Extract sentense problem?

hpl123

Hi all,
Working on a problem and wonder if anyone has any good ideas on how to solve it. I am trying to extract multiple sentenses (one by one) from a plain text attribute.

It would either look like this:
This is sentense 1
This is sentense 2
This is sentense 3

Or it would look like this:
This is sentense1;This is sentense 2;This is sentense 3

I want to extract each sentense via rules i.e the 1 attribute above with the 3 sentenses should populate 3 other attribute sith one sentense in each attribute.

Anyone has an idea on how to solve this in Aware IM? Thanks in advance.

kklosson

Is your question how to parse the string or how to construct the process?

BenHayat

kklosson wrote
Is your question how to parse the string or how to construct the process?

I think he is asking about "Parsing".

kklosson

Does each sentence actually have a semi-colon delimiter?

tford

For more on parsing, see documentation of the SET action in the Rule Language Reference document at C:\AwareIM\docs

hpl123

Thanks guys,
The SET action could possibly work but I doubt it. I was a bit unclear about what i´m really after in my post so here is a bit more information.

The solution I´m trying to build is a "quickadd" function where a user can type in for example tasks:
Call Mike about proposal
Buy catfood
Print travel itinerary

OR
Call Mike about proposal;Buy catfood;Print travel itinerary

and the end result should be 2 seperate BO instances (for example "Task" BO). The problem with SET is the attribute name has to be part of the input BUT in the solution i´m trying to build, each line or sentense should go to the same attribute but in a new/seperate BO instance.

Anyone know how it can be done and if it can, help on rule i.e. how the rule would have to be constructed would be appriciated. I´m drawing a blank here 🙂.

tford

Henrik,

In theory, you should be able to store the string that was input with some delimeter like a semi-colon, then iterate through each time stripping off and storing the value on the left side of the string.

You should be able to use LAST_INDEX_OF, INDEX_OF, and CHARS_FROM_LEFT (Or CHARS_FROM_RIGHT) to accomplish this.

hpl123

Thanks Tom,
Yeah, managed to solve it quite easily actually (kudos Aware IM 🙂). Here is the rule for reference if anyone is interested in doing the same thing:

If Input.Input WAS CHANGED AND LENGTH(Input.Input)>0 Then 
Input.IndexString=INDEX_OF(';',Input.Input) 
 CREATE Task WITH Task.Task=SUBSTRING (Input.Input, 0,Input.IndexString-1) 
Input.Input=REPLACE_PATTERN(Input.Input,SUBSTRING (Input.Input, 0,Input.IndexString),'')

This rule automatically create tasks for rows, strings etc. that are divided with a ; . Example:
This is task 1;
This is task 2;
This is task 3;

The only limitation it seems is if a string is EXACTLY the same (Test1 and Test1), in those cases it doesn´t fully work and I might try to solve that as well but really don´t need to as 2 strings are never the same in my scenario/app.

tford

Very creative! Thanks for sharing Henrik.

hpl123

Improved the solution some and now also works with EXACTly the same string (NB: only 1 instance is created). This new solution also don´t rely on another attribute for the string index:

If Input.Input WAS CHANGED  AND Input.Input CONTAINS ';' Then 
CREATE Task WITH Task.Task=SUBSTRING(Input.Input,0,INDEX_OF(';',Input.Input)-1) 
Input.Input=REPLACE_PATTERN(Input.Input,SUBSTRING(Input.Input,0,INDEX_OF(';',Input.Input)),'')

customaware

Nice solution Henrik.