src/jalview/io/NewickFile.java

   1 /*
   2  * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$)
   3  * Copyright (C) $$Year-Rel$$ The Jalview Authors
   4  *
   5  * This file is part of Jalview.
   6  *
   7  * Jalview is free software: you can redistribute it and/or
   8  * modify it under the terms of the GNU General Public License
   9  * as published by the Free Software Foundation, either version 3
  10  * of the License, or (at your option) any later version.
  11  *
  12  * Jalview is distributed in the hope that it will be useful, but
  13  * WITHOUT ANY WARRANTY; without even the implied warranty
  14  * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
  15  * PURPOSE.  See the GNU General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU General Public License
  18  * along with Jalview.  If not, see <http://www.gnu.org/licenses/>.
  19  * The Jalview Authors are detailed in the 'AUTHORS' file.
  20  */
  21 // NewickFile.java
  22 // Tree I/O
  23 // http://evolution.genetics.washington.edu/phylip/newick_doc.html
  24 // TODO: Implement Basic NHX tag parsing and preservation
  25 // TODO: http://evolution.genetics.wustl.edu/eddy/forester/NHX.html
  26 // TODO: Extended SequenceNodeI to hold parsed NHX strings
  27 package jalview.io;
  28
  29 import jalview.datamodel.SequenceNode;
  30 import jalview.jsdev.RegExp;
  31 import jalview.jsdev.api.RegExpInterface;
  32 import jalview.util.MessageManager;
  33
  34 import java.io.BufferedReader;
  35 import java.io.File;
  36 import java.io.FileReader;
  37 import java.io.IOException;
  38 import java.util.StringTokenizer;
  39
  40 import javajs.J2SIgnoreImport;
  41
  42 //import com.stevesoft.pat.Regex;
  43
  44 /**
  45  * Parse a new hanpshire style tree Caveats: NHX files are NOT supported and the
  46  * tree distances and topology are unreliable when they are parsed. TODO: on
  47  * this: NHX codes are appended in comments beginning with &&NHX. The codes are
  48  * given below (from http://www.phylosoft.org/forester/NHX.html): Element Type
  49  * Description Corresponding phyloXML element (parent element in parentheses) no
  50  * tag string name of this node/clade (MUST BE FIRST, IF ASSIGNED)
  51  * <name>(<clade>) : decimal branch length to parent node (MUST BE SECOND, IF
  52  * ASSIGNED) <branch_length>(<clade>) :GN= string gene name <name>(<sequence>)
  53  * :AC= string sequence accession <accession>(<sequence>) :ND= string node
  54  * identifier - if this is being used, it has to be unique within each phylogeny
  55  * <node_id>(<clade>) :B= decimal confidence value for parent branch
  56  * <confidence>(<clade>) :D= 'T', 'F', or '?' 'T' if this node represents a
  57  * duplication event - 'F' if this node represents a speciation event, '?' if
  58  * this node represents an unknown event (D= tag should be replaced by Ev= tag)
  59  * n/a :Ev=duplications>speciations>gene losses>event type>duplication type int
  60  * int int string string event (replaces the =D tag), number of duplication,
  61  * speciation, and gene loss events, type of event (transfer, fusion, root,
  62  * unknown, other, speciation_duplication_loss, unassigned) <events>(<clade>)
  63  * :E= string EC number at this node <annotation>(<sequence>) :Fu= string
  64  * function at this node <annotation>(<sequence>)
  65  * :DS=protein-length>from>to>support>name>from>... int int int double string
  66  * int ... domain structure at this node <domain_architecture>(<sequence>) :S=
  67  * string species name of the species/phylum at this node <taxonomy>(<clade>)
  68  * :T= integer taxonomy ID of the species/phylum at this node <id>(<taxonomy>)
  69  * :W= integer width of parent branch <width>(<clade>) :C=rrr.ggg.bbb
  70  * integer.integer.integer color of parent branch <color>(<clade>) :Co= 'Y' or
