zlib-based homecooked gzip reader for significantly faster decompression times

Package.swift

@@ -8,14 +8,10 @@ let package = Package(
   ],
   dependencies: [
     .package(url: "https://github.com/apple/swift-argument-parser", from: "1.5.0"),
-    .package(url: "https://github.com/tsolomko/SWCompression", from: "4.8.6"),
   ],
   targets: [
     .target(
       name: "darwin-apk",
-      dependencies: [
-        .product(name: "SWCompression", package: "SWCompression"),
-      ],
       path: "Sources/apk"),
     .executableTarget(
       name: "dpk",

Sources/apk/Index/ApkIndexUpdate.swift

@@ -4,7 +4,6 @@
  */
 
 import Foundation
-import SWCompression
 import CryptoKit
 
 public struct ApkIndexUpdater {
@@ -64,23 +63,46 @@ public struct ApkIndexUpdater {
     let tarSignature: [TarReader.Entry]
     let tarRecords: [TarReader.Entry]
 
-    let tars = try GzipArchive.multiUnarchive(  // Slow...
-      archive: Data(contentsOf: indexURL))
-    assert(tars.count >= 2)
+    print("Archive:    \(indexURL.lastPathComponent)")
 
-    var signatureStream = MemoryInputStream(buffer: tars[0].data)
+    let durFormat = Duration.UnitsFormatStyle(
+      allowedUnits: [ .seconds, .milliseconds ],
+      width: .condensedAbbreviated,
+      fractionalPart: .show(length: 3))
+    let gzipStart = ContinuousClock.now
+
+    var tars = [Data]()
+    do {
+      var file: any InputStream = try FileInputStream(indexURL)
+      //var file: any InputStream = try MemoryInputStream(buffer: try Data(contentsOf: indexURL))
+      tars.append(try GZip.read(inStream: &file))
+      tars.append(try GZip.read(inStream: &file))
+      
+    } catch {
+      fatalError(error.localizedDescription)
+    }
+
+    print("Gzip time:  \((ContinuousClock.now - gzipStart).formatted(durFormat))")
+    let untarStart = ContinuousClock.now
+
+    var signatureStream = MemoryInputStream(buffer: tars[0])
     tarSignature = try TarReader.read(&signatureStream)
-    var recordsStream = MemoryInputStream(buffer: tars[1].data)
+    var recordsStream = MemoryInputStream(buffer: tars[1])
     tarRecords = try TarReader.read(&recordsStream)
 
     guard case .file(let signatureName, _) = tarSignature.first
     else { fatalError("Missing signature") }
-    print(signatureName)
     guard let apkIndexFile = tarRecords.firstFile(name: "APKINDEX")
     else { fatalError("APKINDEX missing") }
     guard let description = tarRecords.firstFile(name: "DESCRIPTION")
     else { fatalError("DESCRIPTION missing") }
 
+    print("TAR time:   \((ContinuousClock.now - untarStart).formatted(durFormat))")
+    let indexStart = ContinuousClock.now
+    defer {
+      print("Index time: \((ContinuousClock.now - indexStart).formatted(durFormat))")
+    }
+
     let reader = TextInputStream(binaryStream: MemoryInputStream(buffer: apkIndexFile))
     return try ApkIndex(raw:
       try ApkRawIndex(lines: reader.lines))

Sources/apk/Utility/FileInputStream.swift

@@ -4,6 +4,8 @@
  */
 
 import Foundation
+import Darwin
+import System
 
 public struct FileInputStream: InputStream {
   private var _hnd: FileHandle
@@ -72,4 +74,12 @@ public struct FileInputStream: InputStream {
       throw .fileHandleError(error)
     }
   }
+
+  public mutating func read(_ buffer: UnsafeMutablePointer<UInt8>, maxLength len: Int) throws(StreamError) -> Int {
+    let res = unistd.read(self._hnd.fileDescriptor, buffer, len)
+    if res < 0 {
+      throw .fileHandleError(Errno(rawValue: errno))
+    }
+    return res
+  }
 }

