pg-seldump 0.2

Program usage

Usage:

pg_seldump [-h] [--version] [--dsn DSN] [--outfile OUTFILE] [--test]
           [-q | -v]
           config [config ...]

Create a selective dump of a PostgreSQL database.

positional arguments:
  config                yaml file describing the data to dump

optional arguments:
  -h, --help            show this help message and exit
  --version             show program's version number and exit
  --dsn DSN             database connection string [default: '']
  --outfile OUTFILE, -o OUTFILE
                        the file where to save the dump [default: stdout]
  --test                test the configuration to verify it works as expected
  -q, --quiet           talk less
  -v, --verbose         talk more

The config files must be YAML files containing a db_objects list of entries. Each entry may have:

Selectors (all the specified ones must match):

• name: name of the db object to dump

• names: list of names or regex of db objects to dump

• schema: schema name of the db object to dump

• schemas: list of schema names or regexp to match schema names of the db object to dump

• kind: kind of object to match. Can be:

  • table

  • sequence

  • paritioned table

  • materialized view

• kinds: list of kind of objects to match (like for kind)

• adjust_score: adjustment for the match score to break rules ties

Data modifiers:

• action: what to do with the matched object:

  • dump: dump the object in the output (default)

  • skip: don’t dump the object

  • error: raise an error in case of match (useful to create strict descriptions where all the db objects must be mentioned explicitly)

• no_columns: list of columns names to omit

• filter: WHERE condition to include only a subset of the records in the dump

• replace: mapping from column names to SQL expressions to replace values into the dump with somethings else

The objects in the database are matched to the rules in the config files. Every match will have a score according to how specific was the selector matched the object.

• name or names list: 1000

• names regexp: 500

• schema or schemas list: 100

• schemas regexp: 50

• kind or kinds: 10

The rule with the highest score will apply. If two rules have exactly the same score the program will report an error: you can use adjust_score to break the tie.