ITTable.h

Go to the documentation of this file.

//$$FILE$$
//$$VERSION$$
//$$DATE$$
//$$LICENSE$$


#ifndef ITTABLE_H
#define ITTABLE_H


#include <float.h>

#include <string>
#include <vector>
#include <map>

#include "GenString.h"
#include "TTable.h"
#include "Serializer.h"


using namespace std;


typedef multimap<string, unsigned int, StringCompare> tIndex;


class ITTable
{
  public:
    enum eOrientation
    {
        eCOLUMN_WISE = 0,
        eROW_WISE
    };

    enum eSearchType
    {
        eEQUAL = 0,
        eLESS_THAN,
        eLESS_THAN_OR_EQUAL,
        eGREATER_THAN,
        eGREATER_THAN_OR_EQUAL
    };

    enum eSearchDir
    {
        eFORWARD = 0,
        eBACKWARD
    };

    static const unsigned char DT_STRING_VAL = 1; 
    static const unsigned char DT_INTEGER_VAL = 2;
    // static const unsigned char DT_DOUBLE_VAL = 3;

    // Sets string comparison case sensitive
    static const unsigned char CASE_SENSE = 0x00;
    // Sets string comparison case insensitive
    static const unsigned char CASE_INSENSE = 0x01;
    // Sets string comparison to be sensitive to whitespace
    static const unsigned char W_SPACE_SENSE = 0x00;
    // Sets string comparison to ignore repeating whitspace.  
    // Also ignores leading and trailing whitespace
    static const unsigned char W_SPACE_INSENSE = 0x02;
    // string datatype
    static const unsigned char DT_STRING  = DT_STRING_VAL  << 4;
    // integer datatype
    static const unsigned char DT_INTEGER = DT_INTEGER_VAL << 4;
    // VLAD FEATURE NOT WORKING double is not working, maybe integer. check it      // double datatype
    // static const unsigned char DT_DOUBLE  = DT_DOUBLE_VAL  << 4;

    ITTable();

    ITTable(eOrientation orient);

    ITTable(const ITTable& inTable);

    ~ITTable();
 
    ITTable& operator=(const ITTable& inTable);

    inline unsigned int GetNumColumns() const;

    void AddColumn(const string& colName, const vector<string>& col =
      vector<string> (0));

    void InsertColumn(const string& colName,
      const string& atColName, const vector<string>& col =
      vector<string> (0));

    void FillColumn(const string& colName, const vector<string>& col);

    void AppendToColumn(const string& colName, const string& cell);

    void AppendToColumn(const string& colName, const vector<string>& col);

    void GetColumn(vector<string>& col, const string& colName);

    void GetColumn(vector<string>& col, const string& colName,
      const unsigned int fromRowIndex, unsigned int toRowIndex);

    void GetColumn(vector<string>& col, const string& colName,
      const vector<unsigned int>& rowIndex);

    void ClearColumn(const string& colName);

    void DeleteColumn(const string& colName);

    inline unsigned int GetNumRows() const;

    unsigned int AddRow(const vector<string>& row = vector<string> (0));

    unsigned int InsertRow(const unsigned int atRowIndex,
      const vector<string>& row = vector<string> (0));

    void FillRow(const unsigned int rowIndex, const vector<string>& row);

    void GetRow(vector<string>& row, const unsigned int rowIndex,
      const string& fromColName = String::Empty, const string& toColName =
      String::Empty);

    void ClearRow(const unsigned int rowIndex);

    void DeleteRow(const unsigned int rowIndex);

    void DeleteRows(const vector<unsigned int>& rows);

    inline unsigned int GetLastRowIndex();

    void UpdateCell(const unsigned int rowIndex, const string& colName,
      const string& value);

    const string& operator()(const unsigned int rowIndex,
      const string& colName) const;

    void SetFlags(const string& colName, const unsigned char flags);

    unsigned char GetDataType(const string& colName);

#ifdef VLAD_SECOND_ITTABLE
    unsigned int FindFirst(const vector<string>& targets,
      const vector<string>& colNames,
      const string& indexName = String::Empty);
#endif