  71  * 'N' collapse this node when drawing the tree (default is not to collapse) n/a
  72  * :XB= string custom data associated with a branch <property>(<clade>) :XN=
  73  * string custom data associated with a node <property>(<clade>) :O= integer
  74  * orthologous to this external node n/a :SN= integer subtree neighbors n/a :SO=
  75  * integer super orthologous (no duplications on paths) to this external node
  76  * n/a
  77  *
  78  * @author Jim Procter
  79  * @version $Revision$
  80  */
  81 @J2SIgnoreImport({java.io.File.class, java.io.FileReader.class})
  82 public class NewickFile extends FileParse
  83 {
  84   SequenceNode root;
  85
  86   private boolean HasBootstrap = false;
  87
  88   private boolean HasDistances = false;
  89
  90   private boolean RootHasDistance = false;
  91
  92   // File IO Flags
  93   boolean ReplaceUnderscores = false;
  94
  95   boolean printRootInfo = true;
  96
  97   private RegExpInterface[] NodeSafeName = new RegExpInterface[]
  98   { RegExp.perlCode("m/[\\[,:'()]/"), // test for
  99       // requiring
 100       // quotes
 101       RegExp.perlCode("s/'/''/"), // escaping quote
 102       // characters
 103       RegExp.perlCode("s/\\/w/_/") // unqoted whitespace
 104   // transformation
 105   };
 106
 107   char QuoteChar = '\'';
 108
 109   /**
 110    * Creates a new NewickFile object.
 111    *
 112    * @param inStr
 113    *          DOCUMENT ME!
 114    *
 115    * @throws IOException
 116    *           DOCUMENT ME!
 117    */
 118   public NewickFile(String inStr) throws IOException
 119   {
 120     super(inStr, "Paste");
 121   }
 122
 123   /**
 124    * Creates a new NewickFile object.
 125    *
 126    * @param inFile
 127    *          DOCUMENT ME!
 128    * @param type
 129    *          DOCUMENT ME!
 130    *
 131    * @throws IOException
 132    *           DOCUMENT ME!
 133    */
 134   public NewickFile(String inFile, String type) throws IOException
 135   {
 136     super(inFile, type);
 137   }
 138
 139   public NewickFile(FileParse source) throws IOException
 140   {
 141     super(source);
 142   }
 143
 144   /**
 145    * Creates a new NewickFile object.
 146    *
 147    * @param newtree
 148    *          DOCUMENT ME!
 149    */
 150   public NewickFile(SequenceNode newtree)
 151   {
 152     root = newtree;
 153   }
 154
 155   /**
 156    * Creates a new NewickFile object.
 157    *
 158    * @param newtree
 159    *          DOCUMENT ME!
 160    * @param bootstrap
 161    *          DOCUMENT ME!
 162    */
 163   public NewickFile(SequenceNode newtree, boolean bootstrap)
 164   {
 165     HasBootstrap = bootstrap;
 166     root = newtree;
 167   }
 168
 169   /**
 170    * Creates a new NewickFile object.
 171    *
 172    * @param newtree
 173    *          DOCUMENT ME!
 174    * @param bootstrap
 175    *          DOCUMENT ME!
 176    * @param distances
 177    *          DOCUMENT ME!
 178    */
 179   public NewickFile(SequenceNode newtree, boolean bootstrap,
 180           boolean distances)
 181   {
 182     root = newtree;
 183     HasBootstrap = bootstrap;
 184     HasDistances = distances;
 185   }
 186
 187   /**
 188    * Creates a new NewickFile object.
 189    *
 190    * @param newtree
 191    *          DOCUMENT ME!
 192    * @param bootstrap
 193    *          DOCUMENT ME!
 194    * @param distances
 195    *          DOCUMENT ME!
 196    * @param rootdistance
 197    *          DOCUMENT ME!
 198    */
 199   public NewickFile(SequenceNode newtree, boolean bootstrap,
 200           boolean distances, boolean rootdistance)
 201   {
 202     root = newtree;
 203     HasBootstrap = bootstrap;
 204     HasDistances = distances;
 205     RootHasDistance = rootdistance;
 206   }
 207
 208   /**
 209    * DOCUMENT ME!
