topic Re: Regex Error (but okay in Regex101) in Alteryx Designer Desktop Discussions

Regex Error (but okay in Regex101)

Bobbins — Thu, 13 Jun 2024 22:26:45 GMT

Hello,

Text String;

"The very Blue Sea is always blue unless its orange when its not BLUE sea"

The idea is to split on every "blue" regardless of case to give:

The very

Blue Sea is always

blue unless its orange when its not

BLUE sea

This works in Reg101 but not in Regex as to parse it gives the error: "Error: Nothing to parse. Enter a complete expression."

Code is:
(?i)\s+(?=\bblue\b)

Thank you

Re: Regex Error (but okay in Regex101)

atcodedog05 — Tue, 25 Jan 2022 17:50:29 GMT

Hi @Bobbins

Reg101 doesn't handle parse hence you are getting this issue in Alteryx.

Hope this helps : )

Re: Regex Error (but okay in Regex101)

atcodedog05 — Tue, 25 Jan 2022 17:54:06 GMT

Hi @Bobbins

Here is how you can achieve the same

Hope this helps : )

Re: Regex Error (but okay in Regex101)

Bobbins — Tue, 25 Jan 2022 17:55:59 GMT

Thanks thats one way to solve it, can you use Regex to do the same so i can keep the case of what i am breaking up?

Re: Regex Error (but okay in Regex101)

atcodedog05 — Tue, 25 Jan 2022 19:51:01 GMT

Hi @Bobbins

In your scenario you are breaking on n number of blues and also first column doesn't have blues hence its bit tricky.

@philipmannering do you have any suggestions on this.

Hope this helps : )

Re: Regex Error (but okay in Regex101)

apathetichell — Wed, 26 Jan 2022 00:09:12 GMT

What you are doing is tokenizing since you are taking one column and splitting it into multiple components. Alteryx tokenizing supports exactly 1 marked group. You have two marked groups - hence this is not doable in Alteryx.

On a purely matched group process - you can see that you have 3 matched groups if you use:

REGEX_CountMatches([Field1],"(?i)\s+(?=\bblue\b)")

I would also note - that regex 101 gives you match information (ie the number of matches) but it does not show you the values of those matches (like you would expect to see from a succesful Regex tokenizing) see the difference between your expression and say:

(\s\w+\s)

where every match in Regex 101 also returns a group.

Re: Regex Error (but okay in Regex101)

apathetichell — Wed, 26 Jan 2022 05:47:36 GMT

I've attached the way to split with case...

The key is tokenize the words, use a multi-row formula to index them by Blue (with a normalized case) and then concatenate in crosstab.

Re: Regex Error (but okay in Regex101)

Bobbins — Wed, 26 Jan 2022 09:00:40 GMT

Thank you kindly, thats a different way to how i was thinking it could be done!

Re: Regex Error (but okay in Regex101)

Bobbins — Wed, 26 Jan 2022 09:01:18 GMT

@atcodedog05 Thank you also for your help :-)

Re: Regex Error (but okay in Regex101)

atcodedog05 — Wed, 26 Jan 2022 09:18:57 GMT

Just curious wouldnt my approach also work.

Happy to help : ) @Bobbins

Cheers and have a nice day!

Re: Regex Error (but okay in Regex101)

Bobbins — Wed, 26 Jan 2022 09:26:22 GMT

The example represents a real life scenario for me where I need the case of the lettering to be kept. If i didn't, then your example would have been perfect. Your solution is valid for someone whom case will not matter and I thank you for noting how it could be done this way. If i can select multiple solutions i will do!

Re: Regex Error (but okay in Regex101)

philipmannering — Wed, 26 Jan 2022 12:35:27 GMT

You can also tokenize on this,

(.+?)\s+(?=blue|$)

No need to worry about the (?i) - you can just have the case insensitive box ticked.

Re: Regex Error (but okay in Regex101)

Bobbins — Wed, 26 Jan 2022 14:02:05 GMT

Interesting @philipmannering

Running this it picks up the first two blues but the all in caps BLUE it fails

Re: Regex Error (but okay in Regex101)

philipmannering — Wed, 26 Jan 2022 21:22:11 GMT

So it does..

This looks about right,

Re: Regex Error (but okay in Regex101)

Bobbins — Thu, 27 Jan 2022 09:04:06 GMT

Wow, thanks for that. Question over the regex part, whats the pipe for in "(?=blue|$)"

Re: Regex Error (but okay in Regex101)

philipmannering — Thu, 27 Jan 2022 10:28:38 GMT

The pipe | means "or". And $ means end of line.

So in this case we're saying that we want to capture text up until a "blue" or "end of the line".

Thinking about it, you'll want to use (?=\bblue\b|$) to split on blue as a word...if you didn't want to split on, say, "bluetooth".

Re: Regex Error (but okay in Regex101)

Bobbins — Thu, 27 Jan 2022 11:02:55 GMT

Perfect, thank you @philipmannering