c++ - Correctly parsing comment in regex -

i creating compiler , having trouble handling commenting when dealing multi-line comments (/* */). issue regex needs fixing. believe looks opening comment token (/*) accept closing comment token (*/) might not part of comment scope.

also issue within string, still try comment out. issue haven't yet implemented appreciated.

the regex using is:

[/][*](.|\n)*[*][/] 

examples:

input:

int main(/* text */) {    int = 0;    /* hello world */    return 1; }  

output:

int main(     return 1; }  

and strings, input be:

 int main() {        printf("/* hi there */\n");        return 1;     }  

output:

int main() {       printf("\n");        return 1; }  

i'm not sure regex library you're using, need what's called non-greedy match.

try this:

\/\*(.|\n)*?\*\/ 

the ? after .* makes match ungreedy.

you can visualize working here.

note perl-compatible regular expression (pcre) syntax, assuming you're using. if you're using posix regular expressions, won't work.

you don't need put / , * inside character class ([...]); need escape them.

you can use pcre_dotall flag make . match \n or \r well, can simplify regex.

pcre_dotall    if  bit set, dot metacharacter in pattern matches char-    acter of value, including 1 indicates newline. however,     ever  matches  1 character, if newlines coded crlf.    without option, dot not match when current position     @ newline. option equivalent perl's /s option, , can    changed within pattern (?s) option setting. negative  class    such [^a] matches newline characters, independent of set-    ting of option. 

then, our regex be:

\/\*.*?\*\/ 

you can make entire regex ungreedy using pcre_ungreedy flag:

pcre_ungreedy     option inverts "greediness" of quantifiers        not greedy default, become greedy if followed "?".    not compatible perl. can set (?u)  option  setting    within pattern. 

in case, this work:

\/\*.*\*\/