1
0
mirror of https://github.com/yt-dlp/yt-dlp.git synced 2024-11-02 17:22:31 +01:00

[Mojvideo] Add new extractor

This commit is contained in:
David Fabijan 2014-08-03 10:50:25 +02:00
parent 493987fefe
commit 5537dce84d
2 changed files with 47 additions and 0 deletions

View File

@ -181,6 +181,7 @@
from .mlb import MLBIE
from .mpora import MporaIE
from .mofosex import MofosexIE
from .mojvideo import MojvideoIE
from .mooshare import MooshareIE
from .morningstar import MorningstarIE
from .motherless import MotherlessIE

View File

@ -0,0 +1,46 @@
import re
from .common import InfoExtractor
class MojvideoIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?mojvideo\.com/video-.*/(?P<id>[a-f0-9]+)'
_TEST = {
'url': 'http://www.mojvideo.com/video-v-avtu-pred-mano-rdecelaska-alfi-nipic/3d1ed4497707730b2906',
'md5': 'f7fd662cc8ce2be107b0d4f2c0483ae7',
'info_dict': {
'id': '3d1ed4497707730b2906',
'ext': 'mp4',
'title': 'V avtu pred mano rdečelaska - Alfi Nipič',
'description':'Video: V avtu pred mano rdečelaska - Alfi Nipič',
'height':378,
'width':480
# TODO more properties, either as:
# * A value
# * MD5 checksum; start the string with md5:
# * A regular expression; start the string with re:
# * Any Python type (for example int or float)
}
}
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
video_id = mobj.group('id')
# TODO more code goes here, for example ...
webpage = self._download_webpage(url, video_id)
title = self._html_search_regex(r'<title>(.*?)</title>', webpage, 'title')
description = self._search_regex(r'<meta name="description" content="(.*)" />', webpage, 'video description')
final_url = self._html_search_regex(r'mp4: \'(.*)\'', webpage, 'video url')
height=int(self._search_regex(r'<meta name="video_height" content="([0-9]*)" />',webpage,"video height"))
width=int(self._search_regex(r'<meta name="video_width" content="([0-9]*)" />',webpage,"video width"))
return {
'id': video_id,
'title': title,
'description': description,
'ext': 'mp4',
'url': final_url,
'height':height,
'width':width
# TODO more properties (see youtube_dl/extractor/common.py)
}