From edf421611928e48ac4fda778d6de8ea83585019d Mon Sep 17 00:00:00 2001 From: =?utf8?q?Sergey=20M=E2=80=A4?= Date: Mon, 20 Apr 2015 22:46:01 +0600 Subject: [PATCH] [bambuser] Modernize and extract more metadata --- youtube_dl/extractor/bambuser.py | 25 +++++++++++++++++++------ 1 file changed, 19 insertions(+), 6 deletions(-) diff --git a/youtube_dl/extractor/bambuser.py b/youtube_dl/extractor/bambuser.py index d52302ebc..93913c3f4 100644 --- a/youtube_dl/extractor/bambuser.py +++ b/youtube_dl/extractor/bambuser.py @@ -4,8 +4,15 @@ import re import itertools from .common import InfoExtractor -from ..compat import compat_urllib_request -from ..utils import ExtractorError +from ..compat import ( + compat_urllib_request, + compat_str, +) +from ..utils import ( + ExtractorError, + int_or_none, + float_or_none, +) class BambuserIE(InfoExtractor): @@ -24,6 +31,9 @@ class BambuserIE(InfoExtractor): 'duration': 3741, 'uploader': 'pixelversity', 'uploader_id': '344706', + 'timestamp': 1382976692, + 'upload_date': '20131028', + 'view_count': int, }, 'params': { # It doesn't respect the 'Range' header, it would download the whole video @@ -51,10 +61,13 @@ class BambuserIE(InfoExtractor): 'title': result['title'], 'url': result['url'], 'thumbnail': result.get('preview'), - 'duration': int(result['length']), - 'view_count': int(result['views_total']), - 'uploader': result['username'], - 'uploader_id': result['owner']['uid'], + 'duration': int_or_none(result.get('length')), + 'uploader': result.get('username'), + 'uploader_id': compat_str(result.get('owner', {}).get('uid')), + 'timestamp': int_or_none(result.get('created')), + 'fps': float_or_none(result.get('framerate')), + 'view_count': int_or_none(result.get('views_total')), + 'comment_count': int_or_none(result.get('comment_count')), } -- 2.40.0