55349845d0
Break monolith source files into focused modules: - stemedb-core/types.rs → types/ directory (assertion, source, gold_standard, etc.) - stemedb-storage: audit_store, quota_store, trust_rank_store, vector_index, vote_store → module directories - stemedb-ingest/worker.rs → worker/ with separate test modules - stemedb-query: engine, materializer, query → module directories - stemedb-lens: epoch_aware, skeptic → module directories - stemedb-sim/lib.rs → agent, arenas/, helpers, runner, strategy, types - stemedb-api/tests: integration_tests → http_basic, http_validation, http_epoch, http_pipeline - stemedb-api/tests: e2e_flow_test → e2e_full_pipeline, e2e_lens_resolution - stemedb-query/tests: e2e_pipeline → e2e_pipeline + e2e_decay Also adds new features: gold standard verification, escalation handlers, admin endpoints, concept hierarchy spec, arena roadmap, and Go SDK. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
69 lines
2.5 KiB
Rust
69 lines
2.5 KiB
Rust
//! Shared configuration constants and limits for StemeDB.
|
|
//!
|
|
//! This module centralizes validation limits and default values used across
|
|
//! the codebase to prevent duplication and ensure consistency.
|
|
|
|
/// Maximum allowed subject (entity ID) length in bytes.
|
|
///
|
|
/// Subjects should be concise identifiers, not full-text descriptions.
|
|
/// This limit prevents unbounded memory growth and ensures reasonable
|
|
/// index performance.
|
|
///
|
|
/// # Example
|
|
/// - Valid: `"Tesla_Inc"` (9 bytes)
|
|
/// - Valid: `"COMPOUND_CID_12345"` (18 bytes)
|
|
/// - Invalid: A 2KB entity description embedded in the subject field
|
|
pub const MAX_SUBJECT_LEN: usize = 1024;
|
|
|
|
/// Maximum allowed predicate (relation ID) length in bytes.
|
|
///
|
|
/// Predicates should be short relation names, not sentences.
|
|
/// This limit ensures efficient indexing on the predicate dimension.
|
|
///
|
|
/// # Example
|
|
/// - Valid: `"has_revenue"` (11 bytes)
|
|
/// - Valid: `"treats_condition"` (16 bytes)
|
|
/// - Invalid: A 512-byte natural language description of the relationship
|
|
pub const MAX_PREDICATE_LEN: usize = 256;
|
|
|
|
/// Maximum allowed object (value) length in bytes.
|
|
///
|
|
/// Objects can contain numbers, strings, or structured data (JSON).
|
|
/// This limit prevents unbounded memory growth while allowing reasonable
|
|
/// values like long text descriptions or JSON payloads.
|
|
///
|
|
/// # Example
|
|
/// - Valid: `"96.7"` (4 bytes)
|
|
/// - Valid: `"oblate_spheroid"` (16 bytes)
|
|
/// - Valid: JSON with nested structure (up to 4096 bytes)
|
|
/// - Invalid: A 10KB embedded document in the object field
|
|
pub const MAX_OBJECT_LEN: usize = 4096;
|
|
|
|
/// Maximum source document size in bytes (10 MB).
|
|
///
|
|
/// Source documents are stored as opaque byte blobs indexed by content hash.
|
|
/// This limit prevents storage exhaustion from maliciously large uploads
|
|
/// while supporting typical PDFs, images, and text documents.
|
|
///
|
|
/// # Rationale
|
|
/// - Average research paper PDF: 1-3 MB
|
|
/// - High-resolution screenshot: 2-5 MB
|
|
/// - 10 MB allows reasonable documents while preventing abuse
|
|
///
|
|
/// # Note
|
|
/// For larger source materials, consider storing a pointer (URL, IPFS CID)
|
|
/// in the source metadata instead of the raw bytes.
|
|
pub const MAX_SOURCE_SIZE: usize = 10 * 1024 * 1024;
|
|
|
|
/// Default limit for paginated query results.
|
|
///
|
|
/// Applied when no explicit limit is provided in the query parameters.
|
|
/// Prevents accidentally fetching unbounded result sets.
|
|
///
|
|
/// # Example
|
|
/// ```ignore
|
|
/// GET /v1/query?subject=Tesla_Inc&limit=50 // explicit limit
|
|
/// GET /v1/query?subject=Tesla_Inc // uses DEFAULT_QUERY_LIMIT
|
|
/// ```
|
|
pub const DEFAULT_QUERY_LIMIT: usize = 100;
|