Skip to main content

internal/infra/pg/entityembedding.go

internal/infra/pg · 734 lines · 29 declarations · source

This file carries the package documentation, rendered on the package page.

Declarations​

const EntityEmbeddingInputVersion​

const EntityEmbeddingInputVersion = "entity-evidence/v1"

source

const MaxEntityEmbeddingSources​

const MaxEntityEmbeddingSources = 4096

source

const MaxEntityEmbeddingPageSources​

const MaxEntityEmbeddingPageSources = 4096

source

const MaxEntityEmbeddingTargets​

const MaxEntityEmbeddingTargets = 1_000_000

source

type EntityEmbeddingGeneration​

type EntityEmbeddingGeneration struct {
	ID          string         `json:"id"`
	Project     string         `json:"project"`
	Key         string         `json:"operation_key"`
	Model       EmbeddingModel `json:"model"`
	TargetCount int64          `json:"target_count"`
	After       string         `json:"after_entity_id,omitempty"`
	State       string         `json:"state"`
	Active      bool           `json:"active"`
	Examined    int64          `json:"examined"`
	Stored      int64          `json:"stored"`
	CreatedAt   time.Time      `json:"created_at"`
}

source

type EntityEmbeddingBuildPage​

type EntityEmbeddingBuildPage struct {
	Examined int    `json:"examined"`
	Stored   int    `json:"stored"`
	Complete bool   `json:"complete"`
	After    string `json:"after_entity_id,omitempty"`
}

source

type EntityCandidateOptions​

type EntityCandidateOptions struct {
	Limit           int
	Candidates      int
	applicationRead bool
}

source

type EntityCandidate​

type EntityCandidate struct {
	EntityID      string  `json:"entity_id"`
	IdentityKind  string  `json:"identity_kind"`
	NamePreview   string  `json:"name_preview"`
	NameBytes     int     `json:"name_bytes"`
	NameTruncated bool    `json:"name_truncated"`
	Similarity    float64 `json:"similarity"`
}

source

type EntityCandidateResult​

type EntityCandidateResult struct {
	Generation  EntityEmbeddingGeneration `json:"generation"`
	Approximate bool                      `json:"approximate"`
	Candidates  []EntityCandidate         `json:"candidates"`
}

source

type EntityEmbeddingStore​

type EntityEmbeddingStore struct {
	pool   *pgxpool.Pool
	schema Schema
}

source

func NewEntityEmbeddingStore​

func NewEntityEmbeddingStore(pool *pgxpool.Pool, schema Schema) *EntityEmbeddingStore

source

func entityEmbeddingModelIdentity​

func entityEmbeddingModelIdentity(model EmbeddingModel) string

source

const entityEmbeddingGenerationSQL​

const entityEmbeddingGenerationSQL = `SELECT g.generation_id::text,g.scope,g.operation_key::text,
 g.model_name,g.model_revision,g.endpoint_hash,g.dimensions,g.target_count,coalesce(b.after_entity_id::text,''),
 b.state,coalesce(a.generation_id=g.generation_id,false),b.examined,b.stored,g.created_at
 FROM {schema}.entity_embedding_generation g JOIN {schema}.entity_embedding_build b USING(scope,generation_id)
 LEFT JOIN {schema}.entity_embedding_active a USING(scope)`

source

func scanEntityEmbeddingGeneration​

func scanEntityEmbeddingGeneration(row pgx.Row) (EntityEmbeddingGeneration, error)

source

func entityEmbeddingPartitionName​

func entityEmbeddingPartitionName(id string) string

source

method EntityEmbeddingStore.Generation​

func (s *EntityEmbeddingStore) Generation(ctx context.Context, scope, id string) (EntityEmbeddingGeneration, error)

source

method EntityEmbeddingStore.Active​

func (s *EntityEmbeddingStore) Active(ctx context.Context, scope string) (EntityEmbeddingGeneration, error)

source

method EntityEmbeddingStore.Start​

func (s *EntityEmbeddingStore) Start(ctx context.Context, scope, key, actor string, model EmbeddingModel) (EntityEmbeddingGeneration, error)

Start materializes the current named-entity IDs behind the project row lock. Named-entity inserts take that lock in their migration trigger, so an insert is either included or makes this finite snapshot stale after Start commits.

source

type entityEmbeddingInput​

type entityEmbeddingInput struct {
	entity       string
	text         string
	digest       []byte
	sources      []string
	sourceDigest []byte
}

source

func digestEntitySources​

func digestEntitySources(sources []string) []byte

source

method EntityEmbeddingStore.loadEntityEmbeddingInput​

func (s *EntityEmbeddingStore) loadEntityEmbeddingInput(ctx context.Context, q interface {
	QueryRow(context.Context, string, ...any) pgx.Row
}, scope, generation, entity string) (entityEmbeddingInput, bool, error)

loadEntityEmbeddingInput bounds transferred aliases, quotes and source IDs before the caller allocates them. Quotes beyond the validation budget cannot affect the 2KB surface.

source

method EntityEmbeddingStore.BuildPage​

func (s *EntityEmbeddingStore) BuildPage(ctx context.Context, scope, id, actor string, model EmbeddingModel, limit int, embed EmbeddingBatch) (EntityEmbeddingBuildPage, error)

source

method EntityEmbeddingStore.commitEntityEmbeddingInputs​

func (s *EntityEmbeddingStore) commitEntityEmbeddingInputs(ctx context.Context, tx pgx.Tx,
	generation EntityEmbeddingGeneration, entries []entityEmbeddingInput, vectors [][]float32) (int, error)

source

method EntityEmbeddingStore.Activate​

func (s *EntityEmbeddingStore) Activate(ctx context.Context, scope, id, actor string) error

source

method EntityEmbeddingStore.Repair​

func (s *EntityEmbeddingStore) Repair(ctx context.Context, scope, id, actor string) error

source

method EntityEmbeddingStore.Cancel​

func (s *EntityEmbeddingStore) Cancel(ctx context.Context, scope, id, actor string) error

source

method EntityEmbeddingStore.Prune​

func (s *EntityEmbeddingStore) Prune(ctx context.Context, scope, id, actor string, limit int) (int, error)

source

method EntityEmbeddingStore.Search​

func (s *EntityEmbeddingStore) Search(ctx context.Context, scope, actor string, model EmbeddingModel,
	vector []float32, options EntityCandidateOptions) (EntityCandidateResult, error)

source

method EntityEmbeddingStore.FindCandidates​

func (s *EntityEmbeddingStore) FindCandidates(ctx context.Context, scope string, model EmbeddingModel,
	vector []float32, options EntityCandidateOptions) (EntityCandidateResult, error)

FindCandidates is for a caller that has already enforced project authorization and credential audit. It returns candidate identities only and cannot merge or mutate them.

source