 210    *
 211    * @param Error
 212    *          DOCUMENT ME!
 213    * @param Er
 214    *          DOCUMENT ME!
 215    * @param r
 216    *          DOCUMENT ME!
 217    * @param p
 218    *          DOCUMENT ME!
 219    * @param s
 220    *          DOCUMENT ME!
 221    *
 222    * @return DOCUMENT ME!
 223    */
 224   private String ErrorStringrange(String Error, String Er, int r, int p,
 225           String s)
 226   {
 227     return ((Error == null) ? "" : Error)
 228             + Er
 229             + " at position "
 230             + p
 231             + " ( "
 232             + s.substring(((p - r) < 0) ? 0 : (p - r),
 233                     ((p + r) > s.length()) ? s.length() : (p + r)) + " )\n";
 234   }
 235
 236   // @tree annotations
 237   // These are set automatically by the reader
 238   public boolean HasBootstrap()
 239   {
 240     return HasBootstrap;
 241   }
 242
 243   /**
 244    * DOCUMENT ME!
 245    *
 246    * @return DOCUMENT ME!
 247    */
 248   public boolean HasDistances()
 249   {
 250     return HasDistances;
 251   }
 252
 253   public boolean HasRootDistance()
 254   {
 255     return RootHasDistance;
 256   }
 257
 258         /**
 259          * parse the filesource as a newick file (new hampshire and/or extended)
 260          *
 261          * @throws IOException
 262          *           with a line number and character position for badly formatted NH
 263          *           strings
 264          */
 265         public void parse() throws IOException {
 266                 String nf;
 267
 268                 { // fill nf with complete tree file
 269
 270                         StringBuffer file = new StringBuffer();
 271
 272                         while ((nf = nextLine()) != null) {
 273                                 file.append(nf);
 274                         }
 275
 276                         nf = file.toString();
 277                 }
 278
 279                 root = new SequenceNode();
 280
 281                 SequenceNode realroot = null;
 282                 SequenceNode c = root;
 283
 284                 int d = -1;
 285                 int cp = 0;
 286                 // int flen = nf.length();
 287
 288                 String Error = null;
 289                 String nodename = null;
 290                 String commentString2 = null; // comments after simple node props
 291
 292                 float DefDistance = (float) 0.001; // @param Default distance for a node -
 293                 // very very small
 294                 int DefBootstrap = -1; // @param Default bootstrap for a node
 295
 296                 float distance = DefDistance;
 297                 int bootstrap = DefBootstrap;
 298
 299                 boolean ascending = false; // flag indicating that we are leaving the
 300                 // current node
 301
 302                 RegExpInterface majorsyms = RegExp.newRegex("[(\\['),;]");
 303
 304                 int nextcp = 0;
 305                 int ncp = cp;
 306                 boolean parsednodename = false;
 307                 while (majorsyms.searchFrom(nf, cp) && (Error == null)) {
 308                         int fcp = majorsyms.matchedFrom();
 309                         char schar;
 310                         switch (schar = nf.charAt(fcp)) {
 311                         case '(':
 312
 313                                 // ascending should not be set
 314                                 // New Internal node
 315                                 if (ascending) {
 316                                         Error = ErrorStringrange(Error, "Unexpected '('", 7, fcp, nf);
 317
 318                                         continue;
 319                                 }
 320
 321                                 ;
 322                                 d++;
 323
 324                                 if (c.right() == null) {
 325                                         c.setRight(new SequenceNode(null, c, null, DefDistance, DefBootstrap,
 326                                                         false));
 327                                         c = (SequenceNode) c.right();
 328                                 } else {
 329                                         if (c.left() != null) {
 330                                                 // Dummy node for polytomy - keeps c.left free for new node
 331                                                 SequenceNode tmpn = new SequenceNode(null, c, null, 0, 0, true);
 332                                                 tmpn.SetChildren(c.left(), c.right());
 333                                                 c.setRight(tmpn);
 334                                         }
 335
 336                                         c.setLeft(new SequenceNode(null, c, null, DefDistance, DefBootstrap,
 337                                                         false));
 338                                         c = (SequenceNode) c.left();
 339                                 }
 340
 341                                 if (realroot == null) {
 342                                         realroot = c;
 343                                 }
 344
 345                                 nodename = null;
 346                                 distance = DefDistance;
 347                                 bootstrap = DefBootstrap;
