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                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  70
  71                 else {
  72                         vector<string> myArray = setParameters();
  73
  74                         OptionParser parser(option);
  75                         map<string, string> parameters = parser.getParameters();
  76
  77                         ValidParameters validParameter;
  78                         map<string, string>::iterator it;
  79
  80                         //check to make sure all parameters are valid for command
  81                         for (it = parameters.begin(); it != parameters.end(); it++) {
  82                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  83                         }
  84
  85                         //initialize outputTypes
  86                         vector<string> tempOutNames;
  87                         outputTypes["list"] = tempOutNames;
  88
  89                         //if the user changes the input directory command factory will send this info to us in the output parameter
  90                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  91                         if (inputDir == "not found"){   inputDir = "";          }
  92                         else {
  93                                 string path;
  94                                 it = parameters.find("list");
  95                                 //user has given a template file
  96                                 if(it != parameters.end()){
  97                                         path = m->hasPath(it->second);
  98                                         //if the user has not given a path then, add inputdir. else leave path alone.
  99                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 100                                 }
 101
 102                                 it = parameters.find("group");
 103                                 //user has given a template file
 104                                 if(it != parameters.end()){
 105                                         path = m->hasPath(it->second);
 106                                         //if the user has not given a path then, add inputdir. else leave path alone.
 107                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 108                                 }
 109                         }
 110
 111
 112                         //if the user changes the output directory command factory will send this info to us in the output parameter
 113                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 114
 115                         //check for required parameters
 116                         listfile = validParameter.validFile(parameters, "list", true);
 117                         if (listfile == "not open") { abort = true; }
 118                         else if (listfile == "not found") {
 119                                 listfile = m->getListFile();
 120                                 if (listfile != "") {  m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 121                                 else {
 122                                         m->mothurOut("No valid current list file. You must provide a list file."); m->mothurOutEndLine();
 123                                         abort = true;
 124
 125                                 }
 126                         }else { m->setListFile(listfile); }
 127
 128                         groupfile = validParameter.validFile(parameters, "group", true);
 129                         if (groupfile == "not open") { abort = true; }
 130                         else if (groupfile == "not found") {
 131                                 groupfile = m->getListFile();
 132                                 if (groupfile != "") {
 133                                         m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine();
 134                                         groupMap = new GroupMap(groupfile);
 135
 136                                         int error = groupMap->readMap();
 137                                         if (error == 1) { abort = true; }
 138                                 }else { m->mothurOut("No valid current group file. You must provide a group file."); m->mothurOutEndLine();  abort = true; }
 139                         }else {
 140                                 m->setGroupFile(groupfile);
 141                                 groupMap = new GroupMap(groupfile);
 142
 143                                 int error = groupMap->readMap();
 144                                 if (error == 1) { abort = true; }
 145                         }
 146
 147                         //do you have all files needed
 148                         if ((listfile == "") || (groupfile == "")) { m->mothurOut("You must enter both a listfile and groupfile for the parse.list command. "); m->mothurOutEndLine(); abort = true;  }
 149
 150                         //check for optional parameter and set defaults
 151                         // ...at some point should added some additional type checking...
 152                         label = validParameter.validFile(parameters, "label", false);
 153                         if (label == "not found") { label = "";  allLines = 1; }
 154                         else {
 155                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 156                                 else { allLines = 1;  }
 157                         }
 158                 }
 159
 160         }
 161         catch(exception& e) {
 162                 m->errorOut(e, "ParseListCommand", "ParseListCommand");
 163                 exit(1);
 164         }
 165 }
 166 //**********************************************************************************************************************
 167 int ParseListCommand::execute(){
 168         try {
 169
 170                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 171
 172                 //set fileroot
 173                 string fileroot = outputDir + m->getRootName(m->getSimpleName(listfile));
 174
 175                 //fill filehandles with neccessary ofstreams
 176                 int i;
 177                 ofstream* temp;
 178                 for (i=0; i<groupMap->namesOfGroups.size(); i++) {
 179                         temp = new ofstream;
 180                         filehandles[groupMap->namesOfGroups[i]] = temp;
 181
 182                         string filename = fileroot +  groupMap->namesOfGroups[i] + ".list";
 183                         outputNames.push_back(filename); outputTypes["list"].push_back(filename);
 184                         m->openOutputFile(filename, *temp);
 185                 }
 186
 187                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 188                 set<string> processedLabels;
 189                 set<string> userLabels = labels;
 190
 191                 input = new InputData(listfile, "list");
 192                 list = input->getListVector();
 193                 string lastLabel = list->getLabel();
 194
 195                 if (m->control_pressed) {
 196                         delete input; delete list; delete groupMap;
 197                         for (i=0; i<groupMap->namesOfGroups.size(); i++) {  (*(filehandles[groupMap->namesOfGroups[i]])).close();  delete filehandles[groupMap->namesOfGroups[i]]; }
 198                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 199                         return 0;
 200                 }
 201
 202                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 203
 204                         if (m->control_pressed) {
 205                                 delete input; delete list; delete groupMap;
 206                                 for (i=0; i<groupMap->namesOfGroups.size(); i++) {  (*(filehandles[groupMap->namesOfGroups[i]])).close();  delete filehandles[groupMap->namesOfGroups[i]]; }
