feat(bell): add immutable event store [T-015]

2026-08-10 23:53:11 +08:00
parent 5b42f58f23
commit 236964d242
28 changed files with 1434 additions and 27 deletions
@@ -0,0 +1,302 @@
+// Package event assembles and validates immutable Bell event facts.
+package event
+
+import (
+	"bytes"
+	"context"
+	"crypto/sha256"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"io"
+	"math"
+	"net"
+	"net/url"
+	"regexp"
+	"strings"
+	"time"
+
+	"github.com/oklog/ulid/v2"
+	jsonschema "github.com/santhosh-tekuri/jsonschema/v6"
+)
+
+const MaxPayloadBytes = 1 << 20
+
+type ErrorCode string
+
+const (
+	CodeInvalidJSON        ErrorCode = "invalid_json"
+	CodePayloadTooLarge    ErrorCode = "payload_too_large"
+	CodeUpstreamID         ErrorCode = "upstream_id_forbidden"
+	CodeSchema             ErrorCode = "schema_invalid"
+	CodeTimeOrder          ErrorCode = "time_order_invalid"
+	CodeLatency            ErrorCode = "latency_inconsistent"
+	CodeConfidence         ErrorCode = "confidence_forbidden"
+	CodeEvidence           ErrorCode = "evidence_unsafe"
+	CodePrimarySensor      ErrorCode = "primary_sensor_invalid"
+	CodePrivacyDenied      ErrorCode = "privacy_denied"
+	CodePrivacyUnavailable ErrorCode = "privacy_unavailable"
+)
+
+// ValidationError exposes a stable code without returning sensitive payloads.
+type ValidationError struct {
+	Code ErrorCode
+	Err  error
+}
+
+func (e *ValidationError) Error() string { return string(e.Code) }
+func (e *ValidationError) Unwrap() error { return e.Err }
+
+func validationError(code ErrorCode, err error) error {
+	return &ValidationError{Code: code, Err: err}
+}
+
+// IDGenerator is owned by Bell. Upstream candidates are not allowed to carry id.
+type IDGenerator interface {
+	NewEventID() (string, error)
+}
+
+type ULIDGenerator struct{}
+
+func (ULIDGenerator) NewEventID() (string, error) {
+	return "evt_" + ulid.Make().String(), nil
+}
+
+// PrivacyPolicy resolves the authoritative device/Area policy. Implementations
+// must fail closed when the mapping is missing or stale.
+type PrivacyPolicy interface {
+	VideoAllowed(ctx context.Context, tenantID, siteID, deviceID int64) (bool, error)
+}
+
+// EvidencePolicy checks every evidence/observation URI before persistence.
+type EvidencePolicy interface {
+	ValidateURI(rawURI string) error
+}
+
+// EvidenceGuard rejects reusable credentials, network endpoints and configured
+// customer/tenant names from persisted evidence URIs.
