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=476c1cd9b381bb0d5d5ec1979cc4928bff9fa9f4;hp=30d0c4a8de8d895292879a074a6bf654ecdec4c5;hb=d7465beff921821bf120a25a30a35ef06bddfc0e;hpb=efeac1885cd0211754b34e173a5efb74b6cbedb2

diff --git a/yarrg/yppedia-ocean-scraper b/yarrg/yppedia-ocean-scraper
index 30d0c4a..476c1cd 100755
--- a/yarrg/yppedia-ocean-scraper
+++ b/yarrg/yppedia-ocean-scraper
@@ -42,6 +42,23 @@ import re as regexp
 from optparse import OptionParser
 from BeautifulSoup import BeautifulSoup
 
+
+# For fuck's sake!
+import codecs
+import locale
+def fix_stdout():
+    sys.stdout = codecs.EncodedFile(sys.stdout, locale.getpreferredencoding())
+    def null_decode(input, errors='strict'):
+        return input, len(input)
+    sys.stdout.decode = null_decode
+# From
+#  http://ewx.livejournal.com/457086.html?thread=3016574
+#  http://ewx.livejournal.com/457086.html?thread=3016574
+# lightly modified.
+# See also Debian #415968.
+fix_stdout()
+
+
 ocean = None
 soup = None
 opts = None
@@ -89,13 +106,13 @@ def parse_chart():
 	debug('s',s)
 	s = regexp.sub(r'\&lt\;', '<', s)
 	s = regexp.sub(r'\&gt\;', '>', s)
+	s = regexp.sub(r'\&quot\;', '"', s)
 	s = regexp.sub(r'\&amp\;', '&', s)
 	debug('s',s)
 	return s
 
 def parse_ocean():
-	firstarch = soup.find('a', attrs = {'title': title_arch_ok})
-	debug('fa',firstarch)
+	content = soup.find('div', attrs = {'id': 'content'})
 
 	def findall_title_arch_ok(t):
 		return t.findAll('a', attrs = {'title': title_arch_ok})
@@ -104,7 +121,7 @@ def parse_ocean():
 		if u.name != 'table': return False
 		return len(findall_title_arch_ok(u)) > 1
 
-	archestable = firstarch.findParent('table', attrs={'border':'1'})
+	archestable = content.findChild('table', attrs={'border':'1'})
 	debug('at',archestable)
 
 	archsoups = []