How can I get "->function('xxx','xxx')" in the end of some text

Question

I want to get one or more ->func(xxx,xxx) at the end of a piece of code.

Their could be like this:

any code any code ->func(xxx)

or

any code any code 
->func()

or

any code any code 
-funcA()->funcB(xxx)

or

any code any code 
->funcA()
->funcB(xxx)

or mix them:

o.start_time = obj.s;
o.repair_type -> obj.r;
o.limit -> obj.l;->god("('\"\"')") ->fox(,'->')
->egg()->dog(,'c')
->cat(,'b')-> banana(,'a"\'\(\)\'->"()')  ->  apple(,'a')

In this code, i want to :

• plan A

  1. get substring apple(,'a')
  2. remove -> apple(,'a')
  3. get substring banana(,'a"\'\(\)\'->"()')
  4. remove -> banana(,'a"\'\(\)\'->"()')
  5. get substring cat(,'b')
  6. remove ->cat(,'b')
  7. get substring dog(,'c')
  8. remove ->dog(,'c')
  9. get egg()
  10. remove ->egg()
  11. get fox(,'->')
  12. remove ->fox(,'->')
  13. get god("('\"\"')")god("('\"\"')")
  14. remove ->god("('\"\"')")
  15. OVER

• plan B:

  1. get and remove ->cat(,'b')-> banana(,'a"\'\(\)\'->"()') -> apple(,'a')
     1. get substring apple(,'a')
     2. remove -> apple(,'a')
     3. get substring banana(,'a"\'\(\)\'->"()')
     4. remove -> banana(,'a"\'\(\)\'->"()')
     5. get substring cat(,'b')
     6. remove ->cat(,'b')
  2. get and remove ->egg()->dog(,'c')
     1. get substring dog(,'c')
     2. remove ->dog(,'c')
     3. get egg()
     4. remove ->egg()
  3. get and remove ->god("('\"\"')") ->fox(,'->')
     1. get fox(,'->')
     2. remove ->fox(,'->')
     3. get god("('\"\"')")god("('\"\"')")
     4. remove ->god("('\"\"')")
  4. OVER

Now, I am trying planB by this two RegEx, but not good enough:

loop
    if match "\R\s*->\s*(.+)$"
        get substring and remove
        loop substring
        if match "(?:(?<=\)).)*\s*->\s*(((?!->).)*)$"
            push substring2 to arr
            remove substring2
        else
            break
    else
        break

Answer 1

I do not think a regex is the final means to match what you need, but it can be used for a one-off task.

In PCRE, we have a recursion support, thus we can match function start and end. If the code does not have any comments, you can match these nested (...) together with single- and double-quoted string literals at the end of the string with

(?:((?(3)\s*|\R*)->\s*([\w.]*(\((?>'[^'\\]*(?:\\.[^'\\]*)*'|"[^"\\]*(?:\\.[^"\\]*)*"‌​|[^"'()]++|(?3))*\)))))+\s*\z

See the regex demo

Explanation:

• (?:((?(3)\s*|\R*)->\s*([\w.]*(\((?>'[^'\\]*(?:\\.[^'\\]*)*'|"[^"\\]*(?:\\.[^"\\]*)*"‌​|[^"'()]++|(?3))*\)))))+ - 1 or more occurrences of the following sequences:
  • (?(3)\s*|\R*) - a conditional that checks if Group 3 matched, and matches 0+ whitespace characters if Group 3 is initialized, and matches 0+ linebreak sequences (with \R*) if Group 3 has not matched yet (thus, we match linebreaks only at the start)
  • ->\s* - -> followed with 0+ whitespace
  • ([\w.]*) - (Group 1, function name) 0+ alphanumeric/underscore/dot characters
  • (\((?>'[^'\\]*(?:\\.[^'\\]*)*'|"[^"\\]*(?:\\.[^"\\]*)*"|[^"'()]++|(?1))*\)) - Group 1 matching
    • \( - literal opening (
    • (?>'[^'\\]*(?:\\.[^'\\]*)*'|"[^"\\]*(?:\\.[^"\\]*)*"|[^"'()]++|(?3))* - single quoted literals ('[^'\\]*(?:\\.[^'\\]*)*') or double-quoted literals ("[^"\\]*(?:\\.[^"\\]*)*") or (...) ([^"'()]++|(?3) where (?3) recurses the whole Group 3 subpattern).
    • \) - literal closing )
• \s*\z - 0+ whitespace \s* right before the very end of the string \z.

Answer 2

The proper answer is, you can't parse a non regular language with a regular expression. You need to either user an existing parser for that language if that exists or write your own.

The now publicly available "Flex & Bison" O'Reilly is a good read.