Merge pull request #177 from p-x9/feature/cfstring

p-x9 · web-flow · commit 6debf60ced7d · 2025-02-03T12:20:10.000+09:00
Support table  of `CFString`
diff --git a/Sources/MachOKit/DyldCacheLoaded.swift b/Sources/MachOKit/DyldCacheLoaded.swift
@@ -350,7 +350,6 @@ extension DyldCacheLoaded {
             return nil
         }
 
-        let text: any SegmentCommandProtocol
         let __objc_opt_ro: any SectionProtocol
 
         if libobjc.is64Bit {
@@ -360,7 +359,6 @@ extension DyldCacheLoaded {
                   }) else {
                 return nil
             }
-            text = _text
             __objc_opt_ro = section
         } else {
             guard let _text = libobjc.loadCommands.text,
@@ -369,16 +367,16 @@ extension DyldCacheLoaded {
                   }) else {
                 return nil
             }
-            text = _text
             __objc_opt_ro = section
         }
 
         guard let start = __objc_opt_ro.startPtr(
-            in: text,
             vmaddrSlide: vmaddrSlide
         ) else { return nil }
+
         let layout: OldObjCOptimization.Layout = start
                 .autoBoundPointee()
+
         return .init(
             layout: layout,
             offset: Int(bitPattern: start) - Int(bitPattern: ptr)
diff --git a/Sources/MachOKit/LoadCommand/Model/Section.swift b/Sources/MachOKit/LoadCommand/Model/Section.swift
@@ -34,21 +34,27 @@ public protocol SectionProtocol: LayoutWrapper {
 
     /// Get the pointer where this section starts
     /// - Parameters:
-    ///   - segment: Segment to which this section belongs
     ///   - vmaddrSlide: slide
     /// - Returns: Pointer where this section starts
-    func startPtr(
-        in segment: any SegmentCommandProtocol,
-        vmaddrSlide: Int
-    ) -> UnsafeRawPointer?
+    func startPtr(vmaddrSlide: Int) -> UnsafeRawPointer?
 
+    /// Get the data in this section as a string table
+    ///
     /// returns nil except when type is `cstring_literals
+    ///
+    /// - Parameters:
+    ///   - vmaddrSlide: slide
+    /// - Returns: string table
     func strings(
-        in segment: any SegmentCommandProtocol,
         vmaddrSlide: Int
     ) -> MachOImage.Strings?
 
+    /// Get the data in this section as a string table
+    ///
     /// returns nil except when type is `cstring_literals
+    ///
+    /// - Parameter machO: MachOFile to which `self` belongs
+    /// - Returns: string table
     func strings(in machO: MachOFile) -> MachOFile.Strings?
 
     /// relocation informations.
@@ -62,11 +68,22 @@ public protocol SectionProtocol: LayoutWrapper {
 }
 
 extension SectionProtocol {
+    /// Get the pointer where this section starts
+    /// - Parameters:
+    ///   - segment: Segment to which this section belongs
+    ///   - vmaddrSlide: slide
+    /// - Returns: Pointer where this section starts
+    @available(*, deprecated, renamed: "startPtr(vmaddrSlide:)", message: "No need to provide segment.")
     public func startPtr(in segment: any SegmentCommandProtocol, vmaddrSlide: Int) -> UnsafeRawPointer? {
         segment.startPtr(vmaddrSlide: vmaddrSlide)?
             .advanced(by: -segment.fileOffset)
             .advanced(by: offset)
     }
+
+    public func startPtr(vmaddrSlide: Int) -> UnsafeRawPointer? {
+        let address = vmaddrSlide + address
+        return UnsafeRawPointer(bitPattern: address)
+    }
 }
 
 public struct Section: SectionProtocol {
@@ -194,13 +211,27 @@ extension Section64 {
 }
 