 348                                 cp = fcp + 1;
 349
 350                                 break;
 351
 352                         // Deal with quoted fields
 353                         case '\'':
 354
 355                                 RegExpInterface qnodename = RegExp.newRegex("'([^']|'')+'");
 356
 357                                 if (qnodename.searchFrom(nf, fcp)) {
 358                                         int nl = qnodename.stringMatched().length();
 359                                         nodename = new String(qnodename.stringMatched().substring(1, nl - 1));
 360                                         // unpack any escaped colons
 361                                         RegExpInterface xpandquotes = RegExp.perlCode("s/''/'/");
 362                                         String widernodename = xpandquotes.replaceAll(nodename);
 363                                         nodename = widernodename;
 364                                         // jump to after end of quoted nodename
 365                                         nextcp = fcp + nl + 1;
 366                                         parsednodename = true;
 367                                 } else {
 368                                         Error = ErrorStringrange(Error, "Unterminated quotes for nodename",
 369                                                         7, fcp, nf);
 370                                 }
 371
 372                                 break;
 373
 374                         default:
 375                                 if (schar == ';') {
 376                                         if (d != -1) {
 377                                                 Error = ErrorStringrange(Error, "Wayward semicolon (depth=" + d
 378                                                                 + ")", 7, fcp, nf);
 379                                         }
 380                                         // cp advanced at the end of default
 381                                 }
 382                                 if (schar == '[') {
 383                                         // node string contains Comment or structured/extended NH format info
 384                                         /*
 385                                          * if ((fcp-cp>1 && nf.substring(cp,fcp).trim().length()>1)) { // will
 386                                          * process in remains System.err.println("skipped text:
 387                                          * '"+nf.substring(cp,fcp)+"'"); }
 388                                          */
 389                                         // verify termination.
 390                                         RegExpInterface comment = RegExp.newRegex("]");
 391                                         if (comment.searchFrom(nf, fcp)) {
 392                                                 // Skip the comment field
 393                                                 nextcp = comment.matchedFrom() + 1;
 394                                                 warningMessage = "Tree file contained comments which may confuse input algorithm.";
 395                                                 break;
 396
 397                                                 // cp advanced at the end of default to nextcp, ncp is unchanged so
 398                                                 // any node info can be read.
 399                                         } else {
 400                                                 Error = ErrorStringrange(Error, "Unterminated comment", 3, fcp, nf);
 401                                         }
 402
 403                                         ;
 404                                 }
 405                                 // Parse simpler field strings
 406                                 String fstring = nf.substring(ncp, fcp);
 407                                 // remove any comments before we parse the node info
 408                                 // TODO: test newick file with quoted square brackets in node name (is
 409                                 // this allowed?)
 410                                 while (fstring.indexOf(']') > -1) {
 411                                         int cstart = fstring.indexOf('[');
 412                                         int cend = fstring.indexOf(']');
 413                                         commentString2 = fstring.substring(cstart + 1, cend);
 414                                         fstring = fstring.substring(0, cstart) + fstring.substring(cend + 1);
 415
 416                                 }
 417                                 RegExpInterface uqnodename = RegExp.newRegex("\\b([^' :;\\](),]+)");
 418                                 RegExpInterface nbootstrap = RegExp.newRegex("\\s*([0-9+]+)\\s*:");
 419                                 RegExpInterface ndist = RegExp.newRegex(":([-0-9Ee.+]+)");
 420
 421                                 if (!parsednodename
 422                                                 && uqnodename.search(fstring)
 423                                                 && ((uqnodename.matchedFromI(1) == 0) || (fstring.charAt(uqnodename
 424                                                                 .matchedFromI(1) - 1) != ':'))) // JBPNote
 425                                 // HACK!
