我需要在c++中加载和使用CSV文件数据。在这一点上,它实际上只是一个以逗号分隔的解析器(即不用担心转义新行和逗号)。主要需要的是逐行解析器,它将在每次调用方法时为下一行返回一个向量。

我发现这篇文章看起来很有前途: http://www.boost.org/doc/libs/1_35_0/libs/spirit/example/fundamental/list_parser.cpp

我从未使用过Boost's Spirit,但我愿意尝试一下。但前提是我忽略了一个更直接的解决方案。


当前回答

下面是Unicode CSV解析器的另一个实现(使用wchar_t)。我写了一部分,乔纳森·莱弗勒写了剩下的部分。

注意:此解析器旨在尽可能地复制Excel的行为,特别是在导入损坏或格式错误的CSV文件时。

这是最初的问题-用多行字段和转义双引号解析CSV文件

这是作为SSCCE(简短,自包含,正确示例)的代码。

#include <stdbool.h>
#include <wchar.h>
#include <wctype.h>

extern const wchar_t *nextCsvField(const wchar_t *p, wchar_t sep, bool *newline);

// Returns a pointer to the start of the next field,
// or zero if this is the last field in the CSV
// p is the start position of the field
// sep is the separator used, i.e. comma or semicolon
// newline says whether the field ends with a newline or with a comma
const wchar_t *nextCsvField(const wchar_t *p, wchar_t sep, bool *newline)
{
    // Parse quoted sequences
    if ('"' == p[0]) {
        p++;
        while (1) {
            // Find next double-quote
            p = wcschr(p, L'"');
            // If we don't find it or it's the last symbol
            // then this is the last field
            if (!p || !p[1])
                return 0;
            // Check for "", it is an escaped double-quote
            if (p[1] != '"')
                break;
            // Skip the escaped double-quote
            p += 2;
        }
    }

    // Find next newline or comma.
    wchar_t newline_or_sep[4] = L"\n\r ";
    newline_or_sep[2] = sep;
    p = wcspbrk(p, newline_or_sep);

    // If no newline or separator, this is the last field.
    if (!p)
        return 0;

    // Check if we had newline.
    *newline = (p[0] == '\r' || p[0] == '\n');

    // Handle "\r\n", otherwise just increment
    if (p[0] == '\r' && p[1] == '\n')
        p += 2;
    else
        p++;

    return p;
}

static wchar_t *csvFieldData(const wchar_t *fld_s, const wchar_t *fld_e, wchar_t *buffer, size_t buflen)
{
    wchar_t *dst = buffer;
    wchar_t *end = buffer + buflen - 1;
    const wchar_t *src = fld_s;

    if (*src == L'"')
    {
        const wchar_t *p = src + 1;
        while (p < fld_e && dst < end)
        {
            if (p[0] == L'"' && p+1 < fld_s && p[1] == L'"')
            {
                *dst++ = p[0];
                p += 2;
            }
            else if (p[0] == L'"')
            {
                p++;
                break;
            }
            else
                *dst++ = *p++;
        }
        src = p;
    }
    while (src < fld_e && dst < end)
        *dst++ = *src++;
    if (dst >= end)
        return 0;
    *dst = L'\0';
    return(buffer);
}

static void dissect(const wchar_t *line)
{
    const wchar_t *start = line;
    const wchar_t *next;
    bool     eol;
    wprintf(L"Input %3zd: [%.*ls]\n", wcslen(line), wcslen(line)-1, line);
    while ((next = nextCsvField(start, L',', &eol)) != 0)
    {
        wchar_t buffer[1024];
        wprintf(L"Raw Field: [%.*ls] (eol = %d)\n", (next - start - eol), start, eol);
        if (csvFieldData(start, next-1, buffer, sizeof(buffer)/sizeof(buffer[0])) != 0)
            wprintf(L"Field %3zd: [%ls]\n", wcslen(buffer), buffer);
        start = next;
    }
}

static const wchar_t multiline[] =
   L"First field of first row,\"This field is multiline\n"
    "\n"
    "but that's OK because it's enclosed in double quotes, and this\n"
    "is an escaped \"\" double quote\" but this one \"\" is not\n"
    "   \"This is second field of second row, but it is not multiline\n"
    "   because it doesn't start \n"
    "   with an immediate double quote\"\n"
    ;

int main(void)
{
    wchar_t line[1024];

    while (fgetws(line, sizeof(line)/sizeof(line[0]), stdin))
        dissect(line);
    dissect(multiline);

    return 0;
}

其他回答

不管怎样,下面是我的实现。它处理wstring输入,但是可以很容易地调整为string。它不处理字段中的换行符(因为我的应用程序也不这样做,但添加它的支持并不太难),它不符合RFC中的“\r\n”行尾(假设您使用std::getline),但它确实正确地处理空格修剪和双引号(希望如此)。

using namespace std;

