haader/csv_reader.h

#ifndef CSVREADER_H
#define CSVREADER_H

#include <string>
#include <vector>
#include <istream>


// A Field in a CSVRow
class CSVField {
public:
    CSVField(int start, int length, bool hasQuotes);

    // byte offset
    int start;

    // length in bytes
    int length;

    // contains quotes (use CSVRow::unquote)
    bool hasQuotes;
};

// A Row in a CSV Document
class CSVRow {
public:
    CSVRow(std::string *row,
           const std::string * const filePath=0,
           long lineNumber=-1,
           char delimiter=',',
           char quote='"');

    // Return a human readable representation of a CSVRow.
    std::string toString() const;

    // Get the value of a field.
    // fieldIndex starts at 0.
    // Returns false if fieldIndex is out of bounds or
    // field could not be converted to the specified type.
    bool getFieldAsString(unsigned int fieldIndex, std::string *out) const;
    bool getFieldAsLong(unsigned int fieldIndex, long *out) const;
    bool getFieldAsInt(unsigned int fieldIndex, int *out) const;
    bool getFieldAsDouble(unsigned int fieldIndex, double *out) const;

    std::string getFilePath() const;

    // Get the line number
    // starting at 1 for the first line.
    // Return a value < 0 if there is no line number specified.
    long getLineNumber() const;

    unsigned int getNumberOfFields() const;

    // Get the original string that was quoted using some of the rules from RFC 4180
    // Examples: unquote("'abc'", "'") == "abc"
    //           unquote("'ab''c'", "'") == "ab'c"
    // Fields containing line breaks are not supported.
    static std::string unquote(const std::string &fieldStr, char quoteChar);

    // Quotes a string using some of the rules from RFC 4180
    // Example: quote("ab'c", "'") == "ab''c"
    // Fields containing line breaks are not supported.
    static std::string quote(const std::string &fieldStr, char quoteChar);

private:
    // fill m_fields
    void parse(char delimiter, char quote);

    // raw bytes from CSV document
    std::string *m_row;

    // each field
    std::vector<CSVField> m_fields;

    // file path of the CSV document or NULL
    const std::string * const m_filePath;

    long m_lineNumber;

    // quote character
    char m_quote;
};


// static functions to parse CSV data
class CSVReader {
public:

    // callback function signature
    typedef bool (*RowCallback) (const CSVRow &row, void *userData);

    // Parse a CSV document given by filePath.
    // Invokes callback function for each row with the supplied userData.
    // Stop parsing if callback returns false.
    // filePath: pointer to a file path or NULL if there is no file specified
    // Returns false if file cannot be opened.
    static bool readFromFile(const std::string filePath,
                             RowCallback callback,
                             void *userData,
                             char delimiter=',',
                             char quote='"');

    // Parse CSV data from stream.
    // Could be a std::stringstream or a std::fstream, etc.
    // Invokes callback function for each row with the supplied userData.
    // Stop parsing and return false if callback returns false.
    static bool readFromStream(std::istream &stream,
                               RowCallback callback,
                               void *userData,
                               const std::string * const filePath=0,
                               char delimiter=',',
                               char quote='"');
};

#endif //CSVREADER_H