How to split QString and keep the separator in Qt?

Question

I have a QString: "{x, c | 0x01}", and I want to split it to 7 tokens as below:

{
x
,
c
|
0x01
}

What's the best way to do it in Qt?

I tried to use QString::split(QRegExp("[\\{\\},|]")), but it DOES NOT keep the separator in the result.

I do not find any logic in the separation, what is the characteristic that separates it? — eyllanesc, Jul 25 '17 at 03:14
I think your best bet here would be to iterate through the string and define conditions according to your particular requirement — Abhishek Agarwal, Jul 25 '17 at 03:36
@eyllanesc I want to split it with deliminator char: '{', '}', ',', '|' — ricky, Jul 25 '17 at 11:50
if you want a general explanation I suggest making yourself familiar with a tokenizer or a lexical scanner. If you need a more complete framework there are also DSL language parser (domain specific language), sadly Qt has no support out of the box, but you can look a the c++ boost library for example. — xander, Jul 25 '17 at 12:32

travisjayday · Answer 1 · 2017-07-25T12:57:25.553

1

Basically, you iterate through the string, check if a deliminator is found, and add the deliminator to the list. If no deliminator is found, a new 'word' is added to the list, and until the next deliminator is found, characters will be added to the word. Take a look:

 //input string
QString str = "{x, c | 0x01}";
QList<QString> out;

//flag used to keep track of whether we're adding a mullti-char word, or just a deliminator
bool insideWord = false;

//remove whitespaces
str = str.simplified();
str = str.replace(" ", "");

//iterate through string, check for delims, populate out list
for (int i = 0; i < str.length(); i++)
{
    QChar c = str.at(i);    //get char at current index
    if (c == '{' || c == '}' || c == ',' || c == '|')
    {
        //append deliminator
        out.append(c);
        insideWord = false;
    }
    else
    {
        //append new word to qlist...
        if (!insideWord)
        {
            out.append(c);
            insideWord = true;
        }
        //but if word already started
        else
        {
            //add 'c' to the word in last index of the qlist
            out.last().append(c);
        }
    }
}

//output as requested by OP
qDebug() << "String is" << out;

edited Jul 25 '17 at 12:57

answered Jul 25 '17 at 03:59

travisjayday

784
6
16

Thanks for your answer. Actually, I use a very similar approach as yours. But I'm wondering if there is a better way to improve the performance by avoid calling str.insert() method? – ricky Jul 25 '17 at 11:54
You're right, using str.insert() is quite foolish. Look at my new code, it uses StringBuilder to concatenate to a temporary string. – travisjayday Jul 25 '17 at 12:11
I don't get it, why do you use a temp string in the first place and split it again later if you can simply add the tokens to the `QStringList` directly (since you already have the token!)!? – xander Jul 25 '17 at 12:26
Edit my code again. What do you think? Looks better, doesn't it ;) – travisjayday Jul 25 '17 at 12:47
1

Already better, I guess you can leave that as an example for the OP, it doesn't have to be perfect I guess. :) Btw you can access a reference to the last element in the `QList` like this `out.last().append(c);` (should work and looks better than `out[(out.length()-1)]` I guess). – xander Jul 25 '17 at 12:55

Nikolai Shalakin · Accepted Answer · 2017-07-25T13:22:25.030

Maybe this solution can serve you task:

int main(void) {
    QString str { "{x, c | 0x01}" };
    QRegExp separators { "[\\{\\},|]" };

    QStringList list;
    str.replace( " ", "" );

    int mem = 0;
    for(int i = 0; i<str.size(); ++i) {
        if(i == str.indexOf(separators, i)) {
            if(mem) list.append(str.mid(mem, i-mem)); // append str before separator
            list.append(str.mid(i, 1));               // append separator
            mem = i+1;
        }
    }

    qDebug() << list;

    return 0;
}

Outputs: ("{", "x", ",", "c", "|", "0x01", "}")

You can eliminate if(mem) but then use list.pop_front(); orlist.removeAll(""); after the for cycle, as first element will be a rubbish "".

score -1 · Answer 3 · answered Oct 03 '17 at 11:27

This can be done in a single regular expression, but has to use look-ahead and look-behind.

The expression specified in the question ([\\{\\},|]) will match a 1-character long string consisting of any of the characters {, }, , or |. QString.split will then remove that 1-character long string.

What is needed is to find the zero-character string immediately before each of those separators, using a look-ahead: (?=[\\{\\},|]) and also to find the zero-character string immediately after the separator (?<=[\\{\\},|]).

Combining these gives:

QString::split(QRegularExpression("(?=[\\{\\},|])|(?<=[\\{\\},|])"))

Which will give the desired output: ("{", "x", ",", "c", "|", "0x01", "}")

How to split QString and keep the separator in Qt?

3 Answers3