// trim whitespaces around field or double-quotes, remove double-quotes and replace escaped double-quotes (double double-quotes)
wstring trimquote(const wstring& str, const wstring& whitespace, const wchar_t quotChar)
{
    wstring ws;
    wstring::size_type strBegin = str.find_first_not_of(whitespace);
    if (strBegin == wstring::npos)
        return L"";

    wstring::size_type strEnd = str.find_last_not_of(whitespace);
    wstring::size_type strRange = strEnd - strBegin + 1;

    if((str[strBegin] == quotChar) && (str[strEnd] == quotChar))
    {
        ws = str.substr(strBegin+1, strRange-2);
        strBegin = 0;
        while((strEnd = ws.find(quotChar, strBegin)) != wstring::npos)
        {
            ws.erase(strEnd, 1);
            strBegin = strEnd+1;
        }

    }
    else
        ws = str.substr(strBegin, strRange);
    return ws;
}

pair<unsigned, unsigned> nextCSVQuotePair(const wstring& line, const wchar_t quotChar, unsigned ofs = 0)
{
    pair<unsigned, unsigned> r;
    r.first = line.find(quotChar, ofs);
    r.second = wstring::npos;
    if(r.first != wstring::npos)
    {
        r.second = r.first;
        while(((r.second = line.find(quotChar, r.second+1)) != wstring::npos)
            && (line[r.second+1] == quotChar)) // WARNING: assumes null-terminated string such that line[r.second+1] always exist
            r.second++;

    }
    return r;
}

unsigned parseLine(vector<wstring>& fields, const wstring& line)
{
    unsigned ofs, ofs0, np;
    const wchar_t delim = L',';
    const wstring whitespace = L" \t\xa0\x3000\x2000\x2001\x2002\x2003\x2004\x2005\x2006\x2007\x2008\x2009\x200a\x202f\x205f";
    const wchar_t quotChar = L'\"';
    pair<unsigned, unsigned> quot;

    fields.clear();

    ofs = ofs0 = 0;
    quot = nextCSVQuotePair(line, quotChar);
    while((np = line.find(delim, ofs)) != wstring::npos)
    {
        if((np > quot.first) && (np < quot.second))
        { // skip delimiter inside quoted field
            ofs = quot.second+1;
            quot = nextCSVQuotePair(line, quotChar, ofs);
            continue;
        }
        fields.push_back( trimquote(line.substr(ofs0, np-ofs0), whitespace, quotChar) );
        ofs = ofs0 = np+1;
    }
    fields.push_back( trimquote(line.substr(ofs0), whitespace, quotChar) );

    return fields.size();
}

你可以在escaped_list_separator中使用Boost Tokenizer。

Escaped_list_separator解析csv的超集。Boost::记号赋予器

这只使用Boost标记器头文件,不需要链接到Boost库。

下面是一个例子,(详情请参阅c++中使用Boost Tokenizer解析CSV文件或Boost:: Tokenizer):

#include <iostream>     // cout, endl
#include <fstream>      // fstream
#include <vector>
#include <string>
#include <algorithm>    // copy
#include <iterator>     // ostream_operator
#include <boost/tokenizer.hpp>

int main()
{
    using namespace std;
    using namespace boost;
    string data("data.csv");

    ifstream in(data.c_str());
    if (!in.is_open()) return 1;

    typedef tokenizer< escaped_list_separator<char> > Tokenizer;
    vector< string > vec;
    string line;

    while (getline(in,line))
    {
        Tokenizer tok(line);
        vec.assign(tok.begin(),tok.end());

        // vector now contains strings from one row, output to cout here
        copy(vec.begin(), vec.end(), ostream_iterator<string>(cout, "|"));

        cout << "\n----------------------" << endl;
    }
}

@sastanin的解决方案的一个小版本,以便它可以处理引号中的换行。

std::vector<std::vector<std::string>> readCSV(std::istream &in) {
    std::vector<std::vector<std::string>> table;

