From: Ian Jackson <ian@liberator.relativity.greenend.org.uk>
Date: Mon, 31 Aug 2009 15:36:42 +0000 (+0100)
Subject: Cope with spaces in arch names as found on Hunter
X-Git-Tag: 3.4~59
X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?p=ypp-sc-tools.db-live.git;a=commitdiff_plain;h=7078fe9eae28c76913e5c14946f9dbf94cfb3458

Cope with spaces in arch names as found on Hunter
---

diff --git a/yarrg/yppedia-ocean-scraper b/yarrg/yppedia-ocean-scraper
index 401de9b..9105f19 100755
--- a/yarrg/yppedia-ocean-scraper
+++ b/yarrg/yppedia-ocean-scraper
@@ -61,7 +61,7 @@ def fetch():
 	soup = BeautifulSoup(dataf)
 
 
-title_arch_re = regexp.compile('(\\S+) Archipelago \\((\\S+)\\)$')
+title_arch_re = regexp.compile('(\\S.*\\S) Archipelago \\((\\S+)\\)$')
 title_any_re = regexp.compile('(\\S.*\\S) \((\\S+)\\)$')
 href_img_re = regexp.compile('\\.png$')
 
@@ -107,8 +107,8 @@ def parse():
 		debug('links',links)
 		if not links: continue
 		(a,o) = title_arch_info(links[0]['title'])
+		debug('arch-ocean', (a,o))
 		assert(o == ocean)
-		debug('arch', a)
 		assert(a not in arches)
 		isles = []
 		for link in links[1:]: