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 vector<string> ParseListCommand::setParameters(){
  14         try {
  15                 CommandParameter plist("list", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(plist);
  16                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pgroup);
  17                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  18                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  19                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  20
  21                 vector<string> myArray;
  22                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  23                 return myArray;
  24         }
  25         catch(exception& e) {
  26                 m->errorOut(e, "ParseListCommand", "setParameters");
  27                 exit(1);
  28         }
  29 }
  30 //**********************************************************************************************************************
  31 string ParseListCommand::getHelpString(){
  32         try {
  33                 string helpString = "";
  34                 helpString += "The parse.list command reads a list and group file and generates a list file for each group in the groupfile. \n";
  35                 helpString += "The parse.list command parameters are list, group and label.\n";
  36                 helpString += "The list and group parameters are required.\n";
  37                 helpString += "The label parameter is used to read specific labels in your input you want to use.\n";
  38                 helpString += "The parse.list command should be used in the following format: parse.list(list=yourListFile, group=yourGroupFile, label=yourLabels).\n";
  39                 helpString += "Example: parse.list(list=abrecovery.fn.list, group=abrecovery.groups, label=0.03).\n";
  40                 helpString += "Note: No spaces between parameter labels (i.e. list), '=' and parameters (i.e.yourListfile).\n";
  41                 return helpString;
  42         }
  43         catch(exception& e) {
  44                 m->errorOut(e, "ParseListCommand", "getHelpString");
  45                 exit(1);
  46         }
  47 }
  48 //**********************************************************************************************************************
  49 ParseListCommand::ParseListCommand(){
  50         try {
  51                 abort = true; calledHelp = true;
  52                 setParameters();
  53                 vector<string> tempOutNames;
  54                 outputTypes["list"] = tempOutNames;
  55         }
  56         catch(exception& e) {
  57                 m->errorOut(e, "ParseListCommand", "ParseListCommand");
  58                 exit(1);
  59         }
  60 }
  61 //**********************************************************************************************************************
  62 ParseListCommand::ParseListCommand(string option)  {
  63         try {
  64                 abort = false; calledHelp = false;
  65                 allLines = 1;
  66
  67                 //allow user to run help
  68                 if(option == "help") { help(); abort = true; calledHelp = true; }
  69
  70                 else {
  71                         vector<string> myArray = setParameters();
  72
  73                         OptionParser parser(option);
  74                         map<string, string> parameters = parser.getParameters();
  75
  76                         ValidParameters validParameter;
  77                         map<string, string>::iterator it;
  78
  79                         //check to make sure all parameters are valid for command
  80                         for (it = parameters.begin(); it != parameters.end(); it++) {
  81                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  82                         }
  83
  84                         //initialize outputTypes
  85                         vector<string> tempOutNames;
  86                         outputTypes["list"] = tempOutNames;
  87
  88                         //if the user changes the input directory command factory will send this info to us in the output parameter
  89                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  90                         if (inputDir == "not found"){   inputDir = "";          }
  91                         else {
  92                                 string path;
  93                                 it = parameters.find("list");
  94                                 //user has given a template file
  95                                 if(it != parameters.end()){
  96                                         path = m->hasPath(it->second);
  97                                         //if the user has not given a path then, add inputdir. else leave path alone.
  98                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
  99                                 }
 100
 101                                 it = parameters.find("group");
 102                                 //user has given a template file
 103                                 if(it != parameters.end()){
 104                                         path = m->hasPath(it->second);
 105                                         //if the user has not given a path then, add inputdir. else leave path alone.
