readtree.cpp

   1 /*
   2  *  readtree.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 1/22/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "readtree.h"
  11
  12 /***********************************************************************/
  13 //Parent Class
  14 // The following functions are used by all reading formats.
  15 /***********************************************************************/
  16 ReadTree::ReadTree() {
  17         try {
  18                 globaldata = GlobalData::getInstance();
  19                 T = new Tree();
  20                 numNodes = T->getNumNodes();
  21                 numLeaves = T->getNumLeaves();
  22         }
  23         catch(exception& e) {
  24                 cout << "Standard Error: " << e.what() << " has occurred in the ReadTree class Function ReadTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  25                 exit(1);
  26         }
  27         catch(...) {
  28                 cout << "An unknown error has occurred in the ReadTree class function ReadTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  29                 exit(1);
  30         }
  31 }
  32 /***********************************************************************/
  33 int ReadTree::readSpecialChar(istream& f, char c, string name) {
  34     try {
  35                 char d;
  36
  37                 while(isspace(d=f.get()))               {;}
  38                 if(d == EOF){
  39                         cerr << "Error: Input file ends prematurely, expecting a " << name << "\n";
  40                         exit(1);
  41                 }
  42                 if(d != c){
  43                         cerr << "Error: Expected " << name << " in input file.  Found " << d << ".\n";
  44                         exit(1);
  45                 }
  46                 if(d == ')' && f.peek() == '\n'){
  47                         while(isspace(d=f.get()))               {;}
  48                         f.putback(d);
  49                 }
  50
  51                 return d;
  52         }
  53         catch(exception& e) {
  54                 cout << "Standard Error: " << e.what() << " has occurred in the ReadTree class Function readSpecialChar. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  55                 exit(1);
  56         }
  57         catch(...) {
  58                 cout << "An unknown error has occurred in the ReadTree class function readSpecialChar. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  59                 exit(1);
  60         }
  61 }
  62 /**************************************************************************************************/
  63
  64 int ReadTree::readNodeChar(istream& f) {
  65         try {
  66                 char d;
  67                 while(isspace(d=f.get()))               {;}
  68                 if(d == EOF){
  69                         cerr << "Error: Input file ends prematurely, expecting a left parenthesis\n";
  70                         exit(1);
  71                 }
  72                 return d;
  73         }
  74         catch(exception& e) {
  75                 cout << "Standard Error: " << e.what() << " has occurred in the ReadTree class Function readNodeChar. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  76                 exit(1);
  77         }
  78         catch(...) {
  79                 cout << "An unknown error has occurred in the ReadTree class function readNodeChar. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  80                 exit(1);
  81         }
  82 }
  83
  84 /**************************************************************************************************/
  85
  86 float ReadTree::readBranchLength(istream& f) {
  87     try {
  88                 float b;
  89
  90                 if(!(f >> b)){
  91                         cerr << "Error: Missing branch length in input tree.\n";
  92                         exit(1);
  93                 }
  94
  95                 return b;
  96         }
  97         catch(exception& e) {
  98                 cout << "Standard Error: " << e.what() << " has occurred in the ReadTree class Function readBranchLength. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  99                 exit(1);
 100         }
 101         catch(...) {
 102                 cout << "An unknown error has occurred in the ReadTree class function readBranchLength. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 103                 exit(1);
 104         }
 105 }
 106
 107
 108 /***********************************************************************/
 109 /***********************************************************************/
 110
 111
 112 //Child Classes Below
 113
 114 /***********************************************************************/
 115 /***********************************************************************/
 116 //This class reads a file in Newick form and stores it in a tree.
 117
 118 void ReadNewickTree::read() {
 119         try {
 120                 int n = 0;
 121                 int lc, rc;
 122
 123                 int rooted = 0;
 124
 125                 int ch = filehandle.peek();
 126
 127                 if(ch == '('){
 128                         n = numLeaves;  //number of leaves / sequences, we want node 1 to start where the leaves left off
 129                         lc = readNewickInt(filehandle, n, T);
 130
 131                         if(filehandle.peek()==','){
 132                                 readSpecialChar(filehandle,',',"comma");
 133                         }
 134                         // ';' means end of tree.
