Filtering pg_dump with only INSERT lines and changing table name using awk - mysql

Question

I want to create a bash script that does a pg_dump of only INSERT lines and changes the table title in the INSERT line.
I have the following bash script:
#!bin/bash
#Create temp files to store the PSQL dump
DUMPFILE='poops.dump.sql' || (echo "make sql dump file failed" 1>&2; exit 1)
TMPFILE=`mktemp` || (echo "mktemp failed" 1>&2; exit 1)
#Tables to dump: api_order, poops_order_dates, poops_price
#Dump as INSERT queries statements
pg_dump --username="poops" --host="localhost" \
--table="api_order" --table="poops_order_dates" --table="poops_price" \
--no-password --column-inserts \
--data-only "poops" | \
awk '/^INSERT/ {i=1} {if(i) print}' \
> "$TMPFILE" \
|| (echo "pg_dump failed" 1>&2; exit 1)
(echo "start transaction; truncate table api_order; "; \
echo "truncate table poops_order_dates; "; \
echo "truncate table poops_price; "; \
cat "$TMPFILE"; echo 'commit;' ) \
> "$DUMPFILE" \
|| (echo "parsing dump file failed" 1>&2; exit 1)
rm "$TMPFILE"
It removes all the junk before the first INSERT INTO line, however there is still the following stuff after all the INSERT INTO queries:
--
-- Name: api_order_id_seq; Type: SEQUENCE SET; Schema: public; Owner: poops
--
SELECT pg_catalog.setval('api_order_id_seq', 33146, true);
--
-- Data for Name: poops_order_dates; Type: TABLE DATA; Schema: public; Owner: poops
--
--
-- Name: poops_order_dates_id_seq; Type: SEQUENCE SET; Schema: public; Owner: poops
--
SELECT pg_catalog.setval('poops_order_dates_id_seq', 1, false);
--
-- Data for Name: poops_price; Type: TABLE DATA; Schema: public; Owner: poops
--
--
-- Name: poops_price_id_seq; Type: SEQUENCE SET; Schema: public; Owner: poops
--
SELECT pg_catalog.setval('poops_price_id_seq', 1, false);
--
-- PostgreSQL database dump complete
--
How do I write the following awk line:
awk '/^INSERT/ {i=1} {if(i) print}' \
To make it so that it only outputs the INSERT INTO queries, and changes api_order to api_order_test.
I did try:
awk '/^INSERT/ {gsub("INSERT INTO api_order", "INSERT INTO api_order_test", $0); print $0}' \
But because of print $0, it cuts off part of the query where it starts on a new line. I need it to filter for, output and change only
"INSERT INTO api_order .... );\r"
I believe at the end it is a carriage return (\r) or does pg_dump output each INSERT INTO query with a \n at the end?
Raw SQL Dump:
--
-- PostgreSQL database dump
--
SET statement_timeout = 0;
SET lock_timeout = 0;
--
-- Data for Name: api_order; Type: TABLE DATA; Schema: public; Owner: poops
--
INSERT INTO api_order (id, order type, …'', 0, NULL);
INSERT INTO api_order (id, order type, …'', 0, NULL);
INSERT INTO api_order (id, order type, …'', 0, NULL);
INSERT INTO api_order (id, order type, …'', 0, NULL);
INSERT INTO api_order (id, order type, …'', 0, NULL);
--
-- Name: api_order_id_seq; Type: SEQUENCE SET; Schema: public; Owner: poops
--
SELECT pg_catalog.setval('api_order_id_seq', 33294, true);
--
-- Data for Name: hoops_price; Type: TABLE DATA; Schema: public; Owner: poops
--
--
-- Name: hoops_price_id_seq; Type: SEQUENCE SET; Schema: public; Owner: poops
--
SELECT pg_catalog.setval('hoops_price_id_seq', 1, false);
--
-- PostgreSQL database dump complete
--

Answer 1

If everything you don't want to print comes after everything you do want to print, you can "turn off" printing in the same way you turn it on. Something like /^INSERT/ {i=1} /^SELECT/ {i=0} i. (Note that the trailing i is equivalent to your { if (i) print }) If you're input is more complicated or there are other considerations, please post a representative sample of the input to awk and your expected output.
– jas