From 786e654dfdbd478678fa8e1a1e671f6c7910f5ea Mon Sep 17 00:00:00 2001
From: Tim Schumacher <timschumi@gmx.de>
Date: Sun, 22 Oct 2023 19:49:41 +0200
Subject: [PATCH] LibCompress: Implement the XZ delta filter

---
 Tests/LibCompress/TestXz.cpp          |  8 +--
 Userland/Libraries/LibCompress/Xz.cpp | 71 +++++++++++++++++++++++++++
 Userland/Libraries/LibCompress/Xz.h   | 26 ++++++++++
 3 files changed, 102 insertions(+), 3 deletions(-)

diff --git a/Tests/LibCompress/TestXz.cpp b/Tests/LibCompress/TestXz.cpp
index 4f4ca826a4..7a70be3507 100644
--- a/Tests/LibCompress/TestXz.cpp
+++ b/Tests/LibCompress/TestXz.cpp
@@ -1183,8 +1183,8 @@ TEST_CASE(xz_utils_good_1_3delta_lzma2)
 
     auto stream = MUST(try_make<FixedMemoryStream>(compressed));
     auto decompressor = MUST(Compress::XzDecompressor::create(move(stream)));
-    // TODO: This uses the currently unimplemented delta filter.
-    (void)decompressor->read_until_eof(PAGE_SIZE);
+    auto buffer = TRY_OR_FAIL(decompressor->read_until_eof(PAGE_SIZE));
+    EXPECT_EQ(buffer.span(), xz_utils_lorem_ipsum.bytes());
 }
 
 TEST_CASE(xz_utils_good_1_arm64_lzma2_1)
@@ -1864,7 +1864,9 @@ TEST_CASE(xz_utils_unsupported_filter_flags_2)
     auto stream = MUST(try_make<FixedMemoryStream>(compressed));
     auto decompressor = MUST(Compress::XzDecompressor::create(move(stream)));
 
-    // TODO: The delta filter is currently unimplemented. However, once it is implemented, we likely want to mirror the recommended behavior of the specification anyways.
+    // TODO: We don't yet check whether the filter chain satisfies the "can't be the last filter"
+    //       requirement. We just happen to get the result right because we try to uncompress the
+    //       test case and fail.
     auto buffer_or_error = decompressor->read_until_eof(PAGE_SIZE);
     EXPECT(buffer_or_error.is_error());
 }
diff --git a/Userland/Libraries/LibCompress/Xz.cpp b/Userland/Libraries/LibCompress/Xz.cpp
index 2615de91e1..f928d1ad81 100644
--- a/Userland/Libraries/LibCompress/Xz.cpp
+++ b/Userland/Libraries/LibCompress/Xz.cpp
@@ -161,6 +161,66 @@ u32 XzFilterLzma2Properties::dictionary_size() const
     return dictionary_size;
 }
 
+u32 XzFilterDeltaProperties::distance() const
+{
+    // "The Properties byte indicates the delta distance, which can be
+    //  1-256 bytes backwards from the current byte: 0x00 indicates
+    //  distance of 1 byte and 0xFF distance of 256 bytes."
+    return encoded_distance + 1;
+}
+
+ErrorOr<NonnullOwnPtr<XzFilterDelta>> XzFilterDelta::create(MaybeOwned<Stream> stream, u32 distance)
+{
+    auto buffer = TRY(CircularBuffer::create_empty(distance));
+    auto filter = TRY(adopt_nonnull_own_or_enomem(new (nothrow) XzFilterDelta(move(stream), move(buffer))));
+    return filter;
+}
+
+XzFilterDelta::XzFilterDelta(MaybeOwned<Stream> stream, CircularBuffer buffer)
+    : m_stream(move(stream))
+    , m_buffer(move(buffer))
+{
+}
+
+ErrorOr<Bytes> XzFilterDelta::read_some(Bytes bytes)
+{
+    bytes = TRY(m_stream->read_some(bytes));
+
+    auto distance = m_buffer.capacity();
+
+    for (auto& byte : bytes) {
+        if (m_buffer.seekback_limit() >= distance) {
+            u8 byte_at_distance { 0 };
+            MUST(m_buffer.read_with_seekback({ &byte_at_distance, 1 }, distance));
+            byte = byte_at_distance + byte;
+        }
+
+        m_buffer.write({ &byte, 1 });
+        MUST(m_buffer.discard(1));
+    }
+
+    return bytes;
+}
+
+ErrorOr<size_t> XzFilterDelta::write_some(ReadonlyBytes)
+{
+    return EBADF;
+}
+
+bool XzFilterDelta::is_eof() const
+{
+    return m_stream->is_eof();
+}
+
+bool XzFilterDelta::is_open() const
+{
+    return m_stream->is_open();
+}
+
+void XzFilterDelta::close()
+{
+}
+
 ErrorOr<NonnullOwnPtr<XzDecompressor>> XzDecompressor::create(MaybeOwned<Stream> stream)
 {
     auto counting_stream = TRY(try_make<CountingStream>(move(stream)));
@@ -353,6 +413,17 @@ ErrorOr<void> XzDecompressor::load_next_block(u8 encoded_block_header_size)
             continue;
         }
 
+        // 5.3.3. Delta
+        if (filter.id == 0x03) {
+            if (filter.properties.size() < sizeof(XzFilterDeltaProperties))
+                return Error::from_string_literal("XZ Delta filter has a smaller-than-needed properties size");
+
+            auto const* properties = reinterpret_cast<XzFilterDeltaProperties*>(filter.properties.data());
+
+            new_block_stream = TRY(XzFilterDelta::create(move(new_block_stream), properties->distance()));
+            continue;
+        }
+
         return Error::from_string_literal("XZ block header contains unknown filter ID");
     }
 
diff --git a/Userland/Libraries/LibCompress/Xz.h b/Userland/Libraries/LibCompress/Xz.h
index 1d07fa956b..8a054eb0b4 100644
--- a/Userland/Libraries/LibCompress/Xz.h
+++ b/Userland/Libraries/LibCompress/Xz.h
@@ -6,6 +6,7 @@
 
 #pragma once
 
+#include <AK/CircularBuffer.h>
 #include <AK/ConstrainedStream.h>
 #include <AK/CountingStream.h>
 #include <AK/Endian.h>
@@ -98,6 +99,31 @@ struct [[gnu::packed]] XzFilterLzma2Properties {
 };
 static_assert(sizeof(XzFilterLzma2Properties) == 1);
 
+// 5.3.3. Delta
+struct [[gnu::packed]] XzFilterDeltaProperties {
+    u8 encoded_distance;
+
+    u32 distance() const;
+};
+static_assert(sizeof(XzFilterDeltaProperties) == 1);
+
+class XzFilterDelta : public Stream {
+public:
+    static ErrorOr<NonnullOwnPtr<XzFilterDelta>> create(MaybeOwned<Stream>, u32 distance);
+
+    virtual ErrorOr<Bytes> read_some(Bytes) override;
+    virtual ErrorOr<size_t> write_some(ReadonlyBytes) override;
+    virtual bool is_eof() const override;
+    virtual bool is_open() const override;
+    virtual void close() override;
+
+private:
+    XzFilterDelta(MaybeOwned<Stream>, CircularBuffer);
+
+    MaybeOwned<Stream> m_stream;
+    CircularBuffer m_buffer;
+};
+
 class XzDecompressor : public Stream {
 public:
     static ErrorOr<NonnullOwnPtr<XzDecompressor>> create(MaybeOwned<Stream>);