diff options
author | emkael <emkael@tlen.pl> | 2014-10-31 23:19:53 +0100 |
---|---|---|
committer | emkael <emkael@tlen.pl> | 2014-10-31 23:19:53 +0100 |
commit | c3048ce08397e24e90688f090f98e8f999acfa6a (patch) | |
tree | 934eac54ced3eeac415c757e1da7e98ba7868e50 /dumps/chicane-f1.py | |
parent | 9eabd017b4e7306bf59a9ef5776ed1d5a297e469 (diff) |
* helper script for fetching qualifying data
Diffstat (limited to 'dumps/chicane-f1.py')
-rw-r--r-- | dumps/chicane-f1.py | 20 |
1 files changed, 20 insertions, 0 deletions
diff --git a/dumps/chicane-f1.py b/dumps/chicane-f1.py new file mode 100644 index 0000000..2d9b4cd --- /dev/null +++ b/dumps/chicane-f1.py @@ -0,0 +1,20 @@ +#!/usr/bin/env python +import urllib +import urllib2 +import urlparse + +from lxml import html + +for year in range(1954,2015): + url = 'http://chicanef1.com/calendar.pl?' + urllib.urlencode({'year':year,'nc':0}) + contents = urllib2.urlopen(url).read() + tree = html.fromstring(contents) + links = tree.xpath('//table[@cellpadding=6]//tr/td[1]//a') + for link in links: + url = urlparse.urlparse(link.attrib['href']) + url = url._replace(path='race.pl') + query = dict(urlparse.parse_qsl(url.query)) + for type in ['qual', 'preq']: + query['type'] = type + url = url._replace(query=urllib.urlencode(query)) + print urlparse.urljoin('http://chicanef1.com', urlparse.urlunparse(url)) |