WasmParser: Stop copying the expression bytes before parsing them

kateinoigakukun · kateinoigakukun · commit c0f96a873df6 · 2024-09-23T09:53:51.000+09:00
diff --git a/Sources/WAT/Encoder.swift b/Sources/WAT/Encoder.swift
@@ -1,4 +1,5 @@
 import WasmParser
+import WasmTypes
 
 struct Encoder {
     var output: [UInt8] = []
diff --git a/Sources/WAT/NameMapping.swift b/Sources/WAT/NameMapping.swift
@@ -1,4 +1,5 @@
 import WasmParser
+import WasmTypes
 
 /// A module field declaration that may have its name
 protocol NamedModuleFieldDecl {
@@ -126,7 +127,7 @@ struct TypesMap {
         return .empty
     }
     private mutating func resolveBlockType(
-        signature: WasmParser.FunctionType,
+        signature: WasmTypes.FunctionType,
         resolveSignatureIndex: (inout TypesMap) -> Int
     ) throws -> BlockType {
         if signature.parameters.isEmpty {
@@ -146,7 +147,7 @@ struct TypesMap {
     }
 
     /// Resolves a block type from a function type signature
-    mutating func resolveBlockType(signature: WasmParser.FunctionType) throws -> BlockType {
+    mutating func resolveBlockType(signature: WasmTypes.FunctionType) throws -> BlockType {
         return try resolveBlockType(
             signature: signature,
             resolveSignatureIndex: {
@@ -171,7 +172,7 @@ struct TypesMap {
         case let (indexOrId?, _):
             return try nameMapping.resolveIndex(use: indexOrId)
         case (nil, let inline):
-            let inline = inline?.signature ?? WasmParser.FunctionType(parameters: [], results: [])
+            let inline = inline?.signature ?? WasmTypes.FunctionType(parameters: [], results: [])
             return addAnonymousSignature(inline)
         }
     }
@@ -196,7 +197,7 @@ struct TypesMap {
             return (found, Int(index))
         case (nil, let inline):
             // If no index and no inline type, then it's a function type with no parameters or results
-            let inline = inline ?? WatParser.FunctionType(signature: WasmParser.FunctionType(parameters: [], results: []), parameterNames: [])
+            let inline = inline ?? WatParser.FunctionType(signature: WasmTypes.FunctionType(parameters: [], results: []), parameterNames: [])
             // Check if the inline type already exists
             if let index = indices[inline.signature] {
                 return (inline, index)
diff --git a/Sources/WAT/ParseInstruction.swift b/Sources/WAT/ParseInstruction.swift
@@ -1,6 +1,9 @@
 // swift-format-ignore-file
-// This file is generated by Utilities/generate_inst_visitor.swift
+//// Automatically generated by Utilities/Sources/WasmGen.swift
+//// DO NOT EDIT DIRECTLY
+
 import WasmParser
+import WasmTypes
 
 /// Parses a text instruction, consuming immediate tokens as necessary.
 /// - Parameters:
diff --git a/Sources/WAT/Parser/ExpressionParser.swift b/Sources/WAT/Parser/ExpressionParser.swift
@@ -1,4 +1,5 @@
 import WasmParser
+import WasmTypes
 
 struct ExpressionParser<Visitor: InstructionVisitor> {
     typealias LocalsMap = NameMapping<WatParser.LocalDecl>
diff --git a/Sources/WAT/Parser/WatParser.swift b/Sources/WAT/Parser/WatParser.swift
@@ -1,4 +1,5 @@
 import WasmParser
+import WasmTypes
 
 struct WatParser {
     var parser: Parser
@@ -40,7 +41,7 @@ struct WatParser {
     }
 
     struct FunctionType {
-        let signature: WasmParser.FunctionType
+        let signature: WasmTypes.FunctionType
         /// Names of the parameters. The number of names must match the number of parameters in `type`.
         let parameterNames: [String?]
     }
@@ -554,7 +555,7 @@ struct WatParser {
         let (params, names) = try params()
         let results = try results()
         try parser.expect(.rightParen)
-        return FunctionType(signature: WasmParser.FunctionType(parameters: params, results: results), parameterNames: names)
+        return FunctionType(signature: WasmTypes.FunctionType(parameters: params, results: results), parameterNames: names)
     }
 
