X-Git-Url: http://git.cielonegro.org/gitweb.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fgrooveshark.py;h=fff74a70a891fc163ff488408e4df348564b8a29;hb=e72c7e41238681338c92427f6731c74947645308;hp=6798addceb1282828c991ec1ce5a4a3f2e19e338;hpb=dd06c95e43ebbef35e2ef8b589bd25a646166553;p=youtube-dl.git diff --git a/youtube_dl/extractor/grooveshark.py b/youtube_dl/extractor/grooveshark.py index 6798addce..fff74a70a 100644 --- a/youtube_dl/extractor/grooveshark.py +++ b/youtube_dl/extractor/grooveshark.py @@ -3,13 +3,19 @@ from __future__ import unicode_literals import time import math +import os.path import re -from urlparse import urlparse, urlunparse, urldefrag -from urllib import quote, urlencode -from os.path import basename + from .common import InfoExtractor -from ..utils import ExtractorError, compat_urllib_request, compat_html_parser +from ..compat import ( + compat_html_parser, + compat_urllib_parse, + compat_urllib_request, + compat_urlparse, +) +from ..utils import ExtractorError + class GroovesharkHtmlParser(compat_html_parser.HTMLParser): def __init__(self): @@ -20,10 +26,10 @@ class GroovesharkHtmlParser(compat_html_parser.HTMLParser): def handle_starttag(self, tag, attrs): attrs = dict((k, v) for k, v in attrs) if tag == 'object': - self._current_object = { 'attrs': attrs, 'params': [] } + self._current_object = {'attrs': attrs, 'params': []} elif tag == 'param': self._current_object['params'].append(attrs) - + def handle_endtag(self, tag): if tag == 'object': self.objects.append(self._current_object) @@ -36,16 +42,17 @@ class GroovesharkHtmlParser(compat_html_parser.HTMLParser): p.close() return p.objects + class GroovesharkIE(InfoExtractor): _VALID_URL = r'https?://(www\.)?grooveshark\.com/#!/s/([^/]+)/([^/]+)' _TEST = { 'url': 'http://grooveshark.com/#!/s/Jolene+Tenth+Key+Remix+Ft+Will+Sessions/6SS1DW?src=5', - 'md5': 'bbccc50b19daca23b8f961152c1dc95b', + 'md5': '7ecf8aefa59d6b2098517e1baa530023', 'info_dict': { 'id': '6SS1DW', 'title': 'Jolene (Tenth Key Remix ft. Will Sessions)', 'ext': 'mp3', - 'duration': 227 + 'duration': 227, } } @@ -53,30 +60,30 @@ class GroovesharkIE(InfoExtractor): do_bootstrap_request = True def _parse_target(self, target): - uri = urlparse(target) + uri = compat_urlparse.urlparse(target) hash = uri.fragment[1:].split('?')[0] - token = basename(hash.rstrip('/')) + token = os.path.basename(hash.rstrip('/')) return (uri, hash, token) def _build_bootstrap_url(self, target): (uri, hash, token) = self._parse_target(target) - query = 'getCommunicationToken=1&hash=%s&%d' % (quote(hash, safe=''), self.ts) - return (urlunparse((uri.scheme, uri.netloc, '/preload.php', None, query, None)), token) + query = 'getCommunicationToken=1&hash=%s&%d' % (compat_urllib_parse.quote(hash, safe=''), self.ts) + return (compat_urlparse.urlunparse((uri.scheme, uri.netloc, '/preload.php', None, query, None)), token) def _build_meta_url(self, target): (uri, hash, token) = self._parse_target(target) - query = 'hash=%s&%d' % (quote(hash, safe=''), self.ts) - return (urlunparse((uri.scheme, uri.netloc, '/preload.php', None, query, None)), token) + query = 'hash=%s&%d' % (compat_urllib_parse.quote(hash, safe=''), self.ts) + return (compat_urlparse.urlunparse((uri.scheme, uri.netloc, '/preload.php', None, query, None)), token) def _build_stream_url(self, meta): - return urlunparse(('http', meta['streamKey']['ip'], '/stream.php', None, None, None)) + return compat_urlparse.urlunparse(('http', meta['streamKey']['ip'], '/stream.php', None, None, None)) def _build_swf_referer(self, target, obj): (uri, _, _) = self._parse_target(target) - return urlunparse((uri.scheme, uri.netloc, obj['attrs']['data'], None, None, None)) + return compat_urlparse.urlunparse((uri.scheme, uri.netloc, obj['attrs']['data'], None, None, None)) def _transform_bootstrap(self, js): - return re.split('^\s*try\s*{', js, flags=re.M)[0] \ + return re.split('(?m)^\s*try\s*{', js)[0] \ .split(' = ', 1)[1].strip().rstrip(';') def _transform_meta(self, js): @@ -86,7 +93,7 @@ class GroovesharkIE(InfoExtractor): (meta_url, token) = self._build_meta_url(target) self.to_screen('Metadata URL: %s' % meta_url) - headers = {'Referer': urldefrag(target)[0]} + headers = {'Referer': compat_urlparse.urldefrag(target)[0]} req = compat_urllib_request.Request(meta_url, headers=headers) res = self._download_json(req, token, transform_source=self._transform_meta) @@ -104,34 +111,47 @@ class GroovesharkIE(InfoExtractor): def _get_bootstrap(self, target): (bootstrap_url, token) = self._build_bootstrap_url(target) - - headers = {'Referer': urldefrag(target)[0]} + + headers = {'Referer': compat_urlparse.urldefrag(target)[0]} req = compat_urllib_request.Request(bootstrap_url, headers=headers) res = self._download_json(req, token, fatal=False, note='Downloading player bootstrap data', errnote='Unable to download player bootstrap data', transform_source=self._transform_bootstrap) return res - + def _get_playerpage(self, target): (_, _, token) = self._parse_target(target) - - res = self._download_webpage( + + webpage = self._download_webpage( target, token, note='Downloading player page', errnote='Unable to download player page', fatal=False) - - if res is not None: - o = GroovesharkHtmlParser.extract_object_tags(res) - return (res, [x for x in o if x['attrs']['id'] == 'jsPlayerEmbed']) - - return (res, None) - + + if webpage is not None: + # Search (for example German) error message + error_msg = self._html_search_regex( + r'