Обсуждение: psql: print values and return the COUNT(*) value to bash?
Currently I do this in a bash script:
Tbl=table_1Fld=field_1
Sql="SELECT CURRENT_TIMESTAMP - MIN($Fld) AS days_ago
Sql="SELECT CURRENT_TIMESTAMP - MIN($Fld) AS days_ago
, MIN($Fld) as oldest_date
, COUNT(*) AS cnt
FROM $Tbl;"
psql -h <host> <db> -Xc "${Sql}"
psql -h <host> <db> -Xc "${Sql}"
days_ago | oldest_date | cnt
----------+-------------+-----
| | 0
(1 row)
| | 0
(1 row)
More stuff happens after this. I want to print out these four lines, but also test the COUNT(*) value and bypass other bash commands if the count == 0.
Is there any relatively simple way to do this (psql print and return count(*))?
Death to <Redacted>, and butter sauce.
Don't boil me, I'm still alive.
<Redacted> lobster!
On Mon, 29 Dec 2025 02:21:41 -0500 Ron Johnson <ronljohnsonjr@gmail.com> wrote: … > More stuff happens after this. I want to print out these four lines, but > also test the COUNT(*) value and bypass other bash commands if the count == > 0. You should probably output your result from psql in a parsable way (CSV ? TSV ? custom ?) for your bash script, then print values from bash. The commands "column" or "printf" could help you format a pretty table if really needed. > Is there any relatively simple way to do this (psql print and return > count(*))? Not in a clean way. To make psql return something else than 0, you need to trigger an error with a failing query to make it return 3 when ON_ERROR_STOP is enabled.
On Mon, Dec 29, 2025 at 02:21:41AM -0500, Ron Johnson wrote:
> Currently I do this in a bash script:
> Tbl=table_1
> Fld=field_1
> Sql="SELECT CURRENT_TIMESTAMP - MIN($Fld) AS days_ago
> , MIN($Fld) as oldest_date
> , COUNT(*) AS cnt
> FROM $Tbl;"
> psql -h <host> <db> -Xc "${Sql}"
> days_ago | oldest_date | cnt
> ----------+-------------+-----
> | | 0
> (1 row)
>
> More stuff happens after this. I want to print out these four lines, but
> also test the COUNT(*) value and bypass other bash commands if the count ==
> 0.
>
> Is there any relatively simple way to do this (psql print and return
> count(*))?
1. Don't interpolate $variables to your sql. Instead use -v "t=$Tbl" -v
"f=$Fld" options to psql, and then make your query use :"t" and :"f"
- it's safer.
2. Catch output of the psql to variable, and check last line, there is
rowcount there.
Example:
=$ cat z.sh
#!/usr/bin/env bash
table=pg_stat_all_tables
field=last_autovacuum
returned="$( psql -d depesz_explain -v "t=$table" -v "f=$field" -X << _SQL_
select current_timestamp - min(:"f") as days_ago
, min(:"f") as oldest_date
, count(*) as cnt
from :"t"
_SQL_
)"
row_count="$( tail -n1 <<< "${returned}" | tr -cd 0-9 )"
echo "Rowcount = $row_count"
echo "Full output:"
echo "${returned}"
# vim: set filetype=bash shiftwidth=4 expandtab smarttab softtabstop=4 tabstop=4 textwidth=132 :
=$ bash z.sh
Rowcount = 1
Full output:
days_ago | oldest_date | cnt
------------------------+-------------------------------+-----
7 days 03:48:07.348247 | 2025-12-22 07:22:32.593814+01 | 212
(1 row)
Best regards,
depesz
On Mon, 29 Dec 2025 11:11:58 +0100
hubert depesz lubaczewski <depesz@depesz.com> wrote:
Hi Ron, Hubert,
>
> Example:
>
> =$ cat z.sh
> #!/usr/bin/env bash
>
> table=pg_stat_all_tables
> field=last_autovacuum
>
> returned="$( psql -d depesz_explain -v "t=$table" -v "f=$field" -X << _SQL_
> select current_timestamp - min(:"f") as days_ago
> , min(:"f") as oldest_date
> , count(*) as cnt
> from :"t"
> _SQL_
> )"
> row_count="$( tail -n1 <<< "${returned}" | tr -cd 0-9 )"
> echo "Rowcount = $row_count"
> echo "Full output:"
> echo "${returned}"
>
> # vim: set filetype=bash shiftwidth=4 expandtab smarttab softtabstop=4 tabstop=4 textwidth=132 :
>
> =$ bash z.sh
> Rowcount = 1
> Full output:
> days_ago | oldest_date | cnt
> ------------------------+-------------------------------+-----
> 7 days 03:48:07.348247 | 2025-12-22 07:22:32.593814+01 | 212
> (1 row)
>
IIUC, Ron wants the 212 number?
Maybe use the -t switch, and 'cut' :
#!/usr/bin/env bash
table=pg_stat_all_tables
field=last_autovacuum
returned="$( psql -t -d vv -v "t=$table" -v "f=$field" -X << _SQL_
select current_timestamp - min(:"f") as days_ago
, min(:"f") as oldest_date
, count(*) as cnt
from :"t"
_SQL_
)"
row_count="$( tail -n1 <<< "${returned}" | cut -d "|" -f 3)"
echo "Rowcount = $row_count"
echo "Full output:"
echo "${returned}"
--
Bien à vous, Vincent Veyron
https://compta.libremen.com
Logiciel libre de comptabilité générale et analytique en partie double
On Mon, Dec 29, 2025 at 05:40:23PM +0100, Vincent Veyron wrote:
> > =$ bash z.sh
> > Rowcount = 1
> > Full output:
> > days_ago | oldest_date | cnt
> > ------------------------+-------------------------------+-----
> > 7 days 03:48:07.348247 | 2025-12-22 07:22:32.593814+01 | 212
> > (1 row)
> >
>
> IIUC, Ron wants the 212 number?
>
> Maybe use the -t switch, and 'cut' :
Ah, that part I misunderstood.
Well, if there is only 1 row returned, always, then the simplest thing
would be to change
> row_count="$( tail -n1 <<< "${returned}" | cut -d "|" -f 3)"
to this:
row_count="$( awk 'NR==3 {print $NF}' <<< "${returned}" )"
format can stay the same, we don't need to psql -t, or -a, and it will
just work :)
Best regards,
depesz