     mutating func optionalFunctionType() throws -> FunctionType? {
@@ -563,7 +564,7 @@ struct WatParser {
         if results.isEmpty, params.isEmpty {
             return nil
         }
-        return FunctionType(signature: WasmParser.FunctionType(parameters: params, results: results), parameterNames: names)
+        return FunctionType(signature: WasmTypes.FunctionType(parameters: params, results: results), parameterNames: names)
     }
 
     mutating func params() throws -> ([ValueType], [String?]) {
diff --git a/Sources/WasmKit/Execution/DispatchInstruction.swift b/Sources/WasmKit/Execution/DispatchInstruction.swift
@@ -1,5 +1,6 @@
 //// Automatically generated by Utilities/Sources/VMGen.swift
 //// DO NOT EDIT DIRECTLY
+
 extension Execution {
     @inline(__always)
     mutating func doExecute(_ instruction: UInt64, sp: inout Sp, pc: inout Pc, md: inout Md, ms: inout Ms) throws {
diff --git a/Sources/WasmKit/Execution/Errors.swift b/Sources/WasmKit/Execution/Errors.swift
@@ -1,4 +1,4 @@
-import WasmParser
+import WasmTypes
 
 public enum Trap: Error {
     // FIXME: for debugging purposes, to be eventually deleted
@@ -10,7 +10,7 @@ public enum Trap: Error {
     /// Stack overflow
     case stackOverflow
     /// The stack value type does not match the expected type
-    case stackValueTypesMismatch(expected: WasmParser.ValueType, actual: WasmParser.ValueType)
+    case stackValueTypesMismatch(expected: WasmTypes.ValueType, actual: WasmTypes.ValueType)
     /// Too deep call stack
     case callStackExhausted
 
diff --git a/Sources/WasmKit/Execution/Function.swift b/Sources/WasmKit/Execution/Function.swift
@@ -191,10 +191,9 @@ struct WasmFunctionEntity {
             functionIndex: index,
             intercepting: runtime.value.interceptor != nil
         )
-        let iseq = try translator.translate(
-            expression: code.expression,
-            instance: instance
-        )
+        let iseq = try code.withValue { code in
+            try translator.translate(code: code, instance: instance)
+        }
         self.code = .compiled(iseq)
         return iseq
     }
diff --git a/Sources/WasmKit/Execution/Instructions/Instruction.swift b/Sources/WasmKit/Execution/Instructions/Instruction.swift
@@ -1,5 +1,6 @@
 //// Automatically generated by Utilities/Sources/VMGen.swift
 //// DO NOT EDIT DIRECTLY
+
 enum Instruction: Equatable {
     case copyStack(Instruction.CopyStackOperand)
     case globalGet(Instruction.GlobalGetOperand)
diff --git a/Sources/WasmKit/Execution/Value.swift b/Sources/WasmKit/Execution/Value.swift
@@ -1,11 +1,11 @@
-import enum WasmParser.ReferenceType
-import enum WasmParser.ValueType
+import enum WasmTypes.ReferenceType
+import enum WasmTypes.ValueType
 
 /// > Note:
 /// <https://webassembly.github.io/spec/core/syntax/types.html#value-types>
 
 
-public typealias ReferenceType = WasmParser.ReferenceType
+public typealias ReferenceType = WasmTypes.ReferenceType
 
 extension Value {
     func maybeAddressOffset(_ isMemory64: Bool) -> UInt64? {
diff --git a/Sources/WasmKit/Translator.swift b/Sources/WasmKit/Translator.swift
@@ -1,4 +1,5 @@
 import WasmParser
+import WasmTypes
 
 class ISeqAllocator {
 
@@ -264,7 +265,7 @@ struct InstructionTranslator<Context: TranslatorContext>: InstructionVisitor {
     typealias Output = Void
 
     typealias LabelRef = Int
-    typealias ValueType = WasmParser.ValueType
+    typealias ValueType = WasmTypes.ValueType
 
     struct ControlStack {
         typealias BlockType = FunctionType
@@ -709,7 +710,7 @@ struct InstructionTranslator<Context: TranslatorContext>: InstructionVisitor {
         funcTypeInterner: Interner<FunctionType>,
         module: Context,
         type: FunctionType,
-        locals: [WasmParser.ValueType],
+        locals: [WasmTypes.ValueType],
         functionIndex: FunctionIndex,
         intercepting: Bool
     ) {
@@ -898,18 +899,14 @@ struct InstructionTranslator<Context: TranslatorContext>: InstructionVisitor {
 
