dsv.js

function objectConverter(columns) {
  return new Function("d", "return {" + columns.map(function(name, i) {
    return JSON.stringify(name) + ": d[" + i + "]";
  }).join(",") + "}");
}

function customConverter(columns, f) {
  var object = objectConverter(columns);
  return function(row, i) {
    return f(object(row), i, columns);
  };
}

// Compute unique columns in order of discovery.
function inferColumns(rows) {
  var columnSet = Object.create(null),
      columns = [];

  rows.forEach(function(row) {
    for (var column in row) {
      if (!(column in columnSet)) {
        columns.push(columnSet[column] = column);
      }
    }
  });

  return columns;
}

export default function(delimiter) {
  var reFormat = new RegExp("[\"" + delimiter + "\n\r]"),
      delimiterCode = delimiter.charCodeAt(0);

  function parse(text, f) {
    var convert, columns, rows = parseRows(text, function(row, i) {
      if (convert) return convert(row, i - 1);
      columns = row, convert = f ? customConverter(row, f) : objectConverter(row);
    });
    rows.columns = columns;
    return rows;
  }

  function parseRows(text, f) {
    var EOL = {}, // sentinel value for end-of-line
        EOF = {}, // sentinel value for end-of-file
        rows = [], // output rows
        N = text.length,
        I = 0, // current character index
        n = 0, // the current line number
        t, // the current token
        eol; // is the current token followed by EOL?

    function token() {
      if (I >= N) return EOF; // special case: end of file
      if (eol) return eol = false, EOL; // special case: end of line

      // special case: quotes
      var j = I, c;
      if (text.charCodeAt(j) === 34) {
        var i = j;
        while (i++ < N) {
          if (text.charCodeAt(i) === 34) {
            if (text.charCodeAt(i + 1) !== 34) break;
            ++i;
          }
        }
        I = i + 2;
        c = text.charCodeAt(i + 1);
        if (c === 13) {
          eol = true;
          if (text.charCodeAt(i + 2) === 10) ++I;
        } else if (c === 10) {
          eol = true;
        }
        return text.slice(j + 1, i).replace(/""/g, "\"");
      }

      // common case: find next delimiter or newline
      while (I < N) {
        var k = 1;
        c = text.charCodeAt(I++);
        if (c === 10) eol = true; // \n
        else if (c === 13) { eol = true; if (text.charCodeAt(I) === 10) ++I, ++k; } // \r|\r\n
        else if (c !== delimiterCode) continue;
        return text.slice(j, I - k);
      }

      // special case: last token before EOF
      return text.slice(j);
    }

    while ((t = token()) !== EOF) {
      var a = [];
      while (t !== EOL && t !== EOF) {
        a.push(t);
        t = token();
      }
      if (f && (a = f(a, n++)) == null) continue;
      rows.push(a);
    }

    return rows;
  }

  function format(rows, columns) {
    if (columns == null) columns = inferColumns(rows);
    return [columns.map(formatValue).join(delimiter)].concat(rows.map(function(row) {
      return columns.map(function(column) {
        return formatValue(row[column]);
      }).join(delimiter);
    })).join("\n");
  }

  function formatRows(rows) {
    return rows.map(formatRow).join("\n");
  }

  function formatRow(row) {
    return row.map(formatValue).join(delimiter);
  }

  function formatValue(text) {
    return text == null ? ""
        : reFormat.test(text += "") ? "\"" + text.replace(/\"/g, "\"\"") + "\""
        : text;
  }

  return {
    parse: parse,
    parseRows: parseRows,
    format: format,
    formatRows: formatRows
  };
}