 135                         else if((ch=filehandle.peek())==';' || ch=='['){
 136                                 rooted = 1;
 137                         }
 138                         if(rooted != 1){
 139                                 rc = readNewickInt(filehandle, n, T);
 140                                 if(filehandle.peek() == ')'){
 141                                         readSpecialChar(filehandle,')',"right parenthesis");
 142                                 }
 143                         }
 144                 }
 145                 //note: treeclimber had the code below added - not sure why?
 146                  else{
 147                         filehandle.putback(ch);
 148                         char name[MAX_LINE];
 149                         filehandle.get(name, MAX_LINE,'\n');
 150                         SKIPLINE(filehandle, ch);
 151
 152
 153                         n = T->getIndex(name);
 154                         if(n!=0){
 155                                 cerr << "Internal error: The only taxon is not taxon 0.\n";
 156                                 exit(1);
 157                         }
 158                         lc = rc = -1;
 159                 }
 160
 161                 while((ch=filehandle.get())!=';'){;}
 162                         if(rooted != 1){
 163                         T->tree[n].setChildren(lc,rc);
 164                         T->tree[n].setBranchLength(0);
 165                         T->tree[n].setParent(-1);
 166                         if(lc!=-1){             T->tree[lc].setParent(n);               }
 167                         if(rc!=-1){             T->tree[rc].setParent(n);               }
 168                 }
 169
 170                 //save tree for later commands
 171                 globaldata->gTree = T;
 172         }
 173         catch(exception& e) {
 174                 cout << "Standard Error: " << e.what() << " has occurred in the ReadNewickTree class Function read. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 175                 exit(1);
 176         }
 177         catch(...) {
 178                 cout << "An unknown error has occurred in the ReadNewickTree class function read. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 179                 exit(1);
 180         }
 181 }
 182 /**************************************************************************************************/
 183
 184 int ReadNewickTree::readNewickInt(istream& f, int& n, Tree* T) {
 185         try {
 186                 int c = readNodeChar(f);
 187
 188                 if(c == '('){
 189                         int lc = readNewickInt(f, n, T);
 190                         readSpecialChar(f,',',"comma");
 191
 192                         int rc = readNewickInt(f, n, T);
 193                         if(f.peek()==')'){
 194                                 readSpecialChar(f,')',"right parenthesis");
 195                         }
 196
 197                         if(f.peek() == ':'){
 198                                 readSpecialChar(f,':',"colon");
 199                                 if(n >= numNodes){      cerr << "Error: Too many nodes in input tree\n";  exit(1); }
 200                                 T->tree[n].setBranchLength(readBranchLength(f));
 201                         }else{T->tree[n].setBranchLength(0.0); }
 202
 203                         T->tree[n].setChildren(lc,rc);
 204                         T->tree[lc].setParent(n);
 205                         T->tree[rc].setParent(n);
 206
 207                         return n++;
 208                 }else{
 209                         f.putback(c);
 210                         string name = "";
 211                         char d=f.get();
 212                         while(d != ':' && d != ',' && d!=')' && d!='\n'){
 213                                 name += d;
 214                                 d=f.get();
 215                         }
 216
 217                         int blen = 0;
 218                         if(d == ':')    {               blen = 1;                       }
 219
 220                         f.putback(d);
 221
 222                         //set group info
 223                         string group = globaldata->gTreemap->getGroup(name);
 224
 225                         //find index in tree of name
 226                         int n1 = T->getIndex(name);
 227
 228                         if(n1 == -1){cerr << "Name: " << name << " not found\n"; exit(1);}
 229
 230                         else T->tree[n1].setGroup(group);
 231
 232                         T->tree[n1].setChildren(-1,-1);
 233
 234                         if(blen == 1){
 235                                 f.get();
 236                                 T->tree[n1].setBranchLength(readBranchLength(f));
 237                         }else{
 238                                 T->tree[n1].setBranchLength(0.0);
 239                         }
 240
 241                         while((c=f.get())!=0 && (c != ':' && c != ',' && c!=')') )              {;}
 242                         f.putback(c);
 243
 244                         return n1;
 245                 }
 246         }
 247         catch(exception& e) {
 248                 cout << "Standard Error: " << e.what() << " has occurred in the ReadNewickTree class Function readNewickInt. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 249                 exit(1);
 250         }
 251         catch(...) {
 252                 cout << "An unknown error has occurred in the ReadNewickTree class function readNewickInt. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 253                 exit(1);
 254         }
 255 }
 256 /**************************************************************************************************/
 257 /**************************************************************************************************/
 258