schema

Documentation

Overview

Package schema holds shared data-schema primitives used by both the catalog reader and (eventually) code-generated clients. The canonical terms encoding in this file MUST remain byte-identical to pipeline/normalize/terms.py — their equality is asserted by testdata/terms_golden.jsonl which is consumed by both test suites.

Index

• func CanonicalizeTerms(t Terms) string
• func TermsHash(t Terms) string
• type Kind
  • func (k Kind) IsLLM() bool
• type Terms

Functions

func CanonicalizeTerms

func CanonicalizeTerms(t Terms) string

CanonicalizeTerms returns the canonical JSON array encoding of Terms. Output: `["commitment","tenancy","os","support_tier","upfront","payment_option"]` with no whitespace, matching json.dumps(..., separators=(",", ":")) on the Python side.

func TermsHash

func TermsHash(t Terms) string

TermsHash returns the 128-bit hex digest (32 chars) of CanonicalizeTerms(t).

Types

type Kind

type Kind string

Kind enumerates the resource kinds understood by this binary. New kinds are appended per §5 and require a binary release; see spec §5 Kind taxonomy. M1 seed ships only LLM kinds because the only live shard is openrouter.

const (
	KindLLMText       Kind = "llm.text"
	KindLLMMultimodal Kind = "llm.multimodal"
	KindLLMEmbedding  Kind = "llm.embedding"
	KindLLMImage      Kind = "llm.image"
	KindLLMAudio      Kind = "llm.audio"
)

Supported Kind values.

func (Kind) IsLLM

func (k Kind) IsLLM() bool

IsLLM reports whether k is one of the llm.* kinds.

type Terms

type Terms struct {
	Commitment    string
	Tenancy       string
	OS            string
	SupportTier   string
	Upfront       string
	PaymentOption string
}

Terms is the six-tuple that participates in terms_hash. Field order in the JSON array is fixed (see CanonicalizeTerms) and a schema_version bump is required to change it.