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                         remove((fileroot + groupMap->namesOfGroups[i] + ".list").c_str());
  39                 }
  40
  41         }
  42         catch(exception& e) {
  43                 errorOut(e, "ParseListCommand", "ParseListCommand");
  44                 exit(1);
  45         }
  46
  47 }
  48 /***********************************************************************/
  49 void ParseListCommand::parse(int index, SharedListVector* list) {
  50         try {
  51                 string member, bin, groupName;
  52                 bin = list->get(index);
  53
  54                 while (bin.find_first_of(',') != -1) {//while you still have sequences
  55                         member = bin.substr(0,bin.find_first_of(','));
  56                         if ((bin.find_first_of(',')+1) <= bin.length()) {  //checks to make sure you don't have comma at end of string
  57                                 bin = bin.substr(bin.find_first_of(',')+1, bin.length());
  58                         }
  59
  60                         groupName = groupMap->getGroup(member);
  61                         if (groupName != "not found") {
  62                                 listGroups[groupName] = listGroups[groupName] + "," + member; //adds prefix to the correct group.
  63                         }else {
  64                                 mothurOut("Error: Sequence '" + toString(member) + "' was not found in the group file, please correct\n");
  65                         }
  66                 }
  67
  68                 //save last name after comma
  69                 groupName = groupMap->getGroup(bin);
  70                 if (groupName != "not found") {
  71                         listGroups[groupName] = listGroups[groupName] + "," + bin; //adds prefix to the correct group.
  72                 }else {
  73                         mothurOut("Error: Sequence '" + toString(bin) + "' was not found in the group file, please correct\n");
  74                 }
  75         }
  76         catch(exception& e) {
  77                 errorOut(e, "ParseListCommand", "parse");
  78                 exit(1);
  79         }
  80 }
  81
  82 //**********************************************************************************************************************
  83
  84 int ParseListCommand::execute(){
  85         try{
  86
  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                         string lastLabel = list->getLabel();
  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                         //parses and sets each groups listvector
 102                         //as long as you are not at the end of the file or done wih the lines you want
 103                         while((list != NULL) && ((globaldata->allLines == 1) || (userLabels.size() != 0) || (userLines.size() != 0))) {
 104
 105                                 if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(lastLabel) == 1){
 106                                         mothurOut(list->getLabel() + "\t" + toString(count)); mothurOutEndLine();
 107                                         process(list);
 108
 109                                         processedLabels.insert(list->getLabel());
 110                                         userLabels.erase(list->getLabel());
 111                                         userLines.erase(count);
 112                                 }
 113
 114                                 if ((anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 115                                         delete list;
 116                                         list = input->getSharedListVector(lastLabel);
 117
 118                                         mothurOut(list->getLabel() + "\t" + toString(count)); mothurOutEndLine();
 119                                         process(list);
 120
 121                                         processedLabels.insert(list->getLabel());
 122                                         userLabels.erase(list->getLabel());
 123                                 }
 124
 125
 126                                 lastLabel = list->getLabel();
 127
 128                                 delete list;
 129                                 list = input->getSharedListVector();
 130                                 count++;
 131                         }
 132
 133                         //output error messages about any remaining user labels
 134                         set<string>::iterator it;
 135                         bool needToRun = false;
 136                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 137                                 mothurOut("Your file does not include the label " + *it);
 138                                 if (processedLabels.count(lastLabel) != 1) {
 139                                         mothurOut(". I will use " + lastLabel + "."); mothurOutEndLine();
 140                                         needToRun = true;
 141                                 }else {
 142                                         mothurOut(". Please refer to " + lastLabel + "."); mothurOutEndLine();
 143                                 }
 144                         }
 145
 146                         //run last line if you need to
 147                         if (needToRun == true)  {
 148                                 delete list;
 149                                 list = input->getSharedListVector(lastLabel);
 150
 151                                 mothurOut(list->getLabel() + "\t" + toString(count)); mothurOutEndLine();
 152                                 process(list);
 153                                 delete list;
 154                         }
 155
 156                         globaldata->gSharedList = NULL;
 157                         //delete list vectors to fill with parsed data
 158                         for (it2 = mapOfLists.begin(); it2 != mapOfLists.end(); it2++) {
 159                                 delete it2->second;
 160                         }
 161                         for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 162                                 delete it2->second;
 163                         }
 164
 165                         delete input;  globaldata->ginput = NULL;
 166                         delete read;
 167
 168
 169                         return 0;
 170         }
 171         catch(exception& e) {
 172                 errorOut(e, "ParseListCommand", "execute");
 173                 exit(1);
 174         }
 175 }
 176 //**********************************************************************************************************************
 177
 178 ParseListCommand::~ParseListCommand(){
 179
 180
 181 }
 182 //**********************************************************************************************************************
 183 void ParseListCommand::process(SharedListVector* thisList) {
 184         try {
 185                         string seq;
 186
 187                         for(int i=0; i<thisList->size(); i++) {
 188                                 parse(i, thisList); //parses data[i] list of sequence names
 189                                 for (it=listGroups.begin(); it != listGroups.end(); it++) {  //loop through map and set new list vectors
 190                                         seq = it->second;
 191                                         seq = seq.substr(1, seq.length()); //rips off extra comma
 192                                         mapOfLists[it->first]->push_back(seq); //sets new listvector for each group
 193                                 }
 194                                 listGroups.clear();
 195                         }
 196                         //prints each new list file
 197                         for (int i=0; i<groupMap->getNumGroups(); i++) {
 198                                 openOutputFileAppend(fileroot + groupMap->namesOfGroups[i] + ".list", *(filehandles[groupMap->namesOfGroups[i]]));
 199                                 mapOfLists[groupMap->namesOfGroups[i]]->setLabel(thisList->getLabel());
 200                                 mapOfLists[groupMap->namesOfGroups[i]]->print(*(filehandles[groupMap->namesOfGroups[i]]));
 201                                 mapOfLists[groupMap->namesOfGroups[i]]->clear();
 202                                 (*(filehandles[groupMap->namesOfGroups[i]])).close();
 203                         }
 204
 205         }
 206         catch(exception& e) {
 207                 errorOut(e, "ParseListCommand", "process");
 208                 exit(1);
 209         }
 210 }