 106                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 107                                 }
 108                         }
 109
 110
 111                         //if the user changes the output directory command factory will send this info to us in the output parameter
 112                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 113
 114                         //check for required parameters
 115                         listfile = validParameter.validFile(parameters, "list", true);
 116                         if (listfile == "not open") { abort = true; }
 117                         else if (listfile == "not found") {
 118                                 listfile = m->getListFile();
 119                                 if (listfile != "") {  m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 120                                 else {
 121                                         m->mothurOut("No valid current list file. You must provide a list file."); m->mothurOutEndLine();
 122                                         abort = true;
 123
 124                                 }
 125                         }
 126
 127                         groupfile = validParameter.validFile(parameters, "group", true);
 128                         if (groupfile == "not open") { abort = true; }
 129                         else if (groupfile == "not found") {
 130                                 groupfile = m->getListFile();
 131                                 if (groupfile != "") {
 132                                         m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine();
 133                                         groupMap = new GroupMap(groupfile);
 134
 135                                         int error = groupMap->readMap();
 136                                         if (error == 1) { abort = true; }
 137                                 }else { m->mothurOut("No valid current group file. You must provide a group file."); m->mothurOutEndLine();  abort = true; }
 138                         }else {
 139                                 groupMap = new GroupMap(groupfile);
 140
 141                                 int error = groupMap->readMap();
 142                                 if (error == 1) { abort = true; }
 143                         }
 144
 145                         //do you have all files needed
 146                         if ((listfile == "") || (groupfile == "")) { m->mothurOut("You must enter both a listfile and groupfile for the parse.list command. "); m->mothurOutEndLine(); abort = true;  }
 147
 148                         //check for optional parameter and set defaults
 149                         // ...at some point should added some additional type checking...
 150                         label = validParameter.validFile(parameters, "label", false);
 151                         if (label == "not found") { label = "";  allLines = 1; }
 152                         else {
 153                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 154                                 else { allLines = 1;  }
 155                         }
 156                 }
 157
 158         }
 159         catch(exception& e) {
 160                 m->errorOut(e, "ParseListCommand", "ParseListCommand");
 161                 exit(1);
 162         }
 163 }
 164 //**********************************************************************************************************************
 165 int ParseListCommand::execute(){
 166         try {
 167
 168                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 169
 170                 //set fileroot
 171                 string fileroot = outputDir + m->getRootName(m->getSimpleName(listfile));
 172
 173                 //fill filehandles with neccessary ofstreams
 174                 int i;
 175                 ofstream* temp;
 176                 for (i=0; i<groupMap->namesOfGroups.size(); i++) {
 177                         temp = new ofstream;
 178                         filehandles[groupMap->namesOfGroups[i]] = temp;
 179
 180                         string filename = fileroot +  groupMap->namesOfGroups[i] + ".list";
 181                         outputNames.push_back(filename); outputTypes["list"].push_back(filename);
 182                         m->openOutputFile(filename, *temp);
 183                 }
 184
 185                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 186                 set<string> processedLabels;
 187                 set<string> userLabels = labels;
 188
 189                 input = new InputData(listfile, "list");
 190                 list = input->getListVector();
 191                 string lastLabel = list->getLabel();
 192
 193                 if (m->control_pressed) {
 194                         delete input; delete list; delete groupMap;
 195                         for (i=0; i<groupMap->namesOfGroups.size(); i++) {  (*(filehandles[groupMap->namesOfGroups[i]])).close();  delete filehandles[groupMap->namesOfGroups[i]]; }
 196                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); } outputTypes.clear();
 197                         return 0;
 198                 }
 199
 200                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 201
 202                         if (m->control_pressed) {
 203                                 delete input; delete list; delete groupMap;
 204                                 for (i=0; i<groupMap->namesOfGroups.size(); i++) {  (*(filehandles[groupMap->namesOfGroups[i]])).close();  delete filehandles[groupMap->namesOfGroups[i]]; }
 205                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); } outputTypes.clear();
 206                                 return 0;
 207                         }
 208
 209                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 210
 211                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 212                                         parse(list);
 213
 214                                         processedLabels.insert(list->getLabel());
 215                                         userLabels.erase(list->getLabel());
 216                         }
 217
 218                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 219                                         string saveLabel = list->getLabel();
 220
 221                                         delete list;
 222                                         list = input->getListVector(lastLabel); //get new list vector to process
 223
 224                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 225                                         parse(list);
 226
 227                                         processedLabels.insert(list->getLabel());
 228                                         userLabels.erase(list->getLabel());
