Seed folder instead of / in addition to seed.sql file (backwards compatible)

[chief-nerd]

We are working as a team together on the supabase backend. One thing that is constantly creating merge conflicts ist the seed.sql file because each feature branch is writing into it.

Solve it similar to migrations: a folder instead of / in addition to a file:

./supabase/seed
├── 0.table1.sql
├── 1.table2.sql
└── 2.table3.sql

Code option A:

• Check if there is a seed.sql -> if so, execute it (current behaviour, backwards compatible)
• Check if there is additionally a /seed FOLDER
• If so execute each .sql file in that folder sorted by name.
  The sorted by name option allows to account for dependencies between seed files (similar to how migrations work).

Code option B:

• Check if seed.sql file or ./seed Folder exist
• If file: -> execute it (current behaviour, backwards compatible) -> stop there
• If folder: execute any SQL file in the /seed FOLDER -> stop there.
• If both: give priority to the file

Code option B has the advantage that if you run supabase db dump > seed.sql it is NOT executing the folder in addition, which would otherwise replicate the data in the db.

[chief-nerd]

Here is the script I am using as a quick fix for now:

#!/bin/bash
echo "Building Seed file ..."
FILE=./supabase/seed.sql

echo "" > $FILE
folders=($(find ./supabase/seeds/* -type f))
for path in "${folders[@]}"
do
	name="${path##*/}"

	echo "Adding $name"
	echo -e "-- $name" >> $FILE 
	echo -e "----------------------\n" >> $FILE
	cat $path >> $FILE
	echo -e "\n\n" >> $FILE
done