From 4e7a3bfd7279669b9040c5265a9c4c4fc4ff1baa Mon Sep 17 00:00:00 2001
From: jprocter <Jim Procter>
Date: Fri, 2 Apr 2010 10:33:54 +0000
Subject: [PATCH] urllength limit and more efficient url construction

---
 src/jalview/util/GroupUrlLink.java |   39 ++++++++++++++++++++++++++----------
 1 file changed, 28 insertions(+), 11 deletions(-)

diff --git a/src/jalview/util/GroupUrlLink.java b/src/jalview/util/GroupUrlLink.java
index 55e3bd9..0401bb2 100644
--- a/src/jalview/util/GroupUrlLink.java
+++ b/src/jalview/util/GroupUrlLink.java
@@ -410,6 +410,8 @@ public class GroupUrlLink
    */
   protected Object[] makeUrlsIf(boolean createFullUrl, Hashtable repstrings, boolean onlyIfMatches)
   {
+    int pass = 0;
+    
     // prepare string arrays in correct order to be assembled into URL input
     String[][] idseq = new String[mtch.length][]; // indexed by pass
     int mins = 0, maxs = 0; // allowed two values, 1 or n-sequences.
@@ -444,7 +446,6 @@ public class GroupUrlLink
                 "Cannot have zero length vector of replacement strings - either 1 value or n values.");
       }
     }
-    int pass = 0;
     // iterate through input, collating segments to be inserted into url
     StringBuffer matched[] = new StringBuffer[idseq.length];
     // and precompile regexes
@@ -462,7 +463,14 @@ public class GroupUrlLink
         rgxs[pass] = null;
       }
     }
-    // record which of the input sequences were actually used to generate the
+    // tot up the invariant lengths for this url
+    int urllength = url_prefix.length();
+    for (pass=0;pass<matched.length; pass++)
+    {
+      urllength+=url_suffix[pass].length();
+    }
+
+    // flags to record which of the input sequences were actually used to generate the
     // url
     boolean[] thismatched = new boolean[maxs];
     int seqsmatched = 0;
@@ -470,10 +478,10 @@ public class GroupUrlLink
     {
       // initialise flag for match
       thismatched[sq] = false;
-      String[] thematches = new String[rgxs.length];
+      StringBuffer[] thematches = new StringBuffer[rgxs.length];
       for (pass = 0; pass < rgxs.length; pass++)
       {
-        thematches[pass] = ""; // initialise - in case there are no more
+        thematches[pass] = new StringBuffer(); // initialise - in case there are no more
         // matches.
         // if a regex is provided, then it must match for all sequences in all
         // tokens for it to be considered.
@@ -491,6 +499,7 @@ public class GroupUrlLink
           {
             rematchat = rg.matchedTo();
             thismatched[sq] |= true;
+            urllength+=rg.charsMatched(); // count length
             if (!createFullUrl)
             {
               continue; // don't bother making the URL replacement text.
@@ -499,7 +508,7 @@ public class GroupUrlLink
             int ns = rg.numSubs();
             if (ns == 0)
             {
-              thematches[pass] += rg.stringMatched();// take whole regex
+              thematches[pass].append(rg.stringMatched());// take whole regex
             }
             /*
              * else if (ns==1) { // take only subgroup match return new String[]
@@ -513,12 +522,12 @@ public class GroupUrlLink
             else
             {
               // debug
-              for (int s = 0; s <= rg.numSubs(); s++)
+              /*for (int s = 0; s <= rg.numSubs(); s++)
               {
                 System.err.println("Sub " + s + " : " + rg.matchedFrom(s)
                         + " : " + rg.matchedTo(s) + " : '"
                         + rg.stringMatched(s) + "'");
-              }
+              }*/
               // try to collate subgroup matches
               StringBuffer subs = new StringBuffer();
               // have to loop through submatches, collating them at top level
@@ -533,12 +542,12 @@ public class GroupUrlLink
                   // s is top level submatch. search for submatches enclosed by
                   // this one
                   int r = s + 1;
-                  String rmtch = "";
+                  StringBuffer rmtch = new StringBuffer();
                   while (r <= ns && rg.matchedTo(r) <= rg.matchedTo(s))
                   {
                     if (rg.matchedFrom(r) > -1)
                     {
-                      rmtch += rg.stringMatched(r);
+                      rmtch.append(rg.stringMatched(r));
                     }
                     r++;
                   }
@@ -557,7 +566,7 @@ public class GroupUrlLink
                   s++;
                 }
               }
-              thematches[pass] += subs.toString();
+              thematches[pass].append(subs);
             }
           }
         }
@@ -567,9 +576,10 @@ public class GroupUrlLink
           if (!onlyIfMatches)
           {
             thismatched[sq] |= true;
+            urllength+=idseq[pass][sq].length(); // tot up length
             if (createFullUrl)
             {
-            thematches[pass] = idseq[pass][sq]; // take whole string -
+            thematches[pass] = new StringBuffer(idseq[pass][sq]); // take whole string -
             // regardless - probably not a
             // good idea!
             /*
@@ -609,6 +619,12 @@ public class GroupUrlLink
       // no matches - no url generated
       return null;
     }
+    // check if we are beyond the feasible command line string limit for this platform
+    if ((urllength+32)>Platform.getMaxCommandLineLength())
+    {
+      System.err.println("URL estimated to be too long "+urllength);
+      return null;
+    }
     if (!createFullUrl)
     {
       // just return the essential info about what the URL would be generated from
@@ -617,6 +633,7 @@ public class GroupUrlLink
                         { seqsmatched }, thismatched};
     }
     // otherwise, create the URL completely.
+    
     StringBuffer submiturl = new StringBuffer();
     submiturl.append(url_prefix);
     for (pass = 0; pass < matched.length; pass++)
-- 
1.7.10.2