From cec9727c7f6a0dad8b10a51f0a6581ac5a1dbe86 Mon Sep 17 00:00:00 2001 From: Yen Chi Hsuan Date: Sun, 15 May 2016 15:35:31 +0800 Subject: [PATCH] [hearthisat] Detect invalid download links (fixes #9440) --- youtube_dl/extractor/hearthisat.py | 38 +++++++++++++++++++++++------- 1 file changed, 29 insertions(+), 9 deletions(-) diff --git a/youtube_dl/extractor/hearthisat.py b/youtube_dl/extractor/hearthisat.py index 7d8698655..ac42ef414 100644 --- a/youtube_dl/extractor/hearthisat.py +++ b/youtube_dl/extractor/hearthisat.py @@ -7,6 +7,7 @@ from ..compat import compat_urlparse from ..utils import ( HEADRequest, + KNOWN_EXTENSIONS, sanitized_Request, str_to_int, urlencode_postdata, @@ -17,7 +18,7 @@ class HearThisAtIE(InfoExtractor): _VALID_URL = r'https?://(?:www\.)?hearthis\.at/(?P[^/]+)/(?P[A-Za-z0-9\-]+)/?$' _PLAYLIST_URL = 'https://hearthis.at/playlist.php' - _TEST = { + _TESTS = [{ 'url': 'https://hearthis.at/moofi/dr-kreep', 'md5': 'ab6ec33c8fed6556029337c7885eb4e0', 'info_dict': { @@ -34,7 +35,25 @@ class HearThisAtIE(InfoExtractor): 'duration': 71, 'categories': ['Experimental'], } - } + }, { + # 'download' link redirects to the original webpage + 'url': 'https://hearthis.at/twitchsf/dj-jim-hopkins-totally-bitchin-80s-dance-mix/', + 'md5': '5980ceb7c461605d30f1f039df160c6e', + 'info_dict': { + 'id': '811296', + 'ext': 'mp3', + 'title': 'TwitchSF - DJ Jim Hopkins - Totally Bitchin\' 80\'s Dance Mix!', + 'description': 'Listen to DJ Jim Hopkins - Totally Bitchin\' 80\'s Dance Mix! by TwitchSF on hearthis.at - Dance', + 'upload_date': '20160328', + 'timestamp': 1459186146, + 'thumbnail': 're:^https?://.*\.jpg$', + 'comment_count': int, + 'view_count': int, + 'like_count': int, + 'duration': 4360, + 'categories': ['Dance'], + }, + }] def _real_extract(self, url): m = re.match(self._VALID_URL, url) @@ -90,13 +109,14 @@ def _real_extract(self, url): ext_handle = self._request_webpage( ext_req, display_id, note='Determining extension') ext = urlhandle_detect_ext(ext_handle) - formats.append({ - 'format_id': 'download', - 'vcodec': 'none', - 'ext': ext, - 'url': download_url, - 'preference': 2, # Usually better quality - }) + if ext in KNOWN_EXTENSIONS: + formats.append({ + 'format_id': 'download', + 'vcodec': 'none', + 'ext': ext, + 'url': download_url, + 'preference': 2, # Usually better quality + }) self._sort_formats(formats) return {