omnidbm

Universal database-to-database transfer engine with an interactive CLI. Transfer any supported source to any supported destination: MongoDB, PostgreSQL, MySQL, SQLite, Redis, CSV, JSONL.

Install

pip install omnidbm
pip install "omnidbm[mysql]"    # MySQL support
pip install "omnidbm[redis]"    # Redis support
pip install "omnidbm[all]"      # everything

CLI

# Transfer all tables
omnidbm transfer --source mongodb://user:pass@host/db --dest postgresql://user:pass@host/db

# Specific tables, batch size, drop destination first
omnidbm transfer -s mongodb://... -d postgresql://... -t users,orders --batch-size 2000 --drop-first

# Count only, no writes
omnidbm transfer -s csv://./data -d mongodb://... --dry-run

# Interactive guided wizard
omnidbm wizard

# Inspect a source
omnidbm inspect mongodb://user:pass@host/db --sample 3

# Connectivity preflight
omnidbm doctor mongodb://... postgresql://...

Options

URI schemes

• mongodb://user:pass@host:27017/db or mongodb+srv://...
• postgresql://user:pass@host:5432/db
• mysql://user:pass@host:3306/db
• sqlite://C:/path/db.sqlite or sqlite://:memory:
• redis://user:pass@host:6379/0 — tables are key prefixes (split on :), (all) for every key
• csv://C:/path/to/file.csv or csv://C:/data/ (directory = multiple tables)
• jsonl://C:/path/out.jsonl or jsonl://C:/data/

Redis documents use the shape {"_key": ..., "_value": ..., "_ttl": ...}: strings, hashes (dict), and lists/sets (list) are supported. Keys missing the table prefix get it prepended on write.

Python API

from omnidbm import transfer, TransferConfig, ConnectorConfig, TableSpec

config = TransferConfig(
    source=ConnectorConfig(uri="mongodb://user:pass@host/db"),
    dest=ConnectorConfig(uri="postgresql://user:pass@host/db"),
    tables=[TableSpec(source="users", dest="users")],
    drop_first=True,
)
results = transfer(config)

Custom connectors: subclass BaseConnector, implement connect, list_tables, read_stream, write_stream, close, then register with @register("scheme").

Notes

• Documents are normalized to JSON-safe values with type tags, so datetime, ObjectId, Binary, and Decimal survive any-to-any transfers losslessly.
• PostgreSQL/MySQL/SQLite destination tables are created automatically with inferred column types.
• MySQL and Redis drivers are optional: pip install omnidbm[mysql], pip install omnidbm[redis].
• CSV destinations require consistent keys across documents; use JSONL for heterogeneous data.
• --conflict overwrite on files behaves like append (files are immutable).