API Surface

This is the stable public shape of the package.

The pattern is consistent across the surface:

• explicit Strawberry type classes define the public GraphQL contract
• Chemist helpers attach SQLAlchemy-aware behavior to those types
• the package stays adaptable by exposing escape hatches where real production schemas need them

Two practical properties are worth calling out:

• Chemist-managed relationship and connection fields are batched through dataloaders and selection-aware SQL shaping, so they avoid per-parent N+1 loading.
• Relationship and connection APIs are intentionally broad enough to cover plain mapped fields, renamed fields, scoped fields, computed fields, root collections, and nested collections.

Surface overview

Area	Main entrypoints	Purpose	Example
Types and mapped fields	@sc.type, sc.attr, @sc.field	Define explicit GraphQL DTOs and computed fields	01_types_and_fields
Relationships	sc.relationship / @sc.relationship	Expose related data or relationship-backed computed fields without N+1 boilerplate	02_relationships
Connections	sc.connection, sc.Connection, sc.OffsetConnection	Expose queryable collections at root or on relationships with batched loading	03_connections_filters_and_ordering
Pagination	sc.CursorPagination, sc.OffsetPagination, sc.PaginationPolicy	Choose flat or nested pagination argument styles and result envelopes	08_nested_pagination_arguments
Filters	@sc.filter, sc.FilterSet, sc.filter_field, sc.manual_filter	Add client-controlled filtering with declarative or manual contracts	03_connections_filters_and_ordering, 06_manual_filters_and_orders
Ordering	@sc.order, sc.order_field, sc.manual_order	Add client-controlled ordering with declarative or manual contracts	03_connections_filters_and_ordering, 06_manual_filters_and_orders
Relay IDs and nodes	sc.Node, sc.node_id, sc.node_field, sc.configure(...), sc.relay.encode_node_id(...), sc.relay.decode_node_id(...)	Define explicit node types, apply package-level defaults, and resolve them through schema-owned relay IDs	04_nodes_and_relay_ids
Resolver node injection	sc.node_lookup	Load an ORM object from a node ID into a query or mutation resolver	07_node_lookup_and_permissions
Resolver contracts	@sc.field(select=...), sc.relationship, sc.connection, sc.node_lookup	Hide Chemist-injected runtime params and expose the rest as GraphQL args	09_resolver_argument_contracts
Schema integration	sc.extensions()	Enable dataloaders and selection caching for Chemist-managed fields	05_context_and_extensions

Practical rule of thumb

• use @sc.type to define the schema you want clients to see
• use sc.Node only for types that should participate in relay/node lookup
• use sc.configure(...) before strawberry.Schema(...) when the application wants package-level defaults
• use sc.relationship(...) for related data and relationship-backed computed fields
• use sc.connection(...) when clients need filtering, ordering, or pagination on root or nested collections
• use the declarative DSLs first
• use manual_filter and manual_order when the GraphQL input shape itself must stay custom