Bug fix 'Detroit Institute of Arts' parser.
This commit is contained in:
parent
e38e1b994d
commit
04091988de
|
@ -6,7 +6,10 @@ import re
|
||||||
re_url = re.compile(r'https?://www.dia.org/art/collection/object/(.+)$')
|
re_url = re.compile(r'https?://www.dia.org/art/collection/object/(.+)$')
|
||||||
|
|
||||||
def get_html(url):
|
def get_html(url):
|
||||||
catalog_id = re_url.search(url).group(1).replace('/', '_')
|
m = re_url.search(url).group(1).replace('/', '_')
|
||||||
|
if not m:
|
||||||
|
return
|
||||||
|
catalog_id = m.group(1).replace('/', '_')
|
||||||
|
|
||||||
filename = f'cache/dia_{catalog_id}.html'
|
filename = f'cache/dia_{catalog_id}.html'
|
||||||
|
|
||||||
|
@ -47,4 +50,6 @@ def parse_html(html):
|
||||||
}
|
}
|
||||||
|
|
||||||
def get_catalog(url):
|
def get_catalog(url):
|
||||||
return parse_html(get_html(url))
|
html = get_html(url)
|
||||||
|
if html:
|
||||||
|
return parse_html(html)
|
||||||
|
|
Loading…
Reference in a new issue