anonymous_namespace{CSVParser.cc} Namespace Reference

Classes
struct	Any
	An any with cached hash. More...
struct	DuplicateKey
struct	InvalidData
struct	MemoryExhausted
struct	NoDataColumns
struct	ShortLine
class	Tokens

Typedefs
using	SliceLabelToken = uint32_t
using	Key = vector< SliceLabelToken, LibCAllocator< SliceLabelToken > >
template<class V >
using	Map = map< Key, V, less< Key >, LibCAllocator< pair< const Key, V > >>

Functions
double	quotedStoD (const string &s, size_t &charsProcd)
string	stripWSAndDecimalSep (const string &s)
size_t	firstNumerical (const vector< string > &v)
bool	emptyTail (const vector< string > &v, size_t start)

Typedef Documentation

Key

using anonymous_namespace{CSVParser.cc}::Key = typedef vector<SliceLabelToken, LibCAllocator<SliceLabelToken> >

Definition at line 225 of file CSVParser.cc.

Map

template<class V >

using anonymous_namespace{CSVParser.cc}::Map = typedef map<Key,V,less<Key>,LibCAllocator<pair<const Key,V> >>

Definition at line 226 of file CSVParser.cc.

SliceLabelToken

using anonymous_namespace{CSVParser.cc}::SliceLabelToken = typedef uint32_t

Definition at line 224 of file CSVParser.cc.

Function Documentation

emptyTail()

bool anonymous_namespace{CSVParser.cc}::emptyTail	(	const vector< string > &	v,
		size_t	start
	)

Definition at line 328 of file CSVParser.cc.

Referenced by minsky::DataSpec::processChunk().

  {
    for (size_t i=start; i<v.size(); ++i)
      if (!v[i].empty()) return false;
    return true;
  }

Here is the caller graph for this function:

firstNumerical()

size_t anonymous_namespace{CSVParser.cc}::firstNumerical

(

const vector< string > &

v

)

Definition at line 305 of file CSVParser.cc.

References quotedStoD(), and stripWSAndDecimalSep().

Referenced by minsky::DataSpec::processChunk().

  {
    size_t r=0;
    for (size_t i=0; i<v.size(); ++i)
      try
        {
          if (!v[i].empty())
            {
              size_t c=0;
              auto s=stripWSAndDecimalSep(v[i]);
              quotedStoD(s,c);
              if (c!=s.size())
                r=i+1;
            }
        }
      catch (...)
        {
          r=i+1;
        }
    return r;
  }

Here is the call graph for this function:

Here is the caller graph for this function:

quotedStoD()

double anonymous_namespace{CSVParser.cc}::quotedStoD	(	const string &	s,
		size_t &	charsProcd
	)

Definition at line 272 of file CSVParser.cc.

Referenced by firstNumerical(), and minsky::isNumerical().

  {
    //strip possible quote characters
    if (!s.empty() && s[0]==s[s.size()-1] && !isalnum(s[0]))
      {
        const double r=quotedStoD(s.substr(1,s.size()-2),charsProcd);
        charsProcd+=2;
        return r;
      }
    if (s.empty()) return nan(""); // treat empty cell as a missing value
    // first try to read the cell as a number
    try {
      const double r=stod(s,&charsProcd);
      if (charsProcd==s.size())
        return r;
    }
    catch (...) {}
    // if not, then strip any leading non-numerical characters ([^0-9.,+-])
    auto n=s.find_first_of("0123456789,.+-");
    return stod(s.substr(n),&charsProcd);
  }

Here is the caller graph for this function:

stripWSAndDecimalSep()

string anonymous_namespace{CSVParser.cc}::stripWSAndDecimalSep

(

const string &

s

)

Definition at line 294 of file CSVParser.cc.

Referenced by firstNumerical(), and minsky::isNumerical().

  {
    string r;
    for (auto c: s)
      if (!isspace(c) && c!=',' && c!='.')
        r+=c;
    return r;
  }

Here is the caller graph for this function: