How to split a CSV input stream in java

Question

I am writing a java application which reads CSV from the standard input. However, I found that I have some troubles in deal with the double quotes.

For example, if i read in a text:

"He said, ""What?"""

the output gives me:

field[0] = `He said, What?"""'

The last two quotes are what I don't want.

Here is my code:

public class Csv{
private BufferedReader fin;
private String fieldsep;
private ArrayList field;

public Csv(){
    this(System.in, ",");
}


public Csv(InputStream in, String sep){
    this.fin = new BufferedReader(new InputStreamReader(in));
    this.fieldsep = sep;
}


// getline: get one line, grow as needed
public String getline() throws IOException {
    String line;

    line = fin.readLine();
    if (line == null)
        return null;

    field = split(line, fieldsep);

    return line;
}

// split: split line into fields
private static ArrayList split(String line, String sep){
    ArrayList list = new ArrayList();
    int i, j;

    if (line.length() == 0)
        return list;

    i = 0;
    do {
        if (i < line.length() && line.charAt(i) == '"') {
            StringBuffer field = new StringBuffer();
            j = advquoted(line, ++i, sep, field);
            list.add(field.toString());
        } 

        else {
            j = line.indexOf(sep, i);
            if (j == -1)
                j = line.length();
            list.add(line.substring(i, j));
        }
        i = j + sep.length();
    } while (j < line.length());

    return list;
}

// advquoted: quoted field; return index of next separator
private static int advquoted(String s, int i, String sep, StringBuffer field){
    field.setLength(0);
    for ( ; i < s.length(); i++) {
        if (s.charAt(i) == '"' && ++i < s.length() && s.charAt(++i) != '"') {
            int j = s.indexOf(sep, i);
            if (j == -1)
                j = s.length();
            field.append(s.substring(i, j));
            i = j;
            break;
        }
        field.append(s.charAt(i));
    }

    return i;
}

I am not sure but to me it's very hard to get through your code. It's seems you have solved the problem very awkward. To get your values from your CSV you can use for example StringTokenizer see https://docs.oracle.com/javase/7/docs/api/java/util/StringTokenizer.html — Paul Wasilewski, May 23 '16 at 04:30
`The last two commas are what I don't want.` ... do you mean the last two _quotes_ are not what you want? — Tim Biegeleisen, May 23 '16 at 04:35
Are you aware that there are good CSV libraries already available? Is there a reason you need to write your own? — Jim Garrison, May 23 '16 at 04:52
Yeh.it should be quotes. this is part of my homework. I just trying to make my own cvs class. — RunningPig, May 23 '16 at 05:15

Bohemian · Answer 1 · 2016-05-23T10:54:56.520

1

Regex and streams to the rescue. You only need one line for the whole thing:

String[] terms = Arrays.stream(csv.split(",(?=(([^\"]*\"){2})*[^\"]*$"))
  .map(s -> s.replace("\"\"", "").toArray(String[]::new);

edited May 23 '16 at 10:54

answered May 23 '16 at 08:46

Bohemian

412,405
93
575
722

score 0 · Answer 2 · answered May 23 '16 at 05:00

0

You have made it very complex simply use StringTokenizer

String testString = "He said, \"\"What?\"\"";
        StringTokenizer st = new StringTokenizer(testString);
         while (st.hasMoreTokens()) {
             System.out.println(st.nextToken());
         }

Output:

He
said,
""What?""

Now you can play with these strings.

answered May 23 '16 at 05:00

Sanjit Kumar Mishra

1,153
13
32

Emm, actually i am trying to make csv class. – RunningPig May 23 '16 at 05:13

score 0 · Answer 3 · answered May 23 '16 at 06:10

As others have suggested, you can make your life easier by using StringTokenizer. The delimiters should be the comma and the double quote, and you want the StringTokenizer to return the delimiters to you. When the delimiter is a comma, the field will be everything up to the next comma. When the delimiter is the ", the field will be everything up to the next ". You may want to trim the fields and remove leading and trailing " from them.

score -1 · Accepted Answer · edited May 23 '17 at 12:33

I guess this should be fine.

public class Csv{

   private BufferedReader fin;
   private String fieldsep;
   private ArrayList<String> field;

   public Csv(){
      this(System.in, ",");
   }


   public Csv(InputStream in, String sep){
      this.fin = new BufferedReader(new InputStreamReader(in));
      this.fieldsep = sep;
   }

   // getline: get one line, grow as needed
   public String getline() throws IOException {
      String line;
      line = fin.readLine();
      if (line == null)
        return null;
      field = split(line, fieldsep);
      return line;
   }

   // split: split line into fields
   private ArrayList split(String line, String sep){
      List<String> list = new ArrayList();
      StringTokenizer tokens = new StringTokenizer(line, sep, false);
      while (tokens.hasMoreElements()) {
            String next = (String) tokens.nextElement();
            next = next.trim().replaceAll("\"\"", "\"");
            list.add(next);
      }
      return list;
   }
}

The result is field = [He said, "What?"].

You should consider that a field in CSV can be enclosed by double qoutes. I don't know if this causes the multiple qoutes of the field "What?" but if so you should know that all fields then should be enclosed by double quotes. For more information about the CSV format see https://www.ietf.org/rfc/rfc4180.txt.

And be aware you never close your input stream! This could cause a memory leak. For more information see Closing BufferedReader and InputStreamReader.

Actually I am making a class. So, can you modify my method. Because I have to follow some instructions. — RunningPig, May 23 '16 at 06:32
@RunningPig I have upated my answer and included my solution in your class. — Paul Wasilewski, May 23 '16 at 06:51

How to split a CSV input stream in java

4 Answers4