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                 if(globaldata->getFileRoot() != ""){
  31                         fileroot = globaldata->getFileRoot();
  32                 }
  33                 else{
  34                         fileroot = getRootName(globaldata->getDistFile());
  35                 }
  36
  37                 //open output list files
  38                 for (i=0; i<groupMap->getNumGroups(); i++) {//opens an output file for each group
  39                         openOutputFile(fileroot + groupMap->namesOfGroups[i] + ".list", *(filehandles[groupMap->namesOfGroups[i]]));
  40                 }
  41         }
  42         catch(exception& e) {
  43                 cout << "Standard Error: " << e.what() << " has occurred in the ParseListCommand class Function ParseListCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  44                 exit(1);
  45         }
  46         catch(...) {
  47                 cout << "An unknown error has occurred in the ParseListCommand class function ParseListCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  48                 exit(1);
  49         }
  50 }
  51 /***********************************************************************/
  52 void ParseListCommand::parse(int index) {
  53         try {
  54                 string prefix, suffix, groupsName;
  55                 suffix = list->get(index);
  56
  57                 while (suffix.find_first_of(',') != -1) {//while you still have sequences
  58                         prefix = suffix.substr(0,suffix.find_first_of(','));
  59                         if ((suffix.find_first_of(',')+1) <= suffix.length()) {  //checks to make sure you don't have comma at end of string
  60                                 suffix = suffix.substr(suffix.find_first_of(',')+1, suffix.length());
  61                         }
  62
  63                         groupsName = groupMap->getGroup(prefix);
  64                         if (groupsName != "not found") {
  65                                 listGroups[groupsName] = listGroups[groupsName] + "," + prefix; //adds prefix to the correct group.
  66                         }else {
  67                                 cerr << "Error: Sequence '" << prefix << "' was not found in the group file, please correct\n";
  68                         }
  69                 }
  70
  71                 //save last name after comma
  72                 groupsName = groupMap->getGroup(suffix);
  73                 if (groupsName != "not found") {
  74                         listGroups[groupsName] = listGroups[groupsName] + "," + suffix; //adds prefix to the correct group.
  75                 }else {
  76                         cerr << "Error: Sequence '" << suffix << "' was not found in the group file, please correct\n";
  77                 }
  78         }
  79         catch(exception& e) {
  80                 cout << "Standard Error: " << e.what() << " has occurred in the ParseListCommand class Function parse. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  81                 exit(1);
  82         }
  83         catch(...) {
  84                 cout << "An unknown error has occurred in the ParseListCommand class function parse. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  85                 exit(1);
  86         }
  87 }
  88
  89 //**********************************************************************************************************************
  90
  91 int ParseListCommand::execute(){
  92         try{
  93                         globaldata = GlobalData::getInstance();
  94
  95                         //read in listfile
  96                         read = new ReadPhilFile(globaldata->inputFileName);
  97                         read->read(&*globaldata);
  98                         input = globaldata->ginput;
  99                         //list = input->getListVector();
 100                         list = globaldata->glist;
 101
 102                         //read in group map info.
 103                         groupMap = new GroupMap(globaldata->getGroupFile());
 104                         groupMap->readMap();
 105
 106                         string seq, label;
 107                         int i;
 108                         //create new list vectors to fill with parsed data
 109                         for (i=0; i<groupMap->getNumGroups(); i++) {
 110                                 groupOfLists[groupMap->namesOfGroups[i]] = new ListVector();
 111                         }
 112
 113                         //parses and sets each groups listvector
 114                         while(list != NULL){
 115                                 label = list->getLabel();
 116                                 for(i=0; i<list->size(); i++) {
 117                                         parse(i); //parses data[i] list of sequence names
 118                                         for (it=listGroups.begin(); it != listGroups.end(); it++) {  //loop through map and set new list vectors
 119                                                 seq = it->second;
 120                                                 seq = seq.substr(1, seq.length()); //rips off extra comma
 121                                                 groupOfLists[it->first]->push_back(seq); //sets new listvector for each group
 122                                         }
 123                                         listGroups.clear();
 124                                 }
 125                                 //prints each new list file
 126                                 for (i=0; i<groupMap->getNumGroups(); i++) {
 127                                         groupOfLists[groupMap->namesOfGroups[i]]->setLabel(label);
 128                                         groupOfLists[groupMap->namesOfGroups[i]]->print(*(filehandles[groupMap->namesOfGroups[i]]));
 129                                         groupOfLists[groupMap->namesOfGroups[i]]->clear();
 130                                 }
 131                                 list = input->getListVector();
 132                         }
 133                         return 0;
 134         }
 135         catch(exception& e) {
 136                 cout << "Standard Error: " << e.what() << " has occurred in the ParseListCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 137                 exit(1);
 138         }
 139         catch(...) {
 140                 cout << "An unknown error has occurred in the ParseListCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 141                 exit(1);
 142         }
 143
 144 }
 145 //**********************************************************************************************************************
 146
 147 ParseListCommand::~ParseListCommand(){
 148         delete list;
 149         delete groupMap;
 150         delete input;
 151         delete read;
 152 }
 153 //**********************************************************************************************************************