 extension SectionProtocol {
+    /// Get the data in this section as a string table
+    ///
+    /// returns nil except when type is `cstring_literals
+    ///
+    /// - Parameters:
+    ///   - segment: sgment to which this section belongs
+    ///   - vmaddrSlide: slide
+    /// - Returns: string table
+    @available(*, deprecated, renamed: "strings(vmaddrSlide:)", message: "No need to provide segment")
     public func strings(
         in segment: any SegmentCommandProtocol,
         vmaddrSlide: Int
+    ) -> MachOImage.Strings? {
+        strings(vmaddrSlide: vmaddrSlide)
+    }
+
+    public func strings(
+        vmaddrSlide: Int
     ) -> MachOImage.Strings? {
         guard flags.type == .cstring_literals else { return nil }
         guard let basePointer = startPtr(
-            in: segment,
             vmaddrSlide: vmaddrSlide
         ) else {
             return nil
diff --git a/Sources/MachOKit/MachOFile.swift b/Sources/MachOKit/MachOFile.swift
@@ -483,6 +483,36 @@ extension MachOFile {
     }
 }
 
+extension MachOFile {
+    public var cfStrings64: DataSequence<CFString64>? {
+        guard let section = sections64.first(where: {
+            $0.sectionName == "__cfstring"
+        }) else { return nil }
+
+        let offset = headerStartOffset + section.offset
+        let count = section.size / CFString64.layoutSize
+
+        return fileHandle.readDataSequence(
+            offset: numericCast(offset),
+            numberOfElements: count
+        )
+    }
+
+    public var cfStrings32: DataSequence<CFString32>? {
+        guard let section = sections32.first(where: {
+            $0.sectionName == "__cfstring"
+        }) else { return nil }
+
+        let offset = headerStartOffset + section.offset
+        let count = section.size / CFString32.layoutSize
+
+        return fileHandle.readDataSequence(
+            offset: numericCast(offset),
+            numberOfElements: count
+        )
+    }
+}
+
 extension MachOFile {
     // https://github.com/apple-oss-distributions/dyld/blob/d552c40cd1de105f0ec95008e0e0c0972de43456/common/MetadataVisitor.cpp#L262
     public func resolveRebase(at offset: UInt64) -> UInt64? {
diff --git a/Sources/MachOKit/MachOImage.swift b/Sources/MachOKit/MachOImage.swift
@@ -283,13 +283,13 @@ extension MachOImage {
                 $0.sectionName == "__cstring"
             }
             guard let cstrings else { return nil }
-            return cstrings.strings(in: text, vmaddrSlide: vmaddrSlide)
+            return cstrings.strings(vmaddrSlide: vmaddrSlide)
         } else if let text = loadCommands.text {
             let cstrings = text.sections(cmdsStart: cmdsStartPtr).first {
                 $0.sectionName == "__cstring"
             }
             guard let cstrings else { return nil }
-            return cstrings.strings(in: text, vmaddrSlide: vmaddrSlide)
+            return cstrings.strings(vmaddrSlide: vmaddrSlide)
         }
         return nil
     }
@@ -301,15 +301,15 @@ extension MachOImage {
             return segments.flatMap { segment in
                 segment.sections(cmdsStart: cmdsStartPtr)
                     .compactMap { section in
-                        section.strings(in: segment, vmaddrSlide: vmaddrSlide)
+                        section.strings(vmaddrSlide: vmaddrSlide)
                     }
             }
         } else {
             let segments = loadCommands.infos(of: LoadCommand.segment)
             return segments.flatMap { segment in
                 segment.sections(cmdsStart: cmdsStartPtr)
                     .compactMap { section in
-                        section.strings(in: segment, vmaddrSlide: vmaddrSlide)
+                        section.strings(vmaddrSlide: vmaddrSlide)
                     }
             }
         }
@@ -698,3 +698,41 @@ extension MachOImage {
         return nil
     }
 }
+
+extension MachOImage {
+    public var cfStrings64: MemorySequence<CFString64>? {
+        guard let section = sections64.first(where: {
+            $0.sectionName == "__cfstring"
+        }) else { return nil }
+        guard let vmaddrSlide else { return nil }
+
+        guard let ptr = section.startPtr(vmaddrSlide: vmaddrSlide) else {
+            return nil
+        }
+        let count = section.size / CFString64.layoutSize
+
+        return .init(
+            basePointer: ptr
+                .assumingMemoryBound(to: CFString64.self),
+            numberOfElements: count
+        )
+    }
+
+    public var cfStrings32: MemorySequence<CFString32>? {
+        guard let section = sections32.first(where: {
+            $0.sectionName == "__cfstring"
+        }) else { return nil }
+        guard let vmaddrSlide else { return nil }
+
+        guard let ptr = section.startPtr(vmaddrSlide: vmaddrSlide) else {
+            return nil
+        }
+        let count = section.size / CFString32.layoutSize
+
+        return .init(
