Cope with some features of the Cobalt WP chart

diff --git a/yarrg/yppedia-chart-parser b/yarrg/yppedia-chart-parser
index a093a0cec965dba77827a3641af83f9563295ec5..b35d8165a016bfe1dbe28a9166ef5ebf008b9543 100755 (executable)
--- a/yarrg/yppedia-chart-parser
+++ b/yarrg/yppedia-chart-parser
@@ -118,7 +118,7 @@ sub yppedia_chart_parse () {
        s/\<--.*--\>//g;
        s/^\s*//; chomp; s/\s+$//; s/\s+/ /g;
        s/\<\/?(?:b|em)\>//g;
-       s/\{\{Chart\ style\|[^{}]*\}\}//g;
+       s/\{\{chart\ style\|[^{}]*\}\}//gi;
        next unless m/\{\{/; # only interested in chart template stuff
 
        my ($x,$y, $arch,$island,$solid,$dirn);
@@ -126,9 +126,9 @@ sub yppedia_chart_parse () {
     
        if (($x,$y,$arch) =
            m/^\{\{ chart\ label \|(\d+)\|(\d+)\| .*
-                   (?: \<big\>)? \'+
+                   (?: \<(?: big|center )\>)* \'+
                    \[\[ [^][\']* \| ([^][\'|]+)\ archipelago \]\]
-                   \'+ (?: \<\/big\>)? \}\}$/xi) {
+                   \'+ (?: \<\/(?: big|center )\>)* \}\}$/xi) {
            printf DEBUG "%2d,%-2d arch %s\n", $x,$y,$arch;
            push @wiarchlabels, [ $x,$y,$arch ];
        } elsif (m/^\{\{ chart\ label \|\d+\|\d+\|

diff --git a/yarrg/yppedia-ocean-scraper b/yarrg/yppedia-ocean-scraper
index 30d0c4a8de8d895292879a074a6bf654ecdec4c5..ad35c4d708a432566941542156e10339d5ed421f 100755 (executable)
--- a/yarrg/yppedia-ocean-scraper
+++ b/yarrg/yppedia-ocean-scraper
@@ -89,6 +89,7 @@ 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