/**
| returns the number of indices added.
*/
unsigned NxsSetReader::InterpretTokenAsIndices(NxsToken &token,
const NxsLabelToIndicesMapper & mapper,
const char * setType,
const char * cmdName,
NxsUnsignedSet * destination)
{
try {
const std::string t = token.GetToken();
if (NxsString::case_insensitive_equals(t.c_str(), "ALL"))
{
unsigned m = mapper.GetMaxIndex();
NxsUnsignedSet s;
for (unsigned i = 0; i <= m; ++i)
s.insert(i);
destination->insert(s.begin(), s.end());
return (unsigned)s.size();
}
return mapper.GetIndicesForLabel(t, destination);
}
catch (const NxsException & x)
{
NxsString errormsg = "Error in the ";
errormsg << setType << " descriptor of a " << cmdName << " command.\n";
errormsg += x.msg;
throw NxsException(errormsg, token);
}
catch (...)
{
NxsString errormsg = "Expecting a ";
errormsg << setType << " descriptor (number or label) in the " << cmdName << ". Encountered ";
errormsg << token.GetToken();
throw NxsException(errormsg, token);
}
}
std::vector<unsigned> NxsSetReader::GetSetAsVector(const NxsUnsignedSet &s)
{
std::vector<unsigned> u;
u.reserve(s.size());
for (NxsUnsignedSet::const_iterator sIt = s.begin(); sIt != s.end(); ++sIt)
u.push_back(*sIt);
return u;
}
void NxsSetReader::ReadSetDefinition(
NxsToken &token,
const NxsLabelToIndicesMapper & mapper,
const char * setType, /* "TAXON" or "CHARACTER" -- for error messages only */
const char * cmdName, /* command name -- "TAXSET" or "EXSET"-- for error messages only */
NxsUnsignedSet * destination, /** to be filled */
const NxsUnsignedSet * taboo)
{
NxsString errormsg;
NxsUnsignedSet tmpset;
NxsUnsignedSet dummy;
if (destination == NULL)
destination = & dummy;
unsigned previousInd = UINT_MAX;
std::vector<unsigned> intersectVec;
while (!token.Equals(";"))
{
if (taboo && token.Equals(","))
return;
if (token.Equals("-"))
{
if (previousInd == UINT_MAX)
{
errormsg = "The '-' must be preceded by number or a ";
errormsg << setType << " label in the " << cmdName << " command.";
throw NxsException(errormsg, token);
}
token.GetNextToken();
if (token.Equals(";") || token.Equals("\\"))
{
errormsg = "Range in the ";
errormsg << setType << " set definition in the " << cmdName << " command must be closed with a number or label.";
throw NxsException(errormsg, token);
}
unsigned endpoint;
if (token.Equals("."))
endpoint = mapper.GetMaxIndex();
else
{
tmpset.clear();
unsigned nAdded = NxsSetReader::InterpretTokenAsIndices(token, mapper, setType, cmdName, &tmpset);
if (nAdded != 1)
{
errormsg = "End of a range in a ";
errormsg << setType << " set definition in the " << cmdName << " command must be closed with a single number or label (not a set).";
throw NxsException(errormsg, token);
}
endpoint = *(tmpset.begin());
if (endpoint < previousInd)
{
errormsg = "End of a range in a ";
errormsg << setType << " set definition in the " << cmdName << " command must be a larger index than the start of the range (found ";
errormsg << previousInd + 1 << " - " << token.GetToken();
throw NxsException(errormsg, token);
}
}
token.GetNextToken();
if (token.Equals("\\"))
{
token.GetNextToken();
NxsString t = token.GetToken();
unsigned stride = 0;
try
{
stride = t.ConvertToUnsigned();
}
catch (const NxsString::NxsX_NotANumber &)
{}
if (stride == 0)
{
errormsg = "Expecting a positive number indicating the 'stride' after the \\ in the ";
errormsg << setType << " set definition in the " << cmdName << " command. Encountered ";
errormsg << t;
throw NxsException(errormsg, token);
}
AddRangeToSet(previousInd, endpoint, stride, destination, taboo, token);
token.GetNextToken();
}
else
AddRangeToSet(previousInd, endpoint, 1, destination, taboo, token);
previousInd = UINT_MAX;
}
else
{
tmpset.clear();
const unsigned nAdded = NxsSetReader::InterpretTokenAsIndices(token, mapper, setType, cmdName, &tmpset);
if (taboo != NULL)
{
set_intersection(taboo->begin(), taboo->end(), tmpset.begin(), tmpset.end(), back_inserter(intersectVec));
if (!intersectVec.empty())
{
errormsg << "Illegal repitition of an index (" << 1 + *(intersectVec.begin()) << ") in multiple subsets.";
throw NxsException(errormsg, token);
}
}
if (nAdded == 1 )
{
previousInd = *(tmpset.begin());
destination->insert(previousInd);
}
else
{
previousInd = UINT_MAX;
destination->insert(tmpset.begin(), tmpset.end());
}
token.GetNextToken();
}
}
}
void NxsSetReader::WriteSetAsNexusValue(const NxsUnsignedSet & nxsset, std::ostream & out)
{
NxsUnsignedSet::const_iterator currIt = nxsset.begin();
const NxsUnsignedSet::const_iterator endIt = nxsset.end();
if (currIt == endIt)
return;
unsigned rangeBegin = 1 + *currIt++;
if (currIt == endIt)
{
out << ' ' << rangeBegin;
return;
}
unsigned prev = 1 + *currIt++;
if (currIt == endIt)
{
out << ' ' << rangeBegin << ' ' << prev;
return;
}
unsigned stride = prev - rangeBegin;
unsigned curr = 1 + *currIt++;
bool inRange = true;
while (currIt != endIt)
{
if (inRange)
{
if (curr - prev != stride)
{
if (prev - rangeBegin == stride)
{
out << ' ' << rangeBegin;
rangeBegin = prev;
stride = curr - prev;
}
else
{
if (stride > 1)
out << ' ' << rangeBegin << '-' << prev << " \\ " << stride;
else
out << ' ' << rangeBegin << '-' << prev ;
inRange = false;
}
}
}
else
{
inRange = true;
rangeBegin = prev;
stride = curr - prev;
}
prev = curr;
curr = 1 + *currIt;
currIt++;
}
if (inRange)
{
if (curr - prev != stride)
{
if (prev - rangeBegin == stride)
out << ' ' << rangeBegin << ' ' << prev;
else
{
if (stride > 1)
out << ' ' << rangeBegin << '-' << prev << " \\ " << stride;
else
out << ' ' << rangeBegin << '-' << prev ;
}
out << ' ' << curr;
}
else
{
if (stride > 1)
out << ' ' << rangeBegin << '-' << curr << " \\ " << stride;
else
out << ' ' << rangeBegin << '-' << curr ;
}
}
else
out << ' ' << prev << ' ' << curr;
}