From fe7710cbccdade0b66ef48a7e2eedad71a0702cc Mon Sep 17 00:00:00 2001 From: Yen Chi Hsuan Date: Tue, 27 Jan 2015 23:55:22 +0800 Subject: [PATCH] [xuite] Add new extractor --- youtube_dl/extractor/__init__.py | 1 + youtube_dl/extractor/xuite.py | 147 +++++++++++++++++++++++++++++++ 2 files changed, 148 insertions(+) create mode 100644 youtube_dl/extractor/xuite.py diff --git a/youtube_dl/extractor/__init__.py b/youtube_dl/extractor/__init__.py index 03c56156a..09070daa4 100644 --- a/youtube_dl/extractor/__init__.py +++ b/youtube_dl/extractor/__init__.py @@ -546,6 +546,7 @@ from .xminus import XMinusIE from .xnxx import XNXXIE from .xvideos import XVideosIE from .xtube import XTubeUserIE, XTubeIE +from .xuite import XuiteIE from .xxxymovies import XXXYMoviesIE from .yahoo import ( YahooIE, diff --git a/youtube_dl/extractor/xuite.py b/youtube_dl/extractor/xuite.py new file mode 100644 index 000000000..943757c27 --- /dev/null +++ b/youtube_dl/extractor/xuite.py @@ -0,0 +1,147 @@ +# -*- coding: utf-8 -*- +from __future__ import unicode_literals + +import base64 +from .common import InfoExtractor +from ..compat import ( + compat_urlparse, + compat_urllib_parse_unquote, + compat_parse_qs +) +from ..utils import ( + ExtractorError, + parse_iso8601, + parse_duration +) + +# ref: http://stackoverflow.com/questions/475074/regex-to-parse-or-validate-base64-data +REGEX_BASE64 = r'(?:[A-Za-z0-9+/]{4})*(?:[A-Za-z0-9+/]{2}==|[A-Za-z0-9+/]{3}=)?' + +CST_ZONE = +8 # China Standard Time + + +class XuiteIE(InfoExtractor): + _VALID_URL = r'http://vlog.xuite.net/play/(?P%s)(/.*)?' % REGEX_BASE64 + _TESTS = [{ + # Audio + 'url': 'http://vlog.xuite.net/play/RGkzc1ZULTM4NjA5MTQuZmx2', + 'md5': '63a42c705772aa53fd4c1a0027f86adf', + 'info_dict': { + 'id': 'RGkzc1ZULTM4NjA5MTQuZmx2', + 'ext': 'mp3', + 'upload_date': '20110902', + 'uploader_id': '15973816', + 'uploader': '阿能', + 'timestamp': 1314932940, + 'title': '孤單南半球-歐德陽' + } + }, { + # Audio with alternative form of url + 'url': 'http://vlog.xuite.net/play/S1dDUjdyLTMyOTc3NjcuZmx2/%E5%AD%AB%E7%87%95%E5%A7%BF-%E7%9C%BC%E6%B7%9A%E6%88%90%E8%A9%A9', + 'md5': 'c91645f324de53d82ebb80930e1a73d2', + 'info_dict': { + 'id': 'S1dDUjdyLTMyOTc3NjcuZmx2', + 'ext': 'mp3', + 'upload_date': '20101226', + 'uploader_id': '10102699', + 'uploader': '蠍', + 'timestamp': 1293367080, + 'title': '孫燕姿-眼淚成詩', + } + }, { + # Video with only one format + 'url': 'http://vlog.xuite.net/play/TkRZNjhULTM0NDE2MjkuZmx2', + 'md5': 'c45737fc8ac5dc8ac2f92ecbcecf505e', + 'info_dict': { + 'id': 'TkRZNjhULTM0NDE2MjkuZmx2', + 'ext': 'mp4', + 'upload_date': '20110306', + 'uploader_id': '10400126', + 'uploader': 'Valen', + 'timestamp': 1299383640, + 'title': '孫燕姿 - 眼淚成詩', + } + }, { + # Video with two formats + 'url': 'http://vlog.xuite.net/play/bWo1N1pLLTIxMzAxMTcwLmZsdg==', + 'md5': '1166e0f461efe55b62e26a2d2a68e6de', + 'info_dict': { + 'id': 'bWo1N1pLLTIxMzAxMTcwLmZsdg==', + 'ext': 'mp4', + 'upload_date': '20150117', + 'uploader_id': '242127761', + 'uploader': '我只是想認真點', + 'timestamp': 1421481240, + 'title': '暗殺教室 02', + } + }] + + def _flv_config(self, media_id): + base64_media_id = base64.b64encode(media_id.encode('utf-8')).decode('utf-8') + flv_config_url = 'http://vlog.xuite.net/flash/player?media=' + base64_media_id + flv_config = self._download_xml(flv_config_url, 'flv config') + + prop_dict = {} + for prop in flv_config.findall('./property'): + prop_id = base64.b64decode(prop.attrib['id']).decode('utf-8') + + if not prop.text: + continue # CDATA may be empty in flv config + + encoded_content = base64.b64decode(prop.text).decode('utf-8') + prop_dict[prop_id] = compat_urllib_parse_unquote(encoded_content) + + return prop_dict + + def _type_string(self, media_url): + query_string = compat_urlparse.urlparse(media_url).query + type_string = compat_parse_qs(query_string)['q'][0] + return type_string + + def _guess_ext(self, media_url): + type_string = self._type_string(media_url) + if type_string == 'mp3': + return 'mp3' + elif type_string == '360' or type_string == '720': + return 'mp4' + else: + raise ExtractorError('Unknown type string %s' % type_string) + + def _real_extract(self, url): + video_id = self._match_id(url) + + page = self._download_webpage(url, video_id) + media_id = self._html_search_regex(r'data-mediaid="(\d+)"', page, 'media id') + flv_config = self._flv_config(media_id) + + timestamp_local = parse_iso8601(flv_config['publish_datetime'], ' ') + timestamp_gmt = timestamp_local - CST_ZONE * 3600 + + ret_attrs = { + 'id': video_id, + 'title': flv_config['title'], + 'thumbnail': flv_config['thumb'], + 'uploader': flv_config['author_name'], + 'timestamp': timestamp_gmt, + 'uploader_id': flv_config['author_id'], + 'duration': parse_duration(flv_config['duration']), + 'categories': [flv_config['category']] + } + + if 'hq_src' in flv_config: + src = flv_config['src'] + src_hq = flv_config['hq_src'] + ret_attrs['formats'] = [{ + 'url': src, + 'ext': self._guess_ext(src), + 'format_id': self._type_string(src) + }, { + 'url': src_hq, + 'ext': self._guess_ext(src_hq), + 'format_id': self._type_string(src_hq) + }] + else: + ret_attrs['url'] = flv_config['src'] + ret_attrs['ext'] = self._guess_ext(flv_config['src']) + + return ret_attrs