[fix] handle missing thumbnail

pull/1/head
Adam Tauber 5 years ago
parent 2dc2e1e8f9
commit 86a378bd01

@ -32,7 +32,7 @@ base_url = 'https://www.ina.fr'
search_url = base_url + '/layout/set/ajax/recherche/result?autopromote=&hf={ps}&b={start}&type=Video&r=&{query}' search_url = base_url + '/layout/set/ajax/recherche/result?autopromote=&hf={ps}&b={start}&type=Video&r=&{query}'
# specific xpath variables # specific xpath variables
results_xpath = '//div[contains(@class,"search-results--list")]/div[@class="media"]' results_xpath = '//div[contains(@class,"search-results--list")]//div[@class="media-body"]'
url_xpath = './/a/@href' url_xpath = './/a/@href'
title_xpath = './/h3[@class="h3--title media-heading"]' title_xpath = './/h3[@class="h3--title media-heading"]'
thumbnail_xpath = './/img/@src' thumbnail_xpath = './/img/@src'
@ -65,8 +65,11 @@ def response(resp):
videoid = result.xpath(url_xpath)[0] videoid = result.xpath(url_xpath)[0]
url = base_url + videoid url = base_url + videoid
title = p.unescape(extract_text(result.xpath(title_xpath))) title = p.unescape(extract_text(result.xpath(title_xpath)))
thumbnail = extract_text(result.xpath(thumbnail_xpath)[0]) try:
if thumbnail[0] == '/': thumbnail = extract_text(result.xpath(thumbnail_xpath)[0])
except:
thumbnail = ''
if thumbnail and thumbnail[0] == '/':
thumbnail = base_url + thumbnail thumbnail = base_url + thumbnail
d = extract_text(result.xpath(publishedDate_xpath)[0]) d = extract_text(result.xpath(publishedDate_xpath)[0])
d = d.split('/') d = d.split('/')

Loading…
Cancel
Save