From: Ian Jackson Date: Wed, 1 Feb 2012 22:20:40 +0000 (+0000) Subject: Merge branch 'master' of /u/ijackson/things/ypp-sc-tools.pctb-dict X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~ijackson/git?a=commitdiff_plain;h=4c8579ae2b99c66fbb9fd2e5ce379fed724462d2;hp=a203bc7861fa98862f6b887d07fe824d790e001b;p=ypp-sc-tools.git Merge branch 'master' of /u/ijackson/things/ypp-sc-tools.pctb-dict --- diff --git a/yarrg/source-info.txt b/yarrg/source-info.txt index d2629ff..fdd5125 100644 --- a/yarrg/source-info.txt +++ b/yarrg/source-info.txt @@ -231,6 +231,13 @@ ocean Hunter Ix Chel Manu Island +ocean Emerald + Crab + The Beaufort Islands + Osprey + Scurvy Reef + Gauntlet Island + ocean Malachite Draco Cetus Island @@ -243,6 +250,16 @@ ocean Sage ocean Viridian +ocean Meridian + Draco + Cetus Island + Threewood Island + Wyvern Island + Basilisk + Zechstein Island + Komodo + Buyan's Vortice + # family oceans ocean Crimson diff --git a/yarrg/yppedia-ocean-scraper b/yarrg/yppedia-ocean-scraper index ba145ea..5e5e090 100755 --- a/yarrg/yppedia-ocean-scraper +++ b/yarrg/yppedia-ocean-scraper @@ -99,7 +99,7 @@ def fetch(): 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):