Applying .sql files without tracking

[timtonk]

Hey.
First of all, thanks guys for this tool, it is extremely simple and helpful for me. Using it since v.2.

I have just noticed PR #246. This one could be useful for me, but I need one small feature on the top of it.
Background: I have a bunch of raw sql files with postgres functions inside. It doesn't make sense to store them in migrations as in case of changes I would need to copy-paste the most part of it to another file. It also doesn't allow me to use diff tool for change tracking. Storing them in files resolves most of these troubles except for one - it is implemented outside of pg-migrate project and requires psql on the machine where it is supposed to be run.
To be more precise, instead of incremental changes I set the actual state of the DB from the provided SQL files. I feel like this approach is extremely reasonable and handy in use for functions and policies.

Thus, I would like to have an option to apply all .sql files from a specified migration dir, but doesn't store any information about them in the migration table. It will make possible to run this process multiple time.

Cheers.

[dolezel]

Hi, thanks for thanks :) and input but I do not think this belongs to migration library as it is completely bypassing it.
I do not know what you have in your sql files but wouldn't something like

  "scripts": {
    "migrateup": "node-pg-migrate up",
    "postmigrateup": "psql $DATABASE_URL < sql_files/*"
  }

in package.json do the stuff?

[timtonk]

It would, but requires psql to be installed on the client machine and ignores pg-migrate settings.

[dolezel]

So something like

cat sql_files/* | node -e '
  let sql = "";
  process.stdin.on("data", data => { sql = `${sql}${data}` });
  process.stdin.on("end", () => {
    const client = new (require("pg").Client)({ connectionString: process.env.DATABASE_URL });
    client.connect()
      .then(() => client.query(sql))
      .catch(() => process.exit(1));
  });
'