LibCompress: Add a PackBits decoder

This compression scheme was quite popular during the 80's, and we can
still find it in use inside file formats such as TIFF or PDF.

Tests/LibCompress/CMakeLists.txt

@@ -3,6 +3,7 @@ set(TEST_SOURCES
     TestDeflate.cpp
     TestGzip.cpp
     TestLzma.cpp
+    TestPackBits.cpp
     TestXz.cpp
     TestZlib.cpp
 )

Tests/LibCompress/TestPackBits.cpp

@@ -0,0 +1,25 @@
+/*
+ * Copyright (c) 2023, Lucas Chollet <lucas.chollet@serenityos.org>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#include <LibTest/TestCase.h>
+
+#include <AK/Array.h>
+#include <LibCompress/PackBitsDecoder.h>
+
+TEST_CASE(pack_bits)
+{
+    Array<u8, 15> const compressed {
+        0xFE, 0xAA, 0x02, 0x80, 0x00, 0x2A, 0xFD, 0xAA, 0x03, 0x80, 0x00, 0x2A, 0x22, 0xF7, 0xAA
+    };
+
+    Array<u8, 24> const raw {
+        0xAA, 0xAA, 0xAA, 0x80, 0x00, 0x2A, 0xAA, 0xAA, 0xAA, 0xAA, 0x80, 0x00,
+        0x2A, 0x22, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA
+    };
+
+    auto unpacked = TRY_OR_FAIL(Compress::PackBits::decode_all(compressed));
+    EXPECT_EQ(unpacked.bytes(), raw);
+}

Userland/Libraries/LibCompress/CMakeLists.txt

@@ -4,6 +4,7 @@ set(SOURCES
     Deflate.cpp
     Lzma.cpp
     Lzma2.cpp
+    PackBitsDecoder.cpp
     Xz.cpp
     Zlib.cpp
     Gzip.cpp

Userland/Libraries/LibCompress/PackBitsDecoder.cpp

@@ -0,0 +1,45 @@
+/*
+ * Copyright (c) 2023, Lucas Chollet <lucas.chollet@serenityos.org>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#include "PackBitsDecoder.h"
+#include <AK/MemoryStream.h>
+
+namespace Compress::PackBits {
+
+ErrorOr<ByteBuffer> decode_all(ReadonlyBytes bytes, Optional<u64> expected_output_size, CompatibilityMode mode)
+{
+    // This implementation uses unsigned values for the selector, as described in the PDF spec.
+    // Note that this remains compatible with other implementations based on signed numbers.
+
+    auto memory_stream = make<FixedMemoryStream>(bytes);
+
+    ByteBuffer decoded_bytes;
+
+    if (expected_output_size.has_value())
+        TRY(decoded_bytes.try_ensure_capacity(*expected_output_size));
+
+    while (memory_stream->remaining() > 0 && decoded_bytes.size() < expected_output_size.value_or(NumericLimits<u64>::max())) {
+        auto const length = TRY(memory_stream->read_value<u8>());
+
+        if (length < 128) {
+            for (u8 i = 0; i <= length; ++i)
+                TRY(decoded_bytes.try_append(TRY(memory_stream->read_value<u8>())));
+        } else if (length > 128) {
+            auto const next_byte = TRY(memory_stream->read_value<u8>());
+
+            for (u8 i = 0; i < 257 - length; ++i)
+                TRY(decoded_bytes.try_append(next_byte));
+        } else {
+            VERIFY(length == 128);
+            if (mode == CompatibilityMode::PDF)
+                break;
+        }
+    }
+
+    return decoded_bytes;
+}
+
+}

Userland/Libraries/LibCompress/PackBitsDecoder.h

@@ -0,0 +1,27 @@
+/*
+ * Copyright (c) 2023, Lucas Chollet <lucas.chollet@serenityos.org>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#pragma once
+
+#include <AK/ByteBuffer.h>
+#include <AK/Optional.h>
+
+namespace Compress::PackBits {
+
+// This implements the PackBits compression scheme, aka run-length compression
+// It is fairly simple and described here: https://web.archive.org/web/20080705155158/http://developer.apple.com/technotes/tn/tn1023.html
+// But also in section:
+//  - 7.4.5 RunLengthDecode Filter of the PDF specification
+//  - Section 9: PackBits Compression of the TIFF specification
+
+enum class CompatibilityMode {
+    Original, // 128 is defined as no-op
+    PDF,      // 128 is defined as end of stream
+};
+
+ErrorOr<ByteBuffer> decode_all(ReadonlyBytes bytes, Optional<u64> expected_output_size = {}, CompatibilityMode mode = CompatibilityMode::Original);
+
+}