binaries/src/clustalw/src/fileInput/GDEFileParser.cpp

   1 /**
   2  * Author: Mark Larkin
   3  *
   4  * Copyright (c) 2007 Des Higgins, Julie Thompson and Toby Gibson.
   5  */
   6 /**
   7  * Changes:
   8  *
   9  * 10-02-07,Nigel Brown(EMBL): changed ifstream to InFileStream to handle
  10  * cross-platform end-of-lines.
  11  */
  12
  13 #ifdef HAVE_CONFIG_H
  14     #include "config.h"
  15 #endif
  16 #include "GDEFileParser.h"
  17
  18 namespace clustalw
  19 {
  20
  21 /*
  22  * Constructor sets up the chartab array.
  23  *
  24  */
  25 GDEFileParser::GDEFileParser(string filePath)
  26 {
  27     fileName = filePath;
  28     fillCharTab();
  29 }
  30
  31 /*
  32  * Nothing to do in destruction of object.
  33  */
  34 GDEFileParser::~GDEFileParser()
  35 {
  36
  37 }
  38
  39
  40     vector<Sequence> GDEFileParser::getSeqRange(int firstSeq, int no, string *offendingSeq)
  41 {
  42     vector<Sequence> seqRangeVector;
  43     int i;
  44
  45     for (i=0; i<no; i++)
  46     {
  47         Sequence tempSeq = getSeq(firstSeq + i, offendingSeq);
  48         if (parseExitCode!=OK) {
  49             seqRangeVector.clear();
  50             return seqRangeVector;
  51         }
  52         seqRangeVector.push_back(tempSeq);
  53     }
  54     return seqRangeVector;
  55 }
  56
  57
  58 /*
  59  * The getSeq function is used to get sequence number seqNum from the file.
  60  */
  61     Sequence GDEFileParser::getSeq(int seqNum, string *offendingSeq)
  62 {
  63     char _line[MAXLINE + 1];
  64     char _sname[MAXNAMES + 1];
  65     string characterSeq = "";
  66     string name = "";
  67     string title = "";
  68     string blank = "";
  69
  70     _line[0] = EOS;
  71
  72     int i;
  73     unsigned char c;
  74     int _currentSeqNum = 0; // Not at any sequence yet!
  75
  76     try
  77     {
  78         _fileIn = new InFileStream;  //nige
  79         _fileIn->open(fileName.c_str());  //nige
  80         _fileIn->seekg(0, std::ios::beg);
  81
  82         bool dnaFlagSet = userParameters->getDNAFlag();
  83         while (_currentSeqNum != seqNum)
  84         {
  85             while((*_line != '#' && dnaFlagSet) ||
  86                   (*_line != '%' && !dnaFlagSet))
  87             {
  88                 if(!_fileIn->getline(_line, MAXLINE + 1))
  89                 {
  90                     _fileIn->close();
  91                     return Sequence(blank, blank, blank);
  92                 }
  93             }
  94             ++_currentSeqNum;
  95             if(_currentSeqNum == seqNum) // Found the sequence
  96             {
  97                 break;
  98             }
  99             // Get next line so that we are past the '#' or '%' line
 100             _fileIn->getline(_line, MAXLINE + 1);  //nige
 101         }
 102
 103         for (i = 1; i <= MAXNAMES; i++)
 104         {
 105             if (_line[i] == '(' || _line[i] == '\n' || _line[i] == '\r')
 106             {
 107                 break;
 108             }
 109             _sname[i - 1] = _line[i];
 110         }
 111         i--;
 112         _sname[i] = EOS;
 113
 114         for (i--; i > 0; i--)
 115         {
 116             if (isspace(_sname[i]))
 117             {
 118                 _sname[i] = EOS;
 119             }
 120             else
 121             {
 122                 break;
 123             }
 124         }
 125         utilityObject->blankToUnderscore(_sname);
 126         name = string(_sname);
 127         title = "";
 128
 129         while (_fileIn->getline(_line, MAXLINE + 1))
 130         {
 131             if (*_line == '%' ||  *_line == '#' ||  *_line == '"')
 132             {
 133                break;
 134             }
 135             for (i = 0; i <= MAXLINE; i++)
 136             {
 137                 c = _line[i];
 138                 if (c == '\n' || c == EOS)
 139                 {
 140                     break;
 141                 }
 142
 143                 c = chartab[c];
 144                 if (c)
 145                 {
 146                     characterSeq += c;
 147                 }
 148             }
 149         }
 150
 151         _fileIn->close();
 152
 153         if ((int)characterSeq.length() > userParameters->getMaxAllowedSeqLength())
 154         {
 155             parseExitCode=SEQUENCETOOBIG;
 156             if (offendingSeq!=NULL)
 157                 offendingSeq->assign(name);
 158             // return empty seq
 159             return Sequence(blank, blank, blank);
 160         }
 161         return Sequence(characterSeq, name, title);
 162     }
 163     catch(...)
