I am working on some SQL scripts which I'm translating and have come across a regexp pattern which works, but I don't know why.

The scene:
I am trying to remove double-qoutes (") from some sql scripts. An example sql line looks like

CREATE OR REPLACE PROCEDURE "proc_name"

I need to 'sed' this to

CREATE OR REPLACE PROCEDURE proc_name

simple enough. I'm not regexp cowboy so I was building my pattern and stumbled across this pattern which works though I don't know why.

sed -e 's/\(PROCEDURE[ ]\)*"/\1/g' sql-file

This produces
CREATE OR REPLACE PROCEDURE proc_name

though I don't know why, can anyone explain? Till then I am using the following pattern which works and I understand why.

sed -e 's/\(PROCEDURE[ ]*\)"\([A-Za-z0-9_]*\)"/\1\2/g' sql-file

-- jpbarto

Hmmm,

Interessting thing, had me thinking for about 30 minutes, but...

It is actually quit easy (as allways)
This on works also, and is clearer why (for me it was)

It says, get everything unil " (if you can find it), and put it in "\1", then only output \1 without the "doubleticks" (gaensefueschen in german), then do it again.
(the g parameter at the end, say repeat until nothing is found anymore)

Nifty, I had not thought about solving something that easy.

Happy Hacking!