airlayer

An in-process semantic engine that compiles .view.yml definitions into dialect-specific SQL. Define your metrics once — dimensions, measures, filters, segments — and compile to Postgres, BigQuery, Snowflake, DuckDB, and 6 other dialects.

import airlayer

result = airlayer.compile(
    views_yaml=[open("orders.view.yml").read()],
    query_json='{"measures": ["orders.total_revenue"], "dimensions": ["orders.status"]}',
    dialect="postgres",
)
print(result["sql"])
print(result["columns"])

No database connection required. No server to deploy. The compiler runs in-process — pass in YAML definitions and a query, get back SQL, bind parameters, and column metadata.

Features

• 10 SQL dialects: Postgres, BigQuery, Snowflake, DuckDB, MySQL, ClickHouse, Redshift, Databricks, SQLite, Domo
• Automatic joins: Declare entities on views, and airlayer infers JOINs via BFS on the entity graph
• 12 built-in motifs: contribution, rank, anomaly detection, YoY/MoM/WoW, trend, moving average, cumulative, and more
• Custom motifs: Define reusable analytical patterns in .motif.yml files
• Validation: Check .view.yml files for errors without compiling a query

API

airlayer.compile(views_yaml, query_json, dialect, ...) → dict with sql, params, columns

airlayer.validate(views_yaml, ...) → True or raises ValueError

See the full documentation for the query format, schema reference, and library usage guide.