bootstrap value parsing/recovery only for nodes with unquoted non-negative integer...

git-svn-id: https://svn.lifesci.dundee.ac.uk/svn/repository/trunk@459 be28352e-c001-0410-b1a7-c7978e42abec

diff --git a/src/uk/ac/vamsas/objects/utils/trees/BinaryNode.java b/src/uk/ac/vamsas/objects/utils/trees/BinaryNode.java
index ca0739a..bc41783 100644 (file)
--- a/src/uk/ac/vamsas/objects/utils/trees/BinaryNode.java
+++ b/src/uk/ac/vamsas/objects/utils/trees/BinaryNode.java
@@ -17,15 +17,15 @@ public class BinaryNode {
 \r
   BinaryNode parent;\r
 \r
-  /** DOCUMENT ME!! */\r
-  public int bootstrap;\r
+  /** bootstrap is non-negative integer */\r
+  public int bootstrap=-1;\r
 \r
   /**\r
    * Creates a new BinaryNode object.\r
    */\r
   public BinaryNode() {\r
     left = right = parent = null;\r
-    bootstrap = 0;\r
+    bootstrap = -1;\r
   }\r
 \r
   /**\r

diff --git a/src/uk/ac/vamsas/objects/utils/trees/NewickFile.java b/src/uk/ac/vamsas/objects/utils/trees/NewickFile.java
index 3c41744..521c616 100644 (file)
--- a/src/uk/ac/vamsas/objects/utils/trees/NewickFile.java
+++ b/src/uk/ac/vamsas/objects/utils/trees/NewickFile.java
@@ -61,7 +61,7 @@ public class NewickFile {
   private Pattern[] NodeSafeName = new Pattern[] {\r
       Pattern.compile("[\\[,:'()]"), // test for requiring quotes\r
       Pattern.compile("'"), // escaping quote characters\r
-      Pattern.compile("/w") // unqoted whitespace transformation\r
+      Pattern.compile("\\s") // unqoted whitespace transformation\r
   };\r
 \r
   char QuoteChar = '\'';\r
@@ -69,13 +69,13 @@ public class NewickFile {
   String newickFile = null;\r
 \r
   /**\r
-   * Creates a new NewickFile object.\r
+   * Creates a new NewickFile object\r
    * \r
    * @param inStr\r
-   *          DOCUMENT ME!\r
+   *          Newick style tree string\r
    * \r
    * @throws IOException\r
-   *           DOCUMENT ME!\r
+   *           if string is not a valid newick file\r
    */\r
   public NewickFile(String inStr) throws IOException {\r
     newickFile = inStr;\r
@@ -223,6 +223,8 @@ public class NewickFile {
 \r
   /**\r
    * call this to convert the newick string into a binary node linked tree\r
+   * Note: this is automatically called by the constructors, so you normally\r
+   * wouldn't need to use this.\r
    * \r
    * @throws IOException\r
    *           if the newick string cannot be parsed.\r
@@ -257,7 +259,7 @@ public class NewickFile {
 \r
     float DefDistance = (float) 0.001; // @param Default distance for a node -\r
     // very very small\r
-    int DefBootstrap = 0; // @param Default bootstrap for a node\r
+    int DefBootstrap = -1; // @param Default bootstrap for a node\r
 \r
     float distance = DefDistance;\r
     int bootstrap = DefBootstrap;\r
@@ -269,7 +271,9 @@ public class NewickFile {
 \r
     Matcher mjsyms = majorsyms.matcher(nf);\r
     char schar;\r
-    while (mjsyms.find(cp) && (Error == null)) {\r
+    int nextcp=0;\r
+    int ncp = cp;\r
+     while (mjsyms.find(cp) && (Error == null)) {\r
       int fcp = mjsyms.start();\r
 \r
       switch (schar = nf.charAt(fcp)) {\r
@@ -329,31 +333,36 @@ public class NewickFile {
         break;\r
 \r
       default:\r
-        int nextcp = 0;\r
-        // Skip Comment or structured/extended NH format info\r
+        // Reached termininating root node label.\r
+        if (schar == ';' && d != -1) {\r
+          Error = ErrorStringrange(Error,\r
+              "Wayward semicolon (depth=" + d + ")", 7, fcp, nf);\r
+        }\r
+\r
+      // Skip Comment or structured/extended NH format info\r
         if (schar == '[') {\r
           if ((nextcp=nf.indexOf(']', fcp)) > -1) {\r
-            // Skip the comment field\r
-            // should advance fcp too here\r
+            // verified that comment is properly terminated.\r
+            // now skip the comment field\r
             nextcp++;\r
-            //fcp = nextcp;\r
-            //schar = nf.charAt(fcp);\r
+            break; // go and search for the next node separator, leaving ncp at beginning of node info\r
           } else {\r
             Error = ErrorStringrange(Error, "Unterminated comment", 3, fcp, nf);\r
