:\w+\.)?youtube(?:-nocookie)?\.com/| + tube\.majestyc\.net/| + youtube\.googleapis\.com/) # the various hostnames, with wildcard subdomains (?:.*?\#/)? # handle anchor (#/) redirect urls (?: # the various things that can precede the ID: (?:(?:v|embed|e)/) # v/ or embed/ or e/ @@ -146,15 +161,18 @@ class YoutubeIE(YoutubeBaseInfoExtractor): (?:.*?&)? # any other preceding param (like /?s=tuff&v=xxxx) v= ) - )? # optional -> youtube.com/xxxx is OK + )) + |youtu\.be/ # just youtu.be/xxxx + ) )? # all until now is optional -> you can pass the naked ID - ([0-9A-Za-z_-]+) # here is it! the YouTube video ID + ([0-9A-Za-z_-]{11}) # here is it! the YouTube video ID (?(1).+)? # if we found the ID, everything can follow $""" _NEXT_URL_RE = r'[\?&]next_url=([^&]+)' # Listed in order of quality - _available_formats = ['38', '37', '46', '22', '45', '35', '44', '34', '18', '43', '6', '5', '17', '13', - '95', '94', '93', '92', '132', '151', + _available_formats = ['38', '37', '46', '22', '45', '35', '44', '34', '18', '43', '6', '5', '36', '17', '13', + # Apple HTTP Live Streaming + '96', '95', '94', '93', '92', '132', '151', # 3D '85', '84', '102', '83', '101', '82', '100', # Dash video @@ -163,8 +181,10 @@ class YoutubeIE(YoutubeBaseInfoExtractor): # Dash audio '141', '172', '140', '171', '139', ] - _available_formats_prefer_free = ['38', '46', '37', '45', '22', '44', '35', '43', '34', '18', '6', '5', '17', '13', - '95', '94', '93', '92', '132', '151', + _available_formats_prefer_free = ['38', '46', '37', '45', '22', '44', '35', '43', '34', '18', '6', '5', '36', '17', '13', + # Apple HTTP Live Streaming + '96', '95', '94', '93', '92', '132', '151', + # 3D '85', '102', '84', '101', '83', '100', '82', # Dash video '138', '248', '137', '247', '136', '246', '245', @@ -172,11 +192,18 @@ class YoutubeIE(YoutubeBaseInfoExtractor): # Dash audio '172', '141', '171', '140', '139', ] + _video_formats_map = { + 'flv': ['35', '34', '6', '5'], + '3gp': ['36', '17', '13'], + 'mp4': ['38', '37', '22', '18'], + 'webm': ['46', '45', '44', '43'], + } _video_extensions = { '13': '3gp', - '17': 'mp4', + '17': '3gp', '18': 'mp4', '22': 'mp4', + '36': '3gp', '37': 'mp4', '38': 'mp4', '43': 'webm', @@ -193,7 +220,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): '101': 'webm', '102': 'webm', - # videos that use m3u8 + # Apple HTTP Live Streaming '92': 'mp4', '93': 'mp4', '94': 'mp4', @@ -234,6 +261,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): '22': '720x1280', '34': '360x640', '35': '480x854', + '36': '240x320', '37': '1080x1920', '38': '3072x4096', '43': '360x640', @@ -335,7 +363,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): u"info_dict": { u"upload_date": u"20120506", u"title": u"Icona Pop - I Love It (feat. Charli XCX) [OFFICIAL VIDEO]", - u"description": u"md5:3e2666e0a55044490499ea45fe9037b7", + u"description": u"md5:5b292926389560516e384ac437c0ec07", u"uploader": u"Icona Pop", u"uploader_id": u"IconaPop" } @@ -352,30 +380,19 @@ class YoutubeIE(YoutubeBaseInfoExtractor): u"uploader_id": u"justintimberlakeVEVO" } }, - { - u'url': u'https://www.youtube.com/watch?v=TGi3HqYrWHE', - u'file': u'TGi3HqYrWHE.mp4', - u'note': u'm3u8 video', - u'info_dict': { - u'title': u'Triathlon - Men - London 2012 Olympic Games', - u'description': u'- Men - TR02 - Triathlon - 07 August 2012 - London 2012 Olympic Games', - u'uploader': u'olympic', - u'upload_date': u'20120807', - u'uploader_id': u'olympic', - }, - u'params': { - u'skip_download': True, - }, - }, ] @classmethod def suitable(cls, url): """Receives a URL and returns True if suitable for this IE.""" - if YoutubePlaylistIE.suitable(url) or YoutubeSubscriptionsIE.suitable(url): return False + if YoutubePlaylistIE.suitable(url): return False return re.match(cls._VALID_URL, url, re.VERBOSE) is not None + def __init__(self, *args, **kwargs): + super(YoutubeIE, self).__init__(*args, **kwargs) + self._player_cache = {} + def report_video_webpage_download(self, video_id): """Report attempt to download video webpage.""" self.to_screen(u'%s: Downloading video webpage' % video_id) @@ -384,19 +401,6 @@ class YoutubeIE(YoutubeBaseInfoExtractor): """Report attempt to download video info webpage.""" self.to_screen(u'%s: Downloading video info webpage' % video_id) - def report_video_subtitles_download(self, video_id): - """Report attempt to download video info webpage.""" - self.to_screen(u'%s: Checking available subtitles' % video_id) - - def report_video_subtitles_request(self, video_id, sub_lang, format): - """Report attempt to download video info webpage.""" - self.to_screen(u'%s: Downloading video subtitles for %s.