CareerCup

This takes care of all required CSV scenarios doing a "yield return" which returns each line to the caller as needed and taking only space as needed.


The input is a IEnumerable<char> so that the function can parse the CSV data as it been loaded so no need to load the entire file into memory in a string form to start parsing the same goes for the caller we will only load the required data from the file no matter the size.

This provides a minimal memory footprint which will be at most a single line of the CSV file.

The memory footprint could be optimize even further by parsing a token at a time but because lines don't tend to be bigger than I did not bother.

public static IEnuerable<List<string>> ParseCSV(IEnumerable<char> S)
{
	if(string.IsNullOrEmpty(S))
		return null;

	List<string> line = new List<string>();
	StringbBuilder token = new StringBuilder();

	string prev = string.Empty;
	bool quote = false;
	foreach(char s in S)
	{
		if(s == '"')
		{

			// On the case of double quotes it will make this to true as previously
			// though that it was the closing quote previously making it false
			// So it quotes is false and it previously was a quote then this is a
			// double quote scenario
			if(quotes == false && prev == '"')
			{
				token.Append('"');
			}

			quotes = ~quotes;
		}

		if(quotes == true)
		{
			token.Append(s);
		}
		else if(s == ',')
		{
			line.Append(token.ToString());
			token.Clear();
		}
		else if(s == '\n')
		{
			line.Add(token.ToString());
			token.Clear();
			yield return line;

			line = new List<string>();
		}
		else
		{
			token.Append(s);
		}

		prev = s;
	}

	// Adding last item left
	line.Add(token.ToString());
	yield return line;
}

#e.g. input
input = "a, \"b, x\", c, \"q, p, r\", n, t"

elem = ""
for x in input.split('\n'):
    inQuotes = False
    for ch in x:
        if ch not in ',"':
            elem += ch
        elif ch == '"':
            inQuotes = not inQuotes
        elif ch == ',':
            if inQuotes:
                elem += ch
            else:
                print elem
                elem = ""      
    print elem + "\n"
    elem = ""

class CSVParser
  attr_accessor :str, :strt_ptr, :elements, :cur_ptr, :partial_word , :open_quotes
  def initialize(str)
    @str = str.split(//)
    @elements = []
    @strt_ptr = 0
    @cur_ptr = 0
    @partial_word = ""
    @open_quotes = false
  end

  def run
    while(@cur_ptr < @str.size)
      slide([',', '"'])
      case @str[@cur_ptr]
      when '"'
        handle_quote
      when ','
        handle_comma
      else
        @cur_ptr += 1
      end
    end

    handle_comma
  end

  def handle_comma
    if @strt_ptr == @cur_ptr
      raise StandardError.new
    else
      elements << build_word(@strt_ptr, @cur_ptr, @partial_word)
      @partial_word = ""
      @cur_ptr += 1
      @strt_ptr = @cur_ptr
    end
  end

  def handle_quote
    if((str[@cur_ptr] == '"') && (str[@cur_ptr+1] == '"'))
      @partial_word = build_word(@strt_ptr, @cur_ptr , @partial_word)
      @cur_ptr += 2
      @strt_ptr = @cur_ptr
    else
      if @open_quotes
        elements << build_word(@strt_ptr, @cur_ptr, @partial_word)
        @partial_word = ""
        @cur_ptr += 1 #removing comma
      else
        @open_quotes = true
      end
      @cur_ptr += 1
      @strt_ptr = @cur_ptr
    end
  end

  private
  def build_word(s, e , init)
    prev_word = str[s..(e-1)].reduce(:+)
    return init + prev_word
  end

  def slide(delimiters)
    while(@cur_ptr < @str.size)
      delimiters.each do |delim|
        return  if @str[@cur_ptr]==delim
      end
      @cur_ptr += 1
    end
  end
end


csv_parser =  CSVParser.new("abc,\"d\"\"e\"\"f\",efc")
csv_parser.run
puts csv_parser.elements