ConfigKit/internal/engine/compose.go

package engine

import "gitea.maximumdirect.net/eric/ConfigKit/internal/document"

// composedDocument is the private handoff between source assembly and later
// precedence layers. Claims remain ordered by source and declaration traversal;
// origins retain every source that contributes to a composite mapping.
type composedDocument struct {
	root               *document.Node
	acquisition        []acquisitionScope
	importGraph        importInclusionGraph
	collectionPolicies collectionPolicyScope
	claims             []compositionClaim
	sequenceRanges     []sequenceRange
	origins            map[string]Origin
	overlays           []overlayRecord
	applications       []historyApplication
	inputBytes         int64
	inputNodes         int
}

type sequenceHistoryMode uint8

const (
	sequenceHistoryReplace sequenceHistoryMode = iota + 1
	sequenceHistoryAppend
	sequenceHistoryMapped
)

type sequenceHistoryContribution struct {
	path         Path
	mode         sequenceHistoryMode
	offset       int
	length       int
	items        []sequenceItemProjection
	replacements []sequenceItemReplacement
}

// assemblyAccumulator owns one mutable working document. Acquired trees held by
// history applications are separate from its root and remain unchanged.
type assemblyAccumulator struct {
	current *composedDocument
	index   mappingIndex
}

func newAssemblyAccumulator(owned *composedDocument) *assemblyAccumulator {
	return &assemblyAccumulator{current: owned}
}

// cloneComposedDocument detaches mutable composition and provenance state;
// history application trees are immutable and retain their existing sharing.
func cloneComposedDocument(base *composedDocument) *composedDocument {
	if base == nil {
		return nil
	}
	return &composedDocument{
		root: document.Clone(base.root), claims: cloneCompositionClaims(base.claims),
		acquisition:        cloneAcquisitionScopes(base.acquisition),
		importGraph:        base.importGraph.clone(),
		collectionPolicies: cloneCollectionPolicyScope(base.collectionPolicies),
		sequenceRanges:     cloneSequenceRanges(base.sequenceRanges),
		origins:            cloneOriginIndex(base.origins), overlays: cloneOverlayRecords(base.overlays),
		applications: cloneHistoryApplications(base.applications),
		inputBytes:   base.inputBytes, inputNodes: base.inputNodes,
	}
}

type compositionClaim struct {
	node        *document.Node // borrowed immutable contribution, not the assembly tree
	path        Path
	kind        document.Kind
	location    Location
	sourceIndex int
}

type compositionClaims struct {
	ordered []compositionClaim
	byPath  map[string][]compositionClaim
}

// sequenceRange records where one source's sequence items were assigned in the
// composed sequence at path. End is exclusive.
type sequenceRange struct {
	sourceIndex int
	path        Path
	start       int
	end         int
	items       []sequenceItemProjection
	projection  sequenceItemProjectionIndex
}

func (assigned sequenceRange) projections() []sequenceItemProjection {
	if assigned.items != nil {
		return append([]sequenceItemProjection{}, assigned.items...)
	}
	return sequenceItemProjections(assigned.end-assigned.start, assigned.start)
}

func (assigned sequenceRange) outputForInput(inputIndex int) (int, bool) {
	if assigned.projection.outputByInput != nil {
		return assigned.projection.outputForInput(inputIndex)
	}
	if assigned.items != nil {
		return newSequenceItemProjectionIndex(assigned.items).outputForInput(inputIndex)
	}
	if inputIndex < 0 || inputIndex >= assigned.end-assigned.start {
		return 0, false
	}
	return assigned.start + inputIndex, true
}

func (assigned sequenceRange) inputForOutput(outputIndex int) (int, bool) {
	if assigned.projection.inputByOutput != nil {
		return assigned.projection.inputForOutput(outputIndex)
	}
	if assigned.items != nil {
		return newSequenceItemProjectionIndex(assigned.items).inputForOutput(outputIndex)
	}
	if outputIndex < assigned.start || outputIndex >= assigned.end {
		return 0, false
	}
	return outputIndex - assigned.start, true
}

func indexSequenceRange(assigned sequenceRange) sequenceRange {
	if assigned.projection.outputByInput == nil && assigned.projection.inputByOutput == nil {
		assigned.projection = newSequenceItemProjectionIndex(assigned.projections())
	}
	return assigned
}

// composeDocuments recursively joins disjoint mapping entries. Source order
// controls only output traversal and diagnostics: every non-mapping overlap is
// rejected before a result is assembled.
func composeDocuments(inputs ...*document.Node) (*composedDocument, []Diagnostic) {
	return composeDocumentsWithPolicy(sequencePolicy{}, inputs...)
}