%s' % (video_id, sub_lang, format)) - - def report_video_subtitles_available(self, video_id, sub_lang_list): - """Report available subtitles.""" - sub_lang = ",".join(list(sub_lang_list.keys())) - self.to_screen(u'%s: Available subtitles for video: %s' % (video_id, sub_lang)) - def report_information_extraction(self, video_id): """Report attempt to extract video information.""" self.to_screen(u'%s: Extracting video information' % video_id) @@ -409,11 +413,664 @@ class YoutubeIE(YoutubeBaseInfoExtractor): """Indicate the download will use the RTMP protocol.""" self.to_screen(u'RTMP download detected') - def _decrypt_signature(self, s): + def _extract_signature_function(self, video_id, player_url, slen): + id_m = re.match(r'.*-(?P[a-zA-Z0-9_-]+)\.(?P[a-z]+)$', + player_url) + player_type = id_m.group('ext') + player_id = id_m.group('id') + + # Read from filesystem cache + func_id = '%s_%s_%d' % (player_type, player_id, slen) + assert os.path.basename(func_id) == func_id + cache_dir = get_cachedir(self._downloader.params) + + cache_enabled = cache_dir is not None + if cache_enabled: + cache_fn = os.path.join(os.path.expanduser(cache_dir), + u'youtube-sigfuncs', + func_id + '.json') + try: + with io.open(cache_fn, 'r', encoding='utf-8') as cachef: + cache_spec = json.load(cachef) + return lambda s: u''.join(s[i] for i in cache_spec) + except IOError: + pass # No cache available + + if player_type == 'js': + code = self._download_webpage( + player_url, video_id, + note=u'Downloading %s player %s' % (player_type, player_id), + errnote=u'Download of %s failed' % player_url) + res = self._parse_sig_js(code) + elif player_type == 'swf': + urlh = self._request_webpage( + player_url, video_id, + note=u'Downloading %s player %s' % (player_type, player_id), + errnote=u'Download of %s failed' % player_url) + code = urlh.read() + res = self._parse_sig_swf(code) + else: + assert False, 'Invalid player type %r' % player_type + + if cache_enabled: + try: + test_string = u''.join(map(compat_chr, range(slen))) + cache_res = res(test_string) + cache_spec = [ord(c) for c in cache_res] + try: + os.makedirs(os.path.dirname(cache_fn)) + except OSError as ose: + if ose.errno != errno.EEXIST: + raise + write_json_file(cache_spec, cache_fn) + except Exception: + tb = traceback.format_exc() + self._downloader.report_warning( + u'Writing cache to %r failed: %s' % (cache_fn, tb)) + + return res + + def _print_sig_code(self, func, slen): + def gen_sig_code(idxs): + def _genslice(start, end, step): + starts = u'' if start == 0 else str(start) + ends = (u':%d' % (end+step)) if end + step >= 0 else u':' + steps = u'' if step == 1 else (u':%d' % step) + return u's[%s%s%s]' % (starts, ends, steps) + + step = None + start = '(Never used)' # Quelch pyflakes warnings - start will be + # set as soon as step is set + for i, prev in zip(idxs[1:], idxs[:-1]): + if step is not None: + if i - prev == step: + continue + yield _genslice(start, prev, step) + step = None + continue + if i - prev in [-1, 1]: + step = i - prev + start = prev + continue + else: + yield u's[%d]' % prev + if step is None: + yield u's[%d]' % i + else: + yield _genslice(start, i, step) + + test_string = u''.join(map(compat_chr, range(slen))) + cache_res = func(test_string) + cache_spec = [ord(c) for c in cache_res] + expr_code = u' + '.join(gen_sig_code(cache_spec)) + code = u'if len(s) == %d:\n return %s\n' % (slen, expr_code) + self.to_screen(u'Extracted signature function:\n' + code) + + def _parse_sig_js(self, jscode): + funcname = self._search_regex( + r'signature=([a-zA-Z]+)', jscode, + u'Initial JS player signature function name') + + functions = {} + + def argidx(varname): + return string.lowercase.index(varname) + + def interpret_statement(stmt, local_vars, allow_recursion=20): + if allow_recursion < 0: + raise ExtractorError(u'Recursion limit reached') + + if stmt.startswith(u'var '): + stmt = stmt[len(u'var '):] + ass_m = re.match(r'^(?P[a-z]+)(?:\[(?P[^\]]+)\])?' + + r'=(?P.*)$', stmt) + if ass_m: + if ass_m.groupdict().get('index'): + def assign(val): + lvar = local_vars[ass_m.group('out')] + idx = interpret_expression(ass_m.group('index'), + local_vars, allow_recursion) + assert isinstance(idx, int) + lvar[idx] = val + return val + expr = ass_m.group('expr') + else: + def assign(val): + local_vars[ass_m.group('out')] = val + return val + expr = ass_m.group('expr') + elif stmt.startswith(u'return '): + assign = lambda v: v + expr = stmt[len(u'return '):] + else: + raise ExtractorError( + u'Cannot determine left side of statement in %r' % stmt) + + v = interpret_expression(expr, local_vars, allow_recursion) + return assign(v) + + def interpret_expression(expr, local_vars, allow_recursion): + if expr.isdigit(): + return int(expr) + + if expr.isalpha(): + return local_vars[expr] + + m = re.match(r'^(?P[a-z]+)\.(?P.*)$', expr) + if m: + member = m.group('member') + val = local_vars[m.group('in')] + if member == 'split("")': + return list(val) + if member == 'join("")': + return u''.join(val) + if member == 'length': + return len(val) + if member == 'reverse()': + return val[::-1] + slice_m = re.match(r'slice\((?P.