Sources/apk/Utility/GZip.swift

@@ -0,0 +1,203 @@
+/*
+ * darwin-apk © 2024 Gay Pizza Specifications
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import Foundation
+import zlib
+
+struct GZip {
+  static let CM_DEFLATE: UInt8 = 8
+
+  static let FTEXT: UInt8    = 1 << 0
+  static let FHCRC: UInt8    = 1 << 1
+  static let FEXTRA: UInt8   = 1 << 2
+  static let FNAME: UInt8    = 1 << 3
+  static let FCOMMENT: UInt8 = 1 << 4
+
+  static let XFL_BEST: UInt8    = 2
+  static let XFL_FASTEST: UInt8 = 4
+
+  private static func skipString(_ stream: inout any InputStream) throws(GZipError) {
+    var c: UInt8?
+    repeat {
+      c = stream.next()
+      guard c != nil else {
+        throw .truncatedStream
+      }
+    } while c != 0
+  }
+
+  static func read(inStream stream: inout any InputStream) throws(GZipError) -> Data {
+    // Check Gzip magic signature
+    guard (try? stream.read(2)) == Data([0x1F, 0x8B]) else {
+      throw .badMagic
+    }
+
+    // Check compression field (should only ever be DEFLATE)
+    guard let compression = stream.next(),
+        compression == Self.CM_DEFLATE else {
+      throw .badHeader
+    }
+
+    guard
+        let flags = stream.next(),
+        let modificationTime = stream.readUInt(),
+        let extraFlags = stream.next(),
+        let operatingSystemID = stream.next() else {
+      throw .truncatedStream
+    }
+
+
+    if flags & Self.FEXTRA != 0 {
+      // Skip "extra" field
+      guard let extraLength = stream.readUShort() else {
+        throw.truncatedStream
+      }
+      do {
+        try stream.seek(.current(Int(extraLength)))
+      } catch {
+        throw .streamError(error)
+      }
+    }
+    if flags & Self.FNAME != 0 {
+      // Skip null-terminated name string
+      try skipString(&stream)
+    }
+    if flags & Self.FCOMMENT != 0 {
+      // Skip null-terminated comment string
+      try skipString(&stream)
+    }
+    if flags & Self.FHCRC != 0 {
+      guard let crc16 = stream.readUShort() else {
+        throw .badField("crc16")
+      }
+    }
+
+    let deflateBegin: Int
+    do {
+      deflateBegin = try stream.tell
+    } catch {
+      throw .streamError(error)
+    }
+
+    var payload = Data()
+    let (streamLength, computedCRC) = try Self.deflate(payload: &payload, stream: &stream)
+
+    // End-of-stream verification fields
+    do {
+      try stream.seek(.set(deflateBegin + streamLength))
+    } catch {
+      throw .streamError(error)
+    }
+    guard
+        let crc = stream.readUInt(),
+        let inputSizeMod32 = stream.readUInt() else {
+      throw .truncatedStream
+    }
+
+    // Perform verification checks
+    guard UInt32(truncatingIfNeeded: computedCRC) == crc else {
+      throw .verificationFailed("CRC32 didn't match")
+    }
+    guard inputSizeMod32 == UInt32(truncatingIfNeeded: payload.count) else {
+      throw .verificationFailed("Bad decompressed size")
+    }
+
+    return payload
+  }
+
+  private static func deflate(payload: inout Data, stream: inout any InputStream) throws(GZipError) -> (Int, UInt) {
+    var zstream = z_stream()
+    var zerr = inflateInit2_(&zstream, -15, ZLIB_VERSION, Int32(MemoryLayout<z_stream>.size))
+    guard zerr == Z_OK else {
+      throw .zlib(zerr)
+    }
+
+    defer {
+      inflateEnd(&zstream)
+    }
+
+    let bufferSize = 0x8000
+    var inputBuffer = [UInt8](repeating: 0, count: bufferSize)
+    var outputBuffer = [UInt8](repeating: 0, count: bufferSize)
+
+    var computeCRC: UInt = crc32(0, nil, 0)
+    var block = 0
+    repeat {
+      if zstream.avail_in == 0 {
+        let read: Int
+        do {
+          read = try stream.read(&inputBuffer, maxLength: inputBuffer.count)
+        } catch {
+          throw .streamError(error)
+        }
+        guard read > 0 else {
+          throw .truncatedStream
+        }
+        zstream.avail_in = UInt32(read)
+        zstream.next_in = inputBuffer.withUnsafeMutableBufferPointer(\.baseAddress!)
+      }
+      zstream.avail_out = UInt32(outputBuffer.count)
+      zstream.next_out = outputBuffer.withUnsafeMutableBufferPointer(\.baseAddress!)
+      zerr = inflate(&zstream, Z_BLOCK)
+
+      let decodedBytes = outputBuffer.count - Int(zstream.avail_out)
+      computeCRC = crc32(computeCRC, outputBuffer, UInt32(decodedBytes))
+      payload += Data(outputBuffer[..<decodedBytes])
+      block += decodedBytes
+
+      if zstream.data_type & (1 << 7) != 0 {
+        // At the end of a deflate block, we're done if it was empty
+        if block == 0 {
+          break
+        }
+        block = 0
+      }
+    } while zerr == Z_OK
+
+    guard zerr == Z_STREAM_END else {
+      throw .zlib(zerr)
+    }
+
+    return (Int(zstream.total_in), computeCRC)
+  }
+}
+
+enum GZipError: LocalizedError {
+  case streamError(_ err: StreamError)
+  case verificationFailed(_ msg: String)
+  case badMagic
+  case badHeader
+  case badField(_ name: String)
+  case truncatedStream
+  case zlib(_ err: Int32)
+
+  var errorDescription: String? {
+    switch self {
+    case .verificationFailed(let msg): msg
+    case .streamError(let err): "Underlying stream error: \(err.localizedDescription)"
+    case .badMagic:             "Not a Gzip file"
+    case .badHeader:            "Malformed Gzip header"
+    case .badField(let name):   "Bad Gzip \(name) field"
+    case .truncatedStream:      "Reached end-of-stream before decoding finished"
+    case .zlib(let err):        "zlib error \(err)"
+    }
+  }
+}
+
+fileprivate extension InputStream {
+  mutating func readUShort() -> UInt16? {
+    guard let buffer = try? self.read(2), buffer.count == 2 else {
+      return nil
+    }
+    return buffer.withUnsafeBytes { $0.load(as: UInt16.self) }.littleEndian
+  }
+
+  mutating func readUInt() -> UInt32? {
+    guard let buffer = try? self.read(4), buffer.count == 4 else {
+      return nil
+    }
+    return buffer.withUnsafeBytes { $0.load(as: UInt32.self) }.littleEndian
+  }
+}

