#include "listseqscommand.h"
#include "sequence.hpp"
#include "listvector.hpp"
+#include "counttable.h"
+
+//**********************************************************************************************************************
+vector<string> ListSeqsCommand::setParameters(){
+ try {
+ CommandParameter pfasta("fasta", "InputTypes", "", "", "FNGLT", "FNGLT", "none","accnos",false,false,true); parameters.push_back(pfasta);
+ CommandParameter pname("name", "InputTypes", "", "", "FNGLT", "FNGLT", "none","accnos",false,false,true); parameters.push_back(pname);
+ CommandParameter pcount("count", "InputTypes", "", "", "FNGLT", "FNGLT", "none","accnos",false,false,true); parameters.push_back(pcount);
+ CommandParameter pgroup("group", "InputTypes", "", "", "FNGLT", "FNGLT", "none","accnos",false,false,true); parameters.push_back(pgroup);
+ CommandParameter plist("list", "InputTypes", "", "", "FNGLT", "FNGLT", "none","accnos",false,false,true); parameters.push_back(plist);
+ CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "FNGLT", "FNGLT", "none","accnos",false,false,true); parameters.push_back(ptaxonomy);
+ CommandParameter palignreport("alignreport", "InputTypes", "", "", "FNGLT", "FNGLT", "none","accnos",false,false); parameters.push_back(palignreport);
+ CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
+ CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
+
+ vector<string> myArray;
+ for (int i = 0; i < parameters.size(); i++) { myArray.push_back(parameters[i].name); }
+ return myArray;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "ListSeqsCommand", "setParameters");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+string ListSeqsCommand::getHelpString(){
+ try {
+ string helpString = "";
+ helpString += "The list.seqs command reads a fasta, name, group, count, list, taxonomy or alignreport file and outputs a .accnos file containing sequence names.\n";
+ helpString += "The list.seqs command parameters are fasta, name, group, count, list, taxonomy and alignreport. You must provide one of these parameters.\n";
+ helpString += "The list.seqs command should be in the following format: list.seqs(fasta=yourFasta).\n";
+ helpString += "Example list.seqs(fasta=amazon.fasta).\n";
+ helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
+ return helpString;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "ListSeqsCommand", "getHelpString");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+string ListSeqsCommand::getOutputPattern(string type) {
+ try {
+ string pattern = "";
+
+ if (type == "accnos") { pattern = "[filename],accnos"; }
+ else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true; }
+
+ return pattern;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "ListSeqsCommand", "getOutputPattern");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+ListSeqsCommand::ListSeqsCommand(){
+ try {
+ abort = true; calledHelp = true;
+ setParameters();
+ vector<string> tempOutNames;
+ outputTypes["accnos"] = tempOutNames;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "ListSeqsCommand", "ListSeqsCommand");
+ exit(1);
+ }
+}
//**********************************************************************************************************************
ListSeqsCommand::ListSeqsCommand(string option) {
try {
- abort = false;
+ abort = false; calledHelp = false;
//allow user to run help
- if(option == "help") { help(); abort = true; }
-
+ if(option == "help") { help(); abort = true; calledHelp = true; }
+ else if(option == "citation") { citation(); abort = true; calledHelp = true;}
else {
- //valid paramters for this command
- string Array[] = {"fasta","name", "group", "alignreport","list","taxonomy","outputdir","inputdir"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+ vector<string> myArray = setParameters();
OptionParser parser(option);
map<string,string> parameters = parser.getParameters();
if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { abort = true; }
}
+ //initialize outputTypes
+ vector<string> tempOutNames;
+ outputTypes["accnos"] = tempOutNames;
+
//if the user changes the output directory command factory will send this info to us in the output parameter
outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){ outputDir = ""; }
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["taxonomy"] = inputDir + it->second; }
}
+
+ it = parameters.find("count");
+ //user has given a template file
+ if(it != parameters.end()){
+ path = m->hasPath(it->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["count"] = inputDir + it->second; }
+ }
}
//check for required parameters
fastafile = validParameter.validFile(parameters, "fasta", true);
if (fastafile == "not open") { abort = true; }
- else if (fastafile == "not found") { fastafile = ""; }
+ else if (fastafile == "not found") { fastafile = ""; }
+ else { m->setFastaFile(fastafile); }
namefile = validParameter.validFile(parameters, "name", true);
if (namefile == "not open") { abort = true; }
else if (namefile == "not found") { namefile = ""; }
+ else { m->setNameFile(namefile); }
groupfile = validParameter.validFile(parameters, "group", true);
if (groupfile == "not open") { abort = true; }
else if (groupfile == "not found") { groupfile = ""; }
+ else { m->setGroupFile(groupfile); }
alignfile = validParameter.validFile(parameters, "alignreport", true);
if (alignfile == "not open") { abort = true; }
listfile = validParameter.validFile(parameters, "list", true);
if (listfile == "not open") { abort = true; }
else if (listfile == "not found") { listfile = ""; }
-
+ else { m->setListFile(listfile); }
+
taxfile = validParameter.validFile(parameters, "taxonomy", true);
if (taxfile == "not open") { abort = true; }
else if (taxfile == "not found") { taxfile = ""; }
+ else { m->setTaxonomyFile(taxfile); }
+
+ countfile = validParameter.validFile(parameters, "count", true);
+ if (countfile == "not open") { abort = true; }
+ else if (countfile == "not found") { countfile = ""; }
+ else { m->setCountTableFile(countfile); }
