From bab92f1e137903283c0b0fb986d01b6f91cdd82e Mon Sep 17 00:00:00 2001
From: gmungoc <g.m.carstairs@dundee.ac.uk>
Date: Mon, 9 May 2016 15:18:12 +0100
Subject: [PATCH] JAL-2098 identify feature file with "<" in feature type name

---
 src/jalview/io/IdentifyFile.java      |   30 +++++++-----------------------
 test/jalview/io/IdentifyFileTest.java |   25 ++++++++++++++++++-------
 2 files changed, 25 insertions(+), 30 deletions(-)

diff --git a/src/jalview/io/IdentifyFile.java b/src/jalview/io/IdentifyFile.java
index 71f4237..889359f 100755
--- a/src/jalview/io/IdentifyFile.java
+++ b/src/jalview/io/IdentifyFile.java
@@ -230,7 +230,6 @@ public class IdentifyFile
                 } catch (IOException ex)
                 {
                 }
-                ;
                 if (dta != null && dta.indexOf("*") > -1)
                 {
                   starterm = true;
@@ -250,34 +249,19 @@ public class IdentifyFile
           // read as a FASTA (probably)
           break;
         }
-        if ((data.indexOf("<") > -1)) // possible Markup Language data i.e HTML,
+        int lessThan = data.indexOf("<");
+        if ((lessThan > -1)) // possible Markup Language data i.e HTML,
                                       // RNAML, XML
         {
-          // FIXME this is nuts - it consumes the rest of the file if no match
-          boolean identified = false;
-          do
-          {
-            if (data.matches("<(?i)html(\"[^\"]*\"|'[^']*'|[^'\">])*>"))
-            {
-              reply = HtmlFile.FILE_DESC;
-              identified = true;
-              break;
-            }
-
-            if (data.matches("<(?i)rnaml (\"[^\"]*\"|'[^']*'|[^'\">])*>"))
-            {
-              reply = "RNAML";
-              identified = true;
-              break;
-            }
-          } while ((data = source.nextLine()) != null);
-
-          if (identified)
+          String upper = data.toUpperCase();
+          if (upper.substring(lessThan).startsWith("<HTML"))
           {
+            reply = HtmlFile.FILE_DESC;
             break;
           }
-          if (data == null)
+          if (upper.substring(lessThan).startsWith("<RNAML"))
           {
+            reply = "RNAML";
             break;
           }
         }
diff --git a/test/jalview/io/IdentifyFileTest.java b/test/jalview/io/IdentifyFileTest.java
index b1efb7a..c00cf06 100644
--- a/test/jalview/io/IdentifyFileTest.java
+++ b/test/jalview/io/IdentifyFileTest.java
@@ -20,6 +20,7 @@
  */
 package jalview.io;
 
+import static org.testng.AssertJUnit.assertEquals;
 import static org.testng.AssertJUnit.assertFalse;
 import static org.testng.AssertJUnit.assertTrue;
 
@@ -41,24 +42,34 @@ public class IdentifyFileTest
   }
 
   /**
-   * Additional tests for (a) Jalview features file with no colour
-   * specifications (old style 'groups' file) and (b) Jalview features file with
-   * embedded GFF
+   * Additional tests for Jalview features file
    */
   @Test(groups = "Functional")
   public void testIdentify_featureFile()
   {
     IdentifyFile ider = new IdentifyFile();
 
-    // Jalview format with features only, no feature colours
+    /*
+     * Jalview format with features only, no feature colours
+     */
     String data = "Iron-sulfur (2Fe-2S)\tFER_CAPAA\t-1\t39\t39\tMETAL\n"
             + "Iron-phosphorus (2Fe-P)\tID_NOT_SPECIFIED\t2\t86\t87\tMETALLIC\n";
-    Assert.assertEquals(IdentifyFile.FeaturesFile, ider.identify(data, AppletFormatAdapter.PASTE));
+    assertEquals(IdentifyFile.FeaturesFile,
+            ider.identify(data, AppletFormatAdapter.PASTE));
 
-    // Jalview feature colour followed by GFF format feature data
+    /*
+     * Jalview feature colour followed by GFF format feature data
+     */
     data = "METAL\tcc9900\n" + "GFF\n"
             + "FER_CAPAA\tuniprot\tMETAL\t44\t45\t4.0\t.\t.\n";
-    Assert.assertEquals(IdentifyFile.FeaturesFile,
+    assertEquals(IdentifyFile.FeaturesFile,
+            ider.identify(data, AppletFormatAdapter.PASTE));
+
+    /*
+     * Feature with '<' in the name (JAL-2098)
+     */
+    data = "kD < 3\tred\n" + "Low kD\tFER_CAPAA\t-1\t39\t39\tkD < 3\n";
+    assertEquals(IdentifyFile.FeaturesFile,
             ider.identify(data, AppletFormatAdapter.PASTE));
   }
 
-- 
1.7.10.2