SRE - group position number

[ripdad]

$n = StringRegExpReplace($s, '(TestA)(TestB)(TestC)', '$1'); <- doesn't work - just for show

I'm trying to get the "group position number" in the SRE.

ie: 1, 2 or 3

The end result would be an offset number for another piece of code i'm working on.

I can do it fairly easy with this...

Local $s = 'TestB'
Local $n = StringReplace(StringReplace(StringReplace($s, 'TestA', 1), 'TestB', 2), 'TestC', 3)
MsgBox(0, '', $n)

I thought perhaps it could be done with an SRE, but I couldn't get it to work.

Any help would be appreciated.

[MrMitchell]

Kind of an interesting problem...maybe here is one way:

Dim $s = 'TestB'
Dim $regex = '(TestA)(TestB)(TestC)'
$array = StringSplit($regex, "(")
For $i = 1 To $array[0]
If StringInStr($regex, $s, 1) Then
  MsgBox(0, "Position of '" & $s & "'", 2)
  ExitLoop
EndIf
Next

[Mat]

So basically you won't to get the index of a string within a set of strings. This is not a problem usually solved by SRE or your (very interesting) StringReplace method.

Depending on the number of string you want to test against really dictates what you do. Typically I would approach it using an array of strings and then searching that.

In the above case, the index is in fact:

Asc(StringRight($s, 1)) - Asc('A') + 1

More information about the set of strings would allow us to help you more, patterns like that make the whole process much faster and easier.

[ripdad]

This was my very first attempt...

Local $n, $s = 'TestB'
Local $a = StringSplit('TestA|TestB|TestC', '|')
For $i = 1 To $a[0]
    If $a[$i] = $s Then
        $n = $i
        ExitLoop
    EndIf
Next
MsgBox(0, '', $n)

And then I was looking for alternatives.. hence the "very interesting" StringReplace.

I didn't know if one could achieve it using an SRE or not. That why I'm looking for other opinions.

@Mat - The "Strings" are as good as I've shown in the examples. It's just one word in each. There is no pattern to them. A,B,C is just to separate them for the purpose of the examples.

Edited April 19, 2012 by ripdad

[Mat]

Ok, then a better approach (although functionally the same) as what you are doing above, is to keep a sorted array of strings, and use _ArrayBinarySearch. That's easier than manually looping and searching yourself, and also scales better (for when you have thousands of words ).

That's just a slightly neater way to write exactly the same as you already have.

Local $n, $s = 'TestB'
Local $a = StringSplit('TestA|TestB|TestC', '|')
$n = _ArrayBinarySearch($a, $s, 1)
MsgBox(0, '', $n)

It will also deal with making sure the item you are looking for does actually exist (it will return -1 in that case).

[jchd]

If we had access to the pcre_extra structure created by PCRE when compiling the pattern, this would be easy using (*MARK:name).

But I don't believe we have access to this structure. I don't even know if it's requested by the internals of the StringRegExp* functions.

I'll look for that anyway.

EDIT: now that I come back to think of it, there is a "simple" way to infer which capturing group fired the match but only when the capturing patterns are exclusive and forced. In the resulting array, only one element will be non-empty. So the index of the matching group is:

Local $i = 0
Do
  If $Result[$i] Then ExitLoop
Until $i = UBound($Result) - 1
$i = Mod($i + 1, UBound($Result))     ; $i is now your index or 0 if no match

I agree it isn't a one-liner, but it has the advantage to work with any regexp pattern, not only literals.

Edited April 19, 2012 by jchd

[ripdad]

@Mat - It's only 3 words, thanks anyways.

@jchd - I have the idea that this would be abnormal for an SRE - judging from your reply. Thanks for looking.

@All - Thanks for your time.

[jchd]

Justed checked: (*MARK:mymark) is flagged as syntax error due to minimalist compile options selected for building AutoIt PCRE library.

Heck!

[ripdad]

I don't know if this is kosher... but this seems to work under these conditions --> strings must be known and hard coded. Any thoughts?

Local $s = 'TestB'
Local $n = UBound(StringRegExp($s, '(TestA)|(TestB)|(TestC)', 3))
MsgBox(0, '', $n)

-EDIT-

For the heck of it - some "Equivalent Statements"?

Local $n = 0, $s = 'TestB'
Switch $s
    Case 'TestA'
        $n = 1
    Case 'TestB'
        $n = 2
    Case 'TestC'
        $n = 3
Endswitch
MsgBox(0, '', $n)

Local $n = 0, $s = 'TestB'
If $s = 'TestA' Then
    $n = 1
ElseIf $s = 'TestB' Then
    $n = 2
ElseIf $s = 'TestC' Then
    $n = 3
EndIf
MsgBox(0, '', $n)

Edited April 21, 2012 by ripdad

[MrMitchell]

Dim $s = 'TestB'
Dim $regex = '(TestA)(TestB)(TestC)'
$array = StringRegExp($regex, "(?U)((.*))", 3)
For $i = 0 To UBound($array) - 1
If StringInStr($array[$i], $s) Then
  MsgBox(0, "Position", $i + 1)
  ExitLoop
EndIf
Next

[jchd]

You've got it. It's simpler than looking for a non-empty string in the resulting array (my post #6).

Contrary to what you say, it works with patterns as well:

Local $s = 'I take this Test with TestB'
Local $n = UBound(StringRegExp($s, '(Tw+A)|(Tw+B)|(Tw+C)', 3))
MsgBox(0, '', $n)

[ripdad]

Well great!

There's always more than one way to skin a cat. Time to get the band-aids.