X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fio%2FNewickFile.java;h=8a464a4c6143fb1341b7c2f7c7ed0331b3c5a165;hb=c0501eaa85c0594f9275766f64de8ea44a59c368;hp=027390a1e8fc6facd4f7ae1e5852df515acacf38;hpb=04c8f7bff663aa469127e9eed4164e02933782f1;p=jalview.git

diff --git a/src/jalview/io/NewickFile.java b/src/jalview/io/NewickFile.java
index 027390a..8a464a4 100755
--- a/src/jalview/io/NewickFile.java
+++ b/src/jalview/io/NewickFile.java
@@ -30,6 +30,7 @@ import java.util.Locale;
 
 import jalview.datamodel.SequenceNode;
 import jalview.util.MessageManager;
+import jalview.util.Platform;
 
 import java.io.BufferedReader;
 import java.io.File;
@@ -39,6 +40,7 @@ import java.util.StringTokenizer;
 
 import com.stevesoft.pat.Regex;
 
+// TODO This class does not conform to Java standards for field name capitalization.
 /**
  * Parse a new hanpshire style tree Caveats: NHX files are NOT supported and the
  * tree distances and topology are unreliable when they are parsed. TODO: on
@@ -78,7 +80,7 @@ import com.stevesoft.pat.Regex;
  */
 public class NewickFile extends FileParse
 {
-  SequenceNode root;
+  private SequenceNode root;
 
   private boolean HasBootstrap = false;
 
@@ -87,21 +89,90 @@ public class NewickFile extends FileParse
   private boolean RootHasDistance = false;
 
   // File IO Flags
-  boolean ReplaceUnderscores = false;
+  private boolean ReplaceUnderscores = false;
 
-  boolean printRootInfo = true;
+  private boolean printRootInfo = true;
 
-  private Regex[] NodeSafeName = new Regex[] {
-      new Regex().perlCode("m/[\\[,:'()]/"), // test for
-      // requiring
-      // quotes
-      new Regex().perlCode("s/'/''/"), // escaping quote
-      // characters
-      new Regex().perlCode("s/\\/w/_/") // unqoted whitespace
-      // transformation
-  };
+  private static final int REGEX_PERL_NODE_REQUIRE_QUOTE = 0;
 
-  char QuoteChar = '\'';
+  private static final int REGEX_PERL_NODE_ESCAPE_QUOTE = 1;
+
+  private static final int REGEX_PERL_NODE_UNQUOTED_WHITESPACE = 2;
+
+  private static final int REGEX_MAJOR_SYMS = 3;
+
+  private static final int REGEX_QNODE_NAME = 4;
+
+  private static final int REGEX_COMMENT = 5;
+
+  private static final int REGEX_UQNODE_NAME = 6;
+
+  private static final int REGEX_NBOOTSTRAP = 7;
+
+  private static final int REGEX_NDIST = 8;
+
+  private static final int REGEX_NO_LINES = 9;
+
+  private static final int REGEX_PERL_EXPAND_QUOTES = 10;
+
+  private static final int REGEX_MAX = 11;
+
+  private static final Regex[] REGEX = new Regex[REGEX_MAX];
+
+  private static Regex getRegex(int id)
+  {
+    if (REGEX[id] == null)
+    {
+      String code = null;
+      String code2 = null;
+      String codePerl = null;
+      switch (id)
+      {
+      case REGEX_PERL_NODE_REQUIRE_QUOTE:
+        codePerl = "m/[\\[,:'()]/";
+        break;
+      case REGEX_PERL_NODE_ESCAPE_QUOTE:
+        codePerl = "s/'/''/";
+        break;
+      case REGEX_PERL_NODE_UNQUOTED_WHITESPACE:
+        codePerl = "s/\\/w/_/";
+        break;
+      case REGEX_PERL_EXPAND_QUOTES:
+        codePerl = "s/''/'/";
+        break;
+      case REGEX_MAJOR_SYMS:
+        code = "[(\\['),;]";
+        break;
+      case REGEX_QNODE_NAME:
+        code = "'([^']|'')+'";
+        break;
+      case REGEX_COMMENT:
+        code = "]";
+        break;
+      case REGEX_UQNODE_NAME:
+        code = "\\b([^' :;\\](),]+)";
+        break;
+      case REGEX_NBOOTSTRAP:
+        code = "\\s*([0-9+]+)\\s*:";
+        break;
+      case REGEX_NDIST:
+        code = ":([-0-9Ee.+]+)";
+        break;
+      case REGEX_NO_LINES:
+        code = "\n+";
+        code2 = "";
+        break;
+      default:
+        return null;
+      }
+      return codePerl == null ? Platform.newRegex(code, code2)
+              : Platform.newRegexPerl(codePerl);
+    }
+    return REGEX[id];
+  }
+
+
+  private char quoteChar = '\'';
 