     /// Translate a Wasm expression into a sequence of instructions.
     mutating func translate(
-        expression: ArraySlice<UInt8>,
+        code: Code,
         instance: InternalInstance
     ) throws -> InstructionSequence {
         if intercepting {
             // Emit `onEnter` instruction at the beginning of the function
             emit(.onEnter(functionIndex))
         }
-        try WasmParser.parseExpression(
-            bytes: Array(expression),
-            features: instance.features, hasDataCount: instance.hasDataCount,
-            visitor: &self
-        )
+        try code.parseExpression(visitor: &self)
         return try finalize()
     }
 
@@ -1304,7 +1301,7 @@ struct InstructionTranslator<Context: TranslatorContext>: InstructionVisitor {
             emit(.select(operand))
         }
     }
-    mutating func visitTypedSelect(type: WasmParser.ValueType) throws -> Output {
+    mutating func visitTypedSelect(type: WasmTypes.ValueType) throws -> Output {
         let condition = try popVRegOperand(.i32)
         let (value1Type, value1) = try popAnyOperand()
         let (_, value2) = try popAnyOperand()
@@ -1521,7 +1518,7 @@ struct InstructionTranslator<Context: TranslatorContext>: InstructionVisitor {
     mutating func visitI64Const(value: Int64) -> Output { visitConst(.i64, .i64(UInt64(bitPattern: value))) }
     mutating func visitF32Const(value: IEEE754.Float32) -> Output { visitConst(.f32, .f32(value.bitPattern)) }
     mutating func visitF64Const(value: IEEE754.Float64) -> Output { visitConst(.f64, .f64(value.bitPattern)) }
-    mutating func visitRefNull(type: WasmParser.ReferenceType) -> Output {
+    mutating func visitRefNull(type: WasmTypes.ReferenceType) -> Output {
         pushEmit(.ref(type), { .refNull(Instruction.RefNullOperand(type: type, result: $0)) })
     }
     mutating func visitRefIsNull() throws -> Output {
diff --git a/Sources/WasmParser/Docs.docc/Docs.md b/Sources/WasmParser/Docs.docc/Docs.md
@@ -48,8 +48,6 @@ while let payload = try parser.parseNext() {
 
 - ``Parser``
 - ``Parser/parseNext()``
-- ``parseExpression(bytes:features:hasDataCount:visitor:)``
-- ``parseExpression(stream:features:hasDataCount:visitor:)``
 - ``NameSectionParser``
 
 ### Visitor
diff --git a/Sources/WasmParser/InstructionVisitor.swift b/Sources/WasmParser/InstructionVisitor.swift
@@ -1,5 +1,9 @@
 // swift-format-ignore-file
-// This file is generated by Utilities/generate_inst_visitor.swift
+//// Automatically generated by Utilities/Sources/WasmGen.swift
+//// DO NOT EDIT DIRECTLY
+
+import WasmTypes
+
 public enum Instruction: Equatable {
     case `unreachable`
     case `nop`
diff --git a/Sources/WasmParser/Stream/ByteStream.swift b/Sources/WasmParser/Stream/ByteStream.swift
@@ -1,10 +1,15 @@
 public protocol ByteStream: Stream where Element == UInt8 {}
 
 public final class StaticByteStream: ByteStream {
-    public let bytes: [UInt8]
+    public let bytes: ArraySlice<UInt8>
     public var currentIndex: Int
 
     public init(bytes: [UInt8]) {
+        self.bytes = ArraySlice(bytes)
+        currentIndex = bytes.startIndex
+    }
+
+    init(bytes: ArraySlice<UInt8>) {
         self.bytes = bytes
         currentIndex = bytes.startIndex
     }
diff --git a/Sources/WasmParser/WasmParser.swift b/Sources/WasmParser/WasmParser.swift
@@ -1,4 +1,5 @@
 import SystemPackage
+import WasmTypes
 
