diff options
Diffstat (limited to 'youtube_dl/extractor/esri.py')
-rw-r--r-- | youtube_dl/extractor/esri.py | 74 |
1 files changed, 74 insertions, 0 deletions
diff --git a/youtube_dl/extractor/esri.py b/youtube_dl/extractor/esri.py new file mode 100644 index 000000000..bf5d2019f --- /dev/null +++ b/youtube_dl/extractor/esri.py @@ -0,0 +1,74 @@ +# coding: utf-8 +from __future__ import unicode_literals + +import re + +from .common import InfoExtractor +from ..compat import compat_urlparse +from ..utils import ( + int_or_none, + parse_filesize, + unified_strdate, +) + + +class EsriVideoIE(InfoExtractor): + _VALID_URL = r'https?://video\.esri\.com/watch/(?P<id>[0-9]+)' + _TEST = { + 'url': 'https://video.esri.com/watch/1124/arcgis-online-_dash_-developing-applications', + 'md5': 'd4aaf1408b221f1b38227a9bbaeb95bc', + 'info_dict': { + 'id': '1124', + 'ext': 'mp4', + 'title': 'ArcGIS Online - Developing Applications', + 'description': 'Jeremy Bartley demonstrates how to develop applications with ArcGIS Online.', + 'thumbnail': 're:^https?://.*\.jpg$', + 'duration': 185, + 'upload_date': '20120419', + } + } + + def _real_extract(self, url): + video_id = self._match_id(url) + + webpage = self._download_webpage(url, video_id) + + formats = [] + for width, height, content in re.findall( + r'(?s)<li><strong>(\d+)x(\d+):</strong>(.+?)</li>', webpage): + for video_url, ext, filesize in re.findall( + r'<a[^>]+href="([^"]+)">([^<]+) \(([^<]+)\)</a>', content): + formats.append({ + 'url': compat_urlparse.urljoin(url, video_url), + 'ext': ext.lower(), + 'format_id': '%s-%s' % (ext.lower(), height), + 'width': int(width), + 'height': int(height), + 'filesize_approx': parse_filesize(filesize), + }) + self._sort_formats(formats) + + title = self._html_search_meta('title', webpage, 'title') + description = self._html_search_meta( + 'description', webpage, 'description', fatal=False) + + thumbnail = self._html_search_meta('thumbnail', webpage, 'thumbnail', fatal=False) + if thumbnail: + thumbnail = re.sub(r'_[st]\.jpg$', '_x.jpg', thumbnail) + + duration = int_or_none(self._search_regex( + [r'var\s+videoSeconds\s*=\s*(\d+)', r"'duration'\s*:\s*(\d+)"], + webpage, 'duration', fatal=False)) + + upload_date = unified_strdate(self._html_search_meta( + 'last-modified', webpage, 'upload date', fatal=None)) + + return { + 'id': video_id, + 'title': title, + 'description': description, + 'thumbnail': thumbnail, + 'duration': duration, + 'upload_date': upload_date, + 'formats': formats + } |