X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fio%2FIdentifyFile.java;h=5a3d700f74c020e3514295215de1fb9d6e3d91ce;hb=caa5930cf71ef14a4af891f2c17aa4a6e0f72bcf;hp=0556e76a20248be0a0ad19fc9e92eaeb4a255689;hpb=35ce4609cc8a776b1d85a1f2c2c70aaa106c1a3b;p=jalview.git

diff --git a/src/jalview/io/IdentifyFile.java b/src/jalview/io/IdentifyFile.java
index 0556e76..5a3d700 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 +33,40 @@ 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 +92,7 @@ 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();
     }
@@ -72,10 +109,11 @@ public class IdentifyFile
     // preserves original behaviour prior to version 2.3
   }
 
-  public FileFormatI identify(AlignmentFileReaderI file, boolean closeSource)
-          throws IOException
+  public FileFormatI identify(AlignmentFileReaderI file,
+          boolean closeSource) throws IOException
   {
-    FileParse fp = new FileParse(file.getInFile(), file.getDataSourceType());
+    FileParse fp = new FileParse(file.getInFile(),
+            file.getDataSourceType());
     return identify(fp, closeSource);
   }
 
@@ -98,12 +136,15 @@ public class IdentifyFile
     boolean lineswereskipped = false;
     boolean isBinary = false; // true if length is non-zero and non-printable
     // characters are encountered
+
     try
     {
       if (!closeSource)
       {
         source.mark();
       }
+      boolean aaIndexHeaderRead = false;
+
       while ((data = source.nextLine()) != null)
       {
         bytesRead += data.length();
@@ -126,21 +167,48 @@ 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;
+        }
+        if (data.startsWith("D ") && aaIndexHeaderRead)
+        {
+          reply = FileFormat.ScoreMatrix;
+          break;
+        }
         if (data.startsWith("##GFF-VERSION"))
         {
           // GFF - possibly embedded in a Jalview features file!
@@ -257,11 +325,16 @@ 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;
@@ -274,11 +347,6 @@ public class IdentifyFile
           }
         }
 
-        if (data.indexOf("{\"") > -1)
-        {
-          reply = FileFormat.Json;
-          break;
-        }
         if ((data.length() < 1) || (data.indexOf("#") == 0))
         {
           lineswereskipped = true;
@@ -292,9 +360,8 @@ public class IdentifyFile
           break;
         }
 
-        if ((data.indexOf("//") == 0)
-                || ((data.indexOf("!!") > -1) && (data.indexOf("!!") < data
-                        .indexOf("_MULTIPLE_ALIGNMENT "))))
+        if ((data.indexOf("//") == 0) || ((data.indexOf("!!") > -1) && (data
+                .indexOf("!!") < data.indexOf("_MULTIPLE_ALIGNMENT "))))
         {
           reply = FileFormat.MSF;
 
@@ -344,8 +411,8 @@ public class IdentifyFile
     }
     if (trimmedLength == 0)
     {
-      System.err
-              .println("File Identification failed! - Empty file was read.");
+      System.err.println(
+              "File Identification failed! - Empty file was read.");
       throw new FileFormatException("EMPTY DATA FILE");
     }
     System.out.println("File format identified as " + reply.toString());
@@ -400,6 +467,11 @@ public class IdentifyFile
     return true;
   }
 
+  /**
+   * 
+   * @param args
+   * @j2sIgnore
+   */
   public static void main(String[] args)
   {
     for (int i = 0; args != null && i < args.length; i++)
@@ -411,9 +483,9 @@ public class IdentifyFile
         type = ider.identify(args[i], DataSourceType.FILE);
       } catch (FileFormatException e)
       {
-        System.err.println(String.format(
-                "Error '%s' identifying file type for %s", args[i],
-                e.getMessage()));
+        System.err.println(
+                String.format("Error '%s' identifying file type for %s",
+                        args[i], e.getMessage()));
       }
       System.out.println("Type of " + args[i] + " is " + type);
     }
@@ -422,4 +494,6 @@ public class IdentifyFile
       System.err.println("Usage: <Filename> [<Filename> ...]");
     }
   }
+
+ 
 }