diff options
Diffstat (limited to 'hypervideo_dl/extractor/myvideoge.py')
| -rw-r--r-- | hypervideo_dl/extractor/myvideoge.py | 68 |
1 files changed, 48 insertions, 20 deletions
diff --git a/hypervideo_dl/extractor/myvideoge.py b/hypervideo_dl/extractor/myvideoge.py index 513d4cb..64cee48 100644 --- a/hypervideo_dl/extractor/myvideoge.py +++ b/hypervideo_dl/extractor/myvideoge.py @@ -1,5 +1,16 @@ +import re + from .common import InfoExtractor -from ..utils import js_to_json +from ..utils import ( + MONTH_NAMES, + clean_html, + get_element_by_class, + get_element_by_id, + int_or_none, + js_to_json, + qualities, + unified_strdate, +) class MyVideoGeIE(InfoExtractor): @@ -11,37 +22,50 @@ class MyVideoGeIE(InfoExtractor): 'id': '3941048', 'ext': 'mp4', 'title': 'The best prikol', + 'upload_date': '20200611', 'thumbnail': r're:^https?://.*\.jpg$', - 'uploader': 'md5:d72addd357b0dd914e704781f7f777d8', - 'description': 'md5:5c0371f540f5888d603ebfedd46b6df3' - } + 'uploader': 'chixa33', + 'description': 'md5:5b067801318e33c2e6eea4ab90b1fdd3', + }, } + _MONTH_NAMES_KA = ['იანვარი', 'თებერვალი', 'მარტი', 'აპრილი', 'მაისი', 'ივნისი', 'ივლისი', 'აგვისტო', 'სექტემბერი', 'ოქტომბერი', 'ნოემბერი', 'დეკემბერი'] + + _quality = staticmethod(qualities(('SD', 'HD'))) def _real_extract(self, url): video_id = self._match_id(url) webpage = self._download_webpage(url, video_id) - title = self._html_search_regex(r'<h1[^>]*>([^<]+)</h1>', webpage, 'title') - description = self._og_search_description(webpage) - thumbnail = self._html_search_meta(['og:image'], webpage) - uploader = self._search_regex(r'<a[^>]+class="mv_user_name"[^>]*>([^<]+)<', webpage, 'uploader', fatal=False) + title = ( + self._og_search_title(webpage, default=None) + or clean_html(get_element_by_class('my_video_title', webpage)) + or self._html_search_regex(r'<title\b[^>]*>([^<]+)</title\b', webpage, 'title')) jwplayer_sources = self._parse_json( self._search_regex( - r"(?s)jwplayer\(\"mvplayer\"\).setup\(.*?sources: (.*?])", webpage, 'jwplayer sources'), - video_id, transform_source=js_to_json) + r'''(?s)jwplayer\s*\(\s*['"]mvplayer['"]\s*\)\s*\.\s*setup\s*\(.*?\bsources\s*:\s*(\[.*?])\s*[,});]''', webpage, 'jwplayer sources', fatal=False) + or '', + video_id, transform_source=js_to_json, fatal=False) + + formats = self._parse_jwplayer_formats(jwplayer_sources or [], video_id) + for f in formats or []: + f['quality'] = self._quality(f['format_id']) - def _formats_key(f): - if f['label'] == 'SD': - return -1 - elif f['label'] == 'HD': - return 1 - else: - return 0 + description = ( + self._og_search_description(webpage) + or get_element_by_id('long_desc_holder', webpage) + or self._html_search_meta('description', webpage)) - jwplayer_sources = sorted(jwplayer_sources, key=_formats_key) + uploader = self._search_regex(r'<a[^>]+class="mv_user_name"[^>]*>([^<]+)<', webpage, 'uploader', fatal=False) - formats = self._parse_jwplayer_formats(jwplayer_sources, video_id) + upload_date = get_element_by_class('mv_vid_upl_date', webpage) + # as ka locale may not be present roll a local date conversion + upload_date = (unified_strdate( + # translate any ka month to an en one + re.sub('|'.join(self._MONTH_NAMES_KA), + lambda m: MONTH_NAMES['en'][self._MONTH_NAMES_KA.index(m.group(0))], + upload_date, re.I)) + if upload_date else None) return { 'id': video_id, @@ -49,5 +73,9 @@ class MyVideoGeIE(InfoExtractor): 'description': description, 'uploader': uploader, 'formats': formats, - 'thumbnail': thumbnail + 'thumbnail': self._og_search_thumbnail(webpage), + 'upload_date': upload_date, + 'view_count': int_or_none(get_element_by_class('mv_vid_views', webpage)), + 'like_count': int_or_none(get_element_by_id('likes_count', webpage)), + 'dislike_count': int_or_none(get_element_by_id('dislikes_count', webpage)), } |