// composeDocumentsWithPolicy composes one strict tier using the registered
// sequence rules when policy contains registrations. The ordinary adapter above
// deliberately preserves the existing unregistered behavior for current callers.
func composeDocumentsWithPolicy(policy sequencePolicy, inputs ...*document.Node) (*composedDocument, []Diagnostic) {
	return composeDocumentsWithCollectionPolicy(policy, collectionPolicyScope{}, nil, inputs...)
}

// composeDocumentsWithCollectionPolicy resolves collection declarations and
// eligible lower policy before strictly composing completed peer documents.
func composeDocumentsWithCollectionPolicy(
	policy sequencePolicy,
	lower collectionPolicyScope,
	declarations []sequenceDeclarationSpec,
	inputs ...*document.Node,
) (*composedDocument, []Diagnostic) {
	if len(inputs) == 0 {
		return nil, []Diagnostic{compositionInvalidDiagnostic("strict composition requires at least one document", nil)}
	}

	collected := compositionClaims{byPath: make(map[string][]compositionClaim)}
	for sourceIndex, input := range inputs {
		if input == nil || input.Kind != document.Mapping {
			var location *Location
			if input != nil {
				converted := publicLocation(input.Location)
				location = &converted
			}
			return nil, []Diagnostic{compositionInvalidDiagnostic("strict composition requires mapping document roots", location)}
		}
		collected.append(input, Path{}, sourceIndex)
	}

	identityFieldsByPath, collectionPolicies, diagnostics := resolveStrictCollectionPolicies(
		collected.ordered, policy, lower, declarations,
	)
	if len(diagnostics) != 0 {
		return nil, diagnostics
	}

	checked := make(map[string]struct{}, len(collected.byPath))
	for _, claim := range collected.ordered {
		pointer := claim.path.String()
		if _, seen := checked[pointer]; seen {
			continue
		}
		checked[pointer] = struct{}{}
		claims := collected.byPath[pointer]
		if len(claims) < 2 || allMappingClaims(claims) || (allSequenceClaims(claims) && len(identityFieldsByPath[pointer]) != 0) {
			continue
		}
		return nil, []Diagnostic{compositionConflictDiagnostic(claim.path, claims)}
	}

	validated, ranges, diagnostics := validateRegisteredSequences(collected.ordered, identityFieldsByPath)
	if len(diagnostics) != 0 {
		return nil, diagnostics
	}

	root := document.Clone(inputs[0])
	index := new(mappingIndex)
	for sourceIndex, input := range inputs[1:] {
		mergeComposedMappings(root, input, Path{}, index, sourceIndex+1, identityFieldsByPath, validated)
	}
	applications := make([]historyApplication, 0, len(inputs))
	seenSequences := make(map[string]bool)
	rangesBySource := make([][]sequenceRange, len(inputs))
	for _, assigned := range ranges {
		rangesBySource[assigned.sourceIndex] = append(rangesBySource[assigned.sourceIndex], assigned)
	}
	for sourceIndex, input := range inputs {
		application := historyApplication{layer: LayerBase, value: input}
		for _, assigned := range rangesBySource[sourceIndex] {
			pointer := assigned.path.String()
			mode := sequenceHistoryAppend
			if !seenSequences[pointer] {
				mode = sequenceHistoryReplace
			}
			seenSequences[pointer] = true
			application.sequences = append(application.sequences, sequenceHistoryContribution{
				path: assigned.path.clone(), mode: mode, offset: assigned.start, length: assigned.end - assigned.start,
			})
		}
		application.sequencesByPath = indexSequenceHistoryContributions(application.sequences)
		applications = append(applications, application)
	}
	origins := projectedCompositionOrigins(inputs, ranges)
	completeCompositionOrigins(root, Path{}, origins)
	return &composedDocument{
		root:               root,
		collectionPolicies: collectionPolicies,
		claims:             cloneCompositionClaims(collected.ordered),
		sequenceRanges:     cloneSequenceRanges(ranges),
		origins:            origins,
		applications:       applications,
	}, nil
}

