getlistcountcommand.cpp

   1 /*
   2  *  getlistcountcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 10/12/09.
   6  *  Copyright 2009 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "getlistcountcommand.h"
  11
  12 //**********************************************************************************************************************
  13 GetListCountCommand::GetListCountCommand(string option)  {
  14         try {
  15                 globaldata = GlobalData::getInstance();
  16                 abort = false;
  17                 allLines = 1;
  18                 labels.clear();
  19
  20                 //allow user to run help
  21                 if(option == "help") { help(); abort = true; }
  22
  23                 else {
  24                         //valid paramters for this command
  25                         string AlignArray[] =  {"list","label","outputdir","inputdir"};
  26                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  27
  28                         OptionParser parser(option);
  29                         map<string, string> parameters = parser.getParameters();
  30
  31                         ValidParameters validParameter;
  32                         map<string, string>::iterator it;
  33
  34                         //check to make sure all parameters are valid for command
  35                         for (it = parameters.begin(); it != parameters.end(); it++) {
  36                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  37                         }
  38
  39                         string ranRead = globaldata->getListFile();
  40
  41                         //if the user changes the input directory command factory will send this info to us in the output parameter
  42                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  43                         if (inputDir == "not found"){   inputDir = "";          }
  44                         else {
  45                                 string path;
  46                                 it = parameters.find("list");
  47                                 //user has given a template file
  48                                 if(it != parameters.end()){
  49                                         path = hasPath(it->second);
  50                                         //if the user has not given a path then, add inputdir. else leave path alone.
  51                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
  52                                 }
  53                         }
  54
  55                         //if the user changes the output directory command factory will send this info to us in the output parameter
  56                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
  57
  58                         //check for required parameters
  59                         listfile = validParameter.validFile(parameters, "list", true);
  60                         if ((listfile == "not found") && (globaldata->getListFile() == ""))  { m->mothurOut("You must read a listfile before running the get.listcount command.");  m->mothurOutEndLine(); abort = true; }
  61                         else if ((listfile == "not found") && (globaldata->getListFile() != "")) { listfile = globaldata->getListFile(); }
  62                         else if (listfile == "not open") { abort = true; }
  63                         else { globaldata->setListFile(listfile); }
  64
  65                         //check for optional parameter and set defaults
  66                         // ...at some point should added some additional type checking...
  67
  68                         label = validParameter.validFile(parameters, "label", false);
  69                         if (label == "not found") { label = ""; }
  70                         else {
  71                                 if(label != "all") {  splitAtDash(label, labels);  allLines = 0;  }
  72                                 else { allLines = 1;  }
  73                         }
  74
  75                         //if the user has not specified any labels use the ones from read.otu
  76                         if ((label == "") && (ranRead != "")) {
  77                                 allLines = globaldata->allLines;
  78                                 labels = globaldata->labels;
  79                         }
  80                 }
  81         }
  82         catch(exception& e) {
  83                 m->errorOut(e, "GetListCountCommand", "GetListCountCommand");
  84                 exit(1);
  85         }
  86 }
  87 //**********************************************************************************************************************
  88
  89 void GetListCountCommand::help(){
  90         try {
  91                 m->mothurOut("The get.listcount command can only be executed after a successful read.otu command of a listfile or providing a list file using the list parameter.\n");
  92                 m->mothurOut("The get.listcount command parameters are list and label.  No parameters are required.\n");
  93                 m->mothurOut("The label parameter allows you to select what distance levels you would like a output files created for, and are separated by dashes.\n");
  94                 m->mothurOut("The get.listcount command should be in the following format: get.listcount(list=yourlistFile, label=yourLabels).\n");
  95                 m->mothurOut("Example get.listcount(list=amazon.fn.list, label=0.10).\n");
  96                 m->mothurOut("The default value for label is all lines in your inputfile.\n");