 207                                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 208                                 return 0;
 209                         }
 210
 211                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 212
 213                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 214                                         parse(list);
 215
 216                                         processedLabels.insert(list->getLabel());
 217                                         userLabels.erase(list->getLabel());
 218                         }
 219
 220                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 221                                         string saveLabel = list->getLabel();
 222
 223                                         delete list;
 224                                         list = input->getListVector(lastLabel); //get new list vector to process
 225
 226                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 227                                         parse(list);
 228
 229                                         processedLabels.insert(list->getLabel());
 230                                         userLabels.erase(list->getLabel());
 231
 232                                         //restore real lastlabel to save below
 233                                         list->setLabel(saveLabel);
 234                         }
 235
 236
 237                         lastLabel = list->getLabel();
 238
 239                         delete list;
 240                         list = input->getListVector(); //get new list vector to process
 241                 }
 242
 243                 if (m->control_pressed) {
 244                                 delete input; delete groupMap;
 245                                 for (i=0; i<groupMap->namesOfGroups.size(); i++) { (*(filehandles[groupMap->namesOfGroups[i]])).close();  delete filehandles[groupMap->namesOfGroups[i]]; }
 246                                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 247                                 return 0;
 248                 }
 249
 250                 //output error messages about any remaining user labels
 251                 set<string>::iterator it;
 252                 bool needToRun = false;
 253                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 254                         m->mothurOut("Your file does not include the label " + *it);
 255                         if (processedLabels.count(lastLabel) != 1) {
 256                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 257                                 needToRun = true;
 258                         }else {
 259                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 260                         }
 261
 262                 }
 263
 264                 if (m->control_pressed) {
 265                                 delete input; delete groupMap;
 266                                 for (i=0; i<groupMap->namesOfGroups.size(); i++) {  (*(filehandles[groupMap->namesOfGroups[i]])).close();  delete filehandles[groupMap->namesOfGroups[i]]; }
 267                                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 268                                 return 0;
 269                 }
 270
 271                 //run last label if you need to
 272                 if (needToRun == true)  {
 273                         if (list != NULL) {     delete list;    }
 274                         list = input->getListVector(lastLabel); //get new list vector to process
 275
 276                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 277                         parse(list);
 278
 279                         delete list;
 280                 }
 281
 282                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 283                         (*(filehandles[it3->first])).close();
 284                         delete it3->second;
 285                 }
 286
 287
 288                 delete groupMap;
 289                 delete input;
 290
 291                 if (m->control_pressed) {
 292                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 293                         return 0;
 294                 }
 295
 296                 //set fasta file as new current fastafile
 297                 string current = "";
 298                 itTypes = outputTypes.find("list");
 299                 if (itTypes != outputTypes.end()) {
 300                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 301                 }
 302
 303                 m->mothurOutEndLine();
 304                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 305                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 306                 m->mothurOutEndLine();
 307
 308                 return 0;
 309         }
 310         catch(exception& e) {
 311                 m->errorOut(e, "ParseListCommand", "execute");
 312                 exit(1);
 313         }
 314 }
 315 /**********************************************************************************************************************/
 316 int ParseListCommand::parse(ListVector* thisList) {
 317         try {
 318
 319                 map<string, string> groupVector;
 320                 map<string, string>::iterator itGroup;
 321                 map<string, int> groupNumBins;
 322
 323                 //print label
 324                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 325                         groupNumBins[it3->first] = 0;
 326                         groupVector[it3->first] = "";
 327                 }
 328
 329
 330                 for (int i = 0; i < thisList->getNumBins(); i++) {
 331                         if (m->control_pressed) { return 0; }
 332
 333                         map<string, string> groupBins;
 334                         string bin = list->get(i);
 335
 336                         vector<string> names;
 337                         m->splitAtComma(bin, names);  //parses bin into individual sequence names
 338
 339                         //parse bin into list of sequences in each group
 340                         for (int j = 0; j < names.size(); j++) {
 341                                 string group = groupMap->getGroup(names[j]);
 342
 343                                 if (group == "not found") { m->mothurOut(names[j] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); exit(1); }
 344
 345                                 itGroup = groupBins.find(group);
 346                                 if(itGroup == groupBins.end()) {
 347                                         groupBins[group] = names[j];  //add first name
 348                                         groupNumBins[group]++;
 349                                 }else{ //add another name
 350                                         groupBins[group] = groupBins[group] + "," + names[j];
 351                                 }
 352                         }
 353
 354                         //print parsed bin info to files
 355                         for (itGroup = groupBins.begin(); itGroup != groupBins.end(); itGroup++) {
 356                                 groupVector[itGroup->first] +=  itGroup->second + '\t';
 357                         }
 358
 359                 }
 360
 361                 //end list vector
 362                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 363                         (*(filehandles[it3->first])) << thisList->getLabel() << '\t' << groupNumBins[it3->first] << '\t' << groupVector[it3->first] << endl;  // label numBins  listvector for that group
 364                 }
 365
 366                 return 0;
 367
 368         }
 369         catch(exception& e) {
 370                 m->errorOut(e, "ParseListCommand", "parse");
 371                 exit(1);
 372         }
 373 }
 374
 375 /**********************************************************************************************************************/
 376
 377