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=5e5e0901f563b097b6554b891e04cf184f89b45a;hp=d55f7acab83d7d0a03ae21655059e4277e1e9068;hb=ac0ff0f496aaa83b013c0102ee8e4e4076d150b4;hpb=1457a9e8d2e5ddda2b05b658e69df1b80a07320d

diff --git a/yarrg/yppedia-ocean-scraper b/yarrg/yppedia-ocean-scraper
index d55f7ac..5e5e090 100755
--- a/yarrg/yppedia-ocean-scraper
+++ b/yarrg/yppedia-ocean-scraper
@@ -88,16 +88,18 @@ def fetch():
 		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 = urllib.urlopen(url)
-	debug('fetched',dataf)
+	url_base = url_base % urllib.quote(ocean,'')
+	if opts.localhtml is None:
+		url = ('http://yppedia.puzzlepirates.com/' + url_base)
+		debug('fetching',url)
+		dataf = urllib.urlopen(url)
+		debug('fetched',dataf)
+	else:
+		dataf = file(opts.localhtml + '/' + url_base, 'r')
 	soup = BeautifulSoup(dataf)
 
-
 title_arch_re = regexp.compile('(\\S.*\\S) Archipelago \\((\\S+)\\)$')
-title_any_re = regexp.compile('(\\S.*\\S) \((\\S+)\\)$')
+title_any_re = regexp.compile('(\\S.*\\S) \((\\S+)\\)(?: \(page does not exist\))?$')
 href_img_re = regexp.compile('\\.png$')
 
 def title_arch_info(t):
@@ -189,6 +191,9 @@ def main():
 		help='print chart source rather than arch/island info')
 	ao('--debug', action='count', dest='debug', default=0,
 		help='enable debugging output')
+	ao('--local-html-dir', action='store', dest='localhtml',
+		help='get yppedia pages from local directory LOCALHTML'+
+			' instead of via HTTP')
 
 	(opts,args) = pa.parse_args()
 	if len(args) != 1: