Skip to main content

internal/infra/pg/reportembedding.go

internal/infra/pg · 728 lines · 29 declarations · source

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

Declarations​

const ReportEmbeddingInputVersion​

const ReportEmbeddingInputVersion = "community-report/v1"

source

const MaxReportEmbeddingSources​

const MaxReportEmbeddingSources = 4096

source

const MaxReportEmbeddingPageSources​

const MaxReportEmbeddingPageSources = 4096

source

const MaxReportEmbeddingTargets​

const MaxReportEmbeddingTargets = 1_000_000

source

type ReportEmbeddingGeneration​

type ReportEmbeddingGeneration 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_report_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 ReportEmbeddingBuildPage​

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

source

type ReportCandidateOptions​

type ReportCandidateOptions struct {
	Limit           int
	Candidates      int
	applicationRead bool
}

source

type ReportCandidate​

type ReportCandidate struct {
	ReportID         string  `json:"report_id"`
	CommunityID      string  `json:"community_id"`
	TitlePreview     string  `json:"title_preview"`
	TitleBytes       int     `json:"title_bytes"`
	TitleTruncated   bool    `json:"title_truncated"`
	SummaryPreview   string  `json:"summary_preview"`
	SummaryBytes     int     `json:"summary_bytes"`
	SummaryTruncated bool    `json:"summary_truncated"`
	Importance       float32 `json:"importance"`
	Similarity       float64 `json:"similarity"`
}

source

type ReportCandidateResult​

type ReportCandidateResult struct {
	Generation  ReportEmbeddingGeneration `json:"generation"`
	Approximate bool                      `json:"approximate"`
	Candidates  []ReportCandidate         `json:"candidates"`
}

source

type ReportEmbeddingStore​

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

source

func NewReportEmbeddingStore​

func NewReportEmbeddingStore(pool *pgxpool.Pool, schema Schema) *ReportEmbeddingStore

source

func reportEmbeddingModelIdentity​

func reportEmbeddingModelIdentity(model EmbeddingModel) string

source

const reportEmbeddingGenerationSQL​

const reportEmbeddingGenerationSQL = `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_report_id::text,''),
 b.state,coalesce(a.generation_id=g.generation_id,false),b.examined,b.stored,g.created_at
 FROM {schema}.report_embedding_generation g JOIN {schema}.report_embedding_build b USING(scope,generation_id)
 LEFT JOIN {schema}.report_embedding_active a USING(scope)`

source

func scanReportEmbeddingGeneration​

func scanReportEmbeddingGeneration(row pgx.Row) (ReportEmbeddingGeneration, error)

source

func reportEmbeddingPartitionName​

func reportEmbeddingPartitionName(id string) string

source

method ReportEmbeddingStore.Generation​

func (s *ReportEmbeddingStore) Generation(ctx context.Context, scope, id string) (ReportEmbeddingGeneration, error)

source

method ReportEmbeddingStore.Active​

func (s *ReportEmbeddingStore) Active(ctx context.Context, scope string) (ReportEmbeddingGeneration, error)

source

method ReportEmbeddingStore.Start​

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

Start materializes current report IDs behind the project row lock. Report writes take that lock in their migration trigger, so a concurrent report is included or makes the finite snapshot stale.

source

type reportEmbeddingInput​

type reportEmbeddingInput struct {
	report       string
	text         string
	digest       []byte
	sources      []string
	sourceDigest []byte
}

source

func digestReportSources​

func digestReportSources(sources []string) []byte

source

method ReportEmbeddingStore.loadReportEmbeddingInput​

func (s *ReportEmbeddingStore) loadReportEmbeddingInput(ctx context.Context, q interface {
	QueryRow(context.Context, string, ...any) pgx.Row
}, scope, generation, reportID string) (reportEmbeddingInput, bool, error)

loadReportEmbeddingInput reads the report and its exact source registrations. The pure report package bounds and validates the semantic input before provider egress.

source

method ReportEmbeddingStore.BuildPage​

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

source

method ReportEmbeddingStore.commitReportEmbeddingInputs​

func (s *ReportEmbeddingStore) commitReportEmbeddingInputs(ctx context.Context, tx pgx.Tx,
	generation ReportEmbeddingGeneration, entries []reportEmbeddingInput, vectors [][]float32) (int, error)

source

method ReportEmbeddingStore.Activate​

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

source

method ReportEmbeddingStore.Repair​

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

source

method ReportEmbeddingStore.Cancel​

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

source

method ReportEmbeddingStore.Prune​

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

source

method ReportEmbeddingStore.Search​

func (s *ReportEmbeddingStore) Search(ctx context.Context, scope, actor string, model EmbeddingModel,
	vector []float32, options ReportCandidateOptions) (ReportCandidateResult, error)

source

method ReportEmbeddingStore.FindCandidates​

func (s *ReportEmbeddingStore) FindCandidates(ctx context.Context, scope string, model EmbeddingModel,
	vector []float32, options ReportCandidateOptions) (ReportCandidateResult, error)

FindCandidates is for a caller that has already enforced project authorization and credential audit. It returns bounded report previews and cannot mutate memory.

source