             nextcp = 0;\r
             break;\r
           }\r
-          ;\r
         }\r
-\r
-        // Reached termininating root node label.\r
-        if (schar == ';' && d != -1) {\r
-          Error = ErrorStringrange(Error,\r
-              "Wayward semicolon (depth=" + d + ")", 7, fcp, nf);\r
+        \r
+        // Parse simpler field strings from substring between ncp and node separator\r
+        String fstring = nf.substring(ncp, fcp);\r
+        // extract any comments from the nodeinfo.\r
+        while (fstring.indexOf(']')>-1)\r
+        {\r
+          int cstart=fstring.indexOf('[');\r
+          int cend=fstring.indexOf(']');\r
+          String comment =  fstring.substring(cstart+1,cend); // TODO: put this somewhere ?\r
+          fstring = fstring.substring(0, cstart)+fstring.substring(cend+1);\r
         }\r
-\r
-        // Parse simpler field strings\r
-        String fstring = nf.substring(cp, fcp);\r
         Matcher uqnodename = Pattern.compile("^([^' :;\\](),]+).*").matcher(\r
             fstring);\r
         if (uqnodename.matches()\r
@@ -372,20 +381,28 @@ public class NewickFile {
           }\r
         }\r
 \r
-        Matcher nbootstrap = Pattern.compile("\\S+([0-9+]+)\\S*:").matcher(\r
+        Matcher nbootstrap = Pattern.compile("\\s*([+0-9]+)\\s*:.*").matcher(\r
             fstring);\r
 \r
-        if (nbootstrap.matches() && (nbootstrap.start(1) > uqnodename.end(1))) {\r
-          try {\r
-            bootstrap = (new Integer(nbootstrap.group(1))).intValue();\r
-            HasBootstrap = true;\r
-          } catch (Exception e) {\r
-            Error = ErrorStringrange(Error, "Can't parse bootstrap value", 4,\r
-                cp + nbootstrap.start(0), nf);\r
+        if (nbootstrap.matches())\r
+        {\r
+          if (nodename!=null && nbootstrap.group(1).equals(nodename)) \r
+          {\r
+              nodename=null; // empty nodename - only bootstrap value\r
+          }\r
+          if ((nodename==null || nodename.length()==0) || nbootstrap.start(1)>=uqnodename.end(1))\r
+          {\r
+            try {\r
+              bootstrap = (new Integer(nbootstrap.group(1))).intValue();\r
+              HasBootstrap = true;\r
+            } catch (Exception e) {\r
+              Error = ErrorStringrange(Error, "Can't parse bootstrap value", 4,\r
+                  ncp + nbootstrap.start(0), nf);\r
+            }\r
           }\r
         }\r
-\r
-        Matcher ndist = Pattern.compile(":([-0-9Ee.+]+)").matcher(fstring);\r
+        \r
+        Matcher ndist = Pattern.compile(".*:([-0-9Ee.+]+)").matcher(fstring);\r
         boolean nodehasdistance = false;\r
 \r
         if (ndist.matches()) {\r
@@ -395,7 +412,7 @@ public class NewickFile {
             nodehasdistance = true;\r
           } catch (Exception e) {\r
             Error = ErrorStringrange(Error, "Can't parse node distance value",\r
-                7, cp + ndist.start(0), nf);\r
+                7, ncp + ndist.start(0), nf);\r
           }\r
         }\r
 \r
@@ -459,18 +476,19 @@ public class NewickFile {
               }\r
             }\r
           }\r
-\r
-          // else : We do nothing if ';' is encountered.\r
         }\r
 \r
         // Reset new node properties to obvious fakes\r
         nodename = null;\r
         distance = DefDistance;\r
         bootstrap = DefBootstrap;\r
-        if (nextcp == 0)\r
-          cp = fcp + 1;\r
-        else\r
-          cp = nextcp;\r
+      }\r
+      // Advance character pointers if necessary\r
+      if (nextcp == 0) {\r
+        ncp = cp = fcp + 1;\r
+      } else {\r
+        cp = nextcp;\r
+        nextcp = 0;\r
       }\r
     }\r
 \r
@@ -640,7 +658,7 @@ public class NewickFile {
   private String printNodeField(SequenceNode c) {\r
     return //c.getNewickNodeName()\r
     ((c.getName() == null) ? "" : nodeName(c.getName()))\r
-        + ((HasBootstrap) ? ((c.getBootstrap() > -1) ? (" " + c.getBootstrap())\r
+        + ((HasBootstrap) ? ((c.getBootstrap() > -1) ? (((c.getName()==null) ? " " : "") + c.getBootstrap())\r
             : "") : "") + ((HasDistances) ? (":" + c.dist) : "");\r
   }\r
 \r