Avro Schema Generator

Apache Avro is a data serialization system used heavily in the Hadoop/Kafka ecosystem. It stores schema with data (or in a schema registry), provides compact binary serialization, and supports schema evolution with backward/forward compatibility. It's the standard format for Kafka messages via the Confluent Schema Registry and for Hadoop/Spark data pipelines.

JSON booleans → Avro boolean. JSON integers → Avro long. JSON floats/decimals → Avro double. JSON strings → Avro string (with datetime/date detection). ISO 8601 datetime strings → Avro timestamp-millis (logical type). Date-only strings (YYYY-MM-DD) → Avro date (logical type). JSON arrays → Avro array with inferred item type. JSON objects → nested Avro record. JSON null values → union ['null', 'string'] with null default.

Logical types annotate a base Avro type with semantic meaning. Common ones: date (stored as int, days since epoch), time-millis (stored as int, ms since midnight), timestamp-millis (stored as long, ms since Unix epoch), decimal (stored as bytes, with precision/scale), uuid (stored as string). The generator detects ISO 8601 datetime and date strings and emits the appropriate logical type.

In Avro, nullable fields are represented as a union: ['null', 'actualType']. When the JSON sample contains null, the generator outputs this union type with a null default. In practice, you can use the Avro union pattern to make any field optional by adding 'null' to its type array.

Avro supports schema evolution with backward/forward compatibility. You can add fields with defaults (backward compatible), add union types, or rename fields with aliases. The key rule: never delete fields without providing a default, and never change field types without using a union. This is why using a Schema Registry (like Confluent) is recommended to track schema versions.

No. All schema generation runs entirely in your browser using JavaScript. Your JSON data — which may contain sensitive field names or data structures — never leaves your machine.