// Conflict claims treat sequences atomically, but inspection needs origins for
// every descendant. Preserve composite mapping claims while filling those gaps.
func completeCompositionOrigins(node *document.Node, path Path, origins map[string]Origin) {
	pointer := path.String()
	if _, present := origins[pointer]; !present {
		origins[pointer] = Origin{Layer: LayerBase, Locations: []Location{publicLocation(node.Location)}}
	}
	switch node.Kind {
	case document.Mapping:
		for _, entry := range node.Mapping {
			completeCompositionOrigins(entry.Value, path.Key(entry.Key), origins)
		}
	case document.Sequence:
		for index, child := range node.Sequence {
			completeCompositionOrigins(child, path.Index(index), origins)
		}
	}
}

func (c *compositionClaims) append(node *document.Node, path Path, sourceIndex int) {
	claim := compositionClaim{
		node:        node,
		path:        path.clone(),
		kind:        node.Kind,
		location:    publicLocation(node.Location),
		sourceIndex: sourceIndex,
	}
	c.ordered = append(c.ordered, claim)
	pointer := path.String()
	c.byPath[pointer] = append(c.byPath[pointer], claim)

	if node.Kind != document.Mapping {
		return
	}
	for _, entry := range node.Mapping {
		c.append(entry.Value, path.Key(entry.Key), sourceIndex)
	}
}

func allSequenceClaims(claims []compositionClaim) bool {
	for _, claim := range claims {
		if claim.kind != document.Sequence {
			return false
		}
	}
	return true
}

type validatedSequence struct {
	items []*document.Node
}

func resolveStrictCollectionPolicies(
	claims []compositionClaim,
	applicationPolicy sequencePolicy,
	lower collectionPolicyScope,
	declarations []sequenceDeclarationSpec,
) (map[string][]string, collectionPolicyScope, []Diagnostic) {
	paths := make([]Path, 0, len(claims)+len(declarations))
	seen := make(map[string]bool, len(claims)+len(declarations))
	sequencePresent := make(map[string]bool)
	declarationsByPath := make(map[string][]sequenceDeclarationSpec, len(declarations))
	for _, claim := range claims {
		if claim.kind != document.Sequence {
			continue
		}
		key := claim.path.String()
		sequencePresent[key] = true
		if !seen[key] {
			seen[key] = true
			paths = append(paths, claim.path.clone())
		}
	}
	for _, declaration := range declarations {
		key := declaration.destination.String()
		declarationsByPath[key] = append(declarationsByPath[key], declaration)
		if !seen[key] {
			seen[key] = true
			paths = append(paths, declaration.destination.clone())
		}
	}

	identityFieldsByPath := make(map[string][]string)
	effectiveLower := cloneCollectionPolicyScope(lower)
	collectionPolicies := cloneCollectionPolicyScope(effectiveLower)
	var diagnostics []Diagnostic
	for _, path := range paths {
		key := path.String()
		currentDeclarations := declarationsByPath[key]

		lowerEntry, hasLower := effectiveLower.collections[key]
		var lowerPolicy *collectionPolicy
		if hasLower {
			copy := cloneCollectionPolicy(lowerEntry.policy)
			lowerPolicy = &copy
		}
		applicationIdentity, _ := applicationPolicy.identityFieldsAt(path)

		// A replacing value drops the old YAML identity before this tier's
		// peers are checked. The explicit operation itself remains applicable.
		selectedMode := sequenceMode(0)
		for _, declaration := range currentDeclarations {
			if declaration.modePresent {
				selectedMode = declaration.mode
				break
			}
		}
		if selectedMode == 0 && lowerPolicy != nil && lowerPolicy.explicitModePresent {
			selectedMode = lowerPolicy.explicitMode
		}
		if sequencePresent[key] && selectedMode == sequenceModeReplace && lowerPolicy != nil {
			lowerPolicy.identityFields = nil
			lowerPolicy.identityPresent = false
			lowerPolicy.identityLocation = nil
			lowerPolicy.recordOrder = nil
			lowerPolicy.recordPolicies = nil
			effectiveLower.retireCollectionPoliciesAtOrBelow(path)
			collectionPolicies.retireCollectionPoliciesAtOrBelow(path)
		}

		resolution, found := resolveCollectionPolicy(path, currentDeclarations, applicationIdentity, lowerPolicy)
		if len(found) != 0 {
			diagnostics = append(diagnostics, found...)
			continue
		}
		if len(resolution.effectiveIdentityFields) != 0 {
			identityFieldsByPath[key] = cloneStrings(resolution.effectiveIdentityFields)
		}

		policyForPath := resolution.policy
		if hasLower && !sequencePresent[key] {
			// A declaration without a new value changes policy metadata but
			// leaves retained record ownership untouched.
			policyForPath = cloneCollectionPolicy(lowerEntry.policy)
			policyForPath.identityFields = cloneStrings(resolution.policy.identityFields)
			policyForPath.identityPresent = resolution.policy.identityPresent
			policyForPath.identityLocation = cloneLocation(resolution.policy.identityLocation)
			policyForPath.explicitMode = resolution.policy.explicitMode
			policyForPath.explicitModePresent = resolution.policy.explicitModePresent
			policyForPath.modeLocation = cloneLocation(resolution.policy.modeLocation)
		}
		if collectionPolicies.collections == nil {
			collectionPolicies.collections = make(map[string]collectionPolicyEntry)
		}
		collectionPolicies.collections[key] = collectionPolicyEntry{destination: path.clone(), policy: policyForPath}
	}
	return identityFieldsByPath, collectionPolicies, canonicalDiagnostics(diagnostics)
}

func validateRegisteredSequences(claims []compositionClaim, identityFieldsByPath map[string][]string) (map[string]map[int]validatedSequence, []sequenceRange, []Diagnostic) {
	itemsByPath := make(map[string]map[int]validatedSequence)
	identitiesByPath := make(map[string]map[string]document.Location)
	lengthsByPath := make(map[string]int)
	var ranges []sequenceRange
	for _, claim := range claims {
		if claim.kind != document.Sequence {
			continue
		}
		identityFields := identityFieldsByPath[claim.path.String()]
		if len(identityFields) == 0 {
			continue
		}
		node := claim.node
		pathKey := claim.path.String()
		identities := identitiesByPath[pathKey]
		if identities == nil {
			identities = make(map[string]document.Location, len(node.Sequence))
			identitiesByPath[pathKey] = identities
		}
		start := lengthsByPath[pathKey]
		contribution := validatedSequence{items: node.Sequence}
		records, diagnostics := validateNormalizedSequenceIdentityFrom(node, claim.path, identityFields, sequenceIdentityDiagnosticCodes{
			invalid: CodeComposeInvalid, conflict: CodeComposeConflict,
		}, start, sequenceStructuralRecord)
		if len(diagnostics) != 0 {
			return nil, nil, diagnostics
		}
		byIndex := make([]sequenceIdentityRecord, len(node.Sequence))
		for _, record := range records {
			byIndex[record.index] = record
		}
		for _, record := range byIndex {
			if first, duplicate := identities[record.tuple]; duplicate {
				current := publicLocation(record.location)
				return nil, nil, []Diagnostic{{
					Code: CodeComposeConflict, Severity: SeverityError,
					Path: claim.path.clone(), Message: "registered sequence contains duplicate identities",
					Primary: ptrLocation(publicLocation(first)), Related: []Location{current},
				}}
			}
			identities[record.tuple] = record.location
		}
		if itemsByPath[pathKey] == nil {
			itemsByPath[pathKey] = make(map[int]validatedSequence)
		}
		itemsByPath[pathKey][claim.sourceIndex] = contribution
		lengthsByPath[pathKey] = start + len(node.Sequence)
		ranges = append(ranges, sequenceRange{
			sourceIndex: claim.sourceIndex, path: claim.path.clone(), start: start, end: start + len(node.Sequence),
		})
	}
	return itemsByPath, ranges, nil
}

func mergeComposedMappings(target, incoming *document.Node, path Path, index *mappingIndex, sourceIndex int, identityFieldsByPath map[string][]string, validated map[string]map[int]validatedSequence) {
	for _, incomingEntry := range incoming.Mapping {
		entryPath := path.Key(incomingEntry.Key)
		position := index.find(target, incomingEntry.Key)
		if position < 0 {
			cloned := incomingEntry
			cloned.Value = document.Clone(incomingEntry.Value)
			index.append(target, cloned)
			continue
		}
		current := target.Mapping[position].Value
		if current.Kind == document.Sequence && incomingEntry.Value.Kind == document.Sequence && len(identityFieldsByPath[entryPath.String()]) != 0 {
			contribution := validated[entryPath.String()][sourceIndex]
			if contribution.items != nil {
				for _, item := range contribution.items {
					current.Sequence = append(current.Sequence, document.Clone(item))
				}
			}
			continue
		}
		mergeDisjointMappingsAtPath(current, incomingEntry.Value, entryPath, index, sourceIndex, identityFieldsByPath, validated)
	}
}

func mergeDisjointMappingsAtPath(target, incoming *document.Node, path Path, index *mappingIndex, sourceIndex int, identityFieldsByPath map[string][]string, validated map[string]map[int]validatedSequence) {
	if target.Kind != document.Mapping || incoming.Kind != document.Mapping {
		return
	}
	mergeComposedMappings(target, incoming, path, index, sourceIndex, identityFieldsByPath, validated)
}

func cloneSequenceRanges(ranges []sequenceRange) []sequenceRange {
	cloned := make([]sequenceRange, len(ranges))
	for i, assigned := range ranges {
		cloned[i] = assigned
		cloned[i].path = assigned.path.clone()
		if assigned.items != nil {
			cloned[i].items = append([]sequenceItemProjection{}, assigned.items...)
		}
		cloned[i].projection.outputByInput = append([]int(nil), assigned.projection.outputByInput...)
		cloned[i].projection.inputByOutput = append([]int(nil), assigned.projection.inputByOutput...)
	}
	return cloned
}

func allMappingClaims(claims []compositionClaim) bool {
	for _, claim := range claims {
		if claim.kind != document.Mapping {
			return false
		}
	}
	return true
}

func mergeDisjointMappings(target, incoming *document.Node, index *mappingIndex) {
	for _, incomingEntry := range incoming.Mapping {
		position := index.find(target, incomingEntry.Key)
		if position < 0 {
			cloned := incomingEntry
			cloned.Value = document.Clone(incomingEntry.Value)
			index.append(target, cloned)
			continue
		}
		// Claim validation guarantees that the only remaining overlap is a
		// mapping pair. Empty mappings naturally contribute no child entries.
		mergeDisjointMappings(target.Mapping[position].Value, incomingEntry.Value, index)
	}
}

func documentEntryIndex(entries []document.Entry, key string) int {
	for index := range entries {
		if entries[index].Key == key {
			return index
		}
	}
	return -1
}

func projectedCompositionOrigins(inputs []*document.Node, ranges []sequenceRange) map[string]Origin {
	origins := make(map[string]Origin)
	sequencesBySource := make([][]sequenceHistoryContribution, len(inputs))
	for _, assigned := range ranges {
		sequencesBySource[assigned.sourceIndex] = append(sequencesBySource[assigned.sourceIndex], sequenceHistoryContribution{
			path: assigned.path.clone(), mode: sequenceHistoryMapped, items: assigned.projections(),
		})
	}
	for sourceIndex, input := range inputs {
		application := historyApplication{value: input, sequences: sequencesBySource[sourceIndex]}
		application.sequencesByPath = indexSequenceHistoryContributions(application.sequences)
		walkHistoryApplication(application, func(_, effective Path, node *document.Node) {
			pointer := effective.String()
			origin := origins[pointer]
			origin.Layer = LayerBase
			origin.Locations = append(origin.Locations, publicLocation(node.Location))
			origins[pointer] = origin
		})
	}
	return origins
}

type sequenceRangeKey struct {
	sourceIndex int
	path        string
}

type sequenceRangeIndex map[sequenceRangeKey]sequenceRange

func indexSequenceRanges(ranges []sequenceRange) sequenceRangeIndex {
	if len(ranges) == 0 {
		return nil
	}
	index := make(sequenceRangeIndex, len(ranges))
	for _, assigned := range ranges {
		assigned = indexSequenceRange(assigned)
		index[sequenceRangeKey{sourceIndex: assigned.sourceIndex, path: assigned.path.String()}] = assigned
	}
	return index
}

func applyCompositionOrigins(decoded *decodedConfiguration, origins map[string]Origin) {
	for pointer, origin := range origins {
		if _, present := decoded.presence[pointer]; present {
			decoded.origins[pointer] = cloneOrigin(origin)
		}
	}
}

func cloneCompositionClaims(claims []compositionClaim) []compositionClaim {
	cloned := make([]compositionClaim, len(claims))
	for index, claim := range claims {
		cloned[index] = claim
		cloned[index].path = claim.path.clone()
	}
	return cloned
}

func compositionInvalidDiagnostic(message string, location *Location) Diagnostic {
	return Diagnostic{
		Code: CodeComposeInvalid, Severity: SeverityError, Message: message,
		Primary: cloneLocation(location),
	}
}

func compositionConflictDiagnostic(path Path, claims []compositionClaim) Diagnostic {
	primary := claims[0].location
	related := make([]Location, 0, len(claims)-1)
	for _, claim := range claims[1:] {
		related = append(related, claim.location)
	}
	return Diagnostic{
		Code:     CodeComposeConflict,
		Severity: SeverityError,
		Path:     path.clone(),
		Message:  "configuration path is defined by multiple sources",
		Primary:  &primary,
		Related:  related,
	}
}

func publicLocation(location document.Location) Location {
	return Location{SourceID: location.SourceID, Line: location.Line, Column: location.Column}
}