Shell Programming and Scripting

Hi All,

I am completely new to Unix Shell Scripting.
I m validating(Basic File Validation) an .HHT file in TIBCO.
After that i need to do Data Validation through UNIX Shell scripting.
Rules in DataValidation:
1.) Need to Check Every field wheather it is a Char or Number?(Fields are seperated bya Delimeter ||)
2.) Need to check the DateFormat. It should be (dd/mm/yyyy hh:mm:ss)
3.) Need to check the Nullabality(null or notNull)
4.) Need to Check Length --> Should be within the range

Now i need to validate all the rows/Records in that File and the
Passed records which satisfies the above criteria should write into another File
the failed records should Write into different File.


Sample File:

Code:

LOCAL_AUDITOR_ID||LOCAL_COLLECTION_PERIOD_CODE||Barcode||START_TIME_DATE
AGDFSCVDFR||PATELKUMAR||1234567890||2/11/2012 09:36:00
AGDFSCVDFR||PATELKUMAR||abcdefghij||2/11/2012 09:36:00
AGD12CVDFR||PATELKUMAR||Abcd123456||2/11/2012 09:36:00
AGDFSCVDFR||PATELKUMAR||12345ndb90||2/11/2012 09:36:00

Note: 1st Record(row)--> List of Columns

Code:

LOCAL_AUDITOR_ID --> Char(15)
LOCAL_COLLECTION_PERIOD_CODE--Char
Barcode--> Number(10)
START_TIME_DATE--Char

Sounds like a PERL, C, C++, JAVA job. First, a file is just an array of char, and you need to parse/validate it sequentially to discover fields, separators, and whatever passes in there for null.

If you take a light weight approach, bash can read your fields into a previously typeset indexed array with "while read -a flds ; do ... done <file". You have to add '|' to $IFS to make it an additional field separator, or "tr '|' ' '" to translate the pipes to spaces (since your data normally has none and space is already in $IFS, along with tab and linefeed). You get to write the ..., and bash has regex tools in the "if [[ ... ]]" form of pattern examination, so you can ensure the field has legal length and the right type of character in the right place. You can capture and save header if you use a form like "tr '|' ' ' <file | ( read -a hdr ; while read -a flds ; do ... done )". A typical regex might be "^[a-zA-Z]\{1,22\}$" to pattern match a field with one to 22 letters only. I think the \{\} is limited to 0-99 bytes, but you can stack them.