 229
 230                                         //restore real lastlabel to save below
 231                                         list->setLabel(saveLabel);
 232                         }
 233
 234
 235                         lastLabel = list->getLabel();
 236
 237                         delete list;
 238                         list = input->getListVector(); //get new list vector to process
 239                 }
 240
 241                 if (m->control_pressed) {
 242                                 delete input; delete groupMap;
 243                                 for (i=0; i<groupMap->namesOfGroups.size(); i++) { (*(filehandles[groupMap->namesOfGroups[i]])).close();  delete filehandles[groupMap->namesOfGroups[i]]; }
 244                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); } outputTypes.clear();
 245                                 return 0;
 246                 }
 247
 248                 //output error messages about any remaining user labels
 249                 set<string>::iterator it;
 250                 bool needToRun = false;
 251                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 252                         m->mothurOut("Your file does not include the label " + *it);
 253                         if (processedLabels.count(lastLabel) != 1) {
 254                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 255                                 needToRun = true;
 256                         }else {
 257                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 258                         }
 259
 260                 }
 261
 262                 if (m->control_pressed) {
 263                                 delete input; delete groupMap;
 264                                 for (i=0; i<groupMap->namesOfGroups.size(); i++) {  (*(filehandles[groupMap->namesOfGroups[i]])).close();  delete filehandles[groupMap->namesOfGroups[i]]; }
 265                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); } outputTypes.clear();
 266                                 return 0;
 267                 }
 268
 269                 //run last label if you need to
 270                 if (needToRun == true)  {
 271                         if (list != NULL) {     delete list;    }
 272                         list = input->getListVector(lastLabel); //get new list vector to process
 273
 274                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 275                         parse(list);
 276
 277                         delete list;
 278                 }
 279
 280                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 281                         (*(filehandles[it3->first])).close();
 282                         delete it3->second;
 283                 }
 284
 285
 286                 delete groupMap;
 287                 delete input;
 288
 289                 if (m->control_pressed) {
 290                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); } outputTypes.clear();
 291                         return 0;
 292                 }
 293
 294                 //set fasta file as new current fastafile
 295                 string current = "";
 296                 itTypes = outputTypes.find("list");
 297                 if (itTypes != outputTypes.end()) {
 298                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 299                 }
 300
 301                 m->mothurOutEndLine();
 302                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 303                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 304                 m->mothurOutEndLine();
 305
 306                 return 0;
 307         }
 308         catch(exception& e) {
 309                 m->errorOut(e, "ParseListCommand", "execute");
 310                 exit(1);
 311         }
 312 }
 313 /**********************************************************************************************************************/
 314 int ParseListCommand::parse(ListVector* thisList) {
 315         try {
 316
 317                 map<string, string> groupVector;
 318                 map<string, string>::iterator itGroup;
 319                 map<string, int> groupNumBins;
 320
 321                 //print label
 322                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 323                         groupNumBins[it3->first] = 0;
 324                         groupVector[it3->first] = "";
 325                 }
 326
 327
 328                 for (int i = 0; i < thisList->getNumBins(); i++) {
 329                         if (m->control_pressed) { return 0; }
 330
 331                         map<string, string> groupBins;
 332                         string bin = list->get(i);
 333
 334                         vector<string> names;
 335                         m->splitAtComma(bin, names);  //parses bin into individual sequence names
 336
 337                         //parse bin into list of sequences in each group
 338                         for (int j = 0; j < names.size(); j++) {
 339                                 string group = groupMap->getGroup(names[j]);
 340
 341                                 if (group == "not found") { m->mothurOut(names[j] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); exit(1); }
 342
 343                                 itGroup = groupBins.find(group);
 344                                 if(itGroup == groupBins.end()) {
 345                                         groupBins[group] = names[j];  //add first name
 346                                         groupNumBins[group]++;
 347                                 }else{ //add another name
 348                                         groupBins[group] = groupBins[group] + "," + names[j];
 349                                 }
 350                         }
 351
 352                         //print parsed bin info to files
 353                         for (itGroup = groupBins.begin(); itGroup != groupBins.end(); itGroup++) {
 354                                 groupVector[itGroup->first] +=  itGroup->second + '\t';
 355                         }
 356
 357                 }
 358
 359                 //end list vector
 360                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 361                         (*(filehandles[it3->first])) << thisList->getLabel() << '\t' << groupNumBins[it3->first] << '\t' << groupVector[it3->first] << endl;  // label numBins  listvector for that group
 362                 }
 363
 364                 return 0;
 365
 366         }
 367         catch(exception& e) {
 368                 m->errorOut(e, "ParseListCommand", "parse");
 369                 exit(1);
 370         }
 371 }
 372
 373 /**********************************************************************************************************************/
 374
 375