 426                                 {
 427                                         if (nodename == null) {
 428                                                 if (ReplaceUnderscores) {
 429                                                         nodename = uqnodename.stringMatchedI(1).replace('_', ' ');
 430                                                 } else {
 431                                                         nodename = uqnodename.stringMatchedI(1);
 432                                                 }
 433                                         } else {
 434                                                 Error = ErrorStringrange(Error,
 435                                                                 "File has broken algorithm - overwritten nodename", 10, fcp, nf);
 436                                         }
 437                                 }
 438                                 // get comment bootstraps
 439
 440                                 if (nbootstrap.search(fstring)) {
 441                                         if (nbootstrap.stringMatchedI(1).equals(uqnodename.stringMatchedI(1))) {
 442                                                 nodename = null; // no nodename here.
 443                                         }
 444                                         if (nodename == null
 445                                                         || nodename.length() == 0
 446                                                         || nbootstrap.matchedFromI(1) > (uqnodename.matchedFromI(1) + uqnodename
 447                                                                         .stringMatched().length())) {
 448                                                 try {
 449                                                         bootstrap = (new Integer(nbootstrap.stringMatchedI(1)))
 450                                                                         .intValue();
 451                                                         HasBootstrap = true;
 452                                                 } catch (Exception e) {
 453                                                         Error = ErrorStringrange(Error, "Can't parse bootstrap value", 4,
 454                                                                         ncp + nbootstrap.matchedFrom(), nf);
 455                                                 }
 456                                         }
 457                                 }
 458
 459                                 boolean nodehasdistance = false;
 460
 461                                 if (ndist.search(fstring)) {
 462                                         try {
 463                                                 distance = (new Float(ndist.stringMatchedI(1))).floatValue();
 464                                                 HasDistances = true;
 465                                                 nodehasdistance = true;
 466                                         } catch (Exception e) {
 467                                                 Error = ErrorStringrange(Error, "Can't parse node distance value",
 468                                                                 7, ncp + ndist.matchedFrom(), nf);
 469                                         }
 470                                 }
 471
 472                                 if (ascending) {
 473                                         // Write node info here
 474                                         c.setName(nodename);
 475                                         // Trees without distances still need a render distance
 476                                         c.dist = (HasDistances) ? distance : DefDistance;
 477                                         // be consistent for internal bootstrap defaults too
 478                                         c.setBootstrap((HasBootstrap) ? bootstrap : DefBootstrap);
 479                                         if (c == realroot) {
 480                                                 RootHasDistance = nodehasdistance; // JBPNote This is really
 481                                                 // UGLY!!! Ensure root node gets
 482                                                 // its given distance
 483                                         }
 484                                         parseNHXNodeProps(c, commentString2);
 485                                         commentString2 = null;
 486                                 } else {
 487                                         // Find a place to put the leaf
 488                                         SequenceNode newnode = new SequenceNode(null, c, nodename,
 489                                                         (HasDistances) ? distance : DefDistance,
 490                                                         (HasBootstrap) ? bootstrap : DefBootstrap, false);
 491                                         parseNHXNodeProps(c, commentString2);
 492                                         commentString2 = null;
 493
 494                                         if (c.right() == null) {
 495                                                 c.setRight(newnode);
 496                                         } else {
 497                                                 if (c.left() == null) {
 498                                                         c.setLeft(newnode);
 499                                                 } else {
 500                                                         // Insert a dummy node for polytomy
 501                                                         // dummy nodes have distances
 502                                                         SequenceNode newdummy = new SequenceNode(null, c, null,
 503                                                                         (HasDistances ? 0 : DefDistance), 0, true);
 504                                                         newdummy.SetChildren(c.left(), newnode);
 505                                                         c.setLeft(newdummy);
