Splitting a string with regex, ignoring delimiters that occur within braces

Question

Suppose I have a string

Max and Bob and Merry and {Jack and Co.} and Lisa.

I need to split it with and being the delimiter, but only if it does not occur within curly braces.

So from the above string I should get 5 strings:
Max, Bob, Merry, Jack and Co., Lisa.

I tried something like this pattern:

[^\\\{.+]\\band\\b[^.+\\\}]

But it doesn't work - Jack and Co. are still split as well (I use C++ so I have to escape special characters twice).

Are lookaheads supported by the qregexp? If so, try [`\\band\\b(?![^{]*})`](https://regex101.com/r/0Fd1tF/1) might need more escaping. — bobble bubble, Dec 11 '16 at 13:21
In C++, you can use a *raw string literal* for regular expressions, enclosed by `R"(` and `)"`. This way, backslashes can be used directly, i.e. `R"(\d*)"` — Meyer, Dec 11 '16 at 13:29
You want to split with too many conditions, that sounds like matching with 2 steps can prove better: 1) extract what is inside braces with `QRegExp("\\{([^{}]*)\\}")` and 2) split with `"\\{[^{}]*\\}|\\s*\\band\\b\\s*"` — Wiktor Stribiżew, Dec 11 '16 at 13:47
bobble bubble, thanks, that seems to work exactly as expected. (Yes, lookaheads are supported in QRegExp, and QRegularExpression supports lookbehinds as well). — Maximko, Dec 11 '16 at 14:03

score 2 · Accepted Answer · answered Dec 11 '16 at 14:39

If lookaheads are supported by the QRegExp you can check if inside braces by looking ahead at the final word boundary if there is a closing } with no opening { in between.

\band\b(?![^{]*})

See this demo at regex101

Need to be escaped as desired or try the raw string literal like @SMeyer commented.

score 1 · Answer 2 · edited May 23 '17 at 12:13

Here is a possible solution, partially based on the comment by bobble-bubble. It will produce the five strings as requested, without surrounding whitespace or curly brackets.

std::string text = "Max and Bob and Merry and {Jack and Co.} and Lisa";
std::regex re(R"(\}? +and +(?![^{]*\})\{?)");

std::sregex_token_iterator it(text.begin(), text.end(), re, -1);
std::sregex_token_iterator end;

while (it != end)
    std::cout << *it++ << std::endl;

I tried to keep it simple, you might want to replace the spaces around and with full whitespace detection. An interactive version is available here.

score 0 · Answer 3 · answered Dec 11 '16 at 13:49

0

Let the {...} part match first. That is, put it on the left side of |.

\{.*?\}|and

That will match {foo and bar} if possible, but if not then it will try to match and.

answered Dec 11 '16 at 13:49

Waxrat

2,075
15
13

This will split on `{...}`, removing it. – Nicolas Dec 11 '16 at 13:51
Yes. So you'll likely want to use a split function that returns what was matched. – Waxrat Dec 11 '16 at 13:57
Don't know what language we're talking about, but in Python you can get the delimiters returned by using parens in the regex: `m=re.split(r'(\{.*?\}|and)',s)` – Waxrat Dec 11 '16 at 14:07
Even in that case it will match the brackets too instead of only what's inside. – Nicolas Dec 11 '16 at 14:13
True. Are you saying you think that's a problem? – Waxrat Dec 11 '16 at 14:18

Splitting a string with regex, ignoring delimiters that occur within braces

3 Answers3