  97                 m->mothurOut("The get.listcount command outputs a .otu file for each distance you specify listing the bin number and the names of the sequences in that bin.\n");
  98                 m->mothurOut("Note: No spaces between parameter labels (i.e. list), '=' and parameters (i.e.yourListFile).\n\n");
  99         }
 100         catch(exception& e) {
 101                 m->errorOut(e, "GetListCountCommand", "help");
 102                 exit(1);
 103         }
 104 }
 105
 106 //**********************************************************************************************************************
 107
 108 GetListCountCommand::~GetListCountCommand(){}
 109
 110 //**********************************************************************************************************************
 111
 112 int GetListCountCommand::execute(){
 113         try {
 114                 if (abort == true) {    return 0;       }
 115
 116                 globaldata->setFormat("list");
 117
 118                 //read list file
 119                 read = new ReadOTUFile(listfile);
 120                 read->read(&*globaldata);
 121
 122                 input = globaldata->ginput;
 123                 list = globaldata->gListVector;
 124                 string lastLabel = list->getLabel();
 125
 126                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 127                 set<string> processedLabels;
 128                 set<string> userLabels = labels;
 129
 130                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 131
 132                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 133
 134                                 process(list);
 135
 136                                 processedLabels.insert(list->getLabel());
 137                                 userLabels.erase(list->getLabel());
 138                         }
 139
 140                         if ((anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 141                                 string saveLabel = list->getLabel();
 142
 143                                 delete list;
 144                                 list = input->getListVector(lastLabel);
 145
 146                                 process(list);
 147
 148                                 processedLabels.insert(list->getLabel());
 149                                 userLabels.erase(list->getLabel());
 150
 151                                 //restore real lastlabel to save below
 152                                 list->setLabel(saveLabel);
 153                         }
 154
 155                         lastLabel = list->getLabel();
 156
 157                         delete list;
 158                         list = input->getListVector();
 159                 }
 160
 161
 162                 //output error messages about any remaining user labels
 163                 set<string>::iterator it;
 164                 bool needToRun = false;
 165                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 166                         m->mothurOut("Your file does not include the label " + *it);
 167                         if (processedLabels.count(lastLabel) != 1) {
 168                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 169                                 needToRun = true;
 170                         }else {
 171                                 m->mothurOut(". Please refer to " + lastLabel + ".");  m->mothurOutEndLine();
 172                         }
 173                 }
 174
 175                 //run last label if you need to
 176                 if (needToRun == true)  {
 177                         if (list != NULL) {             delete list;    }
 178                         list = input->getListVector(lastLabel);
 179
 180                         process(list);
 181                         delete list;
 182                 }
 183
 184                 delete read;
 185                 globaldata->gListVector = NULL;
 186
 187                 m->mothurOutEndLine();
 188                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 189                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 190                 m->mothurOutEndLine();
 191
 192                 return 0;
 193         }
 194         catch(exception& e) {
 195                 m->errorOut(e, "GetListCountCommand", "execute");
 196                 exit(1);
 197         }
 198 }
 199
 200 //**********************************************************************************************************************
 201 //return 1 if error, 0 otherwise
 202 void GetListCountCommand::process(ListVector* list) {
 203         try {
 204                 string binnames, name, sequence;
 205                 if (outputDir == "") { outputDir += hasPath(listfile); }
 206                 string outputFileName = outputDir + getRootName(getSimpleName(listfile)) + list->getLabel() + ".otu";
 207                 openOutputFile(outputFileName, out);
 208                 outputNames.push_back(outputFileName);
 209
 210                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 211
 212                 //for each bin in the list vector
 213                 for (int i = 0; i < list->getNumBins(); i++) {
 214                         binnames = list->get(i);
 215                         out << i+1 << '\t' << binnames << endl;
 216                 }
 217
 218                 out.close();
 219         }
 220         catch(exception& e) {
 221                 m->errorOut(e, "GetListCountCommand", "process");
 222                 exit(1);
 223         }
 224 }
 225 //**********************************************************************************************************************
 226
 227