parselistcommand.cpp

   1 /*
   2  *  parselistcommand.cpp
   3  *  Dotur
   4  *
   5  *  Created by Sarah Westcott on 1/2/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "parselistcommand.h"
  11
  12 //**********************************************************************************************************************
  13 ParseListCommand::ParseListCommand(){
  14         try {
  15                 globaldata = GlobalData::getInstance();
  16
  17                 //read in group map info.
  18                 groupMap = new GroupMap(globaldata->getGroupFile());
  19                 groupMap->readMap();
  20
  21                 //fill filehandles with neccessary ofstreams
  22                 int i;
  23                 ofstream* temp;
  24                 for (i=0; i<groupMap->getNumGroups(); i++) {
  25                         temp = new ofstream;
  26                         filehandles[groupMap->namesOfGroups[i]] = temp;
  27                 }
  28
  29                 //set fileroot
  30                 fileroot = getRootName(globaldata->getListFile());
  31
  32                 //open output list files
  33                 for (i=0; i<groupMap->getNumGroups(); i++) {//opens an output file for each group
  34                         openOutputFile(fileroot + groupMap->namesOfGroups[i] + ".list", *(filehandles[groupMap->namesOfGroups[i]]));
  35                 }
  36         }
  37         catch(exception& e) {
  38                 cout << "Standard Error: " << e.what() << " has occurred in the ParseListCommand class Function ParseListCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  39                 exit(1);
  40         }
  41         catch(...) {
  42                 cout << "An unknown error has occurred in the ParseListCommand class function ParseListCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  43                 exit(1);
  44         }
  45 }
  46 /***********************************************************************/
  47 void ParseListCommand::parse(int index) {
  48         try {
  49                 string prefix, suffix, groupsName;
  50                 suffix = list->get(index);
  51
  52                 while (suffix.find_first_of(',') != -1) {//while you still have sequences
  53                         prefix = suffix.substr(0,suffix.find_first_of(','));
  54                         if ((suffix.find_first_of(',')+1) <= suffix.length()) {  //checks to make sure you don't have comma at end of string
  55                                 suffix = suffix.substr(suffix.find_first_of(',')+1, suffix.length());
  56                         }
  57
  58                         groupsName = groupMap->getGroup(prefix);
  59                         if (groupsName != "not found") {
  60                                 listGroups[groupsName] = listGroups[groupsName] + "," + prefix; //adds prefix to the correct group.
  61                         }else {
  62                                 cerr << "Error: Sequence '" << prefix << "' was not found in the group file, please correct\n";
  63                         }
  64                 }
  65
  66                 //save last name after comma
  67                 groupsName = groupMap->getGroup(suffix);
  68                 if (groupsName != "not found") {
  69                         listGroups[groupsName] = listGroups[groupsName] + "," + suffix; //adds prefix to the correct group.
  70                 }else {
  71                         cerr << "Error: Sequence '" << suffix << "' was not found in the group file, please correct\n";
  72                 }
  73         }
  74         catch(exception& e) {
  75                 cout << "Standard Error: " << e.what() << " has occurred in the ParseListCommand class Function parse. 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 ParseListCommand class function parse. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  80                 exit(1);
  81         }
  82 }
  83
  84 //**********************************************************************************************************************
  85
  86 int ParseListCommand::execute(){
  87         try{
  88                         globaldata = GlobalData::getInstance();
  89                         int count = 1;
  90
  91                         //read in listfile
  92                         read = new ReadPhilFile(globaldata->inputFileName);
  93                         read->read(&*globaldata);
  94                         input = globaldata->ginput;
  95                         list = globaldata->gSharedList;
  96
  97                         //read in group map info.
  98                         groupMap = new GroupMap(globaldata->getGroupFile());
  99                         groupMap->readMap();
 100
 101                         string seq, label;
 102                         int i;
 103                         //create new list vectors to fill with parsed data
 104                         for (i=0; i<groupMap->getNumGroups(); i++) {
 105                                 groupOfLists[groupMap->namesOfGroups[i]] = new SharedListVector();
 106                         }
 107
 108                         //parses and sets each groups listvector
 109                         while(list != NULL){
 110                                 label = list->getLabel();
 111
 112                                 if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(label) == 1){
 113
 114                                         for(i=0; i<list->size(); i++) {
 115                                                 parse(i); //parses data[i] list of sequence names
 116                                                 for (it=listGroups.begin(); it != listGroups.end(); it++) {  //loop through map and set new list vectors
 117                                                         seq = it->second;
 118                                                         seq = seq.substr(1, seq.length()); //rips off extra comma
 119                                                         groupOfLists[it->first]->push_back(seq); //sets new listvector for each group
 120                                                 }
 121                                                 listGroups.clear();
 122                                         }
 123                                         //prints each new list file
 124                                         for (i=0; i<groupMap->getNumGroups(); i++) {
 125                                                 groupOfLists[groupMap->namesOfGroups[i]]->setLabel(label);
 126                                                 groupOfLists[groupMap->namesOfGroups[i]]->print(*(filehandles[groupMap->namesOfGroups[i]]));
 127                                                 groupOfLists[groupMap->namesOfGroups[i]]->clear();
 128                                         }
 129
 130                                         cout << label << '\t' << count << endl;
 131                                 }
 132
 133                                 list = input->getSharedListVector();
 134                                 count++;
 135                         }
 136
 137                         //set groupmap for .shared commands
 138                         //if (globaldata->gGroupmap != NULL) { delete globaldata->gGroupmap; }
 139                         globaldata->gGroupmap = groupMap;
 140
 141                         return 0;
 142         }
 143         catch(exception& e) {
 144                 cout << "Standard Error: " << e.what() << " has occurred in the ParseListCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 145                 exit(1);
 146         }
 147         catch(...) {
 148                 cout << "An unknown error has occurred in the ParseListCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 149                 exit(1);
 150         }
 151
 152 }
 153 //**********************************************************************************************************************
 154
 155 ParseListCommand::~ParseListCommand(){
 156         delete list;
 157         delete input;
 158         delete read;
 159 }
 160 //**********************************************************************************************************************