To capture a specific string based on some condition

vidyaa · New User Joined: 02 May 2008 Posts: 77 Location: chennai

Hi,

I need to capture specific string based on certain conditons from a file. My file is of 10000 bytes and it can have any no of lines. It will look like
"<s> name=abc sometext name=XXX ocr=1 <kl> sometext name=iuk sometext hjlkljk name=hju name=YYY ocr=4 sometext name=text ocr=1"
i need to capture the words before the word ocr and after name= so my output will be like

XXX
YYY
text

is there a way to achieve this using rexx.

enrico-sorichetti · Posted: Mon Aug 09, 2010 5:42 pm

superk · Posted: Mon Aug 09, 2010 5:50 pm

I'm thinking that if you parsed the string, delimited by spaces, into a series of stem variables, then you could use the first occurence of the string "ocr=" as an ending point, then move backwards through the other variables until you hit the closest "name=" string. Mark the start and end points, then loop through again for the next "ocr=" string.

Once you know all of the strarting and ending points, it should be relatively easy to parse out the value for the "name=" keyword.

superk · Posted: Mon Aug 09, 2010 6:11 pm

Actually, it's a bit easier:

where str is the single string to parse:

vidyaa · New User Joined: 02 May 2008 Posts: 77 Location: chennai

Thanks for your valubale suggestions ..i tried the above code this did work well if the file had single line..but am not able to loop them....i tried using

do i = 1 to Rec.0
str = rec.i
followed by the code given above...but this doesnt work for more than one line

I think it is because we have IF LENGTH(STR) = 0 THEN LEAVE
what can be the possible way to fix it.

vidyaa · New User Joined: 02 May 2008 Posts: 77 Location: chennai

I tried this with concatenating the whole lines into single string using
do i = 1 to rec.0
str=str||rec.0
end

am not sure if this is the best option.

Also when i parse the string

Name="xxx yyy" ocr=fghghg
using
PARSE VAR STR 'NAME=' VAL
i get the output as XXX but i need XXX YYY
what is the way to do this

vidyaa · New User Joined: 02 May 2008 Posts: 77 Location: chennai

how to say delimited by some other varibale say '"'

superk · Posted: Tue Aug 10, 2010 5:33 pm

You know, I was going to ask if there would be any reason to consider imbedded blanks or other characters inside of your target strings, but I honestly thought you would've mentioned that.

One thing you can do to assist with parsing is to translate imbedded blanks into some pre-determined character, something that you know the actual data will never contain. I usually go with an ampersand (@) or tilde (~). Then your parsing remains simple, and at the end all you have to do is translate that character back into a blank.

Anyway, at this point, I'd like to see a few records of the actual REAL data the way it's really formatted. Maybe there's a better way, maybe not.

Pedro · Posted: Wed Aug 11, 2010 6:57 am

The example seems to have some tags... it is not clear if there will be more such tags. If your logical records are delimited by start and ending tags, you should parse for them in an outer loop before parsing for the name and ocr fields.

And when you parse for the name and ocr, I suggest to use the PARSE instruction:

vidyaa · New User Joined: 02 May 2008 Posts: 77 Location: chennai

The input file will be like this

NAME="STCUSTUDAPCASHSALEDBTA">20.00</F></S><S NAME="DETAIL, DESC SECTION" OCR="1">
<F NAME="STFTRITEMAMT">.00</F><F NAME="STHDRGROUPCD"
</F><S NAME="IRCS DETAIL RECORD" OCR="1"><F NAME="STDETLITEMAMT">

basically i would like to take the string in the "NAME" field that will come before
the "OCR". So my output looks like

DETAIL, DESC SECTION
IRCS DETAIL RECORD

Pedro · Posted: Wed Aug 11, 2010 6:07 pm

I think you should parse at record boundaries first, then parse for your field delimiters.

Note that the end of the string is a double quote, rather than 'OCR', per your example.