*)\)', member) + if slice_m: + idx = interpret_expression( + slice_m.group('idx'), local_vars, allow_recursion-1) + return val[idx:] + + m = re.match( + r'^(?P[a-z]+)\[(?P.+)\]$', expr) + if m: + val = local_vars[m.group('in')] + idx = interpret_expression(m.group('idx'), local_vars, + allow_recursion-1) + return val[idx] + + m = re.match(r'^(?P.+?)(?P[%])(?P.+?)$', expr) + if m: + a = interpret_expression(m.group('a'), + local_vars, allow_recursion) + b = interpret_expression(m.group('b'), + local_vars, allow_recursion) + return a % b + + m = re.match( + r'^(?P[a-zA-Z]+)\((?P[a-z0-9,]+)\)$', expr) + if m: + fname = m.group('func') + if fname not in functions: + functions[fname] = extract_function(fname) + argvals = [int(v) if v.isdigit() else local_vars[v] + for v in m.group('args').split(',')] + return functions[fname](argvals) + raise ExtractorError(u'Unsupported JS expression %r' % expr) + + def extract_function(funcname): + func_m = re.search( + r'function ' + re.escape(funcname) + + r'\((?P[a-z,]+)\){(?P[^}]+)}', + jscode) + argnames = func_m.group('args').split(',') + + def resf(args): + local_vars = dict(zip(argnames, args)) + for stmt in func_m.group('code').split(';'): + res = interpret_statement(stmt, local_vars) + return res + return resf + + initial_function = extract_function(funcname) + return lambda s: initial_function([s]) + + def _parse_sig_swf(self, file_contents): + if file_contents[1:3] != b'WS': + raise ExtractorError( + u'Not an SWF file; header is %r' % file_contents[:3]) + if file_contents[:1] == b'C': + content = zlib.decompress(file_contents[8:]) + else: + raise NotImplementedError(u'Unsupported compression format %r' % + file_contents[:1]) + + def extract_tags(content): + pos = 0 + while pos < len(content): + header16 = struct.unpack('> 6 + tag_len = header16 & 0x3f + if tag_len == 0x3f: + tag_len = struct.unpack('> 4 + methods = {} + if kind in [0x00, 0x06]: # Slot or Const + u30() # Slot id + u30() # type_name_idx + vindex = u30() + if vindex != 0: + read_byte() # vkind + elif kind in [0x01, 0x02, 0x03]: # Method / Getter / Setter + u30() # disp_id + method_idx = u30() + methods[multinames[trait_name_idx]] = method_idx + elif kind == 0x04: # Class + u30() # slot_id + u30() # classi + elif kind == 0x05: # Function + u30() # slot_id + function_idx = u30() + methods[function_idx] = multinames[trait_name_idx] + else: + raise ExtractorError(u'Unsupported trait kind %d' % kind) + + if attrs & 0x4 != 0: # Metadata present + metadata_count = u30() + for _c3 in range(metadata_count): + u30() # metadata index + + return methods + + # Classes + TARGET_CLASSNAME = u'SignatureDecipher' + searched_idx = multinames.index(TARGET_CLASSNAME) + searched_class_id = None + class_count = u30() + for class_id in range(class_count): + name_idx = u30() + if name_idx == searched_idx: + # We found the class we're looking for! + searched_class_id = class_id + u30() # super_name idx + flags = read_byte() + if flags & 0x08 != 0: # Protected namespace is present + u30() # protected_ns_idx + intrf_count = u30() + for _c2 in range(intrf_count): + u30() + u30() # iinit + trait_count = u30() + for _c2 in range(trait_count): + parse_traits_info() + + if searched_class_id is None: + raise ExtractorError(u'Target class %r not found' % + TARGET_CLASSNAME) + + method_names = {} + method_idxs = {} + for class_id in range(class_count): + u30() # cinit + trait_count = u30() + for _c2 in range(trait_count): + trait_methods = parse_traits_info() + if class_id == searched_class_id: + method_names.update(trait_methods.items()) + method_idxs.update(dict( + (idx, name) + for name, idx in trait_methods.items())) + + # Scripts + script_count = u30() + for _c in range(script_count): + u30() # init + trait_count = u30() + for _c2 in range(trait_count): + parse_traits_info() + + # Method bodies + method_body_count = u30() + Method = collections.namedtuple('Method', ['code', 'local_count']) + methods = {} + for _c in range(method_body_count): + method_idx = u30() + u30() # max_stack + local_count = u30() + u30() # init_scope_depth + u30() # max_scope_depth + code_length = u30() + code = read_bytes(code_length) + if method_idx in method_idxs: + m = Method(code, local_count) + methods[method_idxs[method_idx]] = m + exception_count = u30() + for _c2 in range(exception_count): + u30() # from + u30() # to + u30() # target + u30() # exc_type + u30() # var_name + trait_count = u30() + for _c2 in range(trait_count): + parse_traits_info() + + assert p + code_reader.tell() == len(code_tag) + assert len(methods) == len(method_idxs) + + method_pyfunctions = {} + + def extract_function(func_name): + if func_name in method_pyfunctions: + return method_pyfunctions[func_name] + if func_name not in methods: + raise ExtractorError(u'Cannot find function %r' % func_name) + m = methods[func_name] + + def resfunc(args): + registers = ['(this)'] + list(args) + [None] * m.local_count + stack = [] + coder = io.BytesIO(m.code) + while True: + opcode = struct.unpack('!B', coder.read(1))[0] + if opcode == 36: # pushbyte + v = struct.unpack('!B', coder.read(1))[0] + stack.append(v) + elif opcode == 44: # pushstring + idx = u30(coder) + stack.append(constant_strings[idx]) + elif opcode == 48: # pushscope + # We don't implement the scope register, so we'll just + # ignore the popped value + stack.pop() + elif opcode == 70: # callproperty + index = u30(coder) + mname = multinames[index] + arg_count = u30(coder) + args = list(reversed( + [stack.pop() for _ in range(arg_count)])) + obj = stack.pop() + if mname == u'split': + assert len(args) == 1 + assert isinstance(args[0], compat_str) + assert isinstance(obj, compat_str) + if args[0] == u'': + res = list(obj) + else: + res = obj.split(args[0]) + stack.append(res) + elif mname == u'slice': + assert len(args) == 1 + assert isinstance(args[0], int) + assert isinstance(obj, list) + res = obj[args[0]:] + stack.append(res) + elif mname == u'join': + assert len(args) == 1 + assert isinstance(args[0], compat_str) + assert isinstance(obj, list) + res = args[0].join(obj) + stack.append(res) + elif mname in method_pyfunctions: + stack.append(method_pyfunctions[mname](args)) + else: + raise NotImplementedError( + u'Unsupported property %r on %r' + % (mname, obj)) + elif opcode == 72: # returnvalue + res = stack.pop() + return res + elif opcode == 79: # callpropvoid + index = u30(coder) + mname = multinames[index] + arg_count = u30(coder) + args = list(reversed( + [stack.pop() for _ in range(arg_count)])) + obj = stack.pop() + if mname == u'reverse': + assert isinstance(obj, list) + obj.reverse() + else: + raise NotImplementedError( + u'Unsupported (void) property %r on %r' + % (mname, obj)) + elif opcode == 93: # findpropstrict + index = u30(coder) + mname = multinames[index] + res = extract_function(mname) + stack.append(res) + elif opcode == 97: # setproperty + index = u30(coder) + value = stack.pop() + idx = stack.pop() + obj = stack.pop() + assert isinstance(obj, list) + assert isinstance(idx, int) + obj[idx] = value + elif opcode == 98: # getlocal + index = u30(coder) + stack.append(registers[index]) + elif opcode == 99: # setlocal + index = u30(coder) + value = stack.pop() + registers[index] = value + elif opcode == 102: # getproperty + index = u30(coder) + pname = multinames[index] + if pname == u'length': + obj = stack.pop() + assert isinstance(obj, list) + stack.append(len(obj)) + else: # Assume attribute access + idx = stack.pop() + assert isinstance(idx, int) + obj = stack.pop() + assert isinstance(obj, list) + stack.append(obj[idx]) + elif opcode == 128: # coerce + u30(coder) + elif opcode == 133: # coerce_s + assert isinstance(stack[-1], (type(None), compat_str)) + elif opcode == 164: # modulo + value2 = stack.pop() + value1 = stack.pop() + res = value1 % value2 + stack.append(res) + elif opcode == 208: # getlocal_0 + stack.append(registers[0]) + elif opcode == 209: # getlocal_1 + stack.append(registers[1]) + elif opcode == 210: # getlocal_2 + stack.append(registers[2]) + elif opcode == 211: # getlocal_3 + stack.append(registers[3]) + elif opcode == 214: # setlocal_2 + registers[2] = stack.pop() + elif opcode == 215: # setlocal_3 + registers[3] = stack.pop() + else: + raise NotImplementedError( + u'Unsupported opcode %d' % opcode) + + method_pyfunctions[func_name] = resfunc + return resfunc + + initial_function = extract_function(u'decipher') + return lambda s: initial_function([s]) + + def _decrypt_signature(self, s, video_id, player_url, age_gate=False): """Turn the encrypted s field into a working signature""" - if len(s) == 92: + if player_url is not None: + try: + player_id = (player_url, len(s)) + if player_id not in self._player_cache: + func = self._extract_signature_function( + video_id, player_url, len(s) + ) + self._player_cache[player_id] = func + func = self._player_cache[player_id] + if self._downloader.params.get('youtube_print_sig_code'): + self._print_sig_code(func, len(s)) + return func(s) + except Exception: + tb = traceback.format_exc() + self._downloader.report_warning( + u'Automatic signature extraction failed: ' + tb) + + self._downloader.report_warning( + u'Warning: Falling back to static signature algorithm') + + return self._static_decrypt_signature( + s, video_id, player_url, age_gate) + + def _static_decrypt_signature(self, s, video_id, player_url, age_gate): + if age_gate: + # The videos with age protection use another player, so the + # algorithms can be different. + if len(s) == 86: + return s[2:63] + s[82] + s[64:82] + s[63] + + if len(s) == 93: + return s[86:29:-1] + s[88] + s[28:5:-1] + elif len(s) == 92: return s[25] + s[3:25] + s[0] + s[26:42] + s[79] + s[43:79] + s[91] + s[80:83] + elif len(s) == 91: + return s[84:27:-1] + s[86] + s[26:5:-1] elif len(s) == 90: return