obj-derive

The #[derive(Document)] procedural macro for obj.

Part of obj — the embedded document database. obj-derive provides the #[derive(Document)] macro that generates a type's collection binding, schema, and secondary-index descriptors.

Internal crate — not for direct use. obj-derive is an UNSTABLE implementation detail with no SemVer guarantee. Depend on obj-db instead and use the re-exported derive (obj::Document); it is published only so the public crate can resolve it.

What it generates

#[derive(Document)] reads #[obj(...)] attributes and emits the Document impl — the collection name, schema version, and index set:

#[derive(serde::Serialize, serde::Deserialize, obj::Document)]
#[obj(collection = "orders", version = 1)]
#[obj(index_composite(fields = ("region", "status"), name = "by_region_status"))]
struct Order {
    #[obj(index = unique)]
    email: String,
    #[obj(index)]
    region: String,
    #[obj(index = each)]   // multi-value index; field must be a Vec<...>
    tags: Vec<String>,
    status: String,
}

Attribute	Level	Effect
`collection = "..."`	struct	Override the collection name (defaults to the type).
`version = N`	struct	Set the schema version (defaults to `1`).
`index_composite(fields = (...), name = "...")`	struct	Composite index over ≥ 2 fields (`name` optional).
`index`	field	Standard secondary index on the field.
`index = unique`	field	Unique secondary index on the field.
`index = each`	field	Multi-value index over each element of a `Vec` field.
`name = "..."`	field	Override an index's default name (the field name).

A history(v1 = OldType, ...) struct attribute registers historical schemas for lazy per-document migration.

The encoding is positional and schema-driven — no field names in the bytes — so the generated codec round-trips byte-identically with the Python @obj.document writer for the same logical schema.

License

Dual-licensed under MIT or Apache 2.0, at your option.

obj-derive 1.1.2

obj-derive

What it generates

License