tabularis-postgresql-plugin

A PostgreSQL plugin for Tabularis, the open-source database client.

This plugin connects Tabularis to PostgreSQL via a standalone Rust binary speaking JSON-RPC 2.0 over stdio, replacing what was originally a built-in driver compiled directly into the Tabularis application. It is byte-for-byte behaviorally identical to that built-in driver, proven by an 82-test parity suite that runs both drivers against the same live database and compares every response.

Requires Tabularis v0.20.0 or later. This plugin relies on the plugin runtime introduced in that release; it will not load on earlier versions of Tabularis.

Table of Contents

• Features
• Screenshots
• Connection Configuration
• Supported PostgreSQL Data Types
• Installation
  • From the Tabularium registry
  • Manual Installation (alternative)
• How It Works
• Supported Operations
• Known Limitations
• Building from Source
• Development
  • Contributing: PR Titles & Versioning
• Changelog
• License

Features

• Connection — Host/port or connection-string connections, with SSL (disable, require, verify-ca, verify-full) via rustls.
• Schema Browsing — Databases, schemas, tables, views, materialized views, routines (functions/procedures), and triggers.
• Column & Key Metadata — Column types (including enum labels and pgvector-style extension types via udt_name fallback), indexes (including composite/unique), foreign keys (including cross-schema).
• Query Execution — Arbitrary SQL with pagination, EXPLAIN/EXPLAIN ANALYZE, and multi-statement batches that share a single connection (so BEGIN/COMMIT, temp tables, and SET survive across statements).
• Inline Editing — Insert, update, and delete rows directly from the Tabularis data grid, with type-aware value binding (enum CAST, UUID, JSON/JSONB, arrays, temporal types, BLOB wire format).
• DDL Generation — CREATE TABLE, ADD COLUMN, ALTER COLUMN (including implicit-cast-compatible TYPE changes), CREATE INDEX, ADD CONSTRAINT FOREIGN KEY, plus the corresponding drops.
• View & Trigger Lifecycle — Create/alter/drop views, create/drop triggers, refresh materialized views.
• BLOB Support — Export a bytea column to a file or preview it as a MIME-sniffed data URL.
• Cross-platform — Pre-built binaries for Linux (x86_64/aarch64), macOS (x86_64/aarch64), and Windows (x86_64).

Screenshots

PostgreSQL in the database picker	Connection configuration
Multi-schema browsing	Data grid with enum support

Connection Configuration

Parameter	Description	Required
host	PostgreSQL server hostname	Yes (unless using connection_string)
port	PostgreSQL server port (default 5432)	No
database	Database name to connect to	Yes (unless using connection_string)
username	Database user	Yes (unless using connection_string)
password	Database password	If required by the server
ssl_mode	disable, require, verify-ca, or verify-full	No
ssl_ca	Path to a custom CA bundle PEM file, used to validate the server's certificate under verify-ca/verify-full instead of the system trust store	No
ssl_cert	Path to a client certificate PEM file, for servers requiring mutual TLS (e.g. Google Cloud SQL). Must be set together with ssl_key	No
ssl_key	Path to the private key PEM file matching ssl_cert. Must be set together with ssl_cert	No
connection_string	Full postgres://user:pass@host:port/db URL, as an alternative to the discrete fields above	No
startup_script	SQL run on every new pooled connection (e.g. SET search_path = ...) before it's handed to a query	No

Supported PostgreSQL Data Types

Category	Types
Numeric	SMALLINT, INTEGER, BIGINT, SERIAL, BIGSERIAL, REAL, DOUBLE PRECISION, NUMERIC, DECIMAL, MONEY
String	CHAR, VARCHAR, TEXT
Date/Time	DATE, TIME, TIMESTAMP, TIMESTAMPTZ, INTERVAL
Other	BOOLEAN, UUID, INET, CIDR, MACADDR
JSON	JSON, JSONB
Binary	BYTEA

Installation

From the Tabularium registry

This plugin is published on the Tabularium registry — the same one the DuckDB and Elasticsearch plugins ship through. Install it from Tabularis's in-app plugin browser: Settings → Plugins, search for PostgreSQL, and install.

If you point Tabularis at a different Tabularium instance via tabulariumRegistryUrl in config.json, make sure that registry has ingested this plugin's releases first.

Manual Installation (alternative)

1. Download the latest release for your platform from the Releases page — look for the most recent 1.0.0-beta.N tag; every release published so far, including the first public one, ships on the beta prerelease channel (see Contributing: PR Titles & Versioning).

2. Extract the archive.

3. Copy postgresql-plugin (or postgresql-plugin.exe on Windows) and .tabularium into the Tabularis plugins directory:

   OS	Plugins Directory
   Linux	~/.local/share/tabularis/plugins/postgresql/
   macOS	~/Library/Application Support/com.debba.tabularis/plugins/postgresql/
   Windows	%APPDATA%\debba\tabularis\data\plugins\postgresql\

4. Restart Tabularis.

How It Works

The plugin is a standalone Rust binary that communicates with Tabularis through JSON-RPC 2.0 over stdio:

1. Tabularis spawns the plugin as a child process.
2. Requests are sent as newline-delimited JSON-RPC messages to the plugin's stdin.
3. The plugin connects to PostgreSQL using tokio-postgres / deadpool-postgres and writes responses to stdout.

Connection pools are cached in-process, keyed by host:port:database:user, so repeated calls against the same target reuse an existing pool instead of reconnecting.

Supported Operations

Method	Description
test_connection / ping	Verify connectivity with a lightweight SELECT 1
get_databases	List databases on the server
get_schemas	List schemas in the connected database
get_tables	List tables in a schema
get_columns / get_view_columns / get_materialized_view_columns	Column metadata for tables, views, and materialized views
get_indexes	Index metadata, including composite and unique indexes
get_foreign_keys	Foreign key metadata, including cross-schema references
get_views / get_view_definition / create_view / alter_view / drop_view	View lifecycle
get_materialized_views / refresh_materialized_view	Materialized view lifecycle
get_routines / get_routine_parameters / get_routine_definition	Function/procedure metadata
get_triggers / get_trigger_definition / create_trigger / drop_trigger	Trigger lifecycle
execute_query / execute_query_batch / explain_query	Query execution, multi-statement batches, and query plans
insert_record / update_record / delete_record	Row-level CRUD with type-aware value binding
get_create_table_sql / get_add_column_sql / get_alter_column_sql / get_create_index_sql / get_create_foreign_key_sql / drop_index / drop_foreign_key	DDL generation and execution
save_blob_to_file / fetch_blob_as_data_url	BLOB (bytea) export and preview

Known Limitations

A few RPC methods are registered but not yet implemented — they return a "not implemented" error rather than real data: get_schema_snapshot, get_all_columns_batch, get_all_foreign_keys_batch, get_materialized_view_definition. Tracked in #32, part of the Phase 2 set of planned PostgreSQL-specific features (sequences, JSONB inline editing, extension-aware types, and more).

Building from Source

Prerequisites

• Rust (edition 2021)
• just (optional, wraps the common cargo invocations)
• A running PostgreSQL instance (for integration tests)

Build

just build      # debug build
just release    # release build (what the GitHub Actions workflow ships)

Or directly with cargo:

cargo build --release

The binary will be located at target/release/postgresql-plugin.

Install Locally

just dev-install   # build + copy binary and manifest into the Tabularis plugins dir
just uninstall     # remove the installed plugin

Local Test Database

just demo-db        # postgres:16-alpine in Docker (postgres / password / testdb)
just demo-db-stop

Development

Running Tests

just test    # cargo test — unit tests for SQL builders, parsing, RPC
just lint    # clippy -D warnings
just fmt     # cargo fmt --all

Manual JSON-RPC test via shell

echo '{"jsonrpc":"2.0","method":"test_connection","params":{"params":{"host":"127.0.0.1","port":5432,"username":"postgres","password":"password","database":"testdb"}},"id":1}' \
  | ./target/release/postgresql-plugin

Contributing: PR Titles & Versioning

PR titles must follow Conventional Commits (type: subject, type(scope): subject, or type!: subject for a breaking change) — enforced by CI on every PR. Add a BREAKING CHANGE: footer to the PR description for breaking changes that don't fit cleanly into the title.

Every PR also needs exactly one prerelease:alpha / prerelease:beta / prerelease:rc / prerelease:stable label, so CI knows which release channel to target when suggesting the next version. There's no default — CI fails with a clear error if the label is missing, rather than guessing.

PR title type	Version impact
feat	minor
fix, refactor, perf	patch
docs, style, chore, test, ci, build	none — no release suggested
any type with ! or a BREAKING CHANGE: footer	major

CI posts a comment on the PR suggesting the next tag/version based on the title's type and the prerelease:* label — informational only, nothing is tagged or released automatically (yet). The suggestion updates (and marks the previous suggestion as outdated) only when the underlying classification actually changes, not on every edit to the title text.

Tech Stack

• Language: Rust (edition 2021)
• Database driver: tokio-postgres + deadpool-postgres
• TLS: rustls via tokio-postgres-rustls
• Serialization: serde + serde_json
• Async runtime: tokio
• Protocol: JSON-RPC 2.0 over stdio

Changelog

Maintainers

• @aesslinger

License

Apache-2.0.