#ifdef VLAD_SECOND_ITTABLE
    void Search(vector<unsigned int>& res, const string& target,
      const string& colName, const eSearchType searchType = eEQUAL);
#endif

#ifdef VLAD_SECOND_ITTABLE
    void Search(vector<unsigned int>& res, const vector<string>& targets,
      const vector<string>& colNames, const eSearchType searchType = eEQUAL,
      const string& indexName = String::Empty);
#endif

    void FindDuplicateRows(vector<pair<unsigned int, unsigned int> >& duplRows,
      const vector<string>& colNames, const bool keepDuplRows,
      const eSearchDir searchDir = eFORWARD);

    /* \todo Figure this out. */
    void ValidateOptions(unsigned int colIndex);
    void UpdateIndex(const unsigned int indexIndex,
      const unsigned int rowIndex);
    void InsertIndexEntry(const unsigned int indexIndex,
      const unsigned int rowIndex);
    void DeleteIndexEntry(const unsigned int indexIndex,
      const unsigned int rowIndex);
    void VerifyColumnsIndices(const vector<unsigned int>& colIndices);
    int FindIndex(const vector<unsigned int>& colIndices);
    int SetFlags(const unsigned char newOpts, const unsigned int colIndex);
    unsigned int FindFirst(const vector<string>& targets,
      const vector<unsigned int>& colIndices, const unsigned int indexIndex);

    void Search(vector<unsigned int>& res, const vector<string>& targets,
      const vector<unsigned int>& colIndices, const unsigned int indexIndex,
      const eSearchType searchType = eEQUAL);
    void DeleteIndex(const unsigned int indexIndex);
    void Search(vector<unsigned int>& res, const string& target,
      const unsigned int colIndex, const eSearchType searchType = eEQUAL);
    void FindDuplicateRows(const vector<unsigned int>& colIndices,
      vector<pair<unsigned int, unsigned int> >& duplRows,
      const bool keep, const eSearchDir searchDir = eFORWARD);
    void RebuildIndex(const unsigned int indexIndex);
    void InsertColumn(const unsigned int colIndex,
      const vector<string>& col = vector<string> (0));
    void InsertColumn(const unsigned int colIndex,
      vector<string>::const_iterator colBeg,
      vector<string>::const_iterator colEnd);
    void Clear();
    const string& operator()(const unsigned int rowIndex,
      const unsigned int colIndex) const;
    int UpdateCell(const string& cell, const unsigned int colIndex,
      const unsigned int rowIndex);
    void FillColumn(const vector<string>& col, const unsigned int colIndex);
    void FillColumn(const unsigned int colIndex,
      vector<string>::const_iterator colBeg,
      vector<string>::const_iterator colEnd);
    void AppendToColumn(const unsigned int colIndex, const vector<string>& col);
    void AppendToColumn(const unsigned int colIndex, const string& cell);
    void CreateColumn(const unsigned int atColIndex,
      const vector<string>& col = vector<string> (0));
    void CreateColumn(const unsigned int atColIndex,
      vector<string>::const_iterator colBeg,
      vector<string>::const_iterator colEnd);
    void GetColumn(vector<string>& col, const unsigned int colIndex,
      const unsigned int fromRowIndex, unsigned int toRowIndex);
    void GetColumn(vector<string>& col, const unsigned int colIndex,
      const vector<unsigned int>& rowIndex);
    void ClearColumn(const unsigned int colIndex);
    void DeleteColumn(const unsigned int colIndex);
    void GetColumn(vector<string>& col, const unsigned int colIndex);
    void GetRow(vector<string>& row, const unsigned int rowIndex,
      const unsigned int fromColIndex, unsigned int toColIndex);
    const vector<string>& GetRow(const unsigned int rowIndex);
    eOrientation GetOrientation();
    void CreateIndex(const vector<unsigned int>& colIndices,
      const unsigned int unique = 0);
    
    void SetSerializer(Serializer* ser);

    int WriteObject(Serializer* ser, int& size);

