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                 groupMap = globaldata->gGroupmap;
  21
  22                 //fill filehandles with neccessary ofstreams
  23                 int i;
  24                 ofstream* temp;
  25                 SharedListVector* templist;
  26                 for (i=0; i<groupMap->getNumGroups(); i++) {
  27                         temp = new ofstream;
  28                         templist = new SharedListVector();
  29                         filehandles[groupMap->namesOfGroups[i]] = temp;
  30                         mapOfLists[groupMap->namesOfGroups[i]] = templist;
  31                 }
  32
  33                 //set fileroot
  34                 fileroot = getRootName(globaldata->getListFile());
  35
  36                 //clears file before we start to write to it below
  37                 for (int i=0; i<groupMap->getNumGroups(); i++) {
  38                         openOutputFile(fileroot + groupMap->namesOfGroups[i] + ".list", *(filehandles[groupMap->namesOfGroups[i]]));
  39                         (*(filehandles[groupMap->namesOfGroups[i]])).close();
  40                 }
  41
  42
  43         }
  44         catch(exception& e) {
  45                 cout << "Standard Error: " << e.what() << " has occurred in the ParseListCommand class Function ParseListCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  46                 exit(1);
  47         }
  48         catch(...) {
  49                 cout << "An unknown error has occurred in the ParseListCommand class function ParseListCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  50                 exit(1);
  51         }
  52 }
  53 /***********************************************************************/
  54 void ParseListCommand::parse(int index, SharedListVector* list) {
  55         try {
  56                 string member, bin, groupName;
  57                 bin = list->get(index);
  58
  59                 while (bin.find_first_of(',') != -1) {//while you still have sequences
  60                         member = bin.substr(0,bin.find_first_of(','));
  61                         if ((bin.find_first_of(',')+1) <= bin.length()) {  //checks to make sure you don't have comma at end of string
  62                                 bin = bin.substr(bin.find_first_of(',')+1, bin.length());
  63                         }
  64
  65                         groupName = groupMap->getGroup(member);
  66                         if (groupName != "not found") {
  67                                 listGroups[groupName] = listGroups[groupName] + "," + member; //adds prefix to the correct group.
  68                         }else {
  69                                 cerr << "Error: Sequence '" << member << "' was not found in the group file, please correct\n";
  70                         }
  71                 }
  72
  73                 //save last name after comma
  74                 groupName = groupMap->getGroup(bin);
  75                 if (groupName != "not found") {
  76                         listGroups[groupName] = listGroups[groupName] + "," + bin; //adds prefix to the correct group.
  77                 }else {
  78                         cerr << "Error: Sequence '" << bin << "' was not found in the group file, please correct\n";
  79                 }
  80         }
  81         catch(exception& e) {
  82                 cout << "Standard Error: " << e.what() << " has occurred in the ParseListCommand class Function parse. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  83                 exit(1);
  84         }
  85         catch(...) {
  86                 cout << "An unknown error has occurred in the ParseListCommand class function parse. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  87                 exit(1);
  88         }
  89 }
  90
  91 //**********************************************************************************************************************
  92
  93 int ParseListCommand::execute(){
  94         try{
  95
  96                         int count = 1;
  97
  98                         //read in listfile
  99                         read = new ReadOTUFile(globaldata->inputFileName);
 100                         read->read(&*globaldata);
 101                         input = globaldata->ginput;
 102                         list = globaldata->gSharedList;
 103                         SharedListVector* lastList = list;
 104
 105                         //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 106                         set<string> processedLabels;
 107                         set<string> userLabels = globaldata->labels;
 108                         set<int> userLines = globaldata->lines;
 109
 110                         //parses and sets each groups listvector
 111                         //as long as you are not at the end of the file or done wih the lines you want
 112                         while((list != NULL) && ((globaldata->allLines == 1) || (userLabels.size() != 0) || (userLines.size() != 0))) {
 113
 114                                 if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(list->getLabel()) == 1){
 115                                         cout << list->getLabel() << '\t' << count << endl;
 116                                         process(list);
 117
 118                                         processedLabels.insert(list->getLabel());
 119                                         userLabels.erase(list->getLabel());
 120                                         userLines.erase(count);
 121                                 }
 122
 123                                 if ((anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastList->getLabel()) != 1)) {
 124                                         cout << lastList->getLabel() << '\t' << count << endl;
 125                                         process(lastList);
 126
 127                                         processedLabels.insert(lastList->getLabel());
 128                                         userLabels.erase(lastList->getLabel());
 129                                 }
 130
 131                                 if (count != 1) { delete lastList; }
 132                                 lastList = list;
 133
 134                                 list = input->getSharedListVector();
 135                                 count++;
 136                         }
 137
 138                         //output error messages about any remaining user labels
 139                         set<string>::iterator it;
 140                         bool needToRun = false;
 141                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 142                                 cout << "Your file does not include the label "<< *it;
 143                                 if (processedLabels.count(lastList->getLabel()) != 1) {
 144                                         cout << ". I will use " << lastList->getLabel() << "." << endl;
 145                                         needToRun = true;
 146                                 }else {
 147                                         cout << ". Please refer to " << lastList->getLabel() << "." << endl;
 148                                 }
 149                         }
 150
 151                         //run last line if you need to
 152                         if (needToRun == true)  {
 153                                 cout << lastList->getLabel() << '\t' << count << endl;
 154                                 process(lastList);
 155                         }
 156
 157                         delete lastList;  globaldata->gSharedList = NULL;
 158                         //delete list vectors to fill with parsed data
 159                         for (it2 = mapOfLists.begin(); it2 != mapOfLists.end(); it2++) {
 160                                 delete it2->second;
 161                         }
 162                         for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 163                                 delete it2->second;
 164                         }
 165
 166                         delete input;  globaldata->ginput = NULL;
 167                         delete read;
 168
 169
 170                         return 0;
 171         }
 172         catch(exception& e) {
 173                 cout << "Standard Error: " << e.what() << " has occurred in the ParseListCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 174                 exit(1);
 175         }
 176         catch(...) {
 177                 cout << "An unknown error has occurred in the ParseListCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 178                 exit(1);
 179         }
 180
 181 }
 182 //**********************************************************************************************************************
 183
 184 ParseListCommand::~ParseListCommand(){
 185
 186
 187 }
 188 //**********************************************************************************************************************
 189 void ParseListCommand::process(SharedListVector* thisList) {
 190         try {
 191                         string seq;
 192
 193                         for(int i=0; i<thisList->size(); i++) {
 194                                 parse(i, thisList); //parses data[i] list of sequence names
 195                                 for (it=listGroups.begin(); it != listGroups.end(); it++) {  //loop through map and set new list vectors
 196                                         seq = it->second;
 197                                         seq = seq.substr(1, seq.length()); //rips off extra comma
 198                                         mapOfLists[it->first]->push_back(seq); //sets new listvector for each group
 199                                 }
 200                                 listGroups.clear();
 201                         }
 202                         //prints each new list file
 203                         for (int i=0; i<groupMap->getNumGroups(); i++) {
 204                                 openOutputFileAppend(fileroot + groupMap->namesOfGroups[i] + ".list", *(filehandles[groupMap->namesOfGroups[i]]));
 205                                 mapOfLists[groupMap->namesOfGroups[i]]->setLabel(thisList->getLabel());
 206                                 mapOfLists[groupMap->namesOfGroups[i]]->print(*(filehandles[groupMap->namesOfGroups[i]]));
 207                                 mapOfLists[groupMap->namesOfGroups[i]]->clear();
 208                                 (*(filehandles[groupMap->namesOfGroups[i]])).close();
 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 }