 506                                                 }
 507                                         }
 508                                 }
 509
 510                                 if (ascending) {
 511                                         // move back up the tree from preceding closure
 512                                         c = c.AscendTree();
 513
 514                                         if ((d > -1) && (c == null)) {
 515                                                 Error = ErrorStringrange(
 516                                                                 Error,
 517                                                                 "File broke algorithm: Lost place in tree (is there an extra ')' ?)",
 518                                                                 7, fcp, nf);
 519                                         }
 520                                 }
 521
 522                                 if (nf.charAt(fcp) == ')') {
 523                                         d--;
 524                                         ascending = true;
 525                                 } else {
 526                                         if (nf.charAt(fcp) == ',') {
 527                                                 if (ascending) {
 528                                                         ascending = false;
 529                                                 } else {
 530                                                         // Just advance focus, if we need to
 531                                                         if ((c.left() != null) && (!c.left().isLeaf())) {
 532                                                                 c = (SequenceNode) c.left();
 533                                                         }
 534                                                 }
 535                                         }
 536                                 }
 537
 538                                 // Reset new node properties to obvious fakes
 539                                 nodename = null;
 540                                 distance = DefDistance;
 541                                 bootstrap = DefBootstrap;
 542                                 commentString2 = null;
 543                                 parsednodename = false;
 544                         }
 545                         if (nextcp == 0) {
 546                                 ncp = cp = fcp + 1;
 547                         } else {
 548                                 cp = nextcp;
 549                                 nextcp = 0;
 550                         }
 551                 }
 552
 553                 if (Error != null) {
 554                         throw (new IOException(MessageManager.formatMessage("exception.newfile",
 555                                         new String[] { Error.toString() })));
 556                 }
 557                 if (root == null) {
 558                         throw (new IOException(MessageManager.formatMessage("exception.newfile",
 559                                         new String[] { MessageManager.getString("label.no_tree_read_in") })));
 560                 }
 561                 // THe next line is failing for topali trees - not sure why yet. if
 562                 // (root.right()!=null && root.isDummy())
 563                 root = (SequenceNode) root.right().detach(); // remove the imaginary root.
 564
 565                 if (!RootHasDistance) {
 566                         root.dist = (HasDistances) ? 0 : DefDistance;
 567                 }
 568         }
 569
 570   /**
 571    * parse NHX codes in comment strings and update NewickFile state flags for
 572    * distances and bootstraps, and add any additional properties onto the node.
 573    *
 574    * @param c
 575    * @param commentString
 576    * @param commentString2
 577    */
 578   private void parseNHXNodeProps(SequenceNode c, String commentString)
 579   {
 580     // TODO: store raw comment on the sequenceNode so it can be recovered when
 581     // tree is output
 582     if (commentString != null && commentString.startsWith("&&NHX"))
 583     {
 584       StringTokenizer st = new StringTokenizer(commentString.substring(5),
 585               ":");
 586       while (st.hasMoreTokens())
 587       {
 588         String tok = st.nextToken();
 589         int colpos = tok.indexOf("=");
 590
 591         if (colpos > -1)
 592         {
 593           String code = tok.substring(0, colpos);
 594           String value = tok.substring(colpos + 1);
 595           try
 596           {
 597             // parse out code/value pairs
 598             if (code.toLowerCase().equals("b"))
 599             {
 600               int v = -1;
 601               Float iv = new Float(value);
 602               v = iv.intValue(); // jalview only does integer bootstraps
 603               // currently
 604               c.setBootstrap(v);
 605               HasBootstrap = true;
 606             }
 607             // more codes here.
 608           } catch (Exception e)
 609           {
 610             System.err.println("Couldn't parse code '" + code + "' = '"
 611                     + value + "'");
 612             e.printStackTrace(System.err);
 613           }
 614         }
 615       }
 616     }
 617
 618   }
 619
 620   /**
 621    * DOCUMENT ME!
 622    *
 623    * @return DOCUMENT ME!
 624    */
 625   public SequenceNode getTree()
 626   {
 627     return root;
 628   }
 629
 630   /**
 631    * Generate a newick format tree according to internal flags for bootstraps,
 632    * distances and root distances.
 633    *
 634    * @return new hampshire tree in a single line
 635    */
 636   public String print()
 637   {
 638     synchronized (this)
 639     {
 640       StringBuffer tf = new StringBuffer();
 641       print(tf, root);
 642
 643       return (tf.append(";").toString());
 644     }
 645   }
 646
 647   /**
 648    *
 649    *
 650    * Generate a newick format tree according to internal flags for distances and
 651    * root distances and user specificied writing of bootstraps.
