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
  22                 //fill filehandles with neccessary ofstreams
  23                 int i;
  24                 ofstream* temp;
  25                 for (i=0; i<groupMap->getNumGroups(); i++) {
  26                         temp = new ofstream;
  27                         filehandles[groupMap->namesOfGroups[i]] = temp;
  28                 }
  29
  30                 //set fileroot
  31                 fileroot = getRootName(globaldata->getListFile());
  32
  33                 //open output list files
  34                 for (i=0; i<groupMap->getNumGroups(); i++) {//opens an output file for each group
  35                         openOutputFile(fileroot + groupMap->namesOfGroups[i] + ".list", *(filehandles[groupMap->namesOfGroups[i]]));
  36                 }
  37         }
  38         catch(exception& e) {
  39                 cout << "Standard Error: " << e.what() << " has occurred in the ParseListCommand class Function ParseListCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  40                 exit(1);
  41         }
  42         catch(...) {
  43                 cout << "An unknown error has occurred in the ParseListCommand class function ParseListCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  44                 exit(1);
  45         }
  46 }
  47 /***********************************************************************/
  48 void ParseListCommand::parse(int index, SharedListVector* list) {
  49         try {
  50                 string prefix, suffix, groupsName;
  51                 suffix = list->get(index);
  52
  53                 while (suffix.find_first_of(',') != -1) {//while you still have sequences
  54                         prefix = suffix.substr(0,suffix.find_first_of(','));
  55                         if ((suffix.find_first_of(',')+1) <= suffix.length()) {  //checks to make sure you don't have comma at end of string
  56                                 suffix = suffix.substr(suffix.find_first_of(',')+1, suffix.length());
  57                         }
  58
  59                         groupsName = groupMap->getGroup(prefix);
  60                         if (groupsName != "not found") {
  61                                 listGroups[groupsName] = listGroups[groupsName] + "," + prefix; //adds prefix to the correct group.
  62                         }else {
  63                                 cerr << "Error: Sequence '" << prefix << "' was not found in the group file, please correct\n";
  64                         }
  65                 }
  66
  67                 //save last name after comma
  68                 groupsName = groupMap->getGroup(suffix);
  69                 if (groupsName != "not found") {
  70                         listGroups[groupsName] = listGroups[groupsName] + "," + suffix; //adds prefix to the correct group.
  71                 }else {
  72                         cerr << "Error: Sequence '" << suffix << "' was not found in the group file, please correct\n";
  73                 }
  74         }
  75         catch(exception& e) {
  76                 cout << "Standard Error: " << e.what() << " has occurred in the ParseListCommand class Function parse. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  77                 exit(1);
  78         }
  79         catch(...) {
  80                 cout << "An unknown error has occurred in the ParseListCommand class function parse. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  81                 exit(1);
  82         }
  83 }
  84
  85 //**********************************************************************************************************************
  86
  87 int ParseListCommand::execute(){
  88         try{
  89                         globaldata = GlobalData::getInstance();
  90                         int count = 1;
  91
  92                         //read in listfile
  93                         read = new ReadOTUFile(globaldata->inputFileName);
  94                         read->read(&*globaldata);
  95                         input = globaldata->ginput;
  96                         list = globaldata->gSharedList;
  97                         SharedListVector* lastList = list;
  98
  99                         //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 100                         set<string> processedLabels;
 101                         set<string> userLabels = globaldata->labels;
 102                         set<int> userLines = globaldata->lines;
 103
 104                         //read in group map info.
