[gamestar] Use helper methods to not break if something changes (#3393)

master
Philipp Hagemeister 2014-07-29 05:59:47 +02:00
parent a2a1b0baa2
commit 895ba7d1dd
1 changed files with 20 additions and 14 deletions

View File

@ -4,6 +4,13 @@ from __future__ import unicode_literals
import re import re
from .common import InfoExtractor from .common import InfoExtractor
from ..utils import (
int_or_none,
parse_duration,
str_to_int,
unified_strdate,
)
class GameStarIE(InfoExtractor): class GameStarIE(InfoExtractor):
_VALID_URL = r'http://www\.gamestar\.de/videos/.*,(?P<id>[0-9]+)\.html' _VALID_URL = r'http://www\.gamestar\.de/videos/.*,(?P<id>[0-9]+)\.html'
@ -34,25 +41,24 @@ class GameStarIE(InfoExtractor):
description = self._og_search_description(webpage).strip() description = self._og_search_description(webpage).strip()
og_thumbnail = self._og_search_thumbnail(webpage) thumbnail = self._proto_relative_url(
thumbnail = 'http:' + og_thumbnail self._og_search_thumbnail(webpage), scheme='http:')
upload_date_raw = self._html_search_regex( upload_date = unified_strdate(self._html_search_regex(
r'<span style="float:left;font-size:11px;">Datum: ([0-9]+\.[0-9]+\.[0-9]+)&nbsp;&nbsp;', r'<span style="float:left;font-size:11px;">Datum: ([0-9]+\.[0-9]+\.[0-9]+)&nbsp;&nbsp;',
webpage, 'upload_date').split('.') webpage, 'upload_date', fatal=False))
upload_date = upload_date_raw[2] + upload_date_raw[1] + upload_date_raw[0]
duration_raw = self._html_search_regex( duration = parse_duration(self._html_search_regex(
r'&nbsp;&nbsp;Länge: ([0-9]+:[0-9]+)</span>', webpage, 'duration').split(':') r'&nbsp;&nbsp;Länge: ([0-9]+:[0-9]+)</span>', webpage, 'duration',
duration = int(duration_raw[0])*60 + int(duration_raw[1]) fatal=False))
view_count_raw = self._html_search_regex( view_count = str_to_int(self._html_search_regex(
r'&nbsp;&nbsp;Zuschauer: ([0-9\.]+)&nbsp;&nbsp;', webpage, 'view_count') r'&nbsp;&nbsp;Zuschauer: ([0-9\.]+)&nbsp;&nbsp;', webpage,
view_count = int(view_count_raw.replace('.', '')) 'view_count', fatal=False))
comment_count_raw = self._html_search_regex( comment_count = int_or_none(self._html_search_regex(
r'>Kommentieren \(([0-9]+)\)</a>', webpage, 'comment_count') r'>Kommentieren \(([0-9]+)\)</a>', webpage, 'comment_count',
comment_count = int(comment_count_raw) fatal=False))
return { return {
'id': video_id, 'id': video_id,