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=9105f1963998d96943c08237ad9a36f84deb5d85;hb=87b0d92e050701753d558e49aaae1d433b546170;hpb=7078fe9eae28c76913e5c14946f9dbf94cfb3458

diff --git a/yarrg/yppedia-ocean-scraper b/yarrg/yppedia-ocean-scraper
index 9105f19..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
@@ -53,8 +70,12 @@ def debug(k,v):
 
 def fetch():
 	global soup
-	url = ('http://yppedia.puzzlepirates.com/%s_Ocean' %
-		urllib.quote(ocean,''))
+	if opts.chart:
+		url_base = 'index.php?title=Template:Map:%s_Ocean&action=edit'
+	else:
+		url_base = '%s_Ocean'
+	url = ('http://yppedia.puzzlepirates.com/' +
+			(url_base % urllib.quote(ocean,'')))
 	debug('fetching',url)
 	dataf = urllib2.urlopen(url)
 	debug('fetched',dataf)
@@ -78,9 +99,20 @@ def title_arch_ok(t):
 	if o is None: return False
 	return o == ocean
 
-def parse():
-	firstarch = soup.find('a', attrs = {'title': title_arch_ok})
-	debug('fa',firstarch)
+def parse_chart():
+	ta = soup.find('textarea')
+	debug('ta',ta)
+	s = ta.string
+	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():
+	content = soup.find('div', attrs = {'id': 'content'})
 
 	def findall_title_arch_ok(t):
 		return t.findAll('a', attrs = {'title': title_arch_ok})
@@ -89,7 +121,7 @@ def parse():
 		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 = []
@@ -136,8 +168,11 @@ def main():
 	global opts
 
 	pa = OptionParser(
-		'''usage: .../yppedia-ocean-scraper [--debug] OCEAN''')
+'''usage: .../yppedia-ocean-scraper [--debug] [--chart] OCEAN''')
 	ao = pa.add_option
+
+	ao('--chart', action='store_true', dest='chart',
+		help='print chart source rather than arch/island info')
 	ao('--debug', action='count', dest='debug', default=0,
 		help='enable debugging output')
 
@@ -148,7 +183,10 @@ def main():
 	ocean = args[0]
 
 	fetch()
-	parse()
-	output()
+	if opts.chart:
+		print parse_chart()
+	else:
+		parse_ocean()
+		output()
 
 main()