[RFC 9651] Add support for Display String type to RawStructuredFieldValues (#43)

clintonpi · web-flow · commit c2e79ce44ce4 · 2024-11-22T18:11:50.000Z
Motivation:

RFC 9651 added the Display String Structured Type.

Modifications:

Implement the parser and serializer for Display String in the RawStructuredFieldValues module.

Result:

The RawStructuredFieldValues module will support the Display String type.
diff --git a/Sources/RawStructuredFieldValues/ASCII.swift b/Sources/RawStructuredFieldValues/ASCII.swift
@@ -45,8 +45,12 @@ let asciiSlash = UInt8(ascii: "/")
 let asciiPeriod = UInt8(ascii: ".")
 let asciiComma = UInt8(ascii: ",")
 let asciiCapitalA = UInt8(ascii: "A")
+let asciiCapitalF = UInt8(ascii: "F")
 let asciiCapitalZ = UInt8(ascii: "Z")
 let asciiLowerA = UInt8(ascii: "a")
+let asciiLowerF = UInt8(ascii: "f")
 let asciiLowerZ = UInt8(ascii: "z")
 let asciiCapitals = asciiCapitalA...asciiCapitalZ
 let asciiLowercases = asciiLowerA...asciiLowerZ
+let asciiHexCapitals = asciiCapitalA...asciiCapitalF
+let asciiHexLowercases = asciiLowerA...asciiLowerF
diff --git a/Sources/RawStructuredFieldValues/ComponentTypes.swift b/Sources/RawStructuredFieldValues/ComponentTypes.swift
@@ -110,6 +110,8 @@ extension BareItem {
 
         case .date:
             throw StructuredHeaderError.invalidItem
+        case .displayString:
+            throw StructuredHeaderError.invalidItem
         }
     }
 }
@@ -141,6 +143,9 @@ public enum RFC9651BareItem: Sendable {
 
     /// A date item.
     case date(Int)
+
+    /// A display string item.
+    case displayString(String)
 }
 
 extension RFC9651BareItem: ExpressibleByBooleanLiteral {
diff --git a/Sources/RawStructuredFieldValues/Errors.swift b/Sources/RawStructuredFieldValues/Errors.swift
@@ -27,6 +27,7 @@ public struct StructuredHeaderError: Error, Sendable {
         case invalidBoolean
         case invalidToken
         case invalidDate
+        case invalidDisplayString
         case invalidList
         case invalidDictionary
         case missingKey
@@ -53,6 +54,7 @@ extension StructuredHeaderError {
     public static let invalidBoolean = StructuredHeaderError(.invalidBoolean)
     public static let invalidToken = StructuredHeaderError(.invalidToken)
     public static let invalidDate = StructuredHeaderError(.invalidDate)
+    public static let invalidDisplayString = StructuredHeaderError(.invalidDisplayString)
     public static let invalidList = StructuredHeaderError(.invalidList)
     public static let invalidDictionary = StructuredHeaderError(.invalidDictionary)
     public static let missingKey = StructuredHeaderError(.missingKey)
diff --git a/Sources/RawStructuredFieldValues/FieldParser.swift b/Sources/RawStructuredFieldValues/FieldParser.swift
@@ -224,6 +224,8 @@ extension StructuredFieldValueParser {
             return try self._parseAToken()
         case asciiAt:
             return try self._parseADate()
+        case asciiPercent:
+            return try self._parseADisplayString()
         default:
             throw StructuredHeaderError.invalidItem
         }
@@ -491,6 +493,84 @@ extension StructuredFieldValueParser {
         return try self._parseAnIntegerOrDecimal(isDate: true)
     }
 
+    private mutating func _parseADisplayString() throws -> RFC9651BareItem {
+        assert(self.underlyingData.first == asciiPercent)
+        self.underlyingData.consumeFirst()
+
+        guard self.underlyingData.first == asciiDquote else {
+            throw StructuredHeaderError.invalidDisplayString
+        }
+
+        self.underlyingData.consumeFirst()
+
+        var byteArray = [UInt8]()
+
+        while let char = self.underlyingData.first {
+            self.underlyingData.consumeFirst()
+
+            switch char {
+            case 0x00...0x1F, 0x7F...:
+                throw StructuredHeaderError.invalidDisplayString
+            case asciiPercent:
+                if self.underlyingData.count < 2 {
+                    throw StructuredHeaderError.invalidDisplayString
+                }
+
+                let octetHex = EncodedHex(self.underlyingData.prefix(2))
+
+                self.underlyingData = self.underlyingData.dropFirst(2)
+
+                guard let octet = octetHex.decode() else {
+                    throw StructuredHeaderError.invalidDisplayString
+                }
+
+                byteArray.append(octet)
+            case asciiDquote:
+                #if compiler(>=6.0)
+                if #available(macOS 15.0, iOS 18.0, tvOS 18.0, watchOS 11.0, *) {
+                    let unicodeSequence = String(validating: byteArray, as: UTF8.self)
+
+                    guard let unicodeSequence else {
+                        throw StructuredHeaderError.invalidDisplayString
+                    }
+
+                    return .displayString(unicodeSequence)
+                } else {
+                    return try _decodeDisplayString(byteArray: &byteArray)
+                }
+                #else
+                return try _decodeDisplayString(byteArray: &byteArray)
+                #endif
+            default:
+                byteArray.append(char)
+            }
+        }
+
+        // Fail parsing — reached the end of the string without finding a closing DQUOTE.
+        throw StructuredHeaderError.invalidDisplayString
+    }
+
+    /// This method is called in environments where `String(validating:as:)` is unavailable. It uses
+    /// `String(validatingUTF8:)` which requires `byteArray` to be null terminated. `String(validating:as:)`
+    /// does not require that requirement. Therefore, it does not perform null checks, which makes it more optimal.
+    private func _decodeDisplayString(byteArray: inout [UInt8]) throws -> RFC9651BareItem {
+        // String(validatingUTF8:) requires byteArray to be null-terminated.
+        byteArray.append(0)
+
+        let unicodeSequence = byteArray.withUnsafeBytes {
+            $0.withMemoryRebound(to: CChar.self) {
+                // This force-unwrap is safe, as the buffer must successfully bind to CChar.
+                String(validatingUTF8: $0.baseAddress!)
+            }
+        }
+
+        guard let unicodeSequence else {
+            throw StructuredHeaderError.invalidDisplayString
+        }
+
+        return .displayString(unicodeSequence)
+    }
+
     private mutating func _parseParameters() throws -> OrderedMap<Key, RFC9651BareItem> {
         var parameters = OrderedMap<Key, RFC9651BareItem>()
 
@@ -643,3 +723,39 @@ extension StrippingStringEscapesCollection.Index: Comparable {
         lhs._baseIndex < rhs._baseIndex
     }
 }
+
+/// `EncodedHex` represents a (possibly invalid) hex value in UTF8.
+struct EncodedHex {
+    private(set) var firstChar: UInt8
+    private(set) var secondChar: UInt8
+
+    init<Bytes: RandomAccessCollection>(_ bytes: Bytes) where Bytes.Element == UInt8 {
+        precondition(bytes.count == 2)
+        self.firstChar = bytes[bytes.startIndex]
+        self.secondChar = bytes[bytes.index(after: bytes.startIndex)]
+    }
+
+    /// Validates and converts `EncodedHex` to a base 10 UInt8.
+    ///
+    /// If `EncodedHex` does not represent a valid hex value, the result of this method is nil.
+    fileprivate func decode() -> UInt8? {
+        guard
+            let firstCharAsInteger = self.htoi(self.firstChar),
+            let secondCharAsInteger = self.htoi(self.secondChar)
+        else { return nil }
+
+        return (firstCharAsInteger << 4) + secondCharAsInteger
+    }
+
+    /// Converts a hex character given in UTF8 to its integer value.
+    private func htoi(_ asciiChar: UInt8) -> UInt8? {
+        switch asciiChar {
+        case asciiZero...asciiNine:
+            return asciiChar - asciiZero
+        case asciiLowerA...asciiLowerF:
+            return asciiChar - asciiLowerA + 10
+        default:
+            return nil
+        }
+    }
+}
diff --git a/Sources/RawStructuredFieldValues/FieldSerializer.swift b/Sources/RawStructuredFieldValues/FieldSerializer.swift
@@ -213,6 +213,29 @@ extension StructuredFieldValueSerializer {
             }
 