 164     {
 165         _fileIn->close();
 166         cerr << "There was an exception in the GDEnFileParser::getSeq function.\n"
 167              << "Need to end program\n";
 168         exit(1);
 169     }
 170
 171 }
 172
 173 /*
 174  * The countSeqs function returns the number of sequences in the file.
 175  */
 176 int GDEFileParser::countSeqs()
 177 {
 178     char line[MAXLINE + 1];
 179     int _nseqs = 0;
 180
 181     try
 182     {
 183         _fileIn = new InFileStream;  //nige
 184         _fileIn->open(fileName.c_str());  //nige
 185
 186         if(!_fileIn->is_open())
 187         {
 188             return 0; // No sequences found!
 189         }
 190
 191         while (_fileIn->getline(line, MAXLINE + 1))
 192         {
 193             if ((*line == '%') && (userParameters->getDNAFlag() == false))
 194             {
 195                 _nseqs++;
 196             }
 197             else if ((*line == '#') && (userParameters->getDNAFlag() == true))
 198             {
 199                 _nseqs++;
 200             }
 201         }
 202         _fileIn->close();
 203
 204         return _nseqs;
 205     }
 206     catch(...)
 207     {
 208         _fileIn->close();
 209         cerr << "An exception has occured in the function GDEFileParser::countSeqs()\n"
 210              << "Program needs to terminate.\nPlease contact the Clustal developers\n";
 211         exit(1);
 212     }
 213 }
 214
 215 /*
 216  * getSecStructure gets the secondary structure from the file.
 217  */
 218 void GDEFileParser::getSecStructure(vector<char>& gapPenaltyMask, vector<char>& secStructMask,
 219                                     string& secStructName, int &structPenalties, int length)
 220 {
 221     char _title[MAXLINE + 1];
 222     char _line[MAXLINE + 1];
 223     char _lin2[MAXLINE + 1];
 224     char _sname[MAXNAMES + 1];
 225     int i, len, offset = 0;
 226     unsigned char c;
 227
 228     try
 229     {
 230         _fileIn = new InFileStream;  //nige
 231         _fileIn->open(fileName.c_str());  //nige
 232         _fileIn->seekg(0, std::ios::beg);
 233
 234         // NOTE I think I should empty the masks before pushing onto them!
 235         gapPenaltyMask.clear();
 236         secStructMask.clear();
 237
 238         for (;;)
 239         {
 240             _line[0] = '\0';
 241             // search for the next comment line
 242             while (*_line != '"')
 243             {
 244                 if (!_fileIn->getline(_line, MAXLINE + 1))
 245                 {
 246                     _fileIn->close();
 247                     return;
 248                 }
 249             }
 250
 251             // is it a secondary structure entry?
 252             if (strncmp(&_line[1], "SS_", 3) == 0)
 253             {
 254                 for (i = 1; i <= MAXNAMES - 3; i++)
 255                 {
 256                     if (_line[i + 3] == '(' || _line[i + 3] == '\n' || _line[i + 3] == '\r')
 257                     {
 258                         break;
 259                     }
 260                     _sname[i - 1] = _line[i + 3];
 261                 }
 262                 i--;
 263                 _sname[i] = EOS;
 264
 265                 // NOTE NOTE NOTE
 266                 // Is it possible for this to be executed????????????????
 267                 // if _line contains ( then we break and dont put it into _sname
 268                 // So how can sname have it???????