+            basePointer: ptr
+                .assumingMemoryBound(to: CFString32.self),
+            numberOfElements: count
+        )
+    }
+}
diff --git a/Sources/MachOKit/Model/CoreFoundation/CFString.swift b/Sources/MachOKit/Model/CoreFoundation/CFString.swift
@@ -0,0 +1,120 @@
+//
+//  CFString.swift
+//  MachOKit
+//
+//  Created by p-x9 on 2025/02/01
+//  
+//
+
+import Foundation
+import MachOKitC
+
+public protocol CFStringProtocol {
+    /// Offset at which string data is stored
+    var stringOffset: Int { get }
+    /// Number (in terms of UTF-16 code pairs) of Unicode characters in a string.
+    var stringSize: Int { get }
+
+    /// A Boolean value that indicates whether the String data is stored in unicode or not.
+    var isUnicode: Bool { get }
+    /// A Boolean value that indicates whether the String data is stored in 8-bit or not.
+    var isEightBit: Bool { get }
+    
+    /// Obtain a stored string as a `Swift.String`
+    /// - Parameter machO: MachOFile to which `self` belongs
+    /// - Returns: stored string
+    func string(in machO: MachOFile) -> String?
+
+    /// Obtain a stored string as a `Swift.String`
+    /// - Parameter machO: MachOImage to which `self` belongs
+    /// - Returns: stored string
+    func string(in machO: MachOImage) -> String?
+}
+
+public struct CFString64: LayoutWrapper, CFStringProtocol {
+    public typealias Layout = CF_CONST_STRING64
+
+    public var layout: Layout
+}
+
+public struct CFString32: LayoutWrapper, CFStringProtocol {
+    public typealias Layout = CF_CONST_STRING32
+
+    public var layout: Layout
+}
+
+extension CFString64 {
+    public var stringOffset: Int {
+        numericCast(layout._ptr & 0x7ffffffff)
+    }
+
+    public var stringSize: Int {
+        numericCast(layout._length)
+    }
+
+    // ref: https://github.com/apple-oss-distributions/CF/blob/dc54c6bb1c1e5e0b9486c1d26dd5bef110b20bf3/CFString.c#L208C1-L212C3
+    /* !!! Note: Constant CFStrings use the bit patterns:
+    C8 (11001000 = default allocator, not inline, not freed contents; 8-bit; has NULL byte; doesn't have length; is immutable)
+    D0 (11010000 = default allocator, not inline, not freed contents; Unicode; is immutable)
+    The bit usages should not be modified in a way that would effect these bit patterns.
+    */
+    public var isUnicode: Bool {
+        layout._base._cfinfo.0 == 0xD0 // FIXME: consider byte swapped environment (CF_INFO_BITS)
+    }
+
+    public var isEightBit: Bool {
+        layout._base._cfinfo.0 == 0xC8
+    }
+}
+
+extension CFString32 {
+    public var stringOffset: Int {
+        numericCast(layout._ptr)
+    }
+
+    public var stringSize: Int {
+        numericCast(layout._length)
+    }
+
+    public var isUnicode: Bool {
+        layout._base._cfinfo.0 == 0xD0
+    }
+
+    public var isEightBit: Bool {
+        layout._base._cfinfo.0 == 0xC8
+    }
+}
+
+extension CFStringProtocol {
+    public func string(in machO: MachOFile) -> String? {
+        let offset = machO.headerStartOffset + stringOffset
+        if isUnicode {
+            let data = machO.fileHandle.readData(
+                offset: UInt64(offset),
+                size: numericCast(stringSize) * numericCast(MemoryLayout<UInt16/*UniChar*/>.size)
+            )
+            return String(bytes: data, encoding: .utf16LittleEndian)
+        } else {
+            return machO.fileHandle.readString(
+                offset: UInt64(offset),
+                size: stringSize
+            )
+        }
+    }
+
+    public func string(in machO: MachOImage) -> String? {
+        guard let ptr = UnsafeRawPointer(bitPattern: UInt(stringOffset)) else {
+            return nil
+        }
+
+        if isUnicode {
+            let data = Data(bytes: ptr, count: numericCast(stringSize) * numericCast(MemoryLayout<UInt16/*UniChar*/>.size))
+            return String(bytes: data, encoding: .utf16LittleEndian)
+        } else {
+            return .init(
+                cString: ptr.assumingMemoryBound(to: CChar.self),
+                encoding: .ascii
+            )
+        }
+    }
+}
diff --git a/Sources/MachOKit/Protocol/MachORepresentable.swift b/Sources/MachOKit/Protocol/MachORepresentable.swift
@@ -14,6 +14,8 @@ public protocol MachORepresentable {
     associatedtype Symbols64: RandomAccessCollection<Symbol>
     associatedtype Symbols: RandomAccessCollection<Symbol>
     associatedtype IndirectSymbols: RandomAccessCollection<IndirectSymbol>
+    associatedtype CFStrings32: RandomAccessCollection<CFString32>
+    associatedtype CFStrings64: RandomAccessCollection<CFString64>
     associatedtype RebaseOperations: Sequence<RebaseOperation>
     associatedtype BindOperations: Sequence<BindOperation>
     associatedtype ExportTrie: Sequence<ExportTrieEntry>
@@ -82,6 +84,13 @@ public protocol MachORepresentable {
     /// Symbol strings is not included.
     var allCStrings: [String] { get }
 