             self.data.append(contentsOf: String(date, radix: 10).utf8)
+        case .displayString(let displayString):
+            let bytes = displayString.utf8
+
+            self.data.append(asciiPercent)
+            self.data.append(asciiDquote)
+
+            for byte in bytes {
+                if byte == asciiPercent
+                    || byte == asciiDquote
+                    || (0x00...0x1F).contains(byte)
+                    || (0x7F...).contains(byte)
+                {
+                    self.data.append(asciiPercent)
+
+                    let encodedByte = UInt8.encodeToHex(byte)
+                    self.data.append(encodedByte.firstChar)
+                    self.data.append(encodedByte.secondChar)
+                } else {
+                    self.data.append(byte)
+                }
+            }
+
+            self.data.append(asciiDquote)
         }
     }
 }
@@ -245,3 +268,18 @@ extension String {
         }
     }
 }
+
+extension UInt8 {
+    /// Converts an integer in base 10 to hex of type `EncodedHex`.
+    fileprivate static func encodeToHex(_ int: Self) -> EncodedHex {
+        let firstChar = self.itoh(int >> 4)
+        let secondChar = self.itoh(int & 0x0F)
+
+        return EncodedHex([firstChar, secondChar])
+    }
+
+    /// Converts an integer to its hex character in UTF8.
+    private static func itoh(_ int: Self) -> Self {
+        (int > 9) ? (asciiLowerA + int - 10) : (asciiZero + int)
+    }
+}
diff --git a/Sources/sh-parser/main.swift b/Sources/sh-parser/main.swift
@@ -171,6 +171,8 @@ extension RFC9651BareItem {
             return "decimal \(d)"
         case .date(let date):
             return "date \(date)"
+        case .displayString(let displayString):
+            return "display string \(displayString)"
         }
     }
 }
diff --git a/Tests/StructuredFieldValuesTests/StructuredFieldParserTests.swift b/Tests/StructuredFieldValuesTests/StructuredFieldParserTests.swift
@@ -87,6 +87,24 @@ final class StructuredFieldParserTests: XCTestCase {
 
             XCTAssertEqual(typeName, "date", "\(fixtureName): Expected type date, got type \(typeName)")
             XCTAssertEqual(typeValue, baseDate, "\(fixtureName): Got \(baseDate), expected \(typeValue)")
+        case (.displayString(let baseDisplayString), .dictionary(let typeDictionary)):
+            guard typeDictionary.count == 2, case .string(let typeName) = typeDictionary["__type"],
+                case .string(let typeValue) = typeDictionary["value"]
+            else {
+                XCTFail("\(fixtureName): Unexpected type dict \(typeDictionary)")
+                return
+            }
+
+            XCTAssertEqual(
+                typeName,
+                "displaystring",
+                "\(fixtureName): Expected type displaystring, got type \(typeName)"
+            )
+            XCTAssertEqual(
+                typeValue,
+                baseDisplayString,
+                "\(fixtureName): Got \(baseDisplayString), expected \(typeValue)"
+            )
         default:
             XCTFail("\(fixtureName): Got \(bareItem), expected \(schema)")
         }
diff --git a/Tests/StructuredFieldValuesTests/StructuredFieldSerializerTests.swift b/Tests/StructuredFieldValuesTests/StructuredFieldSerializerTests.swift
@@ -214,6 +214,9 @@ extension RFC9651BareItem {
             case (.some(.string("date")), .some(.integer(let value))):
                 self = .date(value)
 