+type EvidenceGuard struct {
+	forbidden []string
+}
+
+func NewEvidenceGuard(forbiddenNames ...string) (*EvidenceGuard, error) {
+	guard := &EvidenceGuard{}
+	for _, name := range forbiddenNames {
+		name = strings.ToLower(strings.TrimSpace(name))
+		if name == "" {
+			return nil, errors.New("forbidden evidence name cannot be blank")
+		}
+		guard.forbidden = append(guard.forbidden, name)
+	}
+	return guard, nil
+}
+
+var ipv4Like = regexp.MustCompile(`(?:^|[^0-9])(?:[0-9]{1,3}\.){3}[0-9]{1,3}(?:[^0-9]|$)`)
+
+func (g *EvidenceGuard) ValidateURI(rawURI string) error {
+	parsed, err := url.Parse(rawURI)
+	if err != nil || parsed.Scheme == "" {
+		return errors.New("evidence URI is not absolute")
+	}
+	if parsed.User != nil || parsed.Port() != "" || parsed.RawQuery != "" || parsed.Fragment != "" {
+		return errors.New("evidence URI contains reusable connection material")
+	}
+	if host := parsed.Hostname(); host != "" && net.ParseIP(host) != nil {
+		return errors.New("evidence URI contains an IP address")
+	}
+	lower := strings.ToLower(rawURI)
+	for _, marker := range []string{"password", "passwd", "credential", "secret", "token=", "rtsp://"} {
+		if strings.Contains(lower, marker) {
+			return errors.New("evidence URI contains a forbidden marker")
+		}
+	}
+	if ipv4Like.MatchString(lower) {
+		return errors.New("evidence URI contains an IPv4-like value")
+	}
+	for _, name := range g.forbidden {
+		if strings.Contains(lower, name) {
+			return errors.New("evidence URI contains a configured sensitive name")
+		}
+	}
+	return nil
+}
+
+type Sensor struct {
+	DeviceID int64  `json:"device_id"`
+	Modality string `json:"modality"`
+	Role     string `json:"role"`
+}
+
+type storedShape struct {
+	ID             string    `json:"id"`
+	TenantID       int64     `json:"tenant_id"`
+	SiteID         int64     `json:"site_id"`
+	DeviceID       int64     `json:"device_id"`
+	SourceEventID  string    `json:"source_event_id"`
+	Sensors        []Sensor  `json:"sensors"`
+	Kind           string    `json:"kind"`
+	Severity       string    `json:"severity"`
+	Confidence     *float64  `json:"confidence"`
+	OccurredAt     time.Time `json:"occurred_at"`
+	DetectedAt     time.Time `json:"detected_at"`
+	LatencySeconds float64   `json:"latency_seconds"`
+	Observation    *struct {
+		BBoxSeqURI     *string `json:"bbox_seq_uri"`
+		KeypointSeqURI *string `json:"keypoint_seq_uri"`
+		SignalSeqURI   *string `json:"signal_seq_uri"`
+	} `json:"observation"`
+	Evidence struct {
+		SnapshotURIs []string `json:"snapshot_uris"`
+		ClipURI      *string  `json:"clip_uri"`
+	} `json:"evidence"`
+}
+
+// Event is a final, schema-valid immutable fact. JSON returns a defensive copy.
+type Event struct {
+	shape   storedShape
+	payload []byte
+	digest  [sha256.Size]byte
+}
+
+func (e Event) ID() string                { return e.shape.ID }
+func (e Event) TenantID() int64           { return e.shape.TenantID }
+func (e Event) SiteID() int64             { return e.shape.SiteID }
+func (e Event) DeviceID() int64           { return e.shape.DeviceID }
+func (e Event) SourceEventID() string     { return e.shape.SourceEventID }
+func (e Event) Kind() string              { return e.shape.Kind }
+func (e Event) Severity() string          { return e.shape.Severity }
+func (e Event) OccurredAt() time.Time     { return e.shape.OccurredAt }
+func (e Event) DetectedAt() time.Time     { return e.shape.DetectedAt }
+func (e Event) Digest() [sha256.Size]byte { return e.digest }
+func (e Event) JSON() []byte              { return bytes.Clone(e.payload) }
+
+type Factory struct {
+	schema   *jsonschema.Schema
+	ids      IDGenerator
+	privacy  PrivacyPolicy
+	evidence EvidencePolicy
+}
+
+func NewFactory(schemaJSON []byte, ids IDGenerator, privacy PrivacyPolicy, evidence EvidencePolicy) (*Factory, error) {
+	if ids == nil || privacy == nil || evidence == nil {
+		return nil, errors.New("event factory dependencies are required")
+	}
+	schemaDoc, err := jsonschema.UnmarshalJSON(bytes.NewReader(schemaJSON))
+	if err != nil {
+		return nil, fmt.Errorf("parse event schema: %w", err)
+	}
+	compiler := jsonschema.NewCompiler()
+	compiler.AssertFormat()
+	if err := compiler.AddResource("event-v0.1.schema.json", schemaDoc); err != nil {
+		return nil, fmt.Errorf("register event schema: %w", err)
+	}
+	compiled, err := compiler.Compile("event-v0.1.schema.json")
+	if err != nil {
+		return nil, fmt.Errorf("compile event schema: %w", err)
+	}
+	return &Factory{schema: compiled, ids: ids, privacy: privacy, evidence: evidence}, nil
+}
+
+// Create turns a producer candidate into the final stored v0.1 event. The
+// candidate must contain every v0.1 field except the Bell-owned id.
