Regexp matching for assignment operations in C

Question

I'm trying to write a regexp that catches assignments equal signs within conditional statements in C language that I already extracted (using python module re).

My attempt:

exp = re.compile(r'\(\s*[0-9A-Za-z_]+\s*[^!<>=]=[^=]')

While working for a number of cases, it fails to match a simple case like the following string:

'(c=getc(pp)) == EOF'

Can someone please explain why my regexp is not a match for this string, and how could I make it better ? I'm aware that some weird cases might still elude me, but I can treat those manually, the purpose is to do the bulk of the legwork automatically.

the range to match "less than, different" ... prevents the match because it contains = — Jean-François Fabre, Oct 11 '17 at 12:29
@Jean-FrançoisFabre How can I avoid matching for `==` or `!=` then ? — Valentin B., Oct 11 '17 at 12:31
why not just `exp = re.compile(r'\(\s*[0-9A-Za-z_]+\s*=[^=]')` then, that would match a single equal sign, nothing else. — Jean-François Fabre, Oct 11 '17 at 12:33
It does seem to work, I'm testing it a bit, if you answer the question I will probably accept your answer ! Thanks ! — Valentin B., Oct 11 '17 at 12:38

score 1 · Accepted Answer · answered Oct 11 '17 at 12:38

1

[^!<>=] following your identifier prevents = to be matched after c.

If your intention is to match assignments, try to match only one equal sign after the identifier, like this:

exp = re.compile(r'\(\s*[0-9A-Za-z_]+\s*=[^=]')

print(exp.search('(c=getc(pp)) == EOF'))

which results in:

<_sre.SRE_Match object; span=(0, 4), match='(c=g'>

answered Oct 11 '17 at 12:38

Jean-François Fabre

137,073
23
153
219

I do match for more cases like the example I gave with your regexp, but there is something I don't quite get. This regexp is not anchored, yet it does not match any string that starts with something else, say another parenthesis or a blank. For example `' (c=getc(pp)) == EOF'`is not a match. How come ? – Valentin B. Oct 11 '17 at 12:49
because you're using `match`, I'm using `search` – Jean-François Fabre Oct 11 '17 at 12:51
Oh, ok, my bad for not knowing the `re` module enough ! Thank you again, I am accepting your answer, glad to have found help from a fellow Toulousain ;) – Valentin B. Oct 11 '17 at 13:18
I just hope I didn't help a competitor :) (note: I don't care) – Jean-François Fabre Oct 11 '17 at 14:19
Since I work for a subcontractor at CNES we should be fine :) – Valentin B. Oct 12 '17 at 12:10

score 1 · Answer 2 · answered Oct 11 '17 at 13:06

The reason why this does not work is [^!<>=]=, which makes your code look for a character which is not = followed by a character which is =. I can see your intention in doing so, but it's the wrong way.

For the simple, case have a look at the following expression:

[0-9A-Za-z_]+\s*=\s*[0-9A-Za-z_]+(\(\s*[0-9A-Za-z_]*\s*\))?

This matches the c=getc(pp) part of your source, because it looks for a = which is either followed (or preceded) by optional whitespaces and characters or numbers. Already this prevents the regex from matching ==, <=, !=, or >=.

Aside of that it also looks if the right hand side is a function or simply a variable or just a number (optional match through ? for the bracket-part at the end of the expression). Note also the * for the matching part within the braces ([0-9A-Za-z_]*), which enables you to match function calls without parameters.

Works for:

(c=getc(p)) == EOF
(c =getc()) == EOF
(c=getc( )) == EOF
(c = getc( p )) == EOF
(c = i) == EOF
(c=10) == EOF

This will not work for constructs, such as x = y(z()) (and surely many more).

Aside of this being said, I recommend the following link (not exactly your question, but really nice insights): Regular expression to recognize variable declarations in C

You're welcome. In the case that you put more effort into your expression, let us know your solution. — ohmmega, Oct 12 '17 at 05:59
Besides adding `&\->*` to the set corresponding to the variable name to take into account structures, pointers and adresses, I'm affraid my regexp knowledge is very basic so I don't see what else could be done ! — Valentin B., Oct 12 '17 at 12:12

Regexp matching for assignment operations in C

2 Answers2