+    /// List of CFStrings in all segments
+    var cfStrings: [any CFStringProtocol]? { get }
+    /// List of CFStrings in 64-bit architecture segments
+    var cfStrings64: CFStrings64? { get }
+    /// List of CFStrings in 32-bit architecture segments
+    var cfStrings32: CFStrings32? { get }
+
     /// Sequence of rebase operations
     var rebaseOperations: RebaseOperations? { get }
 
@@ -279,6 +288,16 @@ extension MachORepresentable {
     }
 }
 
+extension MachORepresentable {
+    public var cfStrings: [any CFStringProtocol]? {
+        if is64Bit {
+            cfStrings64?.map { $0 as (any CFStringProtocol) }
+        } else {
+            cfStrings32?.map { $0 as (any CFStringProtocol) }
+        }
+    }
+}
+
 extension MachORepresentable {
     public var bindingSymbols: [BindingSymbol] {
         guard let bindOperations else {
diff --git a/Sources/MachOKitC/include/core_foundation.h b/Sources/MachOKitC/include/core_foundation.h
diff --git a/Tests/MachOKitTests/MachOFilePrintTests.swift b/Tests/MachOKitTests/MachOFilePrintTests.swift
diff --git a/Tests/MachOKitTests/MachOPrintTests.swift b/Tests/MachOKitTests/MachOPrintTests.swift
