How to search for new style C++11 for loops in an existing codebase

Question

I'm using the RegexBuddy utility, with the regex

 ^\s*for\s.*:(?!:).*$

Which I would paraphrase as 'search lines with optional leading white-space, the text 'for', a single white space, then the rest of the line containing a colon, but not two colons together'.

Which mostly works, but the two colon exclusion doesn't (they are typically iterator references).

What's the correct regex for this search?

What if there is no space between the `for` and the opening parentheses (i.e. something like `for(x:y)`)? What if there is multiple spaces? — Some programmer dude, Sep 01 '16 at 08:40
You've already got a working answer from @WiktorStribiżew. But just to explain why your doesn't work - the *first* `:` in your regex can also match the second in your input, making the second `:` of the regex **not** match the character after the second in the input. ` for abc::def` matches (not) `:d`. — SamWhan, Sep 01 '16 at 08:57
In theory, [can C++ code be parsed with regular expressions](http://stackoverflow.com/questions/14589346/is-c-context-free-or-context-sensitive)? Can it be partially parsed (e.g. so only `for` loops should be found)? — Ivan Aksamentov - Drop, Sep 01 '16 at 09:04

rustyx · Accepted Answer · 2016-09-01T09:03:53.097

1

The for statement does not have to be preceded by a space, nor is a space required between for and the opening brace. But OK, here's my attempt:

^\s*for\s*$[^)]*[^:]:[^:][^)]*$.*

Note: In general, it is all but impossible to accurately parse C++ code with a regex. You need a stateful parser to filter out nested comments, etc.

edited Sep 01 '16 at 09:03

answered Sep 01 '16 at 08:47

rustyx

80,671
25
200
267

And whats with the literal `()`? – SamWhan Sep 01 '16 at 08:49
This is for parsing C++11-style for-loops only (the ones that contain a colon) – rustyx Sep 01 '16 at 08:51
That *may* be correct, but you're making assumptions and ignoring criteria given in the question, like "the text 'for', **a single white space**" – SamWhan Sep 01 '16 at 09:01
Your answer is better than @WiktorStribiżew as it picked up the one case that had both a colon and a double colon in the same line. – Tony Rietwyk Sep 02 '16 at 00:39

score 0 · Answer 2 · edited May 23 '17 at 10:29

0

I would use this regex (it works with grep, not sure about RegexBuddy):

^\s*for\s*(.\+[^:]:[^:].\+).*$

The essential part is in [^:]:[^:] which only matches a single colon (not immediately preceded or succeeded by another colon) within an arbitrary text inside a pair of parentheses.

EDIT RustyX's comment suggests that a regex that will match only range-fors is not possible, because it can always be tricked by the colon of a conditional (?:) operator.

edited May 23 '17 at 10:29

Community

1
1

answered Sep 01 '16 at 08:55

Leon

31,443
4
72
97

Would this match `for(int i=0;i<10;i++) a = b ? b : (c + 1);` ? – rustyx Sep 01 '16 at 08:59
@RustyX You are right, it will match the colon of the conditional operator. – Leon Sep 01 '16 at 09:01

How to search for new style C++11 for loops in an existing codebase

2 Answers2