Changeset 24944 for gs3-extensions/html-to-expeditee/trunk/src/src/cgi-bin/html-to-expeditee.pl.in

Timestamp:

2012-01-12T15:52:20+13:00 (12 years ago)

Author:

davidb

Message:

More careful recursive traversal to classifier nodes; further developed of CSS to Expeditee attributes (background colour); and slightly less cludgy way to deal with full img URLS

File:

• gs3-extensions/html-to-expeditee/trunk/src/src/cgi-bin/html-to-expeditee.pl.in (modified) (5 diffs)

gs3-extensions/html-to-expeditee/trunk/src/src/cgi-bin/html-to-expeditee.pl.in

@@ -108 +108 @@
           docOIDs = [];
           var outstandingURLs = [];
+          var visitedURLs = {};
+
           outstandingURLs.push(url);
+          visitedURLs[url] = 1;
 
           while (outstandingURLs.length>0) {
@@ -121 +124 @@
           
           
-          /* any links with a=d ... cl=??? => outstandingURLS */
+          /* any links with a=(b|d) ... cl=??? => outstandingURLS */
           /* any links with a=d ... d=???  => docOIDS */
           
           var actionRE = new RegExp("(\\\\?|&)a=(?:d|b)(&|\$)");
-          var clRE     = new RegExp("(\\\\?|&)cl=" + cl + "\\\\.");
+          var clRE     = new RegExp("(\\\\?|&)cl=" + cl + "(\\\\.\\\\d+)+(&|\$)");
           var docRE    = new RegExp("(?:\\\\?|&)d=(.*?)(?:&|\$)");
 
@@ -133 +136 @@
               if (href && href.match(actionRE)) {
               if (href.match(clRE)) {
-                  outstandingURLs.push(href);
+                  if (!visitedURLs[href]) {      
+                      // console.log("found a new CL line: " + href);
+                      outstandingURLs.push(href);
+                  visitedURLs[href] = 1;
+                  } 
               }
               else if (href.match(docRE)) {
@@ -139 +146 @@
                   var docMatch = docRE.exec(href);
                   var docOID = docMatch[1];
+
+                  // console.log("found a new doc line: " + docOID);
 
                   docOIDs.push(docOID);
@@ -151 +160 @@
 
           var docOID = docOIDs.shift();
-          console.log("doc oid = " + docOID);
+          //console.log("doc oid = " + docOID);
           
           var url;