 269                 if (_sname[i - 1] == '(')
 270                 {
 271                     sscanf(&_line[i + 3], "%d", &offset);
 272                 }
 273                 else
 274                 {
 275                     offset = 0;
 276                 }
 277                 for (i--; i > 0; i--)
 278                 {
 279                     if (isspace(_sname[i]))
 280                     {
 281                         _sname[i] = EOS;
 282                     }
 283                     else
 284                     {
 285                         break;
 286                     }
 287                 }
 288
 289                 utilityObject->blankToUnderscore(_sname);
 290                 secStructName = string(_sname);
 291
 292                 if (userParameters->getInteractive())
 293                 {
 294                     strcpy(_title, "Found secondary structure in alignment file: ");
 295                     strcat(_title, _sname);
 296                     (*_lin2) = utilityObject->promptForYesNo(_title,
 297                         "Use it to set local gap penalties ");
 298                 }
 299                 else
 300                 {
 301                     (*_lin2) = 'y';
 302                 }
 303                 if ((*_lin2 != 'n') && (*_lin2 != 'N'))
 304                 {
 305                     structPenalties = SECST;
 306                     for (i = 0; i < length; i++)
 307                     {
 308                         secStructMask.push_back('.');
 309                     }
 310                     len = 0;
 311                     while (_fileIn->getline(_line, MAXLINE + 1))
 312                     {
 313                         if (*_line == '%' ||  *_line == '#' ||  *_line == '"')
 314                         {
 315                             break;
 316                         }
 317                         for (i = offset; i < length; i++)
 318                         {
 319                             c = _line[i];
 320                             if (c == '\n' || c == EOS)
 321                             {
 322                                 break;
 323                             }
 324                             // EOL
 325                             secStructMask[len++] = c;
 326                         }
 327                         if (len >= length) // NOTE i put in >=
 328                         {
 329                             break;
 330                         }
 331                     }
 332                 }
 333             }
 334
 335             // or is it a gap penalty mask entry?
 336             else if (strncmp(&_line[1], "GM_", 3) == 0)
 337             {
 338                 for (i = 1; i <= MAXNAMES - 3; i++)
 339                 {
 340                     if (_line[i + 3] == '(' || _line[i + 3] == '\n')
 341                     {
 342                         break;
 343                     }
 344                     _sname[i - 1] = _line[i + 3];
 345                 }
 346                 i--;
 347                 _sname[i] = EOS;
 348
 349                 // NOTE NOTE
 350                 // Again I dont think it is possible for _sname to have ( !!!!
 351                 if (_sname[i - 1] == '(')
 352                 {
 353                     sscanf(&_line[i + 3], "%d", &offset);
 354                 }
 355                 else
 356                 {
 357                     offset = 0;
 358                 }
 359                 for (i--; i > 0; i--)
 360                 {
 361                     if (isspace(_sname[i]))
 362                     {
 363                         _sname[i] = EOS;
 364                     }
 365                     else
 366                     {
 367                         break;
 368                     }
 369                 }
 370
 371                 utilityObject->blankToUnderscore(_sname);
 372                 secStructName = string(_sname);
 373
 374                 if (userParameters->getInteractive())
 375                 {
 376                     strcpy(_title, "Found gap penalty mask in alignment file: ");
 377                     strcat(_title, _sname);
 378                     (*_lin2) = utilityObject->promptForYesNo(_title,
 379                         "Use it to set local gap penalties ");
 380                 }
 381                 else
 382                 {
 383                     (*_lin2) = 'y';
 384                 }
 385                 if ((*_lin2 != 'n') && (*_lin2 != 'N'))
 386                 {
 387                     structPenalties = GMASK;
 388                     for (i = 0; i < length; i++)
 389                     {
 390                         gapPenaltyMask.push_back('1');
 391                     }
 392                     len = 0;
 393                     while (_fileIn->getline(_line, MAXLINE + 1))
 394                     {
 395                         if (*_line == '%' ||  *_line == '#' ||  *_line == '"')
 396                         {
 397                             break;
 398                         }
 399                         for (i = offset; i < length; i++)
 400                         {
 401                             c = _line[i];
 402                             if (c == '\n' || c == EOS)
 403                             {
 404                                 break;
 405                             }
 406                             // EOL
 407                             gapPenaltyMask[len++] = c;
 408                         }
 409                         if (len >= length) // NOTE I put in >=
 410                         {
 411                             break;
 412                         }
 413                     }
 414                 }
 415             }
 416             if (structPenalties != NONE)
 417             {
 418                 break;
 419             }
 420         }
 421         _fileIn->close();
 422     }
 423     catch(...)
 424     {
 425         _fileIn->close();
 426         cerr << "An exception has occured in the function GDEFileParser::getSecStructure()\n"
 427              << "Program needs to terminate.\nPlease contact the Clustal developers\n";
 428         exit(1);
 429     }
 430 }
 431
 432 }
 433