Question

I am trying to write a regexp to search for for/if/while keywords in a string as read from C++ source code file but exclude any words which include them like:

WhatifYes()
Whatfor()
Waitforwhile()

I have written my regexp like below:

if { [ regexp {(for|while|if)(\s+)(\()} $lineValue ] } { 

But it is not picking up cases like:

while(( int x = 0 ) > 0 );
while(( int x = 0 ) > 0 )
for(int y =0 ; ; )
for(int y =0 ; ; );
if( (int x = 9) > 0 )
if( (int x = 9) > 0 );

Initially I thought cause my regexp is framed to be like:

if/for/while \s+ ( #space or multiple spaces

But I tried including spaces in above example :

while (( int x = 0 ) > 0 );
while (( int x = 0 ) > 0 )
if ( (int x = 9) > 0 )
if ( (int x = 9) > 0 );

Still the regexp is not working – please let me know what regexp I should use to capture them?

Answer 1

Part of your problem is easy to address, and part is very hard.

The easy part is ensuring that you’ve got a whole word: the \m constraint escape only matches at the start of a word, and the \M constraint escape matches at the end, so we can use:

# Nothing capturing; you can add that as necessary
# Ellipsis for the bits I've not talked about yet
regexp {\m(?:while|if|for)\M\s*...} ...

The very hard part is matching the part in parentheses. The problem is that that’s really a “language” (in a theoretical sense) that requires a different kind of parser than a regular expression to match (i.e., a recursive descent parser, which has a more complex state model than the finite automatons used in RE matching). What’s more, using () characters in those expressions is common. The easiest approach is instead match against a close parenthesis that’s at the end of the line, possibly followed by a semicolon, but that’s definitely not properly correct. Alternatively, supporting a limited number of levels of nested parens is also possible.

# Match a few levels...
regexp {\m(?:while|if|for)\M\s*\((?:[^()]|\((?:[^()]|\([^()]*\))*\))*\)} ...

So, let’s break that RE down:

\m                                Word start
(?:while|if|for)                  One of the keywords 
\M                                Word end
\s*                               Optional spaces
\(                                Open paren
  (?:                             Either...
    [^()]                           Non-paren...
  |                               Or...
    \(                              Open paren
      (?:                           Either...
        [^()]                         Non-paren...
      |                             Or...
        \(                            Open paren
          [^()]*                      Non-parens
        \)                            Close paren
      )*                            ... as many of the above as needed
    \)                              Close paren
  )*                              ... as many of the above as needed
\)                                Close paren

If you look at the above, you’ll notice a pattern. Yes, you can keep on nesting to do as deep as you want. What you can’t do is make the RE engine do that nesting for you.