    int GetObject(UInt32 index, Serializer* ser);

    int Read(unsigned int indexInFile, Serializer* ser);

    int Write(Serializer* ser, int& size);

    void RebuildIndices();

    void InsertEntry(const unsigned int rowIndex);
    void DeleteEntry(const unsigned int rowIndex);

    inline unsigned int GetNumIndices();

    void GetColumn(vector<string>& col, const unsigned int colIndex,
      const unsigned int indexIndex);

  private:
    // number of digit DBL_MIN_10_EXP, letter e is not included in size
    static const unsigned int EXPONENT      =  4;
    static const unsigned int MAX_PRECISION = DBL_DIG;
    //???DBL_MANT_DIG;
    static const unsigned int MANTISSA       =  MAX_PRECISION + 2;
    static const unsigned int INT_LIMIT      = 11;

    // datatype mask
    static const unsigned char DT_MASK        = 15 << 4;
    // string comparison sensitivity mask
    static const unsigned char SC_MASK        = 0x01;
    // white space sensitivity mask
    static const unsigned char WS_MASK        = 0x02;
    static const unsigned char LAST_DT_VALUE  = 3;
    static const unsigned int  DEFAULT_PRECISION = MAX_PRECISION;
    static const unsigned char DEFAULT_OPTIONS;

    // static const string _version;

    TTable _ttable;

    eOrientation _orient;

    Serializer* _ser;

    vector<unsigned char> _compare_opts;

    vector<vector<unsigned int> > _listsOfColumns;
    vector<unsigned int> _unique;
    vector<tIndex> _indices;

    bool AreListsOfColumnsValid(const vector<unsigned int>& colIndices);
    void CreateKey(const vector<unsigned int>& colIndices);

    void Init();

    StringCompare::eCompareType
      GetCompareType(const vector<unsigned int>& colIndices);

    string CellValue(const unsigned int colIndex,
      const unsigned int rowIndex);
    string ConvertString(const string& value, const unsigned int colIndex);
    string MultiStringsValue(const vector<string>& values,
      const vector<unsigned int>& colIndices);
    string SubRowValue(const vector<unsigned int>& colIndices,
      const unsigned int rowIndex);
    string AggregateRow(const vector<unsigned int>& colIndices,
      const unsigned int rowIndex);

    inline void AppendToAndDelimit(string& to, const string& appending);

    string CreateInternalIndexName(const unsigned int indexIndex);
    void ClearIndex(const unsigned int indexIndex);

    void UpdateIndices(const unsigned int rowIndex);
    void ClearIndices();

    bool IsColumnInIndex(const unsigned int indexIndex,
      const unsigned int colIndex);

    int FindKeyIndex();

    void UpdateColListOnColInsert(const unsigned int colIndex);
    void UpdateColListOnColDelete(const unsigned int colIndex);
    void UpdateIndicesOnCellUpdate(const unsigned int rowIndex,
      const unsigned int colIndex);

    void ConvertToInt(const string& a, string& ret);
    void ConvertDouble(const string& a, string& ret);
    void ConvertToLowerNoWhiteSpace(const string& a, string& ret);

    void Print(unsigned int indexIndex);
};


ostream& operator<<(ostream& out, const ITTable& isTable);


inline unsigned int ITTable::GetLastRowIndex()
{

    return(GetNumRows() - 1);

}


inline unsigned int ITTable::GetNumIndices()
{

    return(_listsOfColumns.size());

}


inline void ITTable::AppendToAndDelimit(string& to, const string& appending)
{

    to += appending;
    // VLAD HARDCODED CONST
    to += " ";

}


inline unsigned int ITTable::GetNumColumns() const
{
    if (_orient == eCOLUMN_WISE)
        return(_ttable.GetNumTuples());
    else
        return(_ttable.GetNumColumns());
}

inline unsigned int ITTable::GetNumRows() const
{
    if (_orient == eCOLUMN_WISE)
        return(_ttable.GetNumColumns());
    else
        return(_ttable.GetNumTuples());
}

#endif // ITTABLE_H

---