bool constructSafeLabels(const NxsTaxaBlock & tb, std::vector<std::string> * vs, std::set<std::string> & safeLabelSet)
{
if (vs == 0L)
return false;
const unsigned numTaxa = tb.GetNTaxTotal();
bool changed = false;
for (unsigned i = 0; i < numTaxa; ++i)
{
NxsString oldName = tb.GetTaxonLabel(i);
unsigned oldOffset = 0;
unsigned newOffset = 0;
std::string prefix;
for (; newOffset < 10 && oldOffset < oldName.length(); ++oldOffset)
{
char c = oldName[oldOffset];
if ((isgraph(c)) && (strchr("!@#$%^&*.?|()[]{}/\\,;:=*\'\"`-+<>", c) == NULL))
{
if (newOffset == 0 && isdigit(c))
{
prefix.append(1, 't');
newOffset += 1;
}
prefix.append(1, toupper(c));
newOffset += 1;
}
}
std::string newLabel = assignUniqueName(prefix, 10, safeLabelSet);
safeLabelSet.insert(newLabel);
vs->push_back(newLabel);
if (newLabel != oldName)
changed = true;
}
return changed;
}
/**
* Build a Nexus version of the block's data.
*/
NxsTaxaBlock NxsNexmlTaxa::getNexusBlock()const{
NxsTaxaBlock ret = NxsTaxaBlock();
for ( unsigned i = 0; i < this->GetNTaxTotal(); ++i ){
ret.AddTaxonLabel( this->GetTaxonLabel( i ) );
}
return ret;
}
NxsTaxaBlock *NxsTaxaBlockFactory::GetBlockReaderForID(const std::string & idneeded, NxsReader *reader, NxsToken *)
{
if (reader == NULL || idneeded != "TAXA")
return NULL;
NxsTaxaBlock * nb = new NxsTaxaBlock();
nb->SetImplementsLinkAPI(false);
return nb;
}
NxsNexmlTaxa::NxsNexmlTaxa( NxsTaxaBlock nTax ){
vector< Otu* > otus = vector< Otu* >();
for ( unsigned i = 0; i < nTax.GetNTaxTotal(); ++i ){
otus.push_back( new Otu( Glib::ustring( nTax.GetTaxonLabel( i ).c_str() ) ) );
}
this->otus_ = new Otus( otus );
this->block_name_ = "Taxa Block";
//<------data-model-------------------------------->
this->inactive_taxa_ = set< unsigned int >();
this->active_taxa_ = set< unsigned int >();
this->partitions_ = map< string, NxsPartition >();
this->index_sets_ = map< string, NxsUnsignedSet >();
}
bool substituteSafeTaxaLabels(PublicNexusReader &nexusReader, std::string filepath, bool evenTrivial)
{
std::ostream * outStrPtr = 0L;
std::ofstream outStream;
const unsigned nTaxaBlocks = nexusReader.GetNumTaxaBlocks();
bool someChanged = false;
std::set<std::string> safeNameSet;
try
{
for (unsigned t = 0; t < nTaxaBlocks; ++t)
{
NxsTaxaBlock * tb = nexusReader.GetTaxaBlock(t);
std::vector<std::string> safeLabelsVec;
bool changed = constructSafeLabels(*tb, &safeLabelsVec, safeNameSet);
someChanged = someChanged | changed;
if (changed || evenTrivial)
{
if (outStrPtr == 0L)
{
outStream.open(filepath.c_str());
if (!outStream.good())
{
NxsString eMessage;
eMessage << "Could not open the file \"" << filepath << "\"";
throw NxsException(eMessage);
}
outStrPtr = &outStream;
outStream << "#NEXUS\n";
}
tb->WriteAsNexus(*outStrPtr);
std::vector<NxsString> oldNameVec;
const unsigned numTaxa = tb->GetNTaxTotal();
for (unsigned i = 0; i < numTaxa; ++i)
{
NxsString oldName = tb->GetTaxonLabel(i);
oldNameVec.push_back(oldName);
NxsString newName;
newName << safeLabelsVec[i];
tb->ChangeTaxonLabel(i, newName);
}
NxsString title;
title << tb->GetTitle();
NxsString nt;
nt << "Renamed " << title;
tb->SetTitle(nt, false);
tb->WriteAsNexus(*outStrPtr);
// We can't just create a TaxaAssociationBlock instance here,
// and call WriteAsNexus because it first and second taxa block pointers
// would refer to the same instance...
NxsString taTitle;
taTitle << "Renaming " << title;
*outStrPtr << "BEGIN TaxaAssociation;\n Title " << NxsString::GetEscaped(taTitle) << " ;\n";
*outStrPtr << " Taxa " << NxsString::GetEscaped(title) << " , " << NxsString::GetEscaped(nt) << " ;\n";
*outStrPtr << " Associates ";
for (unsigned i = 0; i < numTaxa; ++i)
{
if (i > 0)
*outStrPtr << " ,\n ";
*outStrPtr << NxsString::GetEscaped(oldNameVec[i]) << " / " << NxsString::GetEscaped(safeLabelsVec[i]);
}
*outStrPtr << " ;\nEND;\n" ;
}
}
}
catch (...)
{
if (outStrPtr != 0L)
outStream.close();
throw;
}
if (outStrPtr != 0L)
outStream.close();
return someChanged;
}
void DataMatrix::CreateMatrixFromNCL(GarliReader &reader){
NxsCharactersBlock *charblock = reader.GetCharactersBlock();
// vector<unsigned> reducedToOrigCharMap = charblock->GetOrigIndexVector();
NxsTaxaBlock *taxablock = reader.GetTaxaBlock();
int numOrigTaxa = charblock->GetNTax();
int numActiveTaxa = charblock->GetNumActiveTaxa();
int numOrigChar = charblock->GetNChar();
int numActiveChar = charblock->GetNumActiveChar();
//int num_chars = reducedToOrigCharMap.size();
//cout << num_chars << endl;
NewMatrix( numActiveTaxa, numActiveChar );
// read in the data, including taxon names
int i=0;
if(modSpec.IsAminoAcid() == false){
for( int origTaxIndex = 0; origTaxIndex < numOrigTaxa; origTaxIndex++ ) {
if(charblock->IsActiveTaxon(origTaxIndex)){
SetTaxonLabel( i, taxablock->GetTaxonLabel(origTaxIndex).c_str());
int j = 0;
for( int origIndex = 0; origIndex < numOrigChar; origIndex++ ) {
if(charblock->IsActiveChar(origIndex)){
unsigned char datum = '\0';
if(charblock->IsGapState(origTaxIndex, origIndex) == true) datum = 15;
else if(charblock->IsMissingState(origTaxIndex, origIndex) == true) datum = 15;
else{
int nstates = charblock->GetNumStates(origTaxIndex, origIndex);
for(int s=0;s<nstates;s++){
datum += CharToBitwiseRepresentation(charblock->GetState(origTaxIndex, origIndex, s));
}
}
SetMatrix( i, j++, datum );
}
}
i++;
}
}
}
else{
for( int origTaxIndex = 0; origTaxIndex < numOrigTaxa; origTaxIndex++ ) {
if(charblock->IsActiveTaxon(origTaxIndex)){
SetTaxonLabel( i, taxablock->GetTaxonLabel(origTaxIndex).c_str());
int j = 0;
for( int origIndex = 0; origIndex < numOrigChar; origIndex++ ) {
if(charblock->IsActiveChar(origIndex)){
unsigned char datum = '\0';
if(charblock->IsGapState(origTaxIndex, origIndex) == true) datum = 20;
else if(charblock->IsMissingState(origTaxIndex, origIndex) == true) datum = 20;
else{
int nstates = charblock->GetNumStates(origTaxIndex, origIndex);
assert(nstates == 1);
datum = CharToAminoAcidNumber(charblock->GetState(origTaxIndex, origIndex, 0));
/* for(int s=0;s<nstates;s++){
datum += CharToBitwiseRepresentation(charblock->GetState(origTaxIndex, origIndex, s));
}
*/ }
SetMatrix( i, j++, datum );
}
}
i++;
}
}
}
}
void AminoacidData::CreateMatrixFromNCL(GarliReader &reader){
NxsCharactersBlock *charblock;
int num=0, numNuc = -1;
do{
charblock = reader.GetCharactersBlock(num);
if(charblock->GetDataType() == NxsCharactersBlock::protein){
if(numNuc < 0) numNuc = num;
else{
throw ErrorException("Multiple characters/data blocks containing protein data found in Nexus datafile!\n\tEither combine the blocks or comment one out.");
}
}
else outman.UserMessage("Ignoring non-protein characters block from Nexus datafile");
num++;
}while(num < reader.NumCharBlocks());
if(numNuc < 0) throw ErrorException("No characters/data blocks containing protein data found in Nexus datafile!");
charblock = reader.GetCharactersBlock(numNuc);
if(charblock->GetNumActiveChar() < charblock->GetNChar()){
outman.UserMessageNoCR("Excluded characters:\n\t");
for(int c=0;c<charblock->GetNCharTotal();c++)
if(charblock->IsExcluded(c)) outman.UserMessageNoCR("%d ", c+1);
outman.UserMessage("");
}
// vector<unsigned> reducedToOrigCharMap = charblock->GetOrigIndexVector();
NxsTaxaBlock *taxablock = reader.GetTaxaBlock();
int numOrigTaxa = charblock->GetNTax();