 #if os(Windows)
     import ucrt
@@ -76,78 +77,54 @@ extension Parser where Stream == FileHandleStream {
     }
 }
 
-/// Parse a WebAssembly expression from the given byte stream
-///
-/// - Parameters:
-///   - stream: The byte stream to parse
-///   - features: The enabled WebAssembly features
-///   - hasDataCount: Whether the module has a data count section
-///   - visitor: The instruction visitor to visit the parsed instructions
-/// - Throws: `WasmParserError` if the parsing fails
-@inlinable
-public func parseExpression<V: InstructionVisitor, Stream: ByteStream>(
-    stream: Stream,
-    features: WasmFeatureSet = .default,
-    hasDataCount: Bool = false,
-    visitor: inout V
-) throws {
-    let parser = Parser(stream: stream, features: features, hasDataCount: hasDataCount)
-    var lastCode: InstructionCode?
-    while try !parser.stream.hasReachedEnd() {
-        (lastCode, _) = try parser.parseInstruction(visitor: &visitor)
-    }
-    guard lastCode == .end else {
-        throw WasmParserError.endOpcodeExpected
+extension Code {
+    /// Parse a WebAssembly expression from the given byte stream
+    ///
+    /// - Parameters:
+    ///   - visitor: The instruction visitor to visit the parsed instructions
+    /// - Throws: `WasmParserError` if the parsing fails
+    ///
+    /// The input bytes sequence is usually extracted from a WebAssembly module's code section.
+    ///
+    /// ```swift
+    /// import WasmParser
+    ///
+    /// struct MyVisitor: VoidInstructionVisitor {
+    ///     func visitLocalGet(localIndex: UInt32) {
+    ///         print("local.get \(localIndex)")
+    ///     }
+    /// }
+    ///
+    /// var parser = WasmParser.Parser(bytes: [
+    ///     0x00, 0x61, 0x73, 0x6d, 0x01, 0x00, 0x00, 0x00, 0x01, 0x06, 0x01, 0x60,
+    ///     0x01, 0x7e, 0x01, 0x7e, 0x03, 0x02, 0x01, 0x00, 0x07, 0x07, 0x01, 0x03,
+    ///     0x66, 0x61, 0x63, 0x00, 0x00, 0x0a, 0x17, 0x01, 0x15, 0x00, 0x20, 0x00,
+    ///     0x50, 0x04, 0x7e, 0x42, 0x01, 0x05, 0x20, 0x00, 0x20, 0x00, 0x42, 0x01,
+    ///     0x7d, 0x10, 0x00, 0x7e, 0x0b, 0x0b
+    /// ])
+    /// while let payload = try parser.parseNext() {
+    ///     switch payload {
+    ///     case .codeSection(let section):
+    ///         for code in section {
+    ///             var visitor = MyVisitor()
+    ///             try code.parseExpression(visitor: &visitor)
+    ///         }
+    ///     default: break
+    ///     }
+    /// }
+    /// ````
+    public func parseExpression<V: InstructionVisitor>(visitor: inout V) throws {
+        let parser = Parser(stream: StaticByteStream(bytes: self.expression), features: self.features, hasDataCount: self.hasDataCount)
+        var lastCode: InstructionCode?
+        while try !parser.stream.hasReachedEnd() {
+            (lastCode, _) = try parser.parseInstruction(visitor: &visitor)
+        }
+        guard lastCode == .end else {
+            throw WasmParserError.endOpcodeExpected
+        }
     }
 }
 