- if ((fastafile == "") && (namefile == "") && (listfile == "") && (groupfile == "") && (alignfile == "") && (taxfile == "")) { m->mothurOut("You must provide a file."); m->mothurOutEndLine(); abort = true; }
+ if ((countfile == "") && (fastafile == "") && (namefile == "") && (listfile == "") && (groupfile == "") && (alignfile == "") && (taxfile == "")) { m->mothurOut("You must provide a file."); m->mothurOutEndLine(); abort = true; }
int okay = 1;
if (outputDir != "") { okay++; }
}
//**********************************************************************************************************************
-void ListSeqsCommand::help(){
- try {
- m->mothurOut("The list.seqs command reads a fasta, name, group, list, taxonomy or alignreport file and outputs a .accnos file containing sequence names.\n");
- m->mothurOut("The list.seqs command parameters are fasta, name, group, list, taxonomy and alignreport. You must provide one of these parameters.\n");
- m->mothurOut("The list.seqs command should be in the following format: list.seqs(fasta=yourFasta).\n");
- m->mothurOut("Example list.seqs(fasta=amazon.fasta).\n");
- m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
- }
- catch(exception& e) {
- m->errorOut(e, "ListSeqsCommand", "help");
- exit(1);
- }
-}
-
-//**********************************************************************************************************************
-
int ListSeqsCommand::execute(){
try {
- if (abort == true) { return 0; }
+ if (abort == true) { if (calledHelp) { return 0; } return 2; }
//read functions fill names vector
if (fastafile != "") { inputFileName = fastafile; readFasta(); }
else if (alignfile != "") { inputFileName = alignfile; readAlign(); }
else if (listfile != "") { inputFileName = listfile; readList(); }
else if (taxfile != "") { inputFileName = taxfile; readTax(); }
+ else if (countfile != "") { inputFileName = countfile; readCount(); }
- if (m->control_pressed) { return 0; }
+ if (m->control_pressed) { outputTypes.clear(); return 0; }
//sort in alphabetical order
sort(names.begin(), names.end());
if (outputDir == "") { outputDir += m->hasPath(inputFileName); }
- string outputFileName = outputDir + m->getRootName(m->getSimpleName(inputFileName)) + "accnos";
+ map<string, string> variables;
+ variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(inputFileName));
+ string outputFileName = getOutputFileName("accnos", variables);
ofstream out;
m->openOutputFile(outputFileName, out);
+ outputNames.push_back(outputFileName); outputTypes["accnos"].push_back(outputFileName);
//output to .accnos file
for (int i = 0; i < names.size(); i++) {
- if (m->control_pressed) { out.close(); remove(outputFileName.c_str()); return 0; }
+ if (m->control_pressed) { outputTypes.clear(); out.close(); m->mothurRemove(outputFileName); return 0; }
out << names[i] << endl;
}
out.close();
- if (m->control_pressed) { remove(outputFileName.c_str()); return 0; }
-
+ if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(outputFileName); return 0; }
+
+ m->setAccnosFile(outputFileName);
+
m->mothurOutEndLine();
- m->mothurOut("Output File Name: "); m->mothurOutEndLine();
+ m->mothurOut("Output File Names: "); m->mothurOutEndLine();
m->mothurOut(outputFileName); m->mothurOutEndLine();
m->mothurOutEndLine();
+ //set accnos file as new current accnosfile
+ string current = "";
+ itTypes = outputTypes.find("accnos");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
+ }
+
return 0;
}
ifstream in;
m->openInputFile(fastafile, in);
string name;
-
+
+ //ofstream out;
+ //string newFastaName = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "numsAdded.fasta";
+ //m->openOutputFile(newFastaName, out);
+ int count = 1;
+ //string lastName = "";
+
while(!in.eof()){
if (m->control_pressed) { in.close(); return 0; }
if (name != "") { names.push_back(name); }
m->gobble(in);
+ if (m->debug) { count++; cout << "[DEBUG]: count = " + toString(count) + ", name = " + currSeq.getName() + "\n"; }
}
in.close();
+ //out.close();
return 0;
if (m->control_pressed) { in.close(); return 0; }
- while (binnames.find_first_of(',') != -1) {
- string name = binnames.substr(0,binnames.find_first_of(','));
- binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
- names.push_back(name);
- }
-
- names.push_back(binnames);
+ m->splitAtComma(binnames, names);
}
}
in.close();
in >> secondCol;
//parse second column saving each name
- while (secondCol.find_first_of(',') != -1) {
- name = secondCol.substr(0,secondCol.find_first_of(','));
- secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
- names.push_back(name);
- }
-
- //get name after last ,
- names.push_back(secondCol);
+ m->splitAtComma(secondCol, names);
m->gobble(in);
}
if (m->control_pressed) { in.close(); return 0; }
- in >> name; //read from first column
+ in >> name; m->gobble(in); //read from first column
in >> group; //read from second column
names.push_back(name);
exit(1);
}
}
-
+//**********************************************************************************************************************
+int ListSeqsCommand::readCount(){
+ try {
+ CountTable ct;
+ ct.readTable(countfile, false);
+
+ if (m->control_pressed) { return 0; }
+
+ names = ct.getNamesOfSeqs();
+
+ return 0;
+
+ }
+ catch(exception& e) {
+ m->errorOut(e, "ListSeqsCommand", "readCount");
+ exit(1);
+ }
+}
//**********************************************************************************************************************
//alignreport file has a column header line then all other lines contain 16 columns. we just want the first column since that contains the name
int ListSeqsCommand::readAlign(){
if (!in.eof()) { in >> junk; }
else { break; }
}
-
+ //m->getline(in);
while(!in.eof()){
if (m->control_pressed) { in.close(); return 0; }
in >> name; //read from first column
-
+ //m->getline(in);
//read rest
for (int i = 0; i < 15; i++) {
if (!in.eof()) { in >> junk; }
names.push_back(firstCol);
m->gobble(in);
+
}
in.close();
projects / mothur.git / blobdiff