chiark / gitweb /
Do entity conversion ourselves; lenny's BS won't for some reason
authorIan Jackson <ijackson@chiark.greenend.org.uk>
Tue, 1 Sep 2009 14:39:54 +0000 (15:39 +0100)
committerIan Jackson <Ian.Jackson@eu.citrix.com>
Tue, 1 Sep 2009 14:39:54 +0000 (15:39 +0100)
yarrg/yppedia-ocean-scraper

index 5f9c5f0..30d0c4a 100755 (executable)
@@ -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})