s[25] + s[3:25] + s[2] + s[26:40] + s[77] + s[41:77] + s[89] + s[78:81] elif len(s) == 89: @@ -423,15 +1080,15 @@ class YoutubeIE(YoutubeBaseInfoExtractor): elif len(s) == 87: return s[6:27] + s[4] + s[28:39] + s[27] + s[40:59] + s[2] + s[60:] elif len(s) == 86: - return s[83:36:-1] + s[0] + s[35:2:-1] + return s[80:72:-1] + s[16] + s[71:39:-1] + s[72] + s[38:16:-1] + s[82] + s[15::-1] elif len(s) == 85: - return s[83:34:-1] + s[0] + s[33:27:-1] + s[3] + s[26:19:-1] + s[34] + s[18:3:-1] + s[27] + return s[3:11] + s[0] + s[12:55] + s[84] + s[56:84] elif len(s) == 84: - return s[81:36:-1] + s[0] + s[35:2:-1] + return s[78:70:-1] + s[14] + s[69:37:-1] + s[70] + s[36:14:-1] + s[80] + s[:14][::-1] elif len(s) == 83: - return s[81:64:-1] + s[82] + s[63:52:-1] + s[45] + s[51:45:-1] + s[1] + s[44:1:-1] + s[0] + return s[80:63:-1] + s[0] + s[62:0:-1] + s[63] elif len(s) == 82: - return s[1:19] + s[0] + s[20:68] + s[19] + s[69:82] + return s[80:37:-1] + s[7] + s[36:7:-1] + s[0] + s[6:0:-1] + s[37] elif len(s) == 81: return s[56] + s[79:56:-1] + s[41] + s[55:41:-1] + s[80] + s[40:34:-1] + s[0] + s[33:29:-1] + s[34] + s[28:9:-1] + s[29] + s[8:0:-1] + s[9] elif len(s) == 80: @@ -442,65 +1099,39 @@ class YoutubeIE(YoutubeBaseInfoExtractor): else: raise ExtractorError(u'Unable to decrypt signature, key length %d not supported; retrying might work' % (len(s))) - def _decrypt_signature_age_gate(self, s): - # The videos with age protection use another player, so the algorithms - # can be different. - if len(s) == 86: - return s[2:63] + s[82] + s[64:82] + s[63] - else: - # Fallback to the other algortihms - return self._decrypt_signature(s) - - def _get_available_subtitles(self, video_id): - self.report_video_subtitles_download(video_id) - request = compat_urllib_request.Request('http://video.google.com/timedtext?hl=en&type=list&v=%s' % video_id) try: - sub_list = compat_urllib_request.urlopen(request).read().decode('utf-8') - except (compat_urllib_error.URLError, compat_http_client.HTTPException, socket.error) as err: + sub_list = self._download_webpage( + 'http://video.google.com/timedtext?hl=en&type=list&v=%s' % video_id, + video_id, note=False) + except ExtractorError as err: self._downloader.report_warning(u'unable to download video subtitles: %s' % compat_str(err)) return {} - sub_lang_list = re.findall(r'name="([^"]*)"[^>]+lang_code="([\w\-]+)"', sub_list) - sub_lang_list = dict((l[1], l[0]) for l in sub_lang_list) + lang_list = re.findall(r'name="([^"]*)"[^>]+lang_code="([\w\-]+)"', sub_list) + + sub_lang_list = {} + for l in lang_list: + lang = l[1] + params = compat_urllib_parse.urlencode({ + 'lang': lang, + 'v': video_id, + 'fmt': self._downloader.params.get('subtitlesformat'), + 'name': l[0], + }) + url = u'http://www.youtube.com/api/timedtext?' + params + sub_lang_list[lang] = url if not sub_lang_list: self._downloader.report_warning(u'video doesn\'t have subtitles') return {} return sub_lang_list - def _list_available_subtitles(self, video_id): - sub_lang_list = self._get_available_subtitles(video_id) - self.report_video_subtitles_available(video_id, sub_lang_list) - - def _request_subtitle(self, sub_lang, sub_name, video_id, format): - """ - Return the subtitle as a string or None if they are not found - """ - self.report_video_subtitles_request(video_id, sub_lang, format) - params = compat_urllib_parse.urlencode({ - 'lang': sub_lang, - 'name': sub_name, - 'v': video_id, - 'fmt': format, - }) - url = 'http://www.youtube.com/api/timedtext?' + params - try: - sub = compat_urllib_request.urlopen(url).read().decode('utf-8') - except (compat_urllib_error.URLError, compat_http_client.HTTPException, socket.error) as err: - self._downloader.report_warning(u'unable to download video subtitles for %s: %s' % (sub_lang, compat_str(err))) - return - if not sub: - self._downloader.report_warning(u'Did not fetch video subtitles') - return - return sub - - def _request_automatic_caption(self, video_id, webpage): + def _get_available_automatic_caption(self, video_id, webpage): """We need the webpage for getting the captions url, pass it as an argument to speed up the process.""" - sub_lang = (self._downloader.params.get('subtitleslangs') or ['en'])[0] sub_format = self._downloader.params.get('subtitlesformat') self.to_screen(u'%s: Looking for automatic captions' % video_id) mobj = re.search(r';ytplayer.config = ({.*?