int numActiveTaxa = charblock->GetNumActiveTaxa();
int numOrigChar = charblock->GetNChar();
int numActiveChar = charblock->GetNumActiveChar();
//int num_chars = reducedToOrigCharMap.size();
//cout << num_chars << endl;
NewMatrix( numActiveTaxa, numActiveChar );
// read in the data, including taxon names
int i=0;
for( int origTaxIndex = 0; origTaxIndex < numOrigTaxa; origTaxIndex++ ) {
if(charblock->IsActiveTaxon(origTaxIndex)){
//internally, blanks in taxon names will be stored as underscores
NxsString tlabel = taxablock->GetTaxonLabel(origTaxIndex);
tlabel.BlanksToUnderscores();
SetTaxonLabel( i, tlabel.c_str());
int j = 0;
bool firstAmbig = true;
for( int origIndex = 0; origIndex < numOrigChar; origIndex++ ) {
if(charblock->IsActiveChar(origIndex)){
unsigned char datum = '\0';
if(charblock->IsGapState(origTaxIndex, origIndex) == true) datum = 20;
else if(charblock->IsMissingState(origTaxIndex, origIndex) == true) datum = 20;
else{
int nstates = charblock->GetNumStates(origTaxIndex, origIndex);
//assert(nstates == 1);
//need to deal with the possibility of multiple states represented in matrix
//just convert to full ambiguity
if(nstates == 1)
datum = CharToDatum(charblock->GetState(origTaxIndex, origIndex, 0));
else{
if(firstAmbig){
outman.UserMessageNoCR("Partially ambiguous characters of taxon %s converted to full ambiguity:\n\t", TaxonLabel(origTaxIndex));
firstAmbig = false;
}
outman.UserMessageNoCR("%d ", origIndex+1);
datum = CharToDatum('?');
}
}
SetMatrix( i, j++, datum );
}
}
if(firstAmbig == false) outman.UserMessage("");
i++;
}
}
}
void NucleotideData::CreateMatrixFromNCL(GarliReader &reader){
NxsCharactersBlock *charblock;
int num=0, numNuc = -1;
do{
charblock = reader.GetCharactersBlock(num);
if(charblock->GetDataType() == NxsCharactersBlock::nucleotide ||
charblock->GetDataType() == NxsCharactersBlock::dna ||
charblock->GetDataType() == NxsCharactersBlock::rna){
if(numNuc < 0) numNuc = num;
else{
throw ErrorException("Multiple characters/data blocks containing nucleotide data found in Nexus datafile!\n\tEither combine the blocks or comment one out.");
}
}
else outman.UserMessage("Ignoring non-nucleotide characters block from Nexus datafile");
num++;
}while(num < reader.NumCharBlocks());
if(numNuc < 0) throw ErrorException("No characters/data blocks containing nucleotide data found in Nexus datafile!");
charblock = reader.GetCharactersBlock(numNuc);
if(charblock->GetNumActiveChar() < charblock->GetNChar()){
outman.UserMessageNoCR("Excluded characters:\n\t");
for(int c=0;c<charblock->GetNCharTotal();c++)
if(charblock->IsExcluded(c)) outman.UserMessageNoCR("%d ", c+1);
outman.UserMessage("");
}
// vector<unsigned> reducedToOrigCharMap = charblock->GetOrigIndexVector();
NxsTaxaBlock *taxablock = reader.GetTaxaBlock();
int numOrigTaxa = charblock->GetNTax();
int numActiveTaxa = charblock->GetNumActiveTaxa();
int numOrigChar = charblock->GetNChar();
int numActiveChar = charblock->GetNumActiveChar();
//int num_chars = reducedToOrigCharMap.size();
//cout << num_chars << endl;
NewMatrix( numActiveTaxa, numActiveChar );
// read in the data, including taxon names
int i=0;
for( int origTaxIndex = 0; origTaxIndex < numOrigTaxa; origTaxIndex++ ) {
if(charblock->IsActiveTaxon(origTaxIndex)){
//internally, blanks in taxon names will be stored as underscores
NxsString tlabel = taxablock->GetTaxonLabel(origTaxIndex);
tlabel.BlanksToUnderscores();
SetTaxonLabel( i, tlabel.c_str());
int j = 0;
for( int origIndex = 0; origIndex < numOrigChar; origIndex++ ) {
if(charblock->IsActiveChar(origIndex)){
unsigned char datum = '\0';
if(charblock->IsGapState(origTaxIndex, origIndex) == true) datum = 15;
else if(charblock->IsMissingState(origTaxIndex, origIndex) == true) datum = 15;
else{
int nstates = charblock->GetNumStates(origTaxIndex, origIndex);
for(int s=0;s<nstates;s++){
datum += CharToBitwiseRepresentation(charblock->GetState(origTaxIndex, origIndex, s));
}
}
SetMatrix( i, j++, datum );
}
}
i++;
}
}
}
