Crate doiget_core 

doiget-core

Core library for doiget: an Open Access first paper-fetcher with strict capability gating, fail-closed provenance logging, and a BiblioFetch.jl-compatible store layout.

Phase 0 ships only this skeleton. Real implementations land in Phase 1. See docs/PUBLIC_API.md for the semver-locked surface and docs/ARCHITECTURE.md for the high-level design.

Re-exports

pub use crate::canonical::CanonicalRef;

pub use crate::canonical::SourceType;

Modules

canonical

Canonical-tuple audit identity for fetched papers (ADR-0021 §1, ADR-0024).

dry_run

Dry-run preview shape for --dry-run CLI fetches and the doiget_metadata_only / doiget_fetch_paper MCP tools.

http

Centralized HTTP client wrapper. All Source impls fetch through here.

orchestrator

Cross-source orchestrators that compose multiple Source impls into a single user-facing operation.

provenance

JSON Lines + SHA-256 hash-chained provenance log.

rate_limiter

Process-wide rate limiter for HTTP fetches across all Source impls.

refs

Bibliography input adapters per ADR-0030.

source

Source abstraction. Each Tier 1/2/3 fetcher implements this trait.

sources

Source implementations.

store

Filesystem-backed metadata store.

user_extension

User-extensible capability gate (ADR-0028, #220).

Structs

AlwaysOn

Marker for the always-on Open Access tier. See docs/CAPABILITY.md.

ArxivId

A validated arXiv id string.

CapabilityProfile

Runtime gate for which sources may be invoked. See docs/CAPABILITY.md.

DenialContext

Structured machine-parseable companion to error.message for recoverable denials.

Doi

A validated DOI string.

MetadataAccess

Which Tier 2 metadata sources are enabled this session. See docs/CAPABILITY.md.

RateLimits

Process-wide rate limits. Hard-coded; not configurable.

Safekey

A filesystem-safe key derived deterministically from a Ref.

TdmGrant

A successful TDM grant.

Enums

CapabilityError

Errors that can arise during CapabilityProfile::from_env.

DenialReason

Closed-set reasons a denial-class error envelope can carry on its optional denial_context.reason field.

ErrorCode

The closed set of error codes doiget surfaces.

Ref

A reference to a paper, either by DOI or arXiv id.

RefParseError

Reasons a Doi::parse / ArxivId::parse / Ref::parse call can fail.

Constants

CITATION_CACHE_TTL_DAYS

Time-to-live for entries in ~/.cache/doiget/citations/. See docs/CACHE.md §3.

DOI_SUFFIX_MAX_LEN

Maximum DOI suffix length accepted at validation. See docs/SECURITY.md §1.1.

MAX_BATCH_REFS

Slice 2 alias for MCP_BATCH_MAX_SIZE using the spec-language name (docs/MCP_TOOLS.md §1 / Slice 2 plan). The numeric value MUST equal MCP_BATCH_MAX_SIZE; an internal test pins the equivalence so the two constants cannot drift.

MAX_CONCURRENT_FETCHES

Hard-coded rate limit. See docs/LEGAL.md §6 safeguard 8.

MAX_FETCHES_PER_SECOND

Hard-coded rate limit. See docs/LEGAL.md §6 safeguard 8.

MCP_BATCH_MAX_SIZE

Maximum batch size for doiget batch and doiget_batch_fetch.

MCP_QUEUE_DEPTH_MAX

Maximum queued MCP requests beyond MAX_CONCURRENT_FETCHES. Excess returns ErrorCode::RateLimited. See docs/SECURITY.md §1.4 / docs/MCP_TOOLS.md.

MCP_STDIN_EOF_SHUTDOWN_SEC

MCP server stdin-EOF graceful-shutdown deadline, in seconds. See ADR-0001 and docs/MCP_TOOLS.md §8.

PDF_MAX_BYTES

Maximum PDF body size accepted by the fetcher, in bytes. See docs/SECURITY.md §1.2 (Oversized PDF).

RESOLVER_CACHE_TTL_DAYS

Time-to-live for entries in ~/.cache/doiget/resolver/. See docs/CACHE.md §3.

SCHEMA_VERSION

TOML schema version this build writes. See docs/STORE.md §3.

VERSION

Crate version. Used by doiget-cli --version and doiget_health.