Sources/apk/Utility/InputStream.swift

@@ -5,10 +5,9 @@
 
 import Foundation
 
-public protocol InputStream: Stream, IteratorProtocol {
-  associatedtype Element = UInt8
-
+public protocol InputStream: Stream, IteratorProtocol where Element == UInt8 {
   mutating func read(_ count: Int) throws(StreamError) -> Data
+  mutating func read(_ buffer: UnsafeMutablePointer<UInt8>, maxLength len: Int) throws(StreamError) -> Int
 }
 
 public extension InputStream {

Sources/apk/Utility/MemoryInputStream.swift

@@ -52,6 +52,15 @@ public struct MemoryInputStream: InputStream {
     return bytes
   }
 
+  public mutating func read(_ buffer: UnsafeMutablePointer<UInt8>, maxLength count: Int) throws(StreamError) -> Int {
+    let beg = min(self._idx, self._len)
+    let end = min(self._idx + count, self._len)
+    let len = beg.distance(to: end)
+    let buf = UnsafeMutableRawBufferPointer(start: buffer, count: len)
+    self._idx += len
+    return self._sli.copyBytes(to: buf, from: beg..<end)
+  }
+
   public mutating func next() -> UInt8? {
     if self._idx < self._len {
       let byte = self._sli[self._idx]

Sources/apk/Utility/Stream.swift

@@ -4,6 +4,7 @@
  */
 
 import Foundation
+import System
 
 public protocol Stream {
   mutating func seek(_ whence: StreamWhence) throws(StreamError)
@@ -21,6 +22,7 @@ public enum StreamError: Error, LocalizedError {
   case seekRange
   case overflow
   case fileHandleError(_ error: any Error)
+  case fileDescriptorError(_ error: Errno)
 
   public var errorDescription: String? {
     switch self {
@@ -28,6 +30,7 @@ public enum StreamError: Error, LocalizedError {
     case .seekRange: "Seek out of range"
     case .overflow: "Stream position overflowed"
     case .fileHandleError(let error): "Error from file handle: \(error.localizedDescription)"
+    case .fileDescriptorError(let error): "\(error)"
     }
   }
 }