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

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();
}

其他回答

我的版本只使用标准c++ 11库。它很好地处理Excel CSV引用:

spam eggs,"foo,bar","""fizz buzz"""
1.23,4.567,-8.00E+09

代码是作为有限状态机编写的,每次只消耗一个字符。我认为这更容易解释。

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

enum class CSVState {
    UnquotedField,
    QuotedField,
    QuotedQuote
};

std::vector<std::string> readCSVRow(const std::string &row) {
    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;
                    default:  // end of quote
                              state = CSVState::UnquotedField;
                              break; }
                break;
        }
    }
    return fields;
}

/// Read CSV file, Excel dialect. Accept "quoted fields ""with quotes"""
std::vector<std::vector<std::string>> readCSV(std::istream &in) {
    std::vector<std::vector<std::string>> table;
    std::string row;
    while (!in.eof()) {
        std::getline(in, row);
        if (in.bad() || in.fail()) {
            break;
        }
        auto fields = readCSVRow(row);
        table.push_back(fields);
    }
    return table;
}

使用Spirit来解析csv并不过分。Spirit非常适合微解析任务。例如,使用Spirit 2.1,它就像:

bool r = phrase_parse(first, last,

    //  Begin grammar
    (
        double_ % ','
    )
    ,
    //  End grammar

    space, v);

向量v被值填满了。在刚刚与Boost 1.41一起发布的新的Spirit 2.1文档中,有一系列教程涉及到这一点。

本教程从简单到复杂。CSV解析器呈现在中间的某个位置,并涉及使用Spirit的各种技术。生成的代码与手写代码一样紧凑。检查生成的汇编程序!

如果您所需要的只是加载一个双精度数据文件(没有整数,没有文本),那么这里有一个随时可用的函数。

#include <sstream>
#include <fstream>
#include <iterator>
#include <string>
#include <vector>
#include <algorithm>

using namespace std;

/**
 * Parse a CSV data file and fill the 2d STL vector "data".
 * Limits: only "pure datas" of doubles, not encapsulated by " and without \n inside.
 * Further no formatting in the data (e.g. scientific notation)
 * It however handles both dots and commas as decimal separators and removes thousand separator.
 * 
 * returnCodes[0]: file access 0-> ok 1-> not able to read; 2-> decimal separator equal to comma separator
 * returnCodes[1]: number of records
 * returnCodes[2]: number of fields. -1 If rows have different field size
 * 
 */
vector<int>
readCsvData (vector <vector <double>>& data, const string& filename, const string& delimiter, const string& decseparator){

 int vv[3] = { 0,0,0 };
 vector<int> returnCodes(&vv[0], &vv[0]+3);

 string rowstring, stringtoken;
 double doubletoken;
 int rowcount=0;
 int fieldcount=0;
 data.clear();

 ifstream iFile(filename, ios_base::in);
 if (!iFile.is_open()){
   returnCodes[0] = 1;
   return returnCodes;
 }
 while (getline(iFile, rowstring)) {
    if (rowstring=="") continue; // empty line
    rowcount ++; //let's start with 1
    if(delimiter == decseparator){
      returnCodes[0] = 2;
      return returnCodes;
    }
    if(decseparator != "."){
     // remove dots (used as thousand separators)
     string::iterator end_pos = remove(rowstring.begin(), rowstring.end(), '.');
     rowstring.erase(end_pos, rowstring.end());
     // replace decimal separator with dots.
     replace(rowstring.begin(), rowstring.end(),decseparator.c_str()[0], '.'); 
    } else {
     // remove commas (used as thousand separators)
     string::iterator end_pos = remove(rowstring.begin(), rowstring.end(), ',');
     rowstring.erase(end_pos, rowstring.end());
    }
    // tokenize..
    vector<double> tokens;
    // Skip delimiters at beginning.
    string::size_type lastPos = rowstring.find_first_not_of(delimiter, 0);
    // Find first "non-delimiter".
    string::size_type pos     = rowstring.find_first_of(delimiter, lastPos);
    while (string::npos != pos || string::npos != lastPos){
        // Found a token, convert it to double add it to the vector.
        stringtoken = rowstring.substr(lastPos, pos - lastPos);
        if (stringtoken == "") {
      tokens.push_back(0.0);
    } else {
          istringstream totalSString(stringtoken);
      totalSString >> doubletoken;
      tokens.push_back(doubletoken);
    }     
        // Skip delimiters.  Note the "not_of"
        lastPos = rowstring.find_first_not_of(delimiter, pos);
        // Find next "non-delimiter"
        pos = rowstring.find_first_of(delimiter, lastPos);
    }
    if(rowcount == 1){
      fieldcount = tokens.size();
      returnCodes[2] = tokens.size();
    } else {
      if ( tokens.size() != fieldcount){
    returnCodes[2] = -1;
      }
    }
    data.push_back(tokens);
 }
 iFile.close();
 returnCodes[1] = rowcount;
 return returnCodes;
}

可以使用std::regex。

根据文件大小和可用内存,可以逐行读取,也可以完全在std::string中读取。

读取文件可以使用:

std::ifstream t("file.txt");
std::string sin((std::istreambuf_iterator<char>(t)),
                 std::istreambuf_iterator<char>());

然后你可以和这个相匹配,它实际上是根据你的需要定制的。

std::regex word_regex(",\\s]+");
auto what = 
    std::sregex_iterator(sin.begin(), sin.end(), word_regex);
auto wend = std::sregex_iterator();

std::vector<std::string> v;
for (;what!=wend ; wend) {
    std::smatch match = *what;
    v.push_back(match.str());
}

如果你确实关心正确解析CSV,这将做它…相对较慢,因为它一次只处理一个字符。

 void ParseCSV(const string& csvSource, vector<vector<string> >& lines)
    {
       bool inQuote(false);
       bool newLine(false);
       string field;
       lines.clear();
       vector<string> line;

       string::const_iterator aChar = csvSource.begin();
       while (aChar != csvSource.end())
       {
          switch (*aChar)
          {
          case '"':
             newLine = false;
             inQuote = !inQuote;
             break;

          case ',':
             newLine = false;
             if (inQuote == true)
             {
                field += *aChar;
             }
             else
             {
                line.push_back(field);
                field.clear();
             }
             break;

          case '\n':
          case '\r':
             if (inQuote == true)
             {
                field += *aChar;
             }
             else
             {
                if (newLine == false)
                {
                   line.push_back(field);
                   lines.push_back(line);
                   field.clear();
                   line.clear();
                   newLine = true;
                }
             }
             break;

          default:
             newLine = false;
             field.push_back(*aChar);
             break;
          }

          aChar++;
       }

       if (field.size())
          line.push_back(field);

       if (line.size())
          lines.push_back(line);
    }