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 = globaldata->gGroupmap;
  19
  20                 //fill filehandles with neccessary ofstreams
  21                 int i;
  22                 ofstream* temp;
  23                 for (i=0; i<groupMap->getNumGroups(); i++) {
  24                         temp = new ofstream;
  25                         filehandles[groupMap->namesOfGroups[i]] = temp;
  26                 }
  27
  28                 //set fileroot
  29                 fileroot = getRootName(globaldata->getListFile());
  30
  31                 //open output list files
  32                 for (i=0; i<groupMap->getNumGroups(); i++) {//opens an output file for each group
  33                         openOutputFile(fileroot + groupMap->namesOfGroups[i] + ".list", *(filehandles[groupMap->namesOfGroups[i]]));
  34                 }
  35         }
  36         catch(exception& e) {
  37                 cout << "Standard Error: " << e.what() << " has occurred in the ParseListCommand class Function ParseListCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  38                 exit(1);
  39         }
  40         catch(...) {
  41                 cout << "An unknown error has occurred in the ParseListCommand class function ParseListCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  42                 exit(1);
  43         }
  44 }
  45 /***********************************************************************/
  46 void ParseListCommand::parse(int index, SharedListVector* list) {
  47         try {
  48                 string prefix, suffix, groupsName;
  49                 suffix = list->get(index);
  50
  51                 while (suffix.find_first_of(',') != -1) {//while you still have sequences
  52                         prefix = suffix.substr(0,suffix.find_first_of(','));
  53                         if ((suffix.find_first_of(',')+1) <= suffix.length()) {  //checks to make sure you don't have comma at end of string
  54                                 suffix = suffix.substr(suffix.find_first_of(',')+1, suffix.length());
  55                         }
  56
  57                         groupsName = groupMap->getGroup(prefix);
  58                         if (groupsName != "not found") {
  59                                 listGroups[groupsName] = listGroups[groupsName] + "," + prefix; //adds prefix to the correct group.
  60                         }else {
  61                                 cerr << "Error: Sequence '" << prefix << "' was not found in the group file, please correct\n";
  62                         }
  63                 }
  64
  65                 //save last name after comma
  66                 groupsName = groupMap->getGroup(suffix);
  67                 if (groupsName != "not found") {
  68                         listGroups[groupsName] = listGroups[groupsName] + "," + suffix; //adds prefix to the correct group.
  69                 }else {
  70                         cerr << "Error: Sequence '" << suffix << "' was not found in the group file, please correct\n";
  71                 }
  72         }
  73         catch(exception& e) {
  74                 cout << "Standard Error: " << e.what() << " has occurred in the ParseListCommand class Function parse. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  75                 exit(1);
  76         }
  77         catch(...) {
  78                 cout << "An unknown error has occurred in the ParseListCommand class function parse. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  79                 exit(1);
  80         }
  81 }
  82
  83 //**********************************************************************************************************************
  84
  85 int ParseListCommand::execute(){
  86         try{
  87                         int count = 1;
  88
  89                         //read in listfile
  90                         read = new ReadOTUFile(globaldata->inputFileName);
  91                         read->read(&*globaldata);
  92                         input = globaldata->ginput;
  93                         list = globaldata->gSharedList;
  94                         SharedListVector* lastList = list;
  95
  96                         //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
  97                         set<string> processedLabels;
  98                         set<string> userLabels = globaldata->labels;
  99                         set<int> userLines = globaldata->lines;
 100
 101                         //create new list vectors to fill with parsed data
 102                         for (int i=0; i<groupMap->getNumGroups(); i++) {
 103                                 groupOfLists[groupMap->namesOfGroups[i]] = new SharedListVector();
 104                         }
 105
 106
 107                         //parses and sets each groups listvector
 108                         //as long as you are not at the end of the file or done wih the lines you want
 109                         while((list != NULL) && ((globaldata->allLines == 1) || (userLabels.size() != 0) || (userLines.size() != 0))) {
 110
 111                                 if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(list->getLabel()) == 1){
 112                                         cout << list->getLabel() << '\t' << count << endl;
 113                                         process(list);
 114
 115                                         processedLabels.insert(list->getLabel());
 116                                         userLabels.erase(list->getLabel());
 117                                         userLines.erase(count);
 118                                 }
 119
 120                                 if ((anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastList->getLabel()) != 1)) {
 121                                         cout << lastList->getLabel() << '\t' << count << endl;
 122                                         process(lastList);
 123
 124                                         processedLabels.insert(lastList->getLabel());
 125                                         userLabels.erase(lastList->getLabel());
 126                                 }
 127
 128                                 if (count != 1) { delete lastList; }
 129                                 lastList = list;
 130
 131                                 list = input->getSharedListVector();
 132                                 count++;
 133                         }
 134
 135                         //output error messages about any remaining user labels
 136                         set<string>::iterator it;
 137                         bool needToRun = false;
 138                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 139                                 cout << "Your file does not include the label "<< *it;
 140                                 if (processedLabels.count(lastList->getLabel()) != 1) {
 141                                         cout << ". I will use " << lastList->getLabel() << "." << endl;
 142                                         needToRun = true;
 143                                 }else {
 144                                         cout << ". Please refer to " << lastList->getLabel() << "." << endl;
 145                                 }
 146                         }
 147
 148                         //run last line if you need to
 149                         if (needToRun == true)  {
 150                                 cout << lastList->getLabel() << '\t' << count << endl;
 151                                 process(lastList);
 152                         }
 153                         delete lastList;
 154
 155                         //set groupmap for .shared commands
 156                         if (globaldata->gGroupmap != NULL) { delete globaldata->gGroupmap; }
 157                         globaldata->gGroupmap = groupMap;
 158
 159                         //close files
 160                         for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 161                                 ofstream* temp = it3->second;
 162                                 (*temp).close();
 163                                 delete it3->second;
 164                         }
 165
 166                         //delete list vectors to fill with parsed data
 167                         for (it2 = groupOfLists.begin(); it2 != groupOfLists.end(); it2++) {
 168                                 delete it2->second;
 169                         }
 170
 171                         return 0;
 172         }
 173         catch(exception& e) {
 174                 cout << "Standard Error: " << e.what() << " has occurred in the ParseListCommand class Function execute. 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 ParseListCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 179                 exit(1);
 180         }
 181
 182 }
 183 //**********************************************************************************************************************
 184
 185 ParseListCommand::~ParseListCommand(){
 186         delete list;
 187         delete input;
 188         delete read;
 189 }
 190 //**********************************************************************************************************************
 191 void ParseListCommand::process(SharedListVector* thisList) {
 192         try {
 193                         string seq;
 194
 195                         for(int i=0; i<thisList->size(); i++) {
 196                                 parse(i, thisList); //parses data[i] list of sequence names
 197                                 for (it=listGroups.begin(); it != listGroups.end(); it++) {  //loop through map and set new list vectors
 198                                         seq = it->second;
 199                                         seq = seq.substr(1, seq.length()); //rips off extra comma
 200                                         groupOfLists[it->first]->push_back(seq); //sets new listvector for each group
 201                                 }
 202                                 listGroups.clear();
 203                         }
 204                         //prints each new list file
 205                         for (int i=0; i<groupMap->getNumGroups(); i++) {
 206                                 groupOfLists[groupMap->namesOfGroups[i]]->setLabel(thisList->getLabel());
 207                                 groupOfLists[groupMap->namesOfGroups[i]]->print(*(filehandles[groupMap->namesOfGroups[i]]));
 208                                 groupOfLists[groupMap->namesOfGroups[i]]->clear();
 209                         }
 210
 211         }
 212         catch(exception& e) {
 213                 cout << "Standard Error: " << e.what() << " has occurred in the ParseListCommand class Function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 214                 exit(1);
 215         }
 216         catch(...) {
 217                 cout << "An unknown error has occurred in the ParseListCommand class function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 218                 exit(1);
 219         }
 220 }