PureXML/Sources/Schema/IdentityValidator.swift

extension PureXML.Schema {
    /// Validates XSD identity constraints (`unique`, `key`, `keyref`) over an
    /// instance document. Each constraint is evaluated at every element whose
    /// name carries it: the selector XPath chooses target nodes relative to that
    /// element, and the field XPaths form each target's value tuple. `key` and
    /// `unique` require distinct tuples; `keyref` requires each tuple to match a
    /// key or unique of the named constraint visible in an enclosing scope.
    struct IdentityValidator {
        /// The constraints declared for each element local name.
        let constraints: [String: [IdentityConstraint]]
        /// Simple types for identity-constraint fields keyed by
        /// ``XSDParser/identityFieldKey(selector:field:)``.
        let fieldTypes: [String: SimpleType]
        /// `default`/`fixed` value constraints on identity-constraint field
        /// targets, keyed the same way as ``fieldTypes``. When a field selects an
        /// absent attribute or an empty element whose declaration carries one of
        /// these, the field's identity value is that value, not nothing.
        let fieldConstraints: [String: PureXML.Schema.ValueConstraint]
        /// Keys (``XSDParser/identityFieldKey(constraint:field:)``) of fields whose
        /// declared target is not a simple type, computed at compile time, so the
        /// field violates cvc-identity-constraint.3 (XSTS idZ010).
        let nonSimpleFields: Set<String>
        /// Global element and type declarations for runtime field typing.
        let types: [String: ElementType]
        /// Selector and field XPaths compiled once per run, not per element.
        private let cache = XPathQueryCache()
        /// Same-name sibling positions for error-location paths, indexed once per
        /// parent and name. ``steps`` looks a target's position up here instead of
        /// rescanning the parent's children every call, which made locating each
        /// of a wide list's targets O(n) and the whole `unique`/`key`/`keyref`
        /// pass quadratic.
        let stepIndex = StepSiblingIndex()

        init(
            constraints: [String: [IdentityConstraint]],
            fieldTypes: [String: SimpleType] = [:],
            fieldConstraints: [String: PureXML.Schema.ValueConstraint] = [:],
            nonSimpleFields: Set<String> = [],
            types: [String: ElementType] = [:],
        ) {
            self.constraints = constraints
            self.fieldTypes = fieldTypes
            self.fieldConstraints = fieldConstraints
            self.nonSimpleFields = nonSimpleFields
            self.types = types
        }

        func validate(
            _ root: PureXML.Model.TreeNode,
            at path: [PureXML.Validation.PathKey] = [],
        ) -> [PureXML.Validation.ValidationError] {
            validateTagged(root, at: path).map(\.error)
        }

        /// The tagged form of ``validate(_:at:)``: the same computation, with each
        /// finding labeled by the decomposed identity rule it belongs to. The
        /// public, untagged form is a thin `.map(\.error)` over this so behavior and
        /// output stay identical for every existing caller.
        func validateTagged(
            _ root: PureXML.Model.TreeNode,
            at path: [PureXML.Validation.PathKey] = [],
        ) -> [XSDFinding] {
            guard !constraints.isEmpty else { return [] }
            // A selector or field XPath that does not compile is a schema-author
            // error, reported once up front rather than silently disabling the
            // constraint (the broken query still evaluates as no-match below).
            var issues: [XSDFinding] = compileErrors(at: path)
            walk(root, at: path, scopes: [], into: &issues)
            return issues
        }

        /// One located error per constraint XPath that fails to compile, in a
        /// deterministic order. The failed compilations are cached, so the walk
        /// afterwards treats those constraints as matching nothing.
        private func compileErrors(at path: [PureXML.Validation.PathKey]) -> [XSDFinding] {
            var errors: [XSDFinding] = []
            for key in constraints.keys.sorted() {
                for constraint in constraints[key] ?? [] {
                    if cache.query(constraint.selector) == nil {
                        errors.append(XSDFinding(
                            .identityConstraintXPathsValid,
                            reason: "identity constraint '\(constraint.name)': invalid selector XPath '\(constraint.selector)'",
                            at: path,
                        ))
                    }
                    for field in constraint.fields where cache.query(field) == nil {
                        errors.append(XSDFinding(.identityConstraintXPathsValid, reason: "identity constraint '\(constraint.name)': invalid field XPath '\(field)'", at: path))
                    }
                }
            }
            return errors
        }

        private func walk(
            _ node: PureXML.Model.TreeNode,
            at path: [PureXML.Validation.PathKey],
            scopes: [[String: [[FieldValue?]]]],
            into issues: inout [XSDFinding],
        ) {
            var frame: [String: [[FieldValue?]]] = [:]
            let declared = node.name.flatMap { constraints[$0.localName] } ?? []
            // Collect key and unique tuples first so a keyref on the same element
            // can resolve against them.
            for constraint in declared {
                collect(constraint, at: node, path: path, frame: &frame, into: &issues)
            }
            let visible = scopes + [frame]
            for constraint in declared {
                if case .keyref = constraint.kind {
                    checkKeyref(constraint, at: node, path: path, scopes: visible, into: &issues)
                }
            }
            let elementChildren = node.children.filter { $0.kind == .element }
            for (child, step) in zip(elementChildren, childSteps(elementChildren)) {
                walk(child, at: path + [step], scopes: visible, into: &issues)
            }
        }

        /// The coding-path step for each element child: its name, with a sibling
        /// index only when more than one child shares that name. Mirrors the
        /// complex-type validator's path construction so identity-constraint
        /// errors and content errors locate consistently.
        private func childSteps(_ children: [PureXML.Model.TreeNode]) -> [PureXML.Validation.PathKey] {
            PureXML.Validation.PathKey.steps(forChildNames: children.map { $0.name?.description ?? "" })
        }

        /// The coding-path step from a `target` down to the single field a
        /// constraint names, so an error locates the offending value (`@id`, a
        /// child element) rather than just the target. Empty for a multi-field
        /// constraint or a complex field path, in which case the error stays on the
        /// target.
        private func fieldStep(_ fields: [String], at target: PureXML.Model.TreeNode, namespaces: [String: String]) -> [PureXML.Validation.PathKey] {
            guard fields.count == 1 else { return [] }
            let field = fields[0]
            if field.hasPrefix("@"), !field.dropFirst().contains(where: { $0 == "/" || $0 == "[" }) {
                return [.attribute(String(field.dropFirst()))]
            }
            let isSimpleName = !field.contains(where: { "/@[(".contains($0) })
            if isSimpleName, let child = select(field, at: target, namespaces: namespaces).first, child.kind == .element {
                return steps(from: target, to: child)
            }
            return []
        }

        private func collect(
            _ constraint: IdentityConstraint,
            at node: PureXML.Model.TreeNode,
            path: [PureXML.Validation.PathKey],
            frame: inout [String: [[FieldValue?]]],
            into issues: inout [XSDFinding],
        ) {
            guard isNonRef(constraint) else { return }
            var tuples: [[FieldValue?]] = []
            var seen = TupleSet()
            for target in select(constraint.selector, at: node, namespaces: constraint.namespaceBindings) {
                let targetPath = path + steps(from: node, to: target)
                if let error = fieldError(constraint, at: target, path: targetPath) {
                    issues.append(error)
                    continue
                }
                let tuple = fieldTuple(constraint.fields, at: target, constraint: constraint)
                if constraint.kind == .key, tuple.contains(where: { $0 == nil }) {
                    issues.append(XSDFinding(
                        .identityKeyFieldPresent,
                        reason: "key '\(constraint.name)': a field is missing",
                        at: targetPath + fieldStep(constraint.fields, at: target, namespaces: constraint.namespaceBindings),
                    ))
                    continue
                }
                if tuple.contains(where: { $0 == nil }) {
                    continue
                }
                if seen.contains(tuple) {
                    issues.append(XSDFinding(
                        .identityValueUnique,
                        reason: "\(label(constraint)) '\(constraint.name)': duplicate value",
                        at: targetPath + fieldStep(constraint.fields, at: target, namespaces: constraint.namespaceBindings),
                    ))
                } else {
                    seen.insert(tuple)
                }
                tuples.append(tuple)
            }
            frame[constraint.name] = tuples
        }

        private func checkKeyref(
            _ constraint: IdentityConstraint,
            at node: PureXML.Model.TreeNode,
            path: [PureXML.Validation.PathKey],
            scopes: [[String: [[FieldValue?]]]],
            into issues: inout [XSDFinding],
        ) {
            guard case let .keyref(refer) = constraint.kind else { return }
            let keyTuples = scopes.reversed().compactMap { $0[refer] }.first ?? []
            var keys = TupleSet()
            for keyTuple in keyTuples {
                keys.insert(keyTuple)
            }
            for target in select(constraint.selector, at: node, namespaces: constraint.namespaceBindings) {
                let targetPath = path + steps(from: node, to: target)
                if let error = fieldError(constraint, at: target, path: targetPath) {
                    issues.append(error)
                    continue
                }
                let tuple = fieldTuple(constraint.fields, at: target, constraint: constraint)
                if tuple.contains(where: { $0 == nil }) {
                    continue
                }
                if !keys.contains(tuple) {
                    issues.append(XSDFinding(
                        .identityKeyrefResolves,
                        reason: "keyref '\(constraint.name)': no matching key '\(refer)'",
                        at: targetPath + fieldStep(constraint.fields, at: target, namespaces: constraint.namespaceBindings),
                    ))
                }
            }
        }

        /// The error when a field of `constraint` selects more than one node at
        /// `target`. A field must evaluate to a node-set with at most one member
        /// (Identity-constraint Definition; cvc-identity-constraint.3); selecting
        /// several is a violation, not a multi-valued key.
        private func fieldError(
            _ constraint: IdentityConstraint,
            at target: PureXML.Model.TreeNode,
            path targetPath: [PureXML.Validation.PathKey],
        ) -> XSDFinding? {
            let step = fieldStep(constraint.fields, at: target, namespaces: constraint.namespaceBindings)
            for field in constraint.fields {
                let selected = select(field, at: target, namespaces: constraint.namespaceBindings)
                // cvc-identity-constraint.3: each field must evaluate to at most one node.
                if selected.count > 1 {
                    return XSDFinding(
                        .identityFieldCardinality,
                        reason: "\(label(constraint)) '\(constraint.name)': field '\(field)' must select at most one node",
                        at: targetPath + step,
                    )
                }
                // cvc-identity-constraint.3 / c-fields-xpaths: a field must identify a node
                // with a simple type (or simple content). An element with element children
                // has complex content (checked structurally), and an element whose DECLARED
                // type is a complex type with non-simple content (empty or element-only, so
                // an attributes-only element with no children, XSTS idG006) is equally
                // invalid (checked through the schema type model). A simple-typed or
                // simple-content element, a text-only node, and an attribute are left alone.
                let element = selected.first.flatMap { $0.kind == .element ? $0 : nil }
                let hasElementChildren = element?.children.contains { $0.kind == .element } ?? false
                // A field whose DECLARED target is non-simple (the ur-type anyType or
                // an inline complex type without simple content, resolved at compile
                // time, XSTS idZ010) is invalid unless the instance gives the node a
                // simple type through xsi:type: cvc-identity-constraint.3 reads the
                // node's governing type, so a text-only anyType element validates only
                // when xsi:type names a simple type. With an xsi:type present the
                // runtime value comparison already governs, so the field is left.
                let declaredNonSimple = element.map { node in
                    nonSimpleFields.contains(PureXML.Schema.XSDParser.identityFieldKey(constraint: constraint, field: field))
                        && Self.xsiTypeLocalName(of: node) == nil
                } ?? false
                if element != nil, hasElementChildren || declaredNonSimple || fieldTargetIsComplexNonSimple(field, at: target) {
                    return XSDFinding(
                        .identityFieldMustBeSimple,
                        reason: "\(label(constraint)) '\(constraint.name)': field '\(field)' must identify a node with simple content",
                        at: targetPath + step,
                    )
                }
            }
            return nil
        }

        // MARK: XPath helpers

        private func select(_ xpath: String, at node: PureXML.Model.TreeNode, namespaces: [String: String]) -> [PureXML.Model.TreeNode] {
            guard let query = cache.query(xpath) else { return [] }
            return query.nodes(over: node, namespaces: namespaces)
        }

        private func fieldTuple(_ fields: [String], at node: PureXML.Model.TreeNode, constraint: IdentityConstraint) -> [FieldValue?] {
            // The field XPath's prefixes resolve against the schema (where the
            // constraint is declared); a QName field VALUE resolves against the
            // instance, so the FieldValue keeps the instance bindings.
            let instanceNamespaces = namespaceBindings(at: node)
            return fields.map { field in
                guard let query = cache.query(field),
                      let value = try? query.value(at: node, namespaces: constraint.namespaceBindings)
                else {
                    return nil
                }
                let type = resolveFieldType(field, constraint: constraint, at: node)
                // An absent attribute (empty node-set) or an empty element whose
                // declaration carries a `default`/`fixed` takes that value as its
                // identity component. A present attribute (even value "") and a
                // non-empty element are instance-supplied and never defaulted.
                if let defaulted = defaultedValue(field, value: value, type: type, constraint: constraint, at: node) {
                    return defaulted
                }
                if let nodes = value.nodes, nodes.isEmpty {
                    return nil
                }
                return FieldValue(string: value.string, type: type ?? Self.effectiveType(of: value), namespaceBindings: instanceNamespaces)
            }
        }

        /// Prefix bindings in scope at `node`, gathered from ancestor elements.
        func namespaceBindings(at node: PureXML.Model.TreeNode) -> [String: String] {
            var chain: [PureXML.Model.TreeNode] = []
            var current: PureXML.Model.TreeNode? = node
            while let element = current, element.kind == .element {
                chain.append(element)
                current = element.parent
            }
            var bindings: [String: String] = [:]
            for element in chain.reversed() {
                for attribute in element.attributes {
                    if attribute.name.prefix == "xmlns" {
                        bindings[attribute.name.localName] = attribute.value
                    } else if attribute.name.prefix == nil, attribute.name.localName == "xmlns" {
                        bindings[""] = attribute.value
                    }
                }
            }
            return bindings
        }

        /// The simple type a field value compares in: the built-in named by the
        /// selected node's `xsi:type`, when present. Identity-constraint values
        /// compare in their value space, so two `xsi:type="xsd:decimal"` fields
        /// `3.0` and `3` are the same value and violate a `unique`/`key`. With no
        /// `xsi:type` the comparison falls back to the lexical form (the type
        /// declared in the schema is not threaded into identity validation).
        private func resolveFieldType(_ field: String, constraint: IdentityConstraint, at target: PureXML.Model.TreeNode) -> SimpleType? {
            if field == ".", let local = target.name?.localName {
                if let type = fieldTypes[PureXML.Schema.XSDParser.identityFieldKey(constraint: constraint, field: field, targetLocal: local)] {
                    return type
                }
                guard let name = target.name else { return nil }
                if let declaration = types[PureXML.Schema.XSDParser.elementDeclarationKey(name)] {
                    return simpleType(of: declaration)
                }
                if let declaration = types[PureXML.Schema.XSDParser.elementKey(name.localName)] {
                    return simpleType(of: declaration)
                }
            }
            return fieldTypes[PureXML.Schema.XSDParser.identityFieldKey(constraint: constraint, field: field)]
        }
    }
}