+func (f *Factory) Create(ctx context.Context, candidate []byte) (Event, error) {
+	if len(candidate) > MaxPayloadBytes {
+		return Event{}, validationError(CodePayloadTooLarge, nil)
+	}
+	decoder := json.NewDecoder(bytes.NewReader(candidate))
+	decoder.UseNumber()
+	var object map[string]any
+	if err := decoder.Decode(&object); err != nil || object == nil {
+		return Event{}, validationError(CodeInvalidJSON, err)
+	}
+	var trailing any
+	if err := decoder.Decode(&trailing); !errors.Is(err, io.EOF) {
+		return Event{}, validationError(CodeInvalidJSON, errors.New("multiple JSON values"))
+	}
+	if _, exists := object["id"]; exists {
+		return Event{}, validationError(CodeUpstreamID, nil)
+	}
+	id, err := f.ids.NewEventID()
+	if err != nil {
+		return Event{}, fmt.Errorf("generate Bell event id: %w", err)
+	}
+	object["id"] = id
+	payload, err := json.Marshal(object)
+	if err != nil {
+		return Event{}, validationError(CodeInvalidJSON, err)
+	}
+	if len(payload) > MaxPayloadBytes {
+		return Event{}, validationError(CodePayloadTooLarge, nil)
+	}
+	instance, err := jsonschema.UnmarshalJSON(bytes.NewReader(payload))
+	if err != nil {
+		return Event{}, validationError(CodeInvalidJSON, err)
+	}
+	if err := f.schema.Validate(instance); err != nil {
+		return Event{}, validationError(CodeSchema, nil)
+	}
+	var shape storedShape
+	if err := json.Unmarshal(payload, &shape); err != nil {
+		return Event{}, validationError(CodeInvalidJSON, err)
+	}
+	if err := f.assertSemantics(ctx, shape); err != nil {
+		return Event{}, err
+	}
+	return Event{shape: shape, payload: payload, digest: sha256.Sum256(payload)}, nil
+}
+
+func (f *Factory) assertSemantics(ctx context.Context, shape storedShape) error {
+	if shape.DetectedAt.Before(shape.OccurredAt) {
+		return validationError(CodeTimeOrder, nil)
+	}
+	actual := shape.DetectedAt.Sub(shape.OccurredAt).Seconds()
+	if math.Abs(actual-shape.LatencySeconds) >= 0.1 {
+		return validationError(CodeLatency, nil)
+	}
+	if shape.Confidence != nil {
+		return validationError(CodeConfidence, nil)
+	}
+	primary := 0
+	for _, sensor := range shape.Sensors {
+		if sensor.Role == "primary" {
+			primary++
+			if sensor.DeviceID != shape.DeviceID {
+				return validationError(CodePrimarySensor, nil)
+			}
+		}
+		if sensor.Modality == "video" {
+			allowed, err := f.privacy.VideoAllowed(ctx, shape.TenantID, shape.SiteID, sensor.DeviceID)
+			if err != nil {
+				return validationError(CodePrivacyUnavailable, nil)
+			}
+			if !allowed {
+				return validationError(CodePrivacyDenied, nil)
+			}
+		}
+	}
+	if primary != 1 {
+		return validationError(CodePrimarySensor, nil)
+	}
+	var uris []string
+	uris = append(uris, shape.Evidence.SnapshotURIs...)
+	if shape.Evidence.ClipURI != nil {
+		uris = append(uris, *shape.Evidence.ClipURI)
+	}
+	if shape.Observation != nil {
+		for _, value := range []*string{
+			shape.Observation.BBoxSeqURI,
+			shape.Observation.KeypointSeqURI,
+			shape.Observation.SignalSeqURI,
+		} {
+			if value != nil {
+				uris = append(uris, *value)
+			}
+		}
+	}
+	for _, rawURI := range uris {
+		if err := f.evidence.ValidateURI(rawURI); err != nil {
+			return validationError(CodeEvidence, nil)
+		}
+	}
+	return nil
+}
@@ -0,0 +1,184 @@
+package event_test
+
+import (
+	"context"
+	"encoding/json"
+	"errors"
+	"os"
+	"path/filepath"
+	"testing"
+
+	"yovision/bell/contracts"
+	"yovision/bell/internal/event"
+)
+
+const fixedEventID = "evt_01J8XQ2K7M3P5R9T0V4W6Y8Z2B"
+
+type fixedIDs struct{ id string }
+
+func (f fixedIDs) NewEventID() (string, error) { return f.id, nil }
+
+type privacy struct {
+	allowed bool
+	err     error
+	calls   int
+}
+
+func (p *privacy) VideoAllowed(context.Context, int64, int64, int64) (bool, error) {
+	p.calls++
+	return p.allowed, p.err
+}
+
+func contractPath(name string) string {
+	return filepath.Join("..", "..", "..", "docs", "raw", "contracts", name)
+}
+
+func candidate(t *testing.T, name string) []byte {
+	t.Helper()
+	raw, err := os.ReadFile(contractPath(name))
+	if err != nil {
+		t.Fatal(err)
+	}
+	var object map[string]any
+	if err := json.Unmarshal(raw, &object); err != nil {
+		t.Fatal(err)
+	}
+	delete(object, "id")
+	encoded, err := json.Marshal(object)
+	if err != nil {
+		t.Fatal(err)
+	}
+	return encoded
+}
+
+func mutate(t *testing.T, raw []byte, fn func(map[string]any)) []byte {
+	t.Helper()
+	var object map[string]any
+	if err := json.Unmarshal(raw, &object); err != nil {
+		t.Fatal(err)
+	}
+	fn(object)
+	encoded, err := json.Marshal(object)
+	if err != nil {
+		t.Fatal(err)
+	}
+	return encoded
+}
+
+func factory(t *testing.T, policy *privacy) *event.Factory {
+	t.Helper()
+	guard, err := event.NewEvidenceGuard("private-customer")
+	if err != nil {
+		t.Fatal(err)
+	}
+	value, err := event.NewFactory(contracts.EventV01Schema, fixedIDs{id: fixedEventID}, policy, guard)
+	if err != nil {
+		t.Fatal(err)
+	}
+	return value
+}
+
+func assertCode(t *testing.T, err error, code event.ErrorCode) {
+	t.Helper()
+	var validation *event.ValidationError
+	if !errors.As(err, &validation) || validation.Code != code {
+		t.Fatalf("expected %s, got %v", code, err)
+	}
+}
+
+func TestFrozenContractCopyIsExact(t *testing.T) {
+	raw, err := os.ReadFile(contractPath("event-v0.1.schema.json"))
+	if err != nil {
+		t.Fatal(err)
+	}
+	if string(raw) != string(contracts.EventV01Schema) {
+		t.Fatal("Bell contract copy drifted from the frozen source")
+	}
+}
+
+func TestFactoryAcceptsAllFrozenExamples(t *testing.T) {
+	for _, name := range []string{
+		"event-v0.1.example-current.json",
+		"event-v0.1.example-target.json",
+		"event-v0.1.example-radar.json",
+	} {
+		t.Run(name, func(t *testing.T) {
+			policy := &privacy{allowed: true}
+			created, err := factory(t, policy).Create(context.Background(), candidate(t, name))
+			if err != nil {
+				t.Fatal(err)
+			}
+			if created.ID() != fixedEventID || len(created.JSON()) == 0 {
+				t.Fatal("Bell did not assemble the final event")
+			}
+			if name == "event-v0.1.example-radar.json" && policy.calls != 0 {
+				t.Fatal("non-video event unexpectedly consulted video policy")
+			}
+		})
+	}
+}
+
+func TestFactoryRejectsUpstreamIDAndUnknownField(t *testing.T) {
+	policy := &privacy{allowed: true}
+	base := candidate(t, "event-v0.1.example-current.json")
+	withID := mutate(t, base, func(object map[string]any) { object["id"] = fixedEventID })
+	_, err := factory(t, policy).Create(context.Background(), withID)
+	assertCode(t, err, event.CodeUpstreamID)
+
+	unknown := mutate(t, base, func(object map[string]any) { object["surprise"] = true })
+	_, err = factory(t, policy).Create(context.Background(), unknown)
+	assertCode(t, err, event.CodeSchema)
+}
+
+func TestFactoryEnforcesCrossFieldAssertions(t *testing.T) {
+	base := candidate(t, "event-v0.1.example-current.json")
+	tests := []struct {
+		name string
+		code event.ErrorCode
+		edit func(map[string]any)
+	}{
+		{"time-order", event.CodeTimeOrder, func(v map[string]any) { v["occurred_at"] = "2026-08-03T10:31:23.000Z" }},
+		{"latency", event.CodeLatency, func(v map[string]any) { v["latency_seconds"] = 9.0 }},
+		{"confidence", event.CodeConfidence, func(v map[string]any) { v["confidence"] = 0.9 }},
+		{"primary", event.CodePrimarySensor, func(v map[string]any) {
+			v["sensors"] = []any{
+				map[string]any{"device_id": float64(5012), "modality": "video", "role": "primary"},
+				map[string]any{"device_id": float64(5013), "modality": "radar", "role": "primary"},
+			}
+		}},
+	}
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			_, err := factory(t, &privacy{allowed: true}).Create(context.Background(), mutate(t, base, test.edit))
+			assertCode(t, err, test.code)
+		})
+	}
+}
+
+func TestFactoryFailsClosedForPrivacyAndEvidence(t *testing.T) {
+	base := candidate(t, "event-v0.1.example-current.json")
+	_, err := factory(t, &privacy{err: errors.New("mapping unavailable")}).Create(context.Background(), base)
+	assertCode(t, err, event.CodePrivacyUnavailable)
+
+	_, err = factory(t, &privacy{allowed: false}).Create(context.Background(), base)
+	assertCode(t, err, event.CodePrivacyDenied)
+
+	unsafe := mutate(t, base, func(v map[string]any) {
+		evidence := v["evidence"].(map[string]any)
+		evidence["snapshot_uris"] = []any{"rtsp://user:password@10.0.0.1:554/private-customer.png"}
+	})
+	_, err = factory(t, &privacy{allowed: true}).Create(context.Background(), unsafe)
+	assertCode(t, err, event.CodeEvidence)
+}
+
+func TestFactoryRequiresFailClosedPoliciesAndPayloadLimit(t *testing.T) {
+	guard, err := event.NewEvidenceGuard()
+	if err != nil {
+		t.Fatal(err)
+	}
+	if _, err := event.NewFactory(contracts.EventV01Schema, fixedIDs{id: fixedEventID}, nil, guard); err == nil {
+		t.Fatal("nil privacy policy unexpectedly accepted")
+	}
+	_, err = factory(t, &privacy{allowed: true}).Create(context.Background(), make([]byte, event.MaxPayloadBytes+1))
+	assertCode(t, err, event.CodePayloadTooLarge)
+}