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, SharedListVector* list) {
  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                         int count = 1;
  89
  90                         //read in listfile
  91                         read = new ReadOTUFile(globaldata->inputFileName);
  92                         read->read(&*globaldata);
  93                         input = globaldata->ginput;
  94                         list = globaldata->gSharedList;
  95                         SharedListVector* lastList = list;
  96
  97                         //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
  98                         set<string> processedLabels;
  99                         set<string> userLabels = globaldata->labels;
 100                         set<int> userLines = globaldata->lines;
 101
 102                         //create new list vectors to fill with parsed data
 103                         for (int i=0; i<groupMap->getNumGroups(); i++) {
 104                                 groupOfLists[groupMap->namesOfGroups[i]] = new SharedListVector();
 105                         }
 106
 107
 108                         //parses and sets each groups listvector
 109                         //as long as you are not at the end of the file or done wih the lines you want
 110                         while((list != NULL) && ((globaldata->allLines == 1) || (userLabels.size() != 0) || (userLines.size() != 0))) {
 111
 112                                 if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(list->getLabel()) == 1){
 113                                         cout << list->getLabel() << '\t' << count << endl;
 114                                         process(list);
 115
 116                                         processedLabels.insert(list->getLabel());
 117                                         userLabels.erase(list->getLabel());
 118                                         userLines.erase(count);
 119                                 }
 120
 121                                 if ((anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastList->getLabel()) != 1)) {
 122                                         cout << lastList->getLabel() << '\t' << count << endl;
 123                                         process(lastList);
 124
 125                                         processedLabels.insert(lastList->getLabel());
 126                                         userLabels.erase(lastList->getLabel());
 127                                 }
 128
 129                                 if (count != 1) { delete lastList; }
 130                                 lastList = list;
 131
 132                                 list = input->getSharedListVector();
 133                                 count++;
 134                         }
 135
 136                         //output error messages about any remaining user labels
 137                         set<string>::iterator it;
 138                         bool needToRun = false;
 139                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 140                                 cout << "Your file does not include the label "<< *it;
 141                                 if (processedLabels.count(lastList->getLabel()) != 1) {
 142                                         cout << ". I will use " << lastList->getLabel() << "." << endl;
 143                                         needToRun = true;
 144                                 }else {
 145                                         cout << ". Please refer to " << lastList->getLabel() << "." << endl;
 146                                 }
 147                         }
 148
 149                         //run last line if you need to
 150                         if (needToRun == true)  {
 151                                 cout << lastList->getLabel() << '\t' << count << endl;
 152                                 process(lastList);
 153                         }
 154                         delete lastList;
 155
 156                         //set groupmap for .shared commands
 157                         if (globaldata->gGroupmap != NULL) { delete globaldata->gGroupmap; }
 158                         globaldata->gGroupmap = groupMap;
 159
 160                         //close files
 161                         for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 162                                 ofstream* temp = it3->second;
 163                                 (*temp).close();
 164                                 delete it3->second;
 165                         }
 166
 167                         //delete list vectors to fill with parsed data
 168                         for (it2 = groupOfLists.begin(); it2 != groupOfLists.end(); it2++) {
 169                                 delete it2->second;
 170                         }
 171
 172                         return 0;
 173         }
 174         catch(exception& e) {
 175                 cout << "Standard Error: " << e.what() << " has occurred in the ParseListCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 176                 exit(1);
 177         }
 178         catch(...) {
 179                 cout << "An unknown error has occurred in the ParseListCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 180                 exit(1);
 181         }
 182
 183 }
 184 //**********************************************************************************************************************
 185
 186 ParseListCommand::~ParseListCommand(){
 187
 188                 globaldata->gSharedList = NULL;
 189                 delete input;  globaldata->ginput = NULL;
 190                 delete read;
 191
 192 }
 193 //**********************************************************************************************************************
 194 void ParseListCommand::process(SharedListVector* thisList) {
 195         try {
 196                         string seq;
 197
 198                         for(int i=0; i<thisList->size(); i++) {
 199                                 parse(i, thisList); //parses data[i] list of sequence names
 200                                 for (it=listGroups.begin(); it != listGroups.end(); it++) {  //loop through map and set new list vectors
 201                                         seq = it->second;
 202                                         seq = seq.substr(1, seq.length()); //rips off extra comma
 203                                         groupOfLists[it->first]->push_back(seq); //sets new listvector for each group
 204                                 }
 205                                 listGroups.clear();
 206                         }
 207                         //prints each new list file
 208                         for (int i=0; i<groupMap->getNumGroups(); i++) {
 209                                 groupOfLists[groupMap->namesOfGroups[i]]->setLabel(thisList->getLabel());
 210                                 groupOfLists[groupMap->namesOfGroups[i]]->print(*(filehandles[groupMap->namesOfGroups[i]]));
 211                                 groupOfLists[groupMap->namesOfGroups[i]]->clear();
 212                         }
 213
 214         }
 215         catch(exception& e) {
 216                 cout << "Standard Error: " << e.what() << " has occurred in the ParseListCommand class Function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 217                 exit(1);
 218         }
 219         catch(...) {
 220                 cout << "An unknown error has occurred in the ParseListCommand class function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 221                 exit(1);
 222         }
 223 }