CsvReader.cpp

#include "CsvReader.h"

#include <stdexcept>

using namespace std;

CsvReader::CsvReader(const char *filePath, size_t bufSize) :
        mBufSize(bufSize),
        mBuf(new char[bufSize]) {
    mFile = fopen(filePath, "rt");
    if(!mFile) {
        throw invalid_argument(string("can't open file ") + filePath);
    }
}

CsvReader::~CsvReader() {
    delete []mBuf;
    if(mFile) {
        fclose(mFile);
    }
}

std::vector<std::vector<std::string>> &CsvReader::parse() {
    size_t readRes = 0;
    do {
        readRes = fread(mBuf, 1, mBufSize, mFile);
        parseNextBuf(mBuf, readRes);
    } while(readRes == mBufSize);
    finish();
    return mRes;
}

void CsvReader::parseNextBuf(const char *buf, size_t len) {
    const char *aChar = buf;
    while (aChar != buf + len) {
        switch (*aChar) {
        case '"':
            if(mLastCharWasQuote) {  // Double quote is just one quote
                mFieldInProgress.push_back(*aChar);
                mLastCharWasQuote = false;
            } else {
                mLastCharWasQuote = true;
            }
            mNewLine = false;
            mInQuote = !mInQuote;
            break;

        case ',':
            mNewLine = false;
            if (mInQuote) {
                mFieldInProgress += *aChar;
            } else {
                mLineInProgress.push_back(std::move(mFieldInProgress));
                mFieldInProgress.clear();
            }
            break;

        case '\n':
        case '\r':
            if (mInQuote) {
                mFieldInProgress += *aChar;
            } else {
                if (!mNewLine) {
                    mLineInProgress.push_back(std::move(mFieldInProgress));
                    mRes.push_back(std::move(mLineInProgress));
                    mFieldInProgress.clear();
                    mLineInProgress.clear();
                    mNewLine = true;
                }
            }
            break;

        default:
            mNewLine = false;
            mFieldInProgress.push_back(*aChar);
            break;
        }

        if(*aChar != '"') {
            mLastCharWasQuote = false;
        }
        ++aChar;
    }
}

void CsvReader::finish() {
    if (!mFieldInProgress.empty())
        mLineInProgress.push_back(std::move(mFieldInProgress));

    if (!mLineInProgress.empty())
        mRes.push_back(std::move(mLineInProgress));
}

CsvReader.h

#pragma once

#include <stdio.h>

#include <vector>
#include <string>

class CsvReader {
public:
    CsvReader(const char *filePath, size_t bufSize = 8192);
    ~CsvReader();

    std::vector<std::vector<std::string>> &parse();
private:
    void parseNextBuf(const char *buf, size_t len);
    void finish();

    size_t mBufSize = 0;
    char *mBuf = nullptr;
    FILE *mFile = nullptr;

    std::string mFieldInProgress;
    std::vector<std::string> mLineInProgress;
    std::vector<std::vector<std::string>> mRes;
    bool mInQuote = false;
    bool mLastCharWasQuote = false;
    bool mNewLine = false;
};