});', webpage) - err_msg = u'Couldn\'t find automatic captions for "%s"' % sub_lang + err_msg = u'Couldn\'t find automatic captions for %s' % video_id if mobj is None: self._downloader.report_warning(err_msg) return {} @@ -509,53 +1140,38 @@ class YoutubeIE(YoutubeBaseInfoExtractor): args = player_config[u'args'] caption_url = args[u'ttsurl'] timestamp = args[u'timestamp'] - params = compat_urllib_parse.urlencode({ - 'lang': 'en', - 'tlang': sub_lang, - 'fmt': sub_format, - 'ts': timestamp, - 'kind': 'asr', + # We get the available subtitles + list_params = compat_urllib_parse.urlencode({ + 'type': 'list', + 'tlangs': 1, + 'asrs': 1, }) - subtitles_url = caption_url + '&' + params - sub = self._download_webpage(subtitles_url, video_id, u'Downloading automatic captions') - return {sub_lang: sub} + list_url = caption_url + '&' + list_params + list_page = self._download_webpage(list_url, video_id) + caption_list = xml.etree.ElementTree.fromstring(list_page.encode('utf-8')) + original_lang_node = caption_list.find('track') + if original_lang_node is None or original_lang_node.attrib.get('kind') != 'asr' : + self._downloader.report_warning(u'Video doesn\'t have automatic captions') + return {} + original_lang = original_lang_node.attrib['lang_code'] + + sub_lang_list = {} + for lang_node in caption_list.findall('target'): + sub_lang = lang_node.attrib['lang_code'] + params = compat_urllib_parse.urlencode({ + 'lang': original_lang, + 'tlang': sub_lang, + 'fmt': sub_format, + 'ts': timestamp, + 'kind': 'asr', + }) + sub_lang_list[sub_lang] = caption_url + '&' + params + return sub_lang_list # An extractor error can be raise by the download process if there are # no automatic captions but there are subtitles except (KeyError, ExtractorError): self._downloader.report_warning(err_msg) return {} - - def _extract_subtitles(self, video_id): - """ - Return a dictionary: {language: subtitles} or {} if the subtitles - couldn't be found - """ - available_subs_list = self._get_available_subtitles(video_id) - sub_format = self._downloader.params.get('subtitlesformat') - if not available_subs_list: #There was some error, it didn't get the available subtitles - return {} - if self._downloader.params.get('allsubtitles', False): - sub_lang_list = available_subs_list - else: - if self._downloader.params.get('subtitleslangs', False): - reqested_langs = self._downloader.params.get('subtitleslangs') - elif 'en' in available_subs_list: - reqested_langs = ['en'] - else: - reqested_langs = [list(available_subs_list.keys())[0]] - - sub_lang_list = {} - for sub_lang in reqested_langs: - if not sub_lang in available_subs_list: - self._downloader.report_warning(u'no closed captions found in the specified language "%s"' % sub_lang) - continue - sub_lang_list[sub_lang] = available_subs_list[sub_lang] - subtitles = {} - for sub_lang in sub_lang_list: - subtitle = self._request_subtitle(sub_lang, sub_lang_list[sub_lang].encode('utf-8'), video_id, sub_format) - if subtitle: - subtitles[sub_lang] = subtitle - return subtitles def _print_formats(self, formats): print('Available formats:') @@ -597,13 +1213,25 @@ class YoutubeIE(YoutubeBaseInfoExtractor): video_url_list = [(f, url_map[f]) for f in existing_formats] # All formats else: # Specific formats. We pick the first in a slash-delimeted sequence. - # For example, if '1/2/3/4' is requested and '2' and '4' are available, we pick '2'. + # Format can be specified as itag or 'mp4' or 'flv' etc. We pick the highest quality + # available in the specified format. For example, + # if '1/2/3/4' is requested and '2' and '4' are available, we pick '2'. + # if '1/mp4/3/4' is requested and '1' and '5' (is a mp4) are available, we pick '1'. + # if '1/mp4/3/4' is requested and '4' and '5' (is a mp4) are available, we pick '5'. req_formats = req_format.split('/') video_url_list = None for rf in req_formats: if rf in url_map: video_url_list = [(rf, url_map[rf])] break + if rf in self._video_formats_map: + for srf in self._video_formats_map[rf]: + if srf in url_map: + video_url_list = [(srf, url_map[srf])] + break + else: + continue + break if video_url_list is None: raise ExtractorError(u'requested format not available') return video_url_list @@ -622,10 +1250,11 @@ class YoutubeIE(YoutubeBaseInfoExtractor): url_map[itag] = format_url return url_map - def _real_extract(self, url): - if re.match(r'(?:https?://)?[^/]+/watch\?feature=[a-z_]+$', url): - self._downloader.report_warning(u'Did you forget to quote the URL? Remember that & is a meta-character in most shells, so you want to put the URL in quotes, like youtube-dl \'http://www.youtube.com/watch?feature=foo&v=BaW_jenozKc\' (or simply youtube-dl BaW_jenozKc ).') + def _extract_annotations(self, video_id): + url = 'https://www.youtube.com/annotations_invideo?features=1&legacy=1&video_id=%s' % video_id + return self._download_webpage(url, video_id, note=u'Searching for annotations.', errnote=u'Unable to download video annotations.') + def _real_extract(self, url): # Extract original video URL from URL with redirection, like age verification, using next_url parameter mobj = re.search(self._NEXT_URL_RE, url) if mobj: @@ -644,7 +1273,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): video_webpage = video_webpage_bytes.decode('utf-8', 'ignore') # Attempt to extract SWF player URL - mobj = re.search(r'swfConfig.*?"(http:\\/\\/.*?watch.*?-.*?\.swf)"', video_webpage) + mobj = re.search(r'swfConfig.