Re: I need to take metadata from a shell script.

On 6 May 2010 16:15, Andy Colson <andy(at)squeakycode(dot)net> wrote:

> On 5/6/2010 2:57 AM, Jaume Calm wrote:
>
>> Hi! I was searching for a command like pg_dumpall but with the
>> difference that I don’t want a single file for all databases, i would
>> like to have a file for each one.
>>
>> I couldn’t fins such command, so the only option I see is to write a
>> shell script with a loop for all the DBs. The problem is that I’m unable
>> to find the way to obtain the DBs’ names in a shell script. Can someone
>> help me with this?
>>
>> Best regards and thank you all for your time.
>>
>>
> Depending on what version of PG you are on, try:
>
> psql -ltA
>
> a little read, cut, awk, perl, etc action and you should be good.
>
> -Andy
>
>
Aha, yes, I should really look at the psql options more.

You could extend that to exclude templates and the postrgres database and
database attributes:

psql -ltA | cut -d "|" -f 1 | grep -v "\( template0 \| template1 \| postgres
\| : \)"

And using Scott's loop:

for line in `psql -lt | cut -d "|" -f 1 | grep -v "\( template0 \|
template1 \| postgres \| : \)" | head -n -1 `; do pg_dump -f
/home/backups/`date +\%Y\%m\%d`/"$line".sql; done

Or adapt it to put it into dated directories. Anyone got a tidier way? :S

Thom