   /**
    * Creates a new NewickFile object.
@@ -259,6 +330,7 @@ public class NewickFile extends FileParse
    */
   public void parse() throws IOException
   {
+    Platform.ensureRegex();
     String nf;
 
     { // fill nf with complete tree file
@@ -296,7 +368,7 @@ public class NewickFile extends FileParse
     boolean ascending = false; // flag indicating that we are leaving the
     // current node
 
-    Regex majorsyms = new Regex("[(\\['),;]");
+    Regex majorsyms = getRegex(REGEX_MAJOR_SYMS); // "[(\\['),;]"
 
     int nextcp = 0;
     int ncp = cp;
@@ -355,7 +427,7 @@ public class NewickFile extends FileParse
       // Deal with quoted fields
       case '\'':
 
-        Regex qnodename = new Regex("'([^']|'')+'");
+        Regex qnodename = getRegex(REGEX_QNODE_NAME);// "'([^']|'')+'");
 
         if (qnodename.searchFrom(nf, fcp))
         {
@@ -363,7 +435,7 @@ public class NewickFile extends FileParse
           nodename = new String(
                   qnodename.stringMatched().substring(1, nl - 1));
           // unpack any escaped colons
-          Regex xpandquotes = Regex.perlCode("s/''/'/");
+          Regex xpandquotes = getRegex(REGEX_PERL_EXPAND_QUOTES);
           String widernodename = xpandquotes.replaceAll(nodename);
           nodename = widernodename;
           // jump to after end of quoted nodename
@@ -397,7 +469,7 @@ public class NewickFile extends FileParse
            * '"+nf.substring(cp,fcp)+"'"); }
            */
           // verify termination.
-          Regex comment = new Regex("]");
+          Regex comment = getRegex(REGEX_COMMENT); // "]"
           if (comment.searchFrom(nf, fcp))
           {
             // Skip the comment field
@@ -428,9 +500,9 @@ public class NewickFile extends FileParse
                   + fstring.substring(cend + 1);
 
         }
-        Regex uqnodename = new Regex("\\b([^' :;\\](),]+)");
-        Regex nbootstrap = new Regex("\\s*([0-9+]+)\\s*:");
-        Regex ndist = new Regex(":([-0-9Ee.+]+)");
+        Regex uqnodename = getRegex(REGEX_UQNODE_NAME);// "\\b([^' :;\\](),]+)"
+        Regex nbootstrap = getRegex(REGEX_NBOOTSTRAP);// "\\s*([0-9+]+)\\s*:");
+        Regex ndist = getRegex(REGEX_NDIST);// ":([-0-9Ee.+]+)");
 
         if (!parsednodename && uqnodename.search(fstring)
                 && ((uqnodename.matchedFrom(1) == 0) || (fstring
@@ -785,7 +857,7 @@ public class NewickFile extends FileParse
    */
   char getQuoteChar()
   {
-    return QuoteChar;
+    return quoteChar;
   }
 
   /**
@@ -798,8 +870,8 @@ public class NewickFile extends FileParse
    */
   char setQuoteChar(char c)
   {
-    char old = QuoteChar;
-    QuoteChar = c;
+    char old = quoteChar;
+    quoteChar = c;
 
     return old;
   }
@@ -814,13 +886,15 @@ public class NewickFile extends FileParse
    */
   private String nodeName(String name)
   {
-    if (NodeSafeName[0].search(name))
+    if (getRegex(REGEX_PERL_NODE_REQUIRE_QUOTE).search(name))
     {
-      return QuoteChar + NodeSafeName[1].replaceAll(name) + QuoteChar;
+      return quoteChar
+              + getRegex(REGEX_PERL_NODE_ESCAPE_QUOTE).replaceAll(name)
+              + quoteChar;
     }
     else
     {
-      return NodeSafeName[2].replaceAll(name);
+      return getRegex(REGEX_PERL_NODE_UNQUOTED_WHITESPACE).replaceAll(name);
     }
   }
 
@@ -967,7 +1041,7 @@ public class NewickFile extends FileParse
       trf.parse();
       System.out.println("Original file :\n");
 
-      Regex nonl = new Regex("\n+", "");
+      Regex nonl = getRegex(REGEX_NO_LINES);// "\n+", "");
       System.out.println(nonl.replaceAll(newickfile.toString()) + "\n");
 
       System.out.println("Parsed file.\n");