    while (!in.eof()) {
        CSVState state = CSVState::UnquotedField;
        std::vector<std::string> fields {""};
        size_t i = 0; // index of the current field
        for (char c : row) {
            switch (state) {
                case CSVState::UnquotedField:
                    switch (c) {
                        case ',': // end of field
                                  fields.push_back(""); i++;
                                  break;
                        case '"': state = CSVState::QuotedField;
                                  break;
                        default:  fields[i].push_back(c);
                                  break; }
                    break;
                case CSVState::QuotedField:
                    switch (c) {
                        case '"': state = CSVState::QuotedQuote;
                                  break;
                        default:  fields[i].push_back(c);
                                  break; }
                    break;
                case CSVState::QuotedQuote:
                    switch (c) {
                        case ',': // , after closing quote
                                  fields.push_back(""); i++;
                                  state = CSVState::UnquotedField;
                                  break;
                        case '"': // "" -> "
                                  fields[i].push_back('"');
                                  state = CSVState::QuotedField;
                                  break;
                        case '\n': // newline
                                  table.push_back(fields);
                                  state = CSVState::UnquotedField;
                                  fields = vector<string>{""};
                                  i = 0;
                        default:  // end of quote
                                  state = CSVState::UnquotedField;
                                  break; }
                    break;
            }
        }
    }
    return table;
}

如果您正在使用Visual Studio / MFC,下面的解决方案可能会使您的工作更轻松。它支持Unicode和MBCS,有注释,除了CString之外没有其他依赖项,对我来说工作得很好。它不支持在带引号的字符串中嵌入换行符,但我不在乎,只要它在这种情况下不崩溃,它不会崩溃。

总体策略是,将带引号的字符串和空字符串作为特殊情况处理,其余使用Tokenize。对于带引号的字符串,策略是找到真正的结束引号,跟踪是否遇到了连续的引号对。如果是,则使用Replace将成对转换为单个。毫无疑问,有更有效的方法,但在我的案例中,性能还不够重要,不足以证明进一步优化的合理性。

class CParseCSV {
public:
// Construction
    CParseCSV(const CString& sLine);

// Attributes
    bool    GetString(CString& sDest);

protected:
    CString m_sLine;    // line to extract tokens from
    int     m_nLen;     // line length in characters
    int     m_iPos;     // index of current position
};

CParseCSV::CParseCSV(const CString& sLine) : m_sLine(sLine)
{
    m_nLen = m_sLine.GetLength();
    m_iPos = 0;
}

bool CParseCSV::GetString(CString& sDest)
{
    if (m_iPos < 0 || m_iPos > m_nLen)  // if position out of range
        return false;
    if (m_iPos == m_nLen) { // if at end of string
        sDest.Empty();  // return empty token
        m_iPos = -1;    // really done now
        return true;
    }
    if (m_sLine[m_iPos] == '\"') {  // if current char is double quote
        m_iPos++;   // advance to next char
        int iTokenStart = m_iPos;
        bool    bHasEmbeddedQuotes = false;
        while (m_iPos < m_nLen) {   // while more chars to parse
            if (m_sLine[m_iPos] == '\"') {  // if current char is double quote
                // if next char exists and is also double quote
                if (m_iPos < m_nLen - 1 && m_sLine[m_iPos + 1] == '\"') {
                    // found pair of consecutive double quotes
                    bHasEmbeddedQuotes = true;  // request conversion
                    m_iPos++;   // skip first quote in pair
                } else  // next char doesn't exist or is normal
                    break;  // found closing quote; exit loop
            }
            m_iPos++;   // advance to next char
        }
        sDest = m_sLine.Mid(iTokenStart, m_iPos - iTokenStart);
        if (bHasEmbeddedQuotes) // if string contains embedded quote pairs
            sDest.Replace(_T("\"\""), _T("\""));    // convert pairs to singles
        m_iPos += 2;    // skip closing quote and trailing delimiter if any
    } else if (m_sLine[m_iPos] == ',') {    // else if char is comma
        sDest.Empty();  // return empty token
        m_iPos++;   // advance to next char
    } else {    // else get next comma-delimited token
        sDest = m_sLine.Tokenize(_T(","), m_iPos);
    }
    return true;
}

// calling code should look something like this:

    CStdioFile  fIn(pszPath, CFile::modeRead);
    CString sLine, sToken;
    while (fIn.ReadString(sLine)) { // for each line of input file
        if (!sLine.IsEmpty()) { // ignore blank lines
            CParseCSV   csv(sLine);
            while (csv.GetString(sToken)) {
                // do something with sToken here
            }
        }
    }

如果你不想在你的项目中包含boost(如果你只打算使用它来进行CSV解析,它就相当大了……)

我在这里有幸使用CSV解析:

http://www.zedwood.com/article/112/cpp-csv-parser

它处理带引号的字段-但不处理内联\n字符(这可能对大多数用途都很好)。