Show
Ignore:
Timestamp:
12.01.2012 15:52:20 (8 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

Files:
1 modified

Legend:

Unmodified
Added
Removed
  • gs3-extensions/html-to-expeditee/trunk/src/src/cgi-bin/html-to-expeditee.pl.in

    r24937 r24944  
    108108          docOIDs = []; 
    109109          var outstandingURLs = []; 
     110          var visitedURLs = {}; 
     111 
    110112          outstandingURLs.push(url); 
     113          visitedURLs[url] = 1; 
    111114 
    112115          while (outstandingURLs.length>0) { 
     
    121124           
    122125           
    123           /* any links with a=d ... cl=??? => outstandingURLS */ 
     126          /* any links with a=(b|d) ... cl=??? => outstandingURLS */ 
    124127          /* any links with a=d ... d=???  => docOIDS */ 
    125128           
    126129          var actionRE = new RegExp("(\\\\?|&)a=(?:d|b)(&|\$)"); 
    127           var clRE     = new RegExp("(\\\\?|&)cl=" + cl + "\\\\."); 
     130          var clRE     = new RegExp("(\\\\?|&)cl=" + cl + "(\\\\.\\\\d+)+(&|\$)"); 
    128131          var docRE    = new RegExp("(?:\\\\?|&)d=(.*?)(?:&|\$)"); 
    129132 
     
    133136              if (href && href.match(actionRE)) { 
    134137              if (href.match(clRE)) { 
    135                   outstandingURLs.push(href); 
     138                  if (!visitedURLs[href]) {       
     139                      // console.log("found a new CL line: " + href); 
     140                      outstandingURLs.push(href); 
     141                  visitedURLs[href] = 1; 
     142                  }  
    136143              } 
    137144              else if (href.match(docRE)) { 
     
    139146                  var docMatch = docRE.exec(href); 
    140147                  var docOID = docMatch[1]; 
     148 
     149                  // console.log("found a new doc line: " + docOID); 
    141150 
    142151                  docOIDs.push(docOID); 
     
    151160 
    152161          var docOID = docOIDs.shift(); 
    153           console.log("doc oid = " + docOID); 
     162          //console.log("doc oid = " + docOID); 
    154163           
    155164          var url;