+            case (.some(.string("displaystring")), .some(.string(let value))):
+                self = .displayString(value)
+
             default:
                 preconditionFailure("Unexpected type object \(typeObject)")
             }
diff --git a/Tests/TestFixtures/display-string.json b/Tests/TestFixtures/display-string.json
@@ -0,0 +1,111 @@
+[
+    {
+        "name": "basic display string (ascii content)",
+        "raw": ["%\"foo bar\""],
+        "header_type": "item",
+        "expected": [{"__type": "displaystring", "value": "foo bar"}, {}]
+    },
+    {
+        "name": "all printable ascii",
+        "raw": ["%\" !%22#$%25&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~\""],
+        "header_type": "item",
+        "expected": [{"__type": "displaystring", "value": " !\"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~"}, {}]
+    },
+    {
+        "name": "non-ascii display string (uppercase escaping)",
+        "raw": ["%\"f%C3%BC%C3%BC\""],
+        "canonical": ["%\"f%c3%bc%c3%bc\""],
+        "header_type": "item",
+        "must_fail": true
+    },
+    {
+        "name": "non-ascii display string (lowercase escaping)",
+        "raw": ["%\"f%c3%bc%c3%bc\""],
+        "header_type": "item",
+        "expected": [{"__type": "displaystring", "value": "füü"}, {}]
+    },
+    {
+        "name": "tab in display string",
+        "raw": ["%\"\t\""],
+        "header_type": "item",
+        "must_fail": true
+    },
+    {
+        "name": "newline in display string",
+        "raw": ["%\"\n\""],
+        "header_type": "item",
+        "must_fail": true
+    },
+    {
+        "name": "single quoted display string",
+        "raw": ["%'foo'"],
+        "header_type": "item",
+        "must_fail": true
+    },
+    {
+        "name": "unquoted display string",
+        "raw": ["%foo"],
+        "header_type": "item",
+        "must_fail": true
+    },
+    {
+        "name": "display string missing initial quote",
+        "raw": ["%foo\""],
+        "header_type": "item",
+        "must_fail": true
+    },
+    {
+        "name": "unbalanced display string",
+        "raw": ["%\"foo"],
+        "header_type": "item",
+        "must_fail": true
+    },
+    {
+        "name": "display string quoting",
+        "raw": ["%\"foo %22bar%22 \\ baz\""],
+        "header_type": "item",
+        "expected": [{"__type": "displaystring", "value": "foo \"bar\" \\ baz"}, {}]
+    },
+    {
+        "name": "bad display string escaping",
+        "raw": ["%\"foo %a"],
+        "header_type": "item",
+        "must_fail": true
+    },
+    {
+        "name": "bad display string utf-8 (invalid 2-byte seq)",
+        "raw": ["%\"%c3%28\""],
+        "header_type": "item",
+        "must_fail": true
+    },
+    {
+        "name": "bad display string utf-8 (invalid sequence id)",
+        "raw": ["%\"%a0%a1\""],
+        "header_type": "item",
+        "must_fail": true
+    },
+    {
+        "name": "bad display string utf-8 (invalid hex)",
+        "raw": ["%\"%g0%1w\""],
+        "header_type": "item",
+        "must_fail": true
+    },
+    {
+        "name": "bad display string utf-8 (invalid 3-byte seq)",
+        "raw": ["%\"%e2%28%a1\""],
+        "header_type": "item",
+        "must_fail": true
+    },
+    {
+        "name": "bad display string utf-8 (invalid 4-byte seq)",
+        "raw": ["%\"%f0%28%8c%28\""],
+        "header_type": "item",
+        "must_fail": true
+    },
+    {
+        "name": "BOM in display string",
+        "raw": ["%\"BOM: %ef%bb%bf\""],
+        "header_type": "item",
+        "expected": [{"__type": "displaystring", "value": "BOM: \uFEFF"}, {}]
+    }
+]

Original file line number	Diff line number	Diff line change
`@@ -110,6 +110,8 @@ extension BareItem {`
`110`	`110`
`111`	`111`	`case .date:`
`112`	`112`	`throw StructuredHeaderError.invalidItem`
	`113`	`+ case .displayString:`
	`114`	`+ throw StructuredHeaderError.invalidItem`
`113`	`115`	`}`
`114`	`116`	`}`
`115`	`117`	`}`
`@@ -141,6 +143,9 @@ public enum RFC9651BareItem: Sendable {`
`141`	`143`
`142`	`144`	`/// A date item.`
`143`	`145`	`case date(Int)`
	`146`	`+`
	`147`	`+ /// A display string item.`
	`148`	`+ case displayString(String)`
`144`	`149`	`}`
`145`	`150`
`146`	`151`	`extension RFC9651BareItem: ExpressibleByBooleanLiteral {`
Original file line number	Diff line number	Diff line change
`@@ -171,6 +171,8 @@ extension RFC9651BareItem {`
`171`	`171`	`return "decimal \(d)"`
`172`	`172`	`case .date(let date):`
`173`	`173`	`return "date \(date)"`
	`174`	`+ case .displayString(let displayString):`
	`175`	`+ return "display string \(displayString)"`
`174`	`176`	`}`
`175`	`177`	`}`
`176`	`178`	`}`
Original file line number	Diff line number	Diff line change
`@@ -214,6 +214,9 @@ extension RFC9651BareItem {`
`214`	`214`	`case (.some(.string("date")), .some(.integer(let value))):`
`215`	`215`	`self = .date(value)`
`216`	`216`
	`217`	`+ case (.some(.string("displaystring")), .some(.string(let value))):`
	`218`	`+ self = .displayString(value)`
	`219`	`+`
`217`	`220`	`default:`
`218`	`221`	`preconditionFailure("Unexpected type object \(typeObject)")`
`219`	`222`	`}`