erdify 0.11.0 — generate ERDs from SQL DDL (erdify[sql])

[FabianClemenz]

erdify 0.11.0 is out 🎉

Generate ERDs straight from SQL — no database connection required. Install with pip install -U 'erdify[sql]' (or bump your pre-commit rev to v0.11.0).

New: SQL DDL frontend (erdify[sql])

Point erdify at .sql files and get the same diagrams you already get from your Python models. It parses DDL statically with sqlglot — no DB, no connection string.

pip install 'erdify[sql]'

erdify schema.sql --format mermaid
erdify ./db --include '*.sql' --sql-dialect postgres -o erd.puml

What it understands:

• CREATE TABLE — columns, types, NOT NULL/DEFAULT, primary keys (incl. composite)
• Foreign keys — inline REFERENCES, table-level, and ALTER TABLE ... ADD CONSTRAINT (composite-aware)
• Many-to-many link tables, detected structurally
• CREATE TYPE ... AS ENUM (Postgres) and CREATE INDEX (in --format json)
• Unmodeled statements (INSERT, sequences, GRANT, …) are skipped, so dumps parse cleanly

New --sql-dialect flag / sql_dialect in [tool.erdify], and input now accepts a file as well as a directory.

Zero-dependency core, still

sqlglot ships only with the erdify[sql] extra and is imported lazily — a plain pip install erdify stays dependency-free. Nothing changes for existing Python-model users.

Thanks 🙏

This shipped straight from the community — see the original idea in #71. Cardinality refinement for UNIQUE foreign keys (1:1) is tracked in #97.

📦 Release: https://github.com/devsuit-berlin/erdify/releases/tag/v0.11.0
📖 Docs: https://erdify.devsuit.io/frameworks/sql/
📝 Changelog: https://github.com/devsuit-berlin/erdify/blob/main/CHANGELOG.md