X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fio%2FIdentifyFile.java;h=cab3e60c2163931c436bf55b39091788e3bb6da7;hb=3c7d4cf5ef1331e2fad9c617c228486badf6f6a0;hp=7ad8fcd6dc7833f4e696c1389c797a231defe046;hpb=3d0101179759ef157b088ea135423cd909512d9f;p=jalview.git

diff --git a/src/jalview/io/IdentifyFile.java b/src/jalview/io/IdentifyFile.java
index 7ad8fcd..cab3e60 100755
--- a/src/jalview/io/IdentifyFile.java
+++ b/src/jalview/io/IdentifyFile.java
@@ -20,6 +20,9 @@
  */
 package jalview.io;
 
+import java.util.Locale;
+
+import java.io.File;
 import java.io.IOException;
 
 /**
@@ -30,6 +33,42 @@ import java.io.IOException;
  */
 public class IdentifyFile
 {
+
+  public FileFormatI identify(Object file, DataSourceType protocol)
+          throws FileFormatException
+  {
+    // BH 2018
+    return (file instanceof File ? identify((File) file, protocol)
+            : identify((String) file, protocol));
+
+  }
+
+  public FileFormatI identify(File file, DataSourceType sourceType)
+          throws FileFormatException
+  {
+    // BH 2018
+    String emessage = "UNIDENTIFIED FILE PARSING ERROR";
+    FileParse parser = null;
+    try
+    {
+      parser = new FileParse(file, sourceType);
+      if (parser.isValid())
+      {
+        return identify(parser);
+      }
+    } catch (Exception e)
+    {
+      System.err.println("Error whilst identifying " + file);
+      e.printStackTrace(System.err);
+      emessage = e.getMessage();
+    }
+    if (parser != null)
+    {
+      throw new FileFormatException(parser.errormessage);
+    }
+    throw new FileFormatException(emessage);
+  }
+
   /**
    * Identify a datasource's file content.
    *
@@ -55,7 +94,7 @@ public class IdentifyFile
       }
     } catch (Exception e)
     {
-      System.err.println("Error whilst identifying");
+      System.err.println("Error whilst identifying " + file);
       e.printStackTrace(System.err);
       emessage = e.getMessage();
     }
@@ -130,26 +169,39 @@ public class IdentifyFile
           if (source.inFile != null)
           {
             String fileStr = source.inFile.getName();
-            // possibly a Jalview archive.
-            if (fileStr.lastIndexOf(".jar") > -1
-                    || fileStr.lastIndexOf(".zip") > -1)
+            if (fileStr.contains(".jar") || fileStr.contains(".zip")
+                    || fileStr.contains(".jvp"))
             {
+              // possibly a Jalview archive (but check further)
               reply = FileFormat.Jalview;
             }
           }
           if (!lineswereskipped && data.startsWith("PK"))
           {
-            reply = FileFormat.Jalview; // archive.
+            reply = FileFormat.Jalview; // archive
             break;
           }
         }
-        data = data.toUpperCase();
+        data = data.toUpperCase(Locale.ROOT);
 
         if (data.startsWith(ScoreMatrixFile.SCOREMATRIX))
         {
           reply = FileFormat.ScoreMatrix;
           break;
         }
+        if (data.startsWith("LOCUS"))
+        {
+          reply = FileFormat.GenBank;
+          break;
+        }
+        if (data.startsWith("ID "))
+        {
+          if (data.substring(2).trim().split(";").length == 7)
+          {
+            reply = FileFormat.Embl;
+            break;
+          }
+        }
         if (data.startsWith("H ") && !aaIndexHeaderRead)
         {
           aaIndexHeaderRead = true;
@@ -275,11 +327,16 @@ public class IdentifyFile
           // read as a FASTA (probably)
           break;
         }
+        if (data.indexOf("{\"") > -1)
+        {
+          reply = FileFormat.Json;
+          break;
+        }
         int lessThan = data.indexOf("<");
         if ((lessThan > -1)) // possible Markup Language data i.e HTML,
                              // RNAML, XML
         {
-          String upper = data.toUpperCase();
+          String upper = data.toUpperCase(Locale.ROOT);
           if (upper.substring(lessThan).startsWith("<HTML"))
           {
             reply = FileFormat.Html;
@@ -292,11 +349,6 @@ public class IdentifyFile
           }
         }
 
-        if (data.indexOf("{\"") > -1)
-        {
-          reply = FileFormat.Json;
-          break;
-        }
         if ((data.length() < 1) || (data.indexOf("#") == 0))
         {
           lineswereskipped = true;
@@ -334,15 +386,31 @@ public class IdentifyFile
           reply = FileFormat.Phylip;
           break;
         }
-        else
+        else if (!lineswereskipped && looksLikeJnetData(data))
         {
-          if (!lineswereskipped && looksLikeJnetData(data))
-          {
             reply = FileFormat.Jnet;
             break;
-          }
+
         }
+        else // phylogenetic file
+        {
+          // may also have a <xml header before hand, probably need to check it 
+          if (data.toLowerCase(Locale.ROOT).startsWith("<phyloxml"))
+          {
+            reply = FileFormat.PhyloXML;
+            break;
+          }
+          // commented out, nexus parser is troublesome
 
+          // else if (((identifier.startsWith("nexus"))
+          // || (identifier.startsWith("#nexus"))
+          // || (identifier.startsWith("# nexus"))
+          // || (identifier.startsWith("begin"))))
+          // {
+          // reply = FileFormat.Nexus;
+          // break;
+          // }
+        }
         lineswereskipped = true; // this means there was some junk before any
         // key file signature
       }
@@ -417,6 +485,11 @@ public class IdentifyFile
     return true;
   }
 
+  /**
+   * 
+   * @param args
+   * @j2sIgnore
+   */
   public static void main(String[] args)
   {
     for (int i = 0; args != null && i < args.length; i++)
@@ -439,4 +512,5 @@ public class IdentifyFile
       System.err.println("Usage: <Filename> [<Filename> ...]");
     }
   }
+
 }