binaries/src/clustalw/src/fileInput/MSFFileParser.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 "MSFFileParser.h"
  17
  18 namespace clustalw
  19 {
  20
  21 /**
  22  * MSFFileParser contructor sets up the chartab array.
  23  * @param filePath
  24  * @return
  25  */
  26 MSFFileParser::MSFFileParser(string filePath)
  27 {
  28     fileName = filePath;
  29     fillCharTab();
  30 }
  31
  32
  33
  34     vector<Sequence> MSFFileParser::getSeqRange(int firstSeq, int no, string *offendingSeq)
  35 {
  36     vector<Sequence> seqRangeVector;
  37     int i;
  38
  39     for (i=0; i<no; i++)
  40     {
  41         Sequence tempSeq = getSeq(firstSeq + i);
  42         if (parseExitCode!=OK) {
  43             seqRangeVector.clear();
  44             return seqRangeVector;
  45         }
  46         seqRangeVector.push_back(tempSeq);
  47     }
  48     return seqRangeVector;
  49 }
  50
  51
  52
  53 /**
  54  * The function getSeq finds the sequence seqNum in the file and returns it.
  55  * @param seqNum The number of the sequence in the file to get.
  56  * @return A sequence object containing the seqNum'th sequence from the file.
  57  */
  58 Sequence MSFFileParser::getSeq(int seqNum, string *offendingSeq)
  59 {
  60     char _line[MAXLINE + 1];
  61     char _sname[MAXNAMES + 1];
  62     string characterSeq = "";
  63     string name = "";
  64     string title = "";
  65     string blank = "";
  66
  67     _line[0] = EOS;
  68     int i, j, k;
  69     unsigned char c;
  70
  71     try
  72     {
  73         _fileIn = new InFileStream;  //nige
  74         _fileIn->open(fileName.c_str());  //nige
  75         _fileIn->seekg(0, std::ios::beg);
  76
  77         for (i = 0;; i++)
  78         {
  79             if (!_fileIn->getline(_line, MAXLINE + 1))
  80             {
  81                 _fileIn->close();
  82                 return Sequence(blank, blank, blank);
  83             }
  84             // read the title
  85             if (utilityObject->lineType(_line, "//"))
  86             {
  87                 break;
  88             }
  89             // lines...ignore
  90         }
  91
  92         while (_fileIn->getline(_line, MAXLINE + 1))
  93         {
  94             if (!utilityObject->blankLine(_line))
  95             {
  96                 for (i = 1; i < seqNum; i++)
  97                 {
  98                     _fileIn->getline(_line, MAXLINE + 1);
  99                 }
 100                 for (j = 0; j <= (int)strlen(_line); j++)
 101                 {
 102                     if (_line[j] != ' ')
 103                     {
 104                         break;
 105                     }
 106                 }
 107                 for (k = j; k <= (int)strlen(_line); k++)
 108                 {
 109                     if (_line[k] == ' ')
 110                     {
 111                         break;
 112                     }
 113                 }
 114
 115                 // Get the name of the sequence
 116                 strncpy(_sname, _line + j, utilityObject->MIN(MAXNAMES, k - j));
 117                 _sname[utilityObject->MIN(MAXNAMES, k - j)] = EOS;
 118                 utilityObject->rTrim(_sname);
 119                 utilityObject->blankToUnderscore(_sname);
 120                 name = string(_sname);
 121
 122                 for (i = k; i <= MAXLINE; i++)
 123                 {
 124                     c = _line[i];
 125                     if (c == '.' || c == '~')
 126                     {
 127                         c = '-';
 128                     }
 129                     if (c == '*')
 130                     {
 131                         c = 'X';
 132                     }
 133                     if (c == '\n' || c == EOS)
 134                     {
 135                         break;
 136                     }
 137                     // EOL
 138                     c = chartab[c];
 139                     if (c)
 140                     {
 141                         characterSeq += c;
 142                     }
 143                 }
 144
 145                 for (i = 0;; i++)
 146                 {
 147                     if (!_fileIn->getline(_line, MAXLINE + 1))
 148                     {
 149                         _fileIn->close();
 150                         return Sequence(characterSeq, name, title);
 151                     }
 152                     if (utilityObject->blankLine(_line))
 153                     {
 154                         break;
 155                     }
 156                 }
 157             }
 158         }
 159         _fileIn->close();
 160
 161         if ((int)characterSeq.length() > userParameters->getMaxAllowedSeqLength())
 162         {
 163             parseExitCode=SEQUENCETOOBIG;
 164             if (offendingSeq!=NULL)
 165                 offendingSeq->assign(name);
 166             // return empty seq
 167             return Sequence(blank, blank, blank);
 168         }
 169         return Sequence(characterSeq, name, title);;
 170     }
 171     catch(...)
 172     {
 173         _fileIn->close();
 174         cerr << "An exception has occured in the function MSFFileParser::getSeq()\n"
 175              << "Program needs to terminate.\nPlease contact the Clustal developers\n";
 176         exit(1);
 177     }
 178 }
 179
 180 /**
 181  * The function countSeqs counts the number of sequences in the file.
 182  * @return The number of sequences in the file.
 183  */
 184 int MSFFileParser::countSeqs()
 185 {
 186     char _line[MAXLINE + 1];
 187     int _numSeqs;
 188
 189     try
 190     {
 191         _fileIn = new InFileStream;  //nige
 192         _fileIn->open(fileName.c_str());  //nige
 193
 194         if(!_fileIn->is_open())
 195         {
 196             return 0; // No sequences found!
 197         }
 198
 199         while (_fileIn->getline(_line, MAXLINE + 1))
 200         {
 201             if (utilityObject->lineType(_line, "//"))
 202             {
 203                 break;
 204             }
 205         }
 206
 207         while (_fileIn->getline(_line, MAXLINE + 1))
 208         {
 209             if (!utilityObject->blankLine(_line))
 210             {
 211                 break;
 212             }
 213             // Look for next non- blank line
 214         }
 215         _numSeqs = 1;
 216
 217         while (_fileIn->getline(_line, MAXLINE + 1))
 218         {
 219             if (utilityObject->blankLine(_line))
 220             {
 221                 _fileIn->close();
 222                 return _numSeqs;
 223             }
 224             _numSeqs++;
 225         }
 226
 227         return 0; // if you got to here-funny format/no seqs.
 228     }
 229     catch(...)
 230     {
 231         _fileIn->close();
 232         cerr << "An exception has occured in the function MSFFileParser::countSeqs()\n"
 233              << "Program needs to terminate.\nPlease contact the Clustal developers\n";
 234         exit(1);
 235     }
 236 }
 237
 238 /**
 239  * There is no secondary structure information in MSF files. Set structPenalties to NONE.
 240  * @param gapPenaltyMask
 241  * @param secStructMask
 242  * @param secStructName
 243  * @param structPenalties
 244  * @param length
 245  */
 246 void MSFFileParser::getSecStructure(vector<char>& gapPenaltyMask, vector<char>& secStructMask,
 247                                     string& secStructName, int &structPenalties, int length)
 248 {
 249     structPenalties = NONE;
 250 }
 251
 252
 253 }
 254