X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?p=ypp-sc-tools.db-live.git;a=blobdiff_plain;f=yarrg%2Fyppedia-ocean-scraper;h=30d0c4a8de8d895292879a074a6bf654ecdec4c5;hp=5f9c5f007b46fda4265ccc7492dc2b3d58e079f0;hb=9f77f36acb520531a53fe18db16fae9292f4d09a;hpb=ff016739a385102c8e11e80757d5404745bd9a39

diff --git a/yarrg/yppedia-ocean-scraper b/yarrg/yppedia-ocean-scraper
index 5f9c5f0..30d0c4a 100755
--- a/yarrg/yppedia-ocean-scraper
+++ b/yarrg/yppedia-ocean-scraper
@@ -62,8 +62,7 @@ def fetch():
 	debug('fetching',url)
 	dataf = urllib2.urlopen(url)
 	debug('fetched',dataf)
-	soup = BeautifulSoup(dataf,
-		convertEntities=BeautifulSoup.HTML_ENTITIES)
+	soup = BeautifulSoup(dataf)
 
 
 title_arch_re = regexp.compile('(\\S.*\\S) Archipelago \\((\\S+)\\)$')
@@ -86,9 +85,13 @@ def title_arch_ok(t):
 def parse_chart():
 	ta = soup.find('textarea')
 	debug('ta',ta)
-	rc = ta.renderContents()
-	debug('rc',rc)
-	return rc
+	s = ta.string
+	debug('s',s)
+	s = regexp.sub(r'\&lt\;', '<', s)
+	s = regexp.sub(r'\&gt\;', '>', s)
+	s = regexp.sub(r'\&amp\;', '&', s)
+	debug('s',s)
+	return s
 
 def parse_ocean():
 	firstarch = soup.find('a', attrs = {'title': title_arch_ok})