-/// Parse a WebAssembly expression from the given byte stream
-///
-/// - Parameters:
-///   - bytes: The bytes to parse
-///   - features: The enabled WebAssembly features
-///   - hasDataCount: Whether the module has a data count section
-///   - visitor: The instruction visitor to visit the parsed instructions
-/// - Throws: `WasmParserError` if the parsing fails
-///
-/// See ``parseExpression(stream:features:hasDataCount:visitor:)`` for stream version.
-///
-/// The input bytes sequence is usually extracted from a WebAssembly module's code section.
-///
-/// ```swift
-/// import WasmParser
-///
-/// struct MyVisitor: VoidInstructionVisitor {
-///     func visitLocalGet(localIndex: UInt32) {
-///         print("local.get \(localIndex)")
-///     }
-/// }
-///
-/// var parser = WasmParser.Parser(bytes: [
-///     0x00, 0x61, 0x73, 0x6d, 0x01, 0x00, 0x00, 0x00, 0x01, 0x06, 0x01, 0x60,
-///     0x01, 0x7e, 0x01, 0x7e, 0x03, 0x02, 0x01, 0x00, 0x07, 0x07, 0x01, 0x03,
-///     0x66, 0x61, 0x63, 0x00, 0x00, 0x0a, 0x17, 0x01, 0x15, 0x00, 0x20, 0x00,
-///     0x50, 0x04, 0x7e, 0x42, 0x01, 0x05, 0x20, 0x00, 0x20, 0x00, 0x42, 0x01,
-///     0x7d, 0x10, 0x00, 0x7e, 0x0b, 0x0b
-/// ])
-/// while let payload = try parser.parseNext() {
-///     switch payload {
-///     case .codeSection(let section):
-///         for code in section {
-///             var visitor = MyVisitor()
-///             try parseExpression(bytes: Array(code.expression), visitor: &visitor)
-///         }
-///     default: break
-///     }
-/// }
-/// ````
-public func parseExpression<V: InstructionVisitor>(bytes: [UInt8], features: WasmFeatureSet = .default, hasDataCount: Bool = false, visitor: inout V) throws {
-    try parseExpression(
-        stream: StaticByteStream(bytes: bytes), features: features,
-        hasDataCount: hasDataCount, visitor: &visitor
-    )
-}
-
 /// Flags for enabling/disabling WebAssembly features
 public struct WasmFeatureSet: OptionSet {
     /// The raw value of the feature set
@@ -1048,7 +1025,7 @@ extension Parser {
             let expressionBytes = try stream.consume(
                 count: Int(size) - (stream.currentIndex - bodyStart)
             )
-            return Code(locals: locals, expression: expressionBytes)
+            return Code(locals: locals, expression: expressionBytes, hasDataCount: hasDataCount, features: features)
         }
     }
 
diff --git a/Sources/WasmParser/WasmTypes.swift b/Sources/WasmParser/WasmTypes.swift
diff --git a/Sources/_CWasmKit/include/DirectThreadedCode.inc b/Sources/_CWasmKit/include/DirectThreadedCode.inc
diff --git a/Tests/WasmKitTests/Execution/HostModuleTests.swift b/Tests/WasmKitTests/Execution/HostModuleTests.swift
diff --git a/Utilities/Sources/GeneratedFile.swift b/Utilities/Sources/GeneratedFile.swift
diff --git a/Utilities/Sources/VMGen.swift b/Utilities/Sources/VMGen.swift
diff --git a/Utilities/Sources/WasmGen.swift b/Utilities/Sources/WasmGen.swift

Original file line number	Diff line number	Diff line change
`@@ -1,4 +1,5 @@`
`1`	`1`	`import WasmParser`
	`2`	`+import WasmTypes`
`2`	`3`
`3`	`4`	`struct Encoder {`
`4`	`5`	`var output: [UInt8] = []`
Original file line number	Diff line number	Diff line change
`@@ -191,10 +191,9 @@ struct WasmFunctionEntity {`
`191`	`191`	`functionIndex: index,`
`192`	`192`	`intercepting: runtime.value.interceptor != nil`
`193`	`193`	`)`
`194`		`- let iseq = try translator.translate(`
`195`		`- expression: code.expression,`
`196`		`- instance: instance`
`197`		`- )`
	`194`	`+ let iseq = try code.withValue { code in`
	`195`	`+ try translator.translate(code: code, instance: instance)`
	`196`	`+ }`
`198`	`197`	`self.code = .compiled(iseq)`
`199`	`198`	`return iseq`
`200`	`199`	`}`