在C＃中拆分字符串

Question

I thought this will be trivial but I can't get this to work.

Assume a line in a CSV file: "Barack Obama", 48, "President", "1600 Penn Ave, Washington DC"

string[] tokens = line.split(',')

I expect this:

 "Barack Obama"
 48
 "President"
 "1600 Penn Ave, Washington DC"

but the last token is 'Washington DC' not "1600 Penn Ave, Washington DC" .

Is there an easy way to get the split function to ignore the comma within quotes?

I have no control over the CSV file and it doesn;t get sent to me. Customer A will be using the app to read files provided by an external individual.

Answer 1

You might have to write your own split function.

• Iterate through each char in the string
• When you hit a " character, toggle a boolean
• When you hit a comma, if the bool is true, ignore it, else, you have your token

Here's an example:

public static class StringExtensions
{
    public static string[] SplitQuoted(this string input, char separator, char quotechar)
    {
        List<string> tokens = new List<string>();

        StringBuilder sb = new StringBuilder();
        bool escaped = false;
        foreach (char c in input)
        {
            if (c.Equals(separator) && !escaped)
            {
                // we have a token
                tokens.Add(sb.ToString().Trim());
                sb.Clear();
            }
            else if (c.Equals(separator) && escaped)
            {
                // ignore but add to string
                sb.Append(c);
            }
            else if (c.Equals(quotechar))
            {
                escaped = !escaped;
                sb.Append(c);
            }
            else
            {
                sb.Append(c);
            }
        }
        tokens.Add(sb.ToString().Trim());

        return tokens.ToArray();
    }
}

Then just call:

string[] tokens = line.SplitQuoted(',','\"');

---

Benchmarks

Results of benchmarking my code and Dan Tao's code are below. I'm happy to benchmark any other solutions if people want them?

Code:

string input = "\"Barak Obama\", 48, \"President\", \"1600 Penn Ave, Washington DC\""; // Console.ReadLine()
string[] tokens = null;

// run tests
DateTime start = DateTime.Now;
for (int i = 0; i < 1000000; i++)
    tokens = input.SplitWithQualifier(',', '\"', false);
Console.WriteLine("1,000,000 x SplitWithQualifier = {0}ms", DateTime.Now.Subtract(start).TotalMilliseconds);

start = DateTime.Now;
for (int i = 0; i<1000000;i++)
    tokens = input.SplitQuoted(',', '\"');
Console.WriteLine("1,000,000 x SplitQuoted =        {0}ms", DateTime.Now.Subtract(start).TotalMilliseconds);

Output:

1,000,000 x SplitWithQualifier = 8156.25ms
1,000,000 x SplitQuoted =        2406.25ms

Answer 2

I have a SplitWithQualifier extension method that I use here and there, which utilizes Regex .

I make no claim as to the robustness of this code, but it has worked all right for me for a while.

// mangled code horribly to fit without scrolling
public static class CsvSplitter
{
    public static string[] SplitWithQualifier(this string text,
                                              char delimiter,
                                              char qualifier,
                                              bool stripQualifierFromResult)
    {
        string pattern = string.Format(
            @"{0}(?=(?:[^{1}]*{1}[^{1}]*{1})*(?![^{1}]*{1}))",
            Regex.Escape(delimiter.ToString()),
            Regex.Escape(qualifier.ToString())
        );

        string[] split = Regex.Split(text, pattern);

        if (stripQualifierFromResult)
            return split.Select(s => s.Trim().Trim(qualifier)).ToArray();
        else
            return split;
    }
}

Usage:

string csv = "\"Barak Obama\", 48, \"President\", \"1600 Penn Ave, Washington DC\"";
string[] values = csv.SplitWithQualifier(',', '\"', true);

foreach (string value in values)
    Console.WriteLine(value);

Output:

Barak Obama
48
President
1600 Penn Ave, Washington DC

Answer 3

I see from the bigger picture that you are actually trying to parse CSV input. So instead of advising on how to split the string properly, I would instead recommend you to use a CSV parser to do this kind of thing.

A Fast CSV Reader

One that I would recommend is the library (source code available) that you can get from this CodeProject page: http://www.codeproject.com/KB/database/CsvReader.aspx

I personally use it myself and like it. It's a .NET native code and a lot faster than using OLEDB (which also can do the CSV parsing for you, but believe me, it's slow).

Answer 4

You should be using Microsoft.VisualBasic.FileIO.TextFieldParser for that. It will handle all the CSV stuff correctly for you, see: A similar question with example using the TextFieldParser

PS: Do not fear using the Microsoft.VisualBasic dll in a C# project, it's all .NET :-)

Answer 5

That would be the expected behavior as quotes are just another string character in C#. Looks like what you are after is the quoted tokens or numeric tokens.

I think you might need to use Regex to split the strings unless some one else knows a better way.

Or you could just loop through the string one character at a time building up the string as you go and build the tokens that way. It's old school but may be the most reliable way in your case.

Answer 6

You can't parse a CSV line with a simple Split on commas, because some cell contents will contain commas that aren't meant to delineate data but are actually part of the cell contents themselves.

Here is a link to a simple regex-based C# method that will convert your CSV into a handly DataTable :

http://www.hotblue.com/article0000.aspx?a=0006

Working with DataTables is very easy - let me know if you need a code sample for that.

Answer 7

I would recommend using a regular expression instead. It will allow you to extract more complicated substrings in a much more versatile manner (precisely as you want).

http://www.c-sharpcorner.com/uploadfile/prasad_1/regexppsd12062005021717am/regexppsd.aspx

http://oreilly.com/windows/archive/csharp-regular-expressions.html

Answer 8

Can't you change how the CSV is generated? Using OpenOffice, you can set the char separator (use ;) and how the string is delimited (using " or ').

It would be like this: 'President';'1600 Penn Ave, Washington DC'

Answer 9

string temp = line.Replace( "\\"", "" );

string[] tokens = temp.Split(',')