chiark / gitweb /
docs typo fixes
[ypp-sc-tools.db-live.git] / yarrg / yppedia-ocean-scraper
index d55f7acab83d7d0a03ae21655059e4277e1e9068..ba145eafcc247b6bae90d13244b159d62570928d 100755 (executable)
@@ -88,14 +88,16 @@ 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+)\\)$')
 href_img_re = regexp.compile('\\.png$')
@@ -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: