PureXML/Tests/SchemaContentModelDepthSafetyTests.swift

#if !os(WASI)
    import Foundation
    import Testing
    @testable import PureXML

    /// Several constructs are flat in the document, so they escape the parser's
    /// element-nesting `maxDepth` and let a compiled depth grow without bound: a
    /// `<xs:group ref>` chain (`g0 = (e0, g1)`, ...), a `complexContent` derivation chain
    /// (`B0` extends `B1` extends ...), and an `<xs:attributeGroup ref>` chain (`ag0` refs
    /// `ag1` refs ...). The compiler inlines a group reference, composes a derivation, and
    /// flattens an attribute group once per link, so a long chain recursed until the stack
    /// overflowed while compiling a legal (if hostile) schema. Each is now bounded by
    /// `maximumContentModelExpansionDepth` (512): within it compiles, past it is rejected
    /// with a LOCATED error, none crashes.
    ///
    /// Runs on an 8 MB stack, the native main-thread budget: reaching the bound is hundreds
    /// of levels deep (a couple of MB), more than swift-testing's small Task stack, and WASI
    /// cannot size a thread stack (its budget is a link-time setting), so the cases are gated
    /// off there.
    @Suite("Schema content-model expansion depth safety")
    struct SchemaContentModelDepthSafetyTests {
        @Test("a group-reference chain within the depth bound compiles")
        func test_groupChainWithinBoundCompiles() {
            #expect(compile(groupChain(depth: 400)).compiled)
        }

        @Test("a group-reference chain past the bound is rejected with a located error, not a crash")
        func test_groupChainOverBoundRejectsLocated() {
            // Past the bound and past the depth at which the unbounded compiler overflowed.
            assertRejectedForDepth(compile(groupChain(depth: 2500)))
        }

        @Test("a complexContent derivation chain within the depth bound compiles")
        func test_derivationChainWithinBoundCompiles() {
            #expect(compile(derivationChain(depth: 10)).compiled)
        }

        @Test("a complexContent derivation chain past the bound is rejected with a located error, not a crash")
        func test_derivationChainOverBoundRejectsLocated() {
            assertRejectedForDepth(compile(derivationChain(depth: 600)))
        }

        @Test("an attribute-group reference chain within the depth bound compiles")
        func test_attributeGroupChainWithinBoundCompiles() {
            #expect(compile(attributeGroupChain(depth: 400)).compiled)
        }

        @Test("an attribute-group reference chain past the bound is rejected with a located error, not a crash")
        func test_attributeGroupChainOverBoundRejectsLocated() {
            // Past the bound and past the depth at which flattening used to overflow the stack.
            assertRejectedForDepth(compile(attributeGroupChain(depth: 600)))
        }

        // MARK: Helpers

        private func assertRejectedForDepth(_ outcome: Outcome) {
            #expect(!outcome.compiled)
            let depthFinding = outcome.findings.first { $0.reason.contains("maximum") && $0.reason.contains("depth") }
            #expect(depthFinding != nil) // rejected for depth, not some unrelated reason
            #expect(!(depthFinding?.codingPath.isEmpty ?? true)) // located: an IDE can underline it
        }

        private func attributeGroupChain(depth: Int) -> String {
            var parts: [String] = []
            for index in 0 ..< depth {
                let inner = index < depth - 1
                    ? "<xs:attributeGroup ref=\"ag\(index + 1)\"/>"
                    : "<xs:attribute name=\"a\" type=\"xs:string\"/>"
                parts.append("<xs:attributeGroup name=\"ag\(index)\">\(inner)</xs:attributeGroup>")
            }
            return "<xs:schema xmlns:xs=\"http://www.w3.org/2001/XMLSchema\">\(parts.joined())"
                + "<xs:element name=\"doc\"><xs:complexType>"
                + "<xs:attributeGroup ref=\"ag0\"/></xs:complexType></xs:element></xs:schema>"
        }

        private func groupChain(depth: Int) -> String {
            var parts: [String] = []
            for index in 0 ..< depth {
                let next = index < depth - 1 ? "<xs:group ref=\"g\(index + 1)\"/>" : ""
                parts.append("<xs:group name=\"g\(index)\"><xs:sequence>"
                    + "<xs:element name=\"e\(index)\" type=\"xs:string\"/>\(next)</xs:sequence></xs:group>")
            }
            return "<xs:schema xmlns:xs=\"http://www.w3.org/2001/XMLSchema\">\(parts.joined())"
                + "<xs:element name=\"doc\"><xs:complexType><xs:sequence>"
                + "<xs:group ref=\"g0\"/></xs:sequence></xs:complexType></xs:element></xs:schema>"
        }

        private func derivationChain(depth: Int) -> String {
            var parts: [String] = []
            for index in 0 ..< depth {
                if index < depth - 1 {
                    parts.append("<xs:complexType name=\"B\(index)\"><xs:complexContent>"
                        + "<xs:extension base=\"B\(index + 1)\"><xs:sequence>"
                        + "<xs:element name=\"e\(index)\" type=\"xs:string\"/></xs:sequence>"
                        + "</xs:extension></xs:complexContent></xs:complexType>")
                } else {
                    parts.append("<xs:complexType name=\"B\(index)\"><xs:sequence>"
                        + "<xs:element name=\"leaf\" type=\"xs:string\"/></xs:sequence></xs:complexType>")
                }
            }
            return "<xs:schema xmlns:xs=\"http://www.w3.org/2001/XMLSchema\">\(parts.joined())"
                + "<xs:element name=\"doc\" type=\"B0\"/></xs:schema>"
        }

        private struct Outcome {
            var compiled: Bool
            var findings: [PureXML.Validation.ValidationError]
        }

        /// Compiles `schema` on a native-sized stack and reports whether it compiled and,
        /// if not, the located findings.
        private func compile(_ schema: String) -> Outcome {
            let result = ResultBox()
            runOnNativeStack {
                do {
                    _ = try PureXML.Schema.Document(schema)
                    result.outcome = Outcome(compiled: true, findings: [])
                } catch let error as PureXML.Schema.SchemaError {
                    if case let .inconsistent(findings) = error {
                        result.outcome = Outcome(compiled: false, findings: findings)
                    } else {
                        result.outcome = Outcome(compiled: false, findings: [])
                    }
                } catch {
                    result.outcome = Outcome(compiled: false, findings: [])
                }
            }
            return result.outcome
        }

        /// @unchecked Sendable: written once by the worker thread before its semaphore
        /// signals and read only after `runOnNativeStack`'s wait, so the worker's write
        /// happens-before the read and there is no concurrent access.
        private final class ResultBox: @unchecked Sendable {
            var outcome = Outcome(compiled: false, findings: [])
        }

        /// Runs `work` on a thread with the native main-thread stack budget and waits. The
        /// semaphore orders the worker's write before the read, so the box is safe.
        private func runOnNativeStack(_ work: @escaping @Sendable () -> Void) {
            let done = DispatchSemaphore(value: 0)
            let thread = Thread {
                work()
                done.signal()
            }
            thread.stackSize = 8 << 20
            thread.start()
            done.wait()
        }
    }
#endif