parselistscommand.cpp

   1 /*
   2  *  parselistcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 2/24/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "parselistscommand.h"
  11
  12 //**********************************************************************************************************************
  13 ParseListCommand::ParseListCommand(string option)  {
  14         try {
  15                 abort = false;
  16                 allLines = 1;
  17
  18                 //allow user to run help
  19                 if(option == "help") { help(); abort = true; }
  20
  21                 else {
  22                         //valid paramters for this command
  23                         string Array[] =  {"list","group", "label", "outputdir","inputdir"};
  24                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  25
  26                         OptionParser parser(option);
  27                         map<string, string> parameters = parser.getParameters();
  28
  29                         ValidParameters validParameter;
  30                         map<string, string>::iterator it;
  31
  32                         //check to make sure all parameters are valid for command
  33                         for (it = parameters.begin(); it != parameters.end(); it++) {
  34                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  35                         }
  36
  37                         //if the user changes the input directory command factory will send this info to us in the output parameter
  38                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  39                         if (inputDir == "not found"){   inputDir = "";          }
  40                         else {
  41                                 string path;
  42                                 it = parameters.find("list");
  43                                 //user has given a template file
  44                                 if(it != parameters.end()){
  45                                         path = m->hasPath(it->second);
  46                                         //if the user has not given a path then, add inputdir. else leave path alone.
  47                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
  48                                 }
  49
  50                                 it = parameters.find("group");
  51                                 //user has given a template file
  52                                 if(it != parameters.end()){
  53                                         path = m->hasPath(it->second);
  54                                         //if the user has not given a path then, add inputdir. else leave path alone.
  55                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
  56                                 }
  57                         }
  58
  59
  60                         //if the user changes the output directory command factory will send this info to us in the output parameter
  61                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
  62
  63                         //check for required parameters
  64                         listfile = validParameter.validFile(parameters, "list", true);
  65                         if (listfile == "not open") { abort = true; }
  66                         else if (listfile == "not found") { listfile = ""; }
  67
  68                         groupfile = validParameter.validFile(parameters, "group", true);
  69                         if (groupfile == "not open") { abort = true; }
  70                         else if (groupfile == "not found") { groupfile = ""; }
  71                         else {
  72                                 groupMap = new GroupMap(groupfile);
  73
  74                                 int error = groupMap->readMap();
  75                                 if (error == 1) { abort = true; }
  76                         }
  77
  78                         //do you have all files needed
  79                         if ((listfile == "") || (groupfile == "")) { m->mothurOut("You must enter both a listfile and groupfile for the parse.list command. "); m->mothurOutEndLine(); abort = true;  }
  80
  81                         //check for optional parameter and set defaults
  82                         // ...at some point should added some additional type checking...
  83                         label = validParameter.validFile(parameters, "label", false);
  84                         if (label == "not found") { label = "";  allLines = 1; }
  85                         else {
  86                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
  87                                 else { allLines = 1;  }
  88                         }
  89                 }
  90
  91         }
  92         catch(exception& e) {
  93                 m->errorOut(e, "ParseListCommand", "ParseListCommand");
  94                 exit(1);
  95         }
  96 }
  97 //**********************************************************************************************************************
  98 void ParseListCommand::help(){
  99         try {
 100                 m->mothurOut("The parse.list command reads a list and group file and generates a list file for each group in the groupfile. \n");
 101                 m->mothurOut("The parse.list command parameters are list, group and label.\n");
 102                 m->mothurOut("The list and group parameters are required.\n");
 103                 m->mothurOut("The label parameter is used to read specific labels in your input you want to use.\n");
 104                 m->mothurOut("The parse.list command should be used in the following format: parse.list(list=yourListFile, group=yourGroupFile, label=yourLabels).\n");
 105                 m->mothurOut("Example: parse.list(list=abrecovery.fn.list, group=abrecovery.groups, label=0.03).\n");
 106                 m->mothurOut("Note: No spaces between parameter labels (i.e. list), '=' and parameters (i.e.yourListfile).\n\n");
 107
 108         }
 109         catch(exception& e) {
 110                 m->errorOut(e, "ParseListCommand", "help");
 111                 exit(1);
 112         }
 113 }
 114 //**********************************************************************************************************************
 115 ParseListCommand::~ParseListCommand(){}
 116 //**********************************************************************************************************************
 117 int ParseListCommand::execute(){
 118         try {
 119
 120                 if (abort == true) {    return 0;       }
 121
 122                 //set fileroot
 123                 string fileroot = outputDir + m->getRootName(m->getSimpleName(listfile));
 124                 vector<string> outputNames;
 125
 126                 //fill filehandles with neccessary ofstreams
 127                 int i;
 128                 ofstream* temp;
 129                 for (i=0; i<groupMap->namesOfGroups.size(); i++) {
 130                         temp = new ofstream;
 131                         filehandles[groupMap->namesOfGroups[i]] = temp;
 132
 133                         string filename = fileroot +  groupMap->namesOfGroups[i] + ".list";
 134                         outputNames.push_back(filename);
 135                         m->openOutputFile(filename, *temp);
 136                 }
 137
 138                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 139                 set<string> processedLabels;
 140                 set<string> userLabels = labels;
 141
 142                 input = new InputData(listfile, "list");
 143                 list = input->getListVector();
 144                 string lastLabel = list->getLabel();
 145
 146                 if (m->control_pressed) {
 147                         delete input; delete list; delete groupMap;
 148                         for (i=0; i<groupMap->namesOfGroups.size(); i++) {  (*(filehandles[groupMap->namesOfGroups[i]])).close();  delete filehandles[groupMap->namesOfGroups[i]]; }
 149                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 150                         return 0;
