ceres-core

The foundation of Ceres.

This crate provides the core building blocks: types, errors, configuration, and data models used throughout the project.

What it provides

• Core Types & Errors: Fundamental data structures and error definitions.
• Configuration: Environment settings and database configuration.
• Data Models: Domain models like Dataset and Portal.

lib.rs:

Ceres Core - Domain types, business logic, and services.

This crate provides the core functionality for Ceres, including:

• Domain models: Dataset, SearchResult, Portal, etc.
• Business logic: Delta detection, statistics tracking
• Services: HarvestService for portal synchronization, SearchService for semantic search, ExportService for streaming exports
• Traits: EmbeddingProvider, DatasetStore, PortalClient for dependency injection
• Progress reporting: ProgressReporter trait for decoupled logging/UI

Architecture

This crate is designed to be reusable by different frontends (CLI, server, etc.). Business logic is decoupled from I/O concerns through traits:

Server Development Notes (ceres-server)

TODO(server): Use utoipa for automatic OpenAPI documentation When creating ceres-server, use the utoipa crate (compatible with Axum/Actix) to auto-generate OpenAPI specs from Rust types and handlers via derive macros. Expose /swagger-ui endpoint for interactive API documentation. This makes the API immediately usable by frontend developers without reading code.

• EmbeddingProvider - abstracts embedding generation (e.g., Gemini API)
• DatasetStore - abstracts database operations (e.g., PostgreSQL)
• PortalClient - abstracts portal access (e.g., CKAN API)

Example

use ceres_core::{HarvestService, SearchService};
use ceres_core::progress::TracingReporter;

// Create services with your implementations
let harvest = HarvestService::new(store, embedding, portal_factory);
let reporter = TracingReporter;
let stats = harvest.sync_portal_with_progress("https://data.gov/api/3", &reporter).await?;

// Semantic search
let search = SearchService::new(store, embedding);
let results = search.search("climate data", 10).await?;