 105                         groupMap = new GroupMap(globaldata->getGroupFile());
 106                         groupMap->readMap();
 107
 108                         //create new list vectors to fill with parsed data
 109                         for (int i=0; i<groupMap->getNumGroups(); i++) {
 110                                 groupOfLists[groupMap->namesOfGroups[i]] = new SharedListVector();
 111                         }
 112
 113
 114                         //parses and sets each groups listvector
 115                         //as long as you are not at the end of the file or done wih the lines you want
 116                         while((list != NULL) && ((globaldata->allLines == 1) || (userLabels.size() != 0) || (userLines.size() != 0))) {
 117
 118                                 if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(list->getLabel()) == 1){
 119                                         cout << list->getLabel() << '\t' << count << endl;
 120                                         process(list);
 121
 122                                         processedLabels.insert(list->getLabel());
 123                                         userLabels.erase(list->getLabel());
 124                                         userLines.erase(count);
 125                                 }
 126
 127                                 if ((anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastList->getLabel()) != 1)) {
 128                                         cout << lastList->getLabel() << '\t' << count << endl;
 129                                         process(lastList);
 130
 131                                         processedLabels.insert(lastList->getLabel());
 132                                         userLabels.erase(lastList->getLabel());
 133                                 }
 134
 135                                 if (count != 1) { delete lastList; }
 136                                 lastList = list;
 137
 138                                 list = input->getSharedListVector();
 139                                 count++;
 140                         }
 141
 142                         //output error messages about any remaining user labels
 143                         set<string>::iterator it;
 144                         bool needToRun = false;
 145                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 146                                 cout << "Your file does not include the label "<< *it;
 147                                 if (processedLabels.count(lastList->getLabel()) != 1) {
 148                                         cout << ". I will use " << lastList->getLabel() << "." << endl;
 149                                         needToRun = true;
 150                                 }else {
 151                                         cout << ". Please refer to " << lastList->getLabel() << "." << endl;
 152                                 }
 153                         }
 154
 155                         //run last line if you need to
 156                         if (needToRun == true)  {
 157                                 cout << lastList->getLabel() << '\t' << count << endl;
 158                                 process(lastList);
 159                         }
 160                         delete lastList;
 161
 162                         //set groupmap for .shared commands
 163                         if (globaldata->gGroupmap != NULL) { delete globaldata->gGroupmap; }
 164                         globaldata->gGroupmap = groupMap;
 165
 166                         return 0;
 167         }
 168         catch(exception& e) {
 169                 cout << "Standard Error: " << e.what() << " has occurred in the ParseListCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 170                 exit(1);
 171         }
 172         catch(...) {
 173                 cout << "An unknown error has occurred in the ParseListCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 174                 exit(1);
 175         }
 176
 177 }
 178 //**********************************************************************************************************************
 179
 180 ParseListCommand::~ParseListCommand(){
 181         delete list;
 182         delete input;
 183         delete read;
 184 }
 185 //**********************************************************************************************************************
 186 void ParseListCommand::process(SharedListVector* thisList) {
 187         try {
 188                         string seq;
 189
 190                         for(int i=0; i<thisList->size(); i++) {
 191                                 parse(i, thisList); //parses data[i] list of sequence names
 192                                 for (it=listGroups.begin(); it != listGroups.end(); it++) {  //loop through map and set new list vectors
 193                                         seq = it->second;
 194                                         seq = seq.substr(1, seq.length()); //rips off extra comma
 195                                         groupOfLists[it->first]->push_back(seq); //sets new listvector for each group
 196                                 }
 197                                 listGroups.clear();
 198                         }
 199                         //prints each new list file
 200                         for (int i=0; i<groupMap->getNumGroups(); i++) {
 201                                 groupOfLists[groupMap->namesOfGroups[i]]->setLabel(thisList->getLabel());
 202                                 groupOfLists[groupMap->namesOfGroups[i]]->print(*(filehandles[groupMap->namesOfGroups[i]]));
 203                                 groupOfLists[groupMap->namesOfGroups[i]]->clear();
 204                         }
 205
 206         }
 207         catch(exception& e) {
 208                 cout << "Standard Error: " << e.what() << " has occurred in the ParseListCommand class Function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 209                 exit(1);
 210         }
 211         catch(...) {
 212                 cout << "An unknown error has occurred in the ParseListCommand class function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 213                 exit(1);
 214         }
 215 }