Add a pgroll pull subcommand

[andrew-farries]

Add a new pgroll pull command that pulls the complete migration history for a schema from the migrations table in the target database and dumps the migrations to disk.

Example

Given the migration history in pgroll.migrations after applying all example/ migrations in this repo:

+--------+-----------------------------------------+-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------->
| schema | name                                    | migration                                                                                                                                                                                                                                              >
|--------+-----------------------------------------+-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------->
| public | 01_create_tables                        | {"name": "01_create_tables", "operations": [{"create_table": {"name": "customers", "columns": [{"pk": true, "name": "id", "type": "integer"}, {"name": "name", "type": "varchar(255)", "unique": true}, {"name": "credit_card", "type": "text", "nullab>
| public | 02_create_another_table                 | {"name": "02_create_another_table", "operations": [{"create_table": {"name": "products", "columns": [{"pk": true, "name": "id", "type": "serial"}, {"name": "name", "type": "varchar(255)", "unique": true}, {"name": "price", "type": "decimal(10,2)"}>
| public | 03_add_column_to_products               | {"name": "03_add_column_to_products", "operations": [{"add_column": {"up": "UPPER(name)", "table": "products", "column": {"name": "description", "type": "varchar(255)", "nullable": true}}}, {"add_column": {"table": "products", "column": {"name": ">
| public | 04_rename_table                         | {"name": "04_rename_table", "operations": [{"rename_table": {"to": "clients", "from": "customers"}}]}                                                                                                                                                  >
| public | 05_sql                                  | {"name": "05_sql", "operations": [{"sql": {"up": "CREATE TABLE users (id INTEGER PRIMARY KEY, name TEXT)", "down": "DROP TABLE users"}}]}                                                                                                              >
| public | 06_add_column_to_sql_table              | {"name": "06_add_column_to_sql_table", "operations": [{"add_column": {"up": "UPPER(name)", "table": "users", "column": {"name": "description", "type": "varchar(255)", "nullable": true}}}]}                                                           >
...

Run:

$ pgroll pull migrations/

This will pull all migrations from the pgroll.migrations table into the migrations/ directory:

$ ls migrations/

01_create_tables.json
02_create_another_table.json
03_add_column_to_products.json
04_rename_table.json
05_sql.json
06_add_column_to_sql_table.json
...

The optional --with-prefixes flag prefixes each migration name with its position in the schema history:

$ pgroll pull --with-prefixes migrations/

This produces the following files:

$ ls migrations/

0001_01_create_tables.json
0002_02_create_another_table.json
0003_03_add_column_to_products.json
0004_04_rename_table.json
0005_05_sql.json
0006_06_add_column_to_sql_table.json
...

The --with-prefixes flag ensures that files are sorted lexicographically by their time of application.

[exekias]

nice! should we add something to the docs?

[andrew-farries]

nice! should we add something to the docs?

If we're happy that pgroll pull is the right command name, yes.

I've added documentation in the most recent commit on the branch.

[exekias]

Awesome! I wonder if we should (not in this PR) update our examples folder to use these prefixes, while taking the numbers out of the migration names

[andrew-farries]

Awesome! I wonder if we should (not in this PR) update our examples folder to use these prefixes, while taking the numbers out of the migration names

Taking migration numbers out of the migration names is a good idea I think 👍