 652    *
 653    * @param withbootstraps
 654    *          controls if bootstrap values are explicitly written.
 655    *
 656    * @return new hampshire tree in a single line
 657    */
 658   public String print(boolean withbootstraps)
 659   {
 660     synchronized (this)
 661     {
 662       boolean boots = this.HasBootstrap;
 663       this.HasBootstrap = withbootstraps;
 664
 665       String rv = print();
 666       this.HasBootstrap = boots;
 667
 668       return rv;
 669     }
 670   }
 671
 672   /**
 673    *
 674    * Generate newick format tree according to internal flags for writing root
 675    * node distances.
 676    *
 677    * @param withbootstraps
 678    *          explicitly write bootstrap values
 679    * @param withdists
 680    *          explicitly write distances
 681    *
 682    * @return new hampshire tree in a single line
 683    */
 684   public String print(boolean withbootstraps, boolean withdists)
 685   {
 686     synchronized (this)
 687     {
 688       boolean dists = this.HasDistances;
 689       this.HasDistances = withdists;
 690
 691       String rv = print(withbootstraps);
 692       this.HasDistances = dists;
 693
 694       return rv;
 695     }
 696   }
 697
 698   /**
 699    * Generate newick format tree according to user specified flags
 700    *
 701    * @param withbootstraps
 702    *          explicitly write bootstrap values
 703    * @param withdists
 704    *          explicitly write distances
 705    * @param printRootInfo
 706    *          explicitly write root distance
 707    *
 708    * @return new hampshire tree in a single line
 709    */
 710   public String print(boolean withbootstraps, boolean withdists,
 711           boolean printRootInfo)
 712   {
 713     synchronized (this)
 714     {
 715       boolean rootinfo = printRootInfo;
 716       this.printRootInfo = printRootInfo;
 717
 718       String rv = print(withbootstraps, withdists);
 719       this.printRootInfo = rootinfo;
 720
 721       return rv;
 722     }
 723   }
 724
 725   /**
 726    * DOCUMENT ME!
 727    *
 728    * @return DOCUMENT ME!
 729    */
 730   char getQuoteChar()
 731   {
 732     return QuoteChar;
 733   }
 734
 735   /**
 736    * DOCUMENT ME!
 737    *
 738    * @param c
 739    *          DOCUMENT ME!
 740    *
 741    * @return DOCUMENT ME!
 742    */
 743   char setQuoteChar(char c)
 744   {
 745     char old = QuoteChar;
 746     QuoteChar = c;
 747
 748     return old;
 749   }
 750
 751   /**
 752    * DOCUMENT ME!
 753    *
 754    * @param name
 755    *          DOCUMENT ME!
 756    *
 757    * @return DOCUMENT ME!
 758    */
 759   private String nodeName(String name)
 760   {
 761     if (NodeSafeName[0].search(name))
 762     {
 763       return QuoteChar + NodeSafeName[1].replaceAll(name) + QuoteChar;
 764     }
 765     else
 766     {
 767       return NodeSafeName[2].replaceAll(name);
 768     }
 769   }
 770
 771   /**
 772    * DOCUMENT ME!
 773    *
 774    * @param c
 775    *          DOCUMENT ME!
 776    *
 777    * @return DOCUMENT ME!
 778    */
 779   private String printNodeField(SequenceNode c)
 780   {
 781     return ((c.getName() == null) ? "" : nodeName(c.getName()))
 782             + ((HasBootstrap) ? ((c.getBootstrap() > -1) ? ((c.getName() != null ? " "
 783                     : "") + c.getBootstrap())
 784                     : "")
 785                     : "") + ((HasDistances) ? (":" + c.dist) : "");
 786   }
 787
 788   /**
 789    * DOCUMENT ME!
 790    *
 791    * @param root
 792    *          DOCUMENT ME!
 793    *
 794    * @return DOCUMENT ME!