*?"(https?:\\/\\/.*?watch.*?-.*?\.swf)"', video_webpage) if mobj is not None: player_url = re.sub(r'\\(.)', r'\1', mobj.group(1)) else: @@ -708,9 +1337,11 @@ class YoutubeIE(YoutubeBaseInfoExtractor): self._downloader.report_warning(u'unable to extract uploader nickname') # title - if 'title' not in video_info: - raise ExtractorError(u'Unable to extract video title') - video_title = compat_urllib_parse.unquote_plus(video_info['title'][0]) + if 'title' in video_info: + video_title = compat_urllib_parse.unquote_plus(video_info['title'][0]) + else: + self._downloader.report_warning(u'Unable to extract video title') + video_title = u'_' # thumbnail image # We try first to get a high quality image: @@ -720,7 +1351,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): video_thumbnail = m_thumb.group(1) elif 'thumbnail_url' not in video_info: self._downloader.report_warning(u'unable to extract video thumbnail') - video_thumbnail = '' + video_thumbnail = None else: # don't panic if we can't find it video_thumbnail = compat_urllib_parse.unquote_plus(video_info['thumbnail_url'][0]) @@ -743,15 +1374,10 @@ class YoutubeIE(YoutubeBaseInfoExtractor): video_description = u'' # subtitles - video_subtitles = None - - if self._downloader.params.get('writesubtitles', False) or self._downloader.params.get('allsubtitles', False): - video_subtitles = self._extract_subtitles(video_id) - elif self._downloader.params.get('writeautomaticsub', False): - video_subtitles = self._request_automatic_caption(video_id, video_webpage) + video_subtitles = self.extract_subtitles(video_id, video_webpage) if self._downloader.params.get('listsubtitles', False): - self._list_available_subtitles(video_id) + self._list_available_subtitles(video_id, video_webpage) return if 'length_seconds' not in video_info: @@ -760,6 +1386,11 @@ class YoutubeIE(YoutubeBaseInfoExtractor): else: video_duration = compat_urllib_parse.unquote_plus(video_info['length_seconds'][0]) + # annotations + video_annotations = None + if self._downloader.params.get('writeannotations', False): + video_annotations = self._extract_annotations(video_id) + # Decide which formats to download try: @@ -770,6 +1401,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor): args = info['args'] # Easy way to know if the 's' value is in url_encoded_fmt_stream_map # this signatures are encrypted + if 'url_encoded_fmt_stream_map' not in args: + raise ValueError(u'No stream_map present') # caught below m_s = re.search(r'[&,]s=', args['url_encoded_fmt_stream_map']) if m_s is not None: self.to_screen(u'%s: Encrypted signatures detected.' % video_id) @@ -802,24 +1435,34 @@ class YoutubeIE(YoutubeBaseInfoExtractor): if 'sig' in url_data: url += '&signature=' + url_data['sig'][0] elif 's' in url_data: + encrypted_sig = url_data['s'][0] if self._downloader.params.get('verbose'): - s = url_data['s'][0] if age_gate: - player_version = self._search_regex(r'ad3-(.+?)\.swf', - video_info['ad3_module'][0] if 'ad3_module' in video_info else 'NOT FOUND', - 'flash player', fatal=False) - player = 'flash player %s' % player_version + if player_url is None: + player_version = 'unknown' + else: + player_version = self._search_regex( + r'-(.+)\.swf$', player_url, + u'flash player', fatal=False) + player_desc = 'flash player %s' % player_version else: - player = u'html5 player %s' % self._search_regex(r'html5player-(.+?)\.js', video_webpage, + player_version = self._search_regex( + r'html5player-(.+?)\.js', video_webpage, 'html5 player', fatal=False) - parts_sizes = u'.'.join(compat_str(len(part)) for part in s.split('.')) + player_desc = u'html5 player %s' % player_version + + parts_sizes = u'.'.join(compat_str(len(part)) for part in encrypted_sig.split('.')) self.to_screen(u'encrypted signature length %d (%s), itag %s, %s' % - (len(s), parts_sizes, url_data['itag'][0], player)) - encrypted_sig = url_data['s'][0] - if age_gate: - signature = self._decrypt_signature_age_gate(encrypted_sig) - else: - signature = self._decrypt_signature(encrypted_sig) + (len(encrypted_sig), parts_sizes, url_data['itag'][0], player_desc)) + + if not age_gate: + jsplayer_url_json = self._search_regex( + r'"assets":.