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