How to prepare for Converting UNIX shell scripts to PL/SQL

Question

Hi All
I was said, that i may have to convert a lot of unix shell script to PL/SQL, what are the concepts i need to know to do it efficently,
what are the options PL/SQL is having to best do that.
I know the question is little unclear, but I too dont have much inputs about that i'm sorry for that, just its a question of how
to prepare myself to do it the best way. What are the concepts i have to be familiar with.
Many Thanks
MJ

Answer 1

Just how much work is involved, is hard to say. Many years ago I also wrote (more than once) a complete ETL system using a combination of shell scripts, SQL*Plus and PL/SQL.
If the PL/SQL code is fairly clean, uses bind variables and not substitution variables, then it should be relatively easy to convert that PL/SQL code in the script to a formal stored procedure in the database.
There is however bits and pieces that will be difficult to move into the PL/SQL layer as it requires new software - like for example FTP'ing a file from the production server to the ETL server. This can be done using external o/s calls from within PL/SQL. Or, you can install a FTP API library in PL/SQL and FTP that file directly into a CLOB and parse and process the CLOB.
Think of Oracle as an o/s in its own right. In Oracle we have a mail client, a web browser, IPC methods like pipes and messages queues, cron, file systems, web servers and services, etc. And PL/SQL is the "shell scripting" (times a thousand) language of this Oracle o/s .
In some cases you will find it fairly easy to map a Unix o/s feature or command to one in Oracle. For example, a Unix wget to fetch a HTML CSV file can easily be replaced in Oracle using a UTL_HTTP call.
On the other hand, techniques used in Unix like creating a pipe to process data, grep for certain stuff and awk certain tokens for sed to process further... in Oracle this will look and work a lot different and use SQL.