 795    */
 796   private String printRootField(SequenceNode root)
 797   {
 798     return (printRootInfo) ? (((root.getName() == null) ? ""
 799             : nodeName(root.getName()))
 800             + ((HasBootstrap) ? ((root.getBootstrap() > -1) ? ((root
 801                     .getName() != null ? " " : "") + +root.getBootstrap())
 802                     : "") : "") + ((RootHasDistance) ? (":" + root.dist)
 803             : "")) : "";
 804   }
 805
 806   // Non recursive call deals with root node properties
 807   public void print(StringBuffer tf, SequenceNode root)
 808   {
 809     if (root != null)
 810     {
 811       if (root.isLeaf() && printRootInfo)
 812       {
 813         tf.append(printRootField(root));
 814       }
 815       else
 816       {
 817         if (root.isDummy())
 818         {
 819           _print(tf, (SequenceNode) root.right());
 820           _print(tf, (SequenceNode) root.left());
 821         }
 822         else
 823         {
 824           tf.append("(");
 825           _print(tf, (SequenceNode) root.right());
 826
 827           if (root.left() != null)
 828           {
 829             tf.append(",");
 830           }
 831
 832           _print(tf, (SequenceNode) root.left());
 833           tf.append(")" + printRootField(root));
 834         }
 835       }
 836     }
 837   }
 838
 839   // Recursive call for non-root nodes
 840   public void _print(StringBuffer tf, SequenceNode c)
 841   {
 842     if (c != null)
 843     {
 844       if (c.isLeaf())
 845       {
 846         tf.append(printNodeField(c));
 847       }
 848       else
 849       {
 850         if (c.isDummy())
 851         {
 852           _print(tf, (SequenceNode) c.left());
 853           if (c.left() != null)
 854           {
 855             tf.append(",");
 856           }
 857           _print(tf, (SequenceNode) c.right());
 858         }
 859         else
 860         {
 861           tf.append("(");
 862           _print(tf, (SequenceNode) c.right());
 863
 864           if (c.left() != null)
 865           {
 866             tf.append(",");
 867           }
 868
 869           _print(tf, (SequenceNode) c.left());
 870           tf.append(")" + printNodeField(c));
 871         }
 872       }
 873     }
 874   }
 875
 876   // Test
 877   /**
 878    * @J2SIgnore
 879    * @param args
 880    */
 881   public static void main(String[] args)
 882   {
 883     try
 884     {
 885       if (args == null || args.length != 1)
 886       {
 887         System.err
 888                 .println("Takes one argument - file name of a newick tree file.");
 889         System.exit(0);
 890       }
 891
 892       File fn = new File(args[0]);
 893
 894       StringBuffer newickfile = new StringBuffer();
 895       BufferedReader treefile = new BufferedReader(new FileReader(fn));
 896       String l;
 897
 898       while ((l = treefile.readLine()) != null)
 899       {
 900         newickfile.append(l);
 901       }
 902
 903       treefile.close();
 904       System.out.println("Read file :\n");
 905
 906       NewickFile trf = new NewickFile(args[0], "File");
 907       trf.parse();
 908       System.out.println("Original file :\n");
 909
 910       RegExpInterface nonl = RegExp.newRegex("\n+", "");
 911       System.out.println(nonl.replaceAll(newickfile.toString()) + "\n");
 912
 913       System.out.println("Parsed file.\n");
 914       System.out.println("Default output type for original input.\n");
 915       System.out.println(trf.print());
 916       System.out.println("Without bootstraps.\n");
 917       System.out.println(trf.print(false));
 918       System.out.println("Without distances.\n");
 919       System.out.println(trf.print(true, false));
 920       System.out.println("Without bootstraps but with distanecs.\n");
 921       System.out.println(trf.print(false, true));
 922       System.out.println("Without bootstraps or distanecs.\n");
 923       System.out.println(trf.print(false, false));
 924       System.out.println("With bootstraps and with distances.\n");
 925       System.out.println(trf.print(true, true));
 926     } catch (java.io.IOException e)
 927     {
 928       System.err.println("Exception\n" + e);
 929       e.printStackTrace();
 930     }
 931   }
 932 }