+?"js":\s*("[^"]+")', + video_webpage, u'JS player URL') + player_url = json.loads(jsplayer_url_json) + + signature = self._decrypt_signature( + encrypted_sig, video_id, player_url, age_gate) url += '&signature=' + signature if 'ratebypass' not in url: url += '&ratebypass=yes' @@ -835,7 +1478,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): return else: - raise ExtractorError(u'no conn or url_encoded_fmt_stream_map information found in video info') + raise ExtractorError(u'no conn, hlsvp or url_encoded_fmt_stream_map information found in video info') results = [] for format_param, video_real_url in video_url_list: @@ -859,7 +1502,9 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 'description': video_description, 'player_url': player_url, 'subtitles': video_subtitles, - 'duration': video_duration + 'duration': video_duration, + 'age_limit': 18 if age_gate else 0, + 'annotations': video_annotations }) return results @@ -893,9 +1538,19 @@ class YoutubePlaylistIE(InfoExtractor): mobj = re.match(self._VALID_URL, url, re.VERBOSE) if mobj is None: raise ExtractorError(u'Invalid URL: %s' % url) + playlist_id = mobj.group(1) or mobj.group(2) + + # Check if it's a video-specific URL + query_dict = compat_urlparse.parse_qs(compat_urlparse.urlparse(url).query) + if 'v' in query_dict: + video_id = query_dict['v'][0] + if self._downloader.params.get('noplaylist'): + self.to_screen(u'Downloading just video %s because of --no-playlist' % video_id) + return self.url_result('https://www.youtube.com/watch?v=' + video_id, 'Youtube') + else: + self.to_screen(u'Downloading playlist PL%s - add --no-playlist to just download video %s' % (playlist_id, video_id)) # Download playlist videos from API - playlist_id = mobj.group(1) or mobj.group(2) videos = [] for page_num in itertools.count(1): @@ -920,8 +1575,11 @@ class YoutubePlaylistIE(InfoExtractor): for entry in response['feed']['entry']: index = entry['yt$position']['$t'] - if 'media$group' in entry and 'media$player' in entry['media$group']: - videos.append((index, entry['media$group']['media$player']['url'])) + if 'media$group' in entry and 'yt$videoid' in entry['media$group']: + videos.append(( + index, + 'https://www.youtube.com/watch?v=' + entry['media$group']['yt$videoid']['$t'] + )) videos = [v[1] for v in sorted(videos)] @@ -987,13 +1645,20 @@ class YoutubeChannelIE(InfoExtractor): class YoutubeUserIE(InfoExtractor): IE_DESC = u'YouTube.com user videos (URL or "ytuser" keyword)' - _VALID_URL = r'(?:(?:(?:https?://)?(?:\w+\.)?youtube\.com/user/)|ytuser:)([A-Za-z0-9_-]+)' + _VALID_URL = r'(?:(?:(?:https?://)?(?:\w+\.)?youtube\.com/(?:user/)?(?!(?:attribution_link|watch)(?:$|[^a-z_A-Z0-9-])))|ytuser:)(?!feed/)([A-Za-z0-9_-]+)' _TEMPLATE_URL = 'http://gdata.youtube.com/feeds/api/users/%s' _GDATA_PAGE_SIZE = 50 - _GDATA_URL = 'http://gdata.youtube.com/feeds/api/users/%s/uploads?max-results=%d&start-index=%d' - _VIDEO_INDICATOR = r'/watch\?v=(.+?)[\<&]' + _GDATA_URL = 'http://gdata.youtube.com/feeds/api/users/%s/uploads?max-results=%d&start-index=%d&alt=json' IE_NAME = u'youtube:user' + @classmethod + def suitable(cls, url): + # Don't return True if the url can be extracted with other youtube + # extractor, the regex would is too permissive and it would match. + other_ies = iter(klass for (name, klass) in globals().items() if name.endswith('IE') and klass is not cls) + if any(ie.suitable(url) for ie in other_ies): return False + else: return super(YoutubeUserIE, cls).suitable(url) + def _real_extract(self, url): # Extract username mobj = re.match(self._VALID_URL, url) @@ -1016,13 +1681,18 @@ class YoutubeUserIE(InfoExtractor): page = self._download_webpage(gdata_url, username, u'Downloading video ids from %d to %d' % (start_index, start_index + self._GDATA_PAGE_SIZE)) + try: + response = json.loads(page) + except ValueError as err: + raise ExtractorError(u'Invalid JSON in API response: ' + compat_str(err)) + if 'entry' not in response['feed']: + # Number of videos is a multiple of self._MAX_RESULTS + break + # Extract video identifiers ids_in_page = [] - - for mobj in re.finditer(self._VIDEO_INDICATOR, page): - if mobj.group(1) not in ids_in_page: - ids_in_page.append(mobj.group(1)) - + for entry in response['feed']['entry']: + ids_in_page.append(entry['id']['$t'].split('/')[-1]) video_ids.extend(ids_in_page) # A little optimization - if current page is not @@ -1161,10 +1831,25 @@ class YoutubeWatchLaterIE(YoutubeFeedsInfoExtractor): class YoutubeFavouritesIE(YoutubeBaseInfoExtractor): IE_NAME = u'youtube:favorites' IE_DESC = u'YouTube.com favourite videos, "ytfav" keyword (requires authentication)' - _VALID_URL = r'https?://www\.youtube\.com/my_favorites|:ytfav(?:o?rites)?' + _VALID_URL = r'https?://www\.youtube\.com/my_favorites|:ytfav(?:ou?rites)?' _LOGIN_REQUIRED = True def _real_extract(self, url): webpage = self._download_webpage('https://www.youtube.com/my_favorites', 'Youtube Favourites videos') playlist_id = self._search_regex(r'list=(.+?)["&]', webpage, u'favourites playlist id') return self.url_result(playlist_id, 'YoutubePlaylist') + + +class YoutubeTruncatedURLIE(InfoExtractor): + IE_NAME = 'youtube:truncated_url' + IE_DESC = False # Do not list + _VALID_URL = r'(?:https?://)?[^/]+/watch\?feature=[a-z_]+$' + + def _real_extract(self, url): + raise ExtractorError( + u'Did you forget to quote the URL? Remember that & is a meta ' + u'character in most shells, so you want to put the URL in quotes, ' + u'like youtube-dl ' + u'\'http://www.youtube.com/watch?feature=foo&v=BaW_jenozKc\'' + u' (or simply youtube-dl BaW_jenozKc ).', + expected=True)