RegExp - Block comments to single comment optimization - regex

Question

Optimize?
I am having trouble with the following regular expression:
/^\s*(/?\*{1,2}(\s*(\b.*\b)\s*(\*/)?)?|\*/?\s*)$/g
I am wondering if I can improve this expression? Also, if anyone can find a problem with this expression, could you take note of it. Here is my live demo in action. It works for all the conditions that I have set up below
Test constraints
These Match
/**
/*
*
*/
/** Javadoc */
/* Block */
* Multi-line
/* Single Line */
/** A
/** A */
/* A
/* A */
These Shouldn't
7 * 8
// Regular comment
Results
After replacing the match with: // $3
I successfully converted them, despite some of them having trailing white-space:
//
//
//
// Javadoc
// Block
// Multi-line
// Single Line
// A
// A
// A
// A
Regex explained
/
^ Line start
\s* 0 or more white-space
( Start group 1
/? forward-slash (OPTIONAL)
\*{1,2} 1 to 2 asterisks
( Start group 2
\s* 0 or more white-space
( Start group 3
\b Start word boundry
.* 0 or more of anything
\b End word boundry
) End group 3
\s* 0 or more white-space
( Start group 4 (OPTIONAL)
\* 0 or more asterisks
/ Forward-slash
)? End group 4
)? End group 2 (OPTIONAL)
| OR
\* Asterisk
/? Forward-slash (OPTIONAL)
\s* 0 or more white-space
) End group 1
$ Line end
/
g Global; match all

Answer 1

Parsing C/C++ style comments are a little more complex when you have to take into account string embedding, escaping, and line continuation.
This also requires a single character consumption pass-thru method (after all else is checked). That being the case, of the many possible match alternations, there is only one item you're interrested in - the C-style comment /* */.
So, in the case of a global search and a lot of matches, only one interrests you, say a capture group 1 match. Of course, in your case a simple replace won't cut it.
So, you have to sit in a global find (but not replace) loop. Each match will be a part of the source string, no part is skipped, so each one will be appended to a destination string.
During the loop, when capture group 1 matches (C-style comment), you can do your not-so-simple substitution's to make it C++ comment, then append that result to the destination string.
Thats the jist of it. If this can't be done in whatever language you use, then it can't be done right, that is for sure!
Here is a list of regex's to perform the conversion to // when you'vee captured a C-style comment, they have to be performed in the order of appearance (the notation is Perl for example only):
# s{ ^ /\* (?: [^\S\n] | \*)* }{// }xmg;
# s{ ^ (?: [^\S\n] | \*)* \*/ $ }{// }xmg;
# s{ ^ $ }{// }xmg;
# s{ ^ (?: [^\S\n] | \*)+ }{// }xmg;
# s{ (?<![\s*]) (?: [^\S\n] | \*)+ $ }{}xmg;
# s{ (?<![\s*]) (?: [^\S\n] | \*)* \*/ $ }{}xmg;
This is the regex to use in your find loop (as explained above). This regex can be found on any Perl news group, its part of the FAQ.
# (?:(/\*[^*]*\*+(?:[^/*][^*]*\*+)*/)|//(?:[^\\]|\\\n?)*?\n)|(?:"(?:\\[\S\s]|[^"\\])*"|'(?:\\[\S\s]|[^'\\])*'|[\S\s][^/"'\\]*)
# -----------------------------------------------------------------
(?: # Comments
( # (1 start)
/\* # Start /* .. */ comment
[^*]* \*+
(?: [^/*] [^*]* \*+ )*
/ # End /* .. */ comment
) # (1 end)
|
// # Start // comment
(?: [^\\] | \\ \n? )*? # Possible line-continuation
\n # End // comment
)
|
(?: # Non - comments
"
(?: \\ [\S\s] | [^"\\] )* # Double quoted text
"
| '
(?: \\ [\S\s] | [^'\\] )* # Single quoted text
'
| [\S\s] # Any other char
[^/"'\\]* # Chars which doesn't start a comment, string, escape,
# or line continuation (escape + newline)
)