 151                 }
 152
 153                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 154
 155                         if (m->control_pressed) {
 156                                 delete input; delete list; delete groupMap;
 157                                 for (i=0; i<groupMap->namesOfGroups.size(); i++) {  (*(filehandles[groupMap->namesOfGroups[i]])).close();  delete filehandles[groupMap->namesOfGroups[i]]; }
 158                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 159                                 return 0;
 160                         }
 161
 162                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 163
 164                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 165                                         parse(list);
 166
 167                                         processedLabels.insert(list->getLabel());
 168                                         userLabels.erase(list->getLabel());
 169                         }
 170
 171                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 172                                         string saveLabel = list->getLabel();
 173
 174                                         delete list;
 175                                         list = input->getListVector(lastLabel); //get new list vector to process
 176
 177                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 178                                         parse(list);
 179
 180                                         processedLabels.insert(list->getLabel());
 181                                         userLabels.erase(list->getLabel());
 182
 183                                         //restore real lastlabel to save below
 184                                         list->setLabel(saveLabel);
 185                         }
 186
 187
 188                         lastLabel = list->getLabel();
 189
 190                         delete list;
 191                         list = input->getListVector(); //get new list vector to process
 192                 }
 193
 194                 if (m->control_pressed) {
 195                                 delete input; delete groupMap;
 196                                 for (i=0; i<groupMap->namesOfGroups.size(); i++) { (*(filehandles[groupMap->namesOfGroups[i]])).close();  delete filehandles[groupMap->namesOfGroups[i]]; }
 197                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 198                                 return 0;
 199                 }
 200
 201                 //output error messages about any remaining user labels
 202                 set<string>::iterator it;
 203                 bool needToRun = false;
 204                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 205                         m->mothurOut("Your file does not include the label " + *it);
 206                         if (processedLabels.count(lastLabel) != 1) {
 207                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 208                                 needToRun = true;
 209                         }else {
 210                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 211                         }
 212
 213                 }
 214
 215                 if (m->control_pressed) {
 216                                 delete input; delete groupMap;
 217                                 for (i=0; i<groupMap->namesOfGroups.size(); i++) {  (*(filehandles[groupMap->namesOfGroups[i]])).close();  delete filehandles[groupMap->namesOfGroups[i]]; }
 218                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 219                                 return 0;
 220                 }
 221
 222                 //run last label if you need to
 223                 if (needToRun == true)  {
 224                         if (list != NULL) {     delete list;    }
 225                         list = input->getListVector(lastLabel); //get new list vector to process
 226
 227                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 228                         parse(list);
 229
 230                         delete list;
 231                 }
 232
 233                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 234                         (*(filehandles[it3->first])).close();
 235                         delete it3->second;
 236                 }
 237
 238
 239                 delete groupMap;
 240                 delete input;
 241
 242                 if (m->control_pressed) {
 243                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 244                         return 0;
 245                 }
 246
 247                 m->mothurOutEndLine();
 248                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 249                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 250                 m->mothurOutEndLine();
 251
 252                 return 0;
 253         }
 254         catch(exception& e) {
 255                 m->errorOut(e, "ParseListCommand", "execute");
 256                 exit(1);
 257         }
 258 }
 259 /**********************************************************************************************************************/
 260 int ParseListCommand::parse(ListVector* thisList) {
 261         try {
 262
 263                 map<string, string> groupVector;
 264                 map<string, string>::iterator itGroup;
 265                 map<string, int> groupNumBins;
 266
 267                 //print label
 268                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 269                         groupNumBins[it3->first] = 0;
 270                         groupVector[it3->first] = "";
 271                 }
 272
 273
 274                 for (int i = 0; i < thisList->getNumBins(); i++) {
 275                         if (m->control_pressed) { return 0; }
 276
 277                         map<string, string> groupBins;
 278                         string bin = list->get(i);
 279
 280                         vector<string> names;
 281                         m->splitAtComma(bin, names);  //parses bin into individual sequence names
 282
 283                         //parse bin into list of sequences in each group
 284                         for (int j = 0; j < names.size(); j++) {
 285                                 string group = groupMap->getGroup(names[j]);
 286
 287                                 if (group == "not found") { m->mothurOut(names[j] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); exit(1); }
 288
 289                                 itGroup = groupBins.find(group);
 290                                 if(itGroup == groupBins.end()) {
 291                                         groupBins[group] = names[j];  //add first name
 292                                         groupNumBins[group]++;
 293                                 }else{ //add another name
 294                                         groupBins[group] = groupBins[group] + "," + names[j];
 295                                 }
 296                         }
 297
 298                         //print parsed bin info to files
 299                         for (itGroup = groupBins.begin(); itGroup != groupBins.end(); itGroup++) {
 300                                 groupVector[itGroup->first] +=  itGroup->second + '\t';
 301                         }
 302
 303                 }
 304
 305                 //end list vector
 306                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 307                         (*(filehandles[it3->first])) << thisList->getLabel() << '\t' << groupNumBins[it3->first] << '\t' << groupVector[it3->first] << endl;  // label numBins  listvector for that group
 308                 }
 309
 310                 return 0;
 311
 312         }
 313         catch(exception& e) {
 314                 m->errorOut(e, "ParseListCommand", "parse");
 315                 exit(1);
 316         }
 317 }
 318
 319 /**********************************************************************************************************************/
 320
 321