CODEWITHSUNDEEP
.netregex
Daniel Bişar
Daniel Bişar

Reputation: 2773

Regex matching closing bracket not in quotes

I am trying to build a regex for matching strings like

1.) $(Something)
2.) $(SomethingElse, ")")
3.) $(SomethingElse, $(SomethingMore), Bla)
4.) $$(NoMatch) <-- should not match
5.) $$$(ShouldMatch) <-- so basically $$ will produce $

in a text.

EDIT: The words Something, SomethingElse, NoMatch, ShouldMatch can be even other words - they are names of macros. The strings i try to match are "macro calls" which can occur in a text and should be replaced by their result. I need the regex just for syntax highlighting. A complete macro call should be highlighted. Number 3 is currently not so import. Number 1 and 2 are required to work. It's fine if number 4 and 5 will not work like written above but that any $( after a $ will not match.

Currently I have 

(?<!\$)+\$\(([^)]*)\)

Which matches any $( if there is no leading $, which could be fine if I will not find another way to apply the $$ structure.

The next step I would like to get done is to ignore the closing bracket if it is in quotes. How could I achieve this?

EDIT So that if I have an input like 

Some text, doesn't matter what. And a $(MyMacro, ")") which will be replaced.

The complete '$(MyMacro, ")")' will get highlighted.

I already have this expression

"(?:\\\\|\\"|[^"])*"

for quotes including escaping of quotes. But I don't know how to apply this in a way to ignore everything between them...

P.S. I am using .NET to apply the regular expressions. So balanced groups will be supported. I just don't know how to apply all this.

Upvotes: 6

Views: 2710

Answers (5)

CSᵠ
CSᵠ

Reputation: 10169

Things like this are complicated... so don't get scared of the following:

RegEx: (?<!\$)(?:\$\$)*(\$\((?:[\w, ]+|(?>"(?:(?<=\\)"|[^"])+")|(?1)+)*\))
Explained demo here: http://regex101.com/r/yZ5dI7

This follows all your 5 points, will match the first 3 macro types and even deeper variations with multiple " or macro-inside-macro only when the number of $ prefixing it is odd.

Upvotes: 1

Qtax
Qtax

Reputation: 33918

You can use an expression like this:

(?<! \$ )                     # not preceded by $
\$ (?: \$\$ )?                # $ or $$$
\(                            # opening (

(?>                           # non-backtracking atomic group
  (?>                         # non-backtracking atomic group
    [^"'()]+                  # literals, spaces, etc
  | " (?: [^"\\]+ | \\. )* "  # double quoted string with escapes
  | ' (?: [^'\\]+ | \\. )* '  # single quoted string with escapes
  | (?<open>       \( )       # open += 1
  | (?<close-open> \) )       # open -= 1, only if open > 0 (balancing group)
  )*
)

(?(open) (?!) )               # fail if open > 0

\)                            # final )

Which can be quoted as above. For example in C#:

var regex = new Regex(@"(?x)    # enable eXtended mode (ignore spaces, comments)
(?<! \$ )                       # not preceded by $
\$ (?: \$\$ )                   # $ or $$$
\(                              # opening (

(?>                             # non-backtracking atomic group
  (?>                           # non-backtracking atomic group
    [^""'()]+                   # literals, spaces, etc
  | "" (?: [^""\\]+ | \\. )* "" # double quoted string with escapes
  | '  (?: [^'\\]+ | \\. )*  '  # single quoted string with escapes
  | (?<open>       \( )         # open += 1
  | (?<close-open> \) )         # open -= 1, only if open > 0 (balancing group)
  )*
)

(?(open) (?!) )                 # fail if open > 0

\)                              # final )
");

Upvotes: 5

Pilou
Pilou

Reputation: 1478

For the part whitout the macro as a param (1 and 2) you can do :

(?<!\$)+\$\(([^)]*?("[^"]*?")?)+\)

You can see here here

In the case whith the macro (3) you can do :

(?<!\$)+\$\(([^)]*?("[^"]*?")?(\$\([^)]*?\))?)+\)

But this will not work for macro containing string whith parenthesis.

You can see the result here

Upvotes: 1

dqm
dqm

Reputation: 1350

You mean something like this?

\$\(SomethingElse, \"|[^"\)]?\"\)

Upvotes: 0

Nikola Sivkov
Nikola Sivkov

Reputation: 2852

I recently was in search of similar regex , but decided that it would be faster to parse the text with C# than a regex as my regex skills are bad ... so i wrote this method to remove Razor code blocks.

you can easily modify it to match your needs without complex regex expressions

Upvotes: 0

Related Questions