From 35a7618ad5dcfa8a9db6a630af1c679f1ee54057 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?V=C3=B5=20V=C4=83n=20Khoa?= Date: Sun, 24 Sep 2017 20:54:40 +0700 Subject: [PATCH 1/2] [zingmp3] Update extractor --- youtube_dl/extractor/zingmp3.py | 49 +++++++++++++++++---------------- 1 file changed, 25 insertions(+), 24 deletions(-) diff --git a/youtube_dl/extractor/zingmp3.py b/youtube_dl/extractor/zingmp3.py index adfdcaabf..4be31f951 100644 --- a/youtube_dl/extractor/zingmp3.py +++ b/youtube_dl/extractor/zingmp3.py @@ -23,7 +23,8 @@ class ZingMp3BaseInfoExtractor(InfoExtractor): expected=True) formats = [] - for quality, source_url in zip(item.get('qualities') or item.get('quality', []), item.get('source_list') or item.get('source', [])): + + for quality, source_url in item['source'].items(): if not source_url or source_url == 'require vip': continue if not re.match(r'https?://', source_url): @@ -46,24 +47,30 @@ class ZingMp3BaseInfoExtractor(InfoExtractor): }) formats.append(f) - cover = item.get('cover') + #cover = item.get('cover') return { 'title': (item.get('name') or item.get('title')).strip(), 'formats': formats, - 'thumbnail': 'http:/' + cover if cover else None, + #'thumbnail': 'http:/' + cover if cover else None, 'artist': item.get('artist'), } def _extract_player_json(self, player_json_url, id, page_type, playlist_title=None): player_json = self._download_json(player_json_url, id, 'Downloading Player JSON') - items = player_json['data'] - if 'item' in items: - items = items['item'] - if len(items) == 1: + ''' if page_type == 'audio' or page_type == 'video': + items = player_json['data']['source'] + else: + items = player_json['data']['items'] + ''' + items = player_json['data'] + if 'items' in items: + items = items['items'] + + if page_type == 'audio' or page_type == 'video': # one single song - data = self._extract_item(items[0], page_type) + data = self._extract_item(items, page_type) data['id'] = id return data @@ -95,25 +102,23 @@ class ZingMp3IE(ZingMp3BaseInfoExtractor): 'id': 'ZWZB9WAB', 'title': 'Xa Mãi Xa', 'ext': 'mp3', - 'thumbnail': r're:^https?://.*\.jpg$', }, }, { 'url': 'http://mp3.zing.vn/video-clip/Let-It-Go-Frozen-OST-Sungha-Jung/ZW6BAEA0.html', - 'md5': '870295a9cd8045c0e15663565902618d', + 'md5': 'c04f2c8c6400d90b43dd0fa6485e2e32', 'info_dict': { 'id': 'ZW6BAEA0', 'title': 'Let It Go (Frozen OST)', 'ext': 'mp4', }, }, { - 'url': 'http://mp3.zing.vn/album/Lau-Dai-Tinh-Ai-Bang-Kieu-Minh-Tuyet/ZWZBWDAF.html', + 'url': 'http://mp3.zing.vn/album/I-Lab-You-Single-Tien-Tien/ZO6976C0.html', 'info_dict': { '_type': 'playlist', - 'id': 'ZWZBWDAF', - 'title': 'Lâu Đài Tình Ái - Bằng Kiều,Minh Tuyết | Album 320 lossless', + 'id': 'ZO6976C0', + 'title': 'I Lab You (Single)', }, - 'playlist_count': 10, - 'skip': 'removed at the request of the owner', + 'playlist_count': 2, }, { 'url': 'http://mp3.zing.vn/playlist/Duong-Hong-Loan-apollobee/IWCAACCB.html', 'only_matching': True, @@ -126,18 +131,14 @@ class ZingMp3IE(ZingMp3BaseInfoExtractor): webpage = self._download_webpage(url, page_id) - player_json_url = self._search_regex([ + player_json_url = 'http://mp3.zing.vn/xhr' + self._search_regex([ r'data-xml="([^"]+)', r'&xmlURL=([^&]+)&' ], webpage, 'player xml url') - playlist_title = None - page_type = self._search_regex(r'/(?:html5)?xml/([^/-]+)', player_json_url, 'page type') - if page_type == 'video': - player_json_url = update_url_query(player_json_url, {'format': 'json'}) - else: - player_json_url = player_json_url.replace('/xml/', '/html5xml/') - if page_type == 'album': - playlist_title = self._og_search_title(webpage) + playlist_title = self._html_search_regex(r'

([^>]+)

', webpage, 'title') + if not playlist_title: + playlist_title = self._og_search_title(webpage) + page_type = self._search_regex(r'type=([^&]+)', player_json_url, 'page type') return self._extract_player_json(player_json_url, page_id, page_type, playlist_title) From d626eb6fc05a6347636d91307d954b6350fb1612 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?V=C3=B5=20V=C4=83n=20Khoa?= Date: Sun, 24 Sep 2017 23:07:08 +0700 Subject: [PATCH 2/2] [zingmp3] Update extractor --- youtube_dl/extractor/zingmp3.py | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) diff --git a/youtube_dl/extractor/zingmp3.py b/youtube_dl/extractor/zingmp3.py index 4be31f951..149326b7d 100644 --- a/youtube_dl/extractor/zingmp3.py +++ b/youtube_dl/extractor/zingmp3.py @@ -116,7 +116,7 @@ class ZingMp3IE(ZingMp3BaseInfoExtractor): 'info_dict': { '_type': 'playlist', 'id': 'ZO6976C0', - 'title': 'I Lab You (Single)', + 'title': 'I Lab You (Single) - Tiên Tiên | Zing MP3', }, 'playlist_count': 2, }, { @@ -136,9 +136,8 @@ class ZingMp3IE(ZingMp3BaseInfoExtractor): r'&xmlURL=([^&]+)&' ], webpage, 'player xml url') - playlist_title = self._html_search_regex(r'

([^>]+)

', webpage, 'title') - if not playlist_title: - playlist_title = self._og_search_title(webpage) + #playlist_title = self._html_search_regex(r'

([^>]+)

', webpage, 'title') + playlist_title = self._og_search_title(webpage) page_type = self._search_regex(r'type=([^&]+)', player_json_url, 'page type') return self._extract_player_json(player_json_url, page_id, page_type, playlist_title)