LibCompress: Don't assume zlib header is available right away

Instead of checking the header in ZlibDecompressor::create(), we now
check it in read_some() when it is called for the first time. This
resolves a FIXME in the new DecompressionStream implementation.

Libraries/LibCompress/Zlib.cpp

@@ -17,7 +17,19 @@ namespace Compress {
 
 ErrorOr<NonnullOwnPtr<ZlibDecompressor>> ZlibDecompressor::create(MaybeOwned<Stream> stream)
 {
-    auto header = TRY(stream->read_value<ZlibHeader>());
+    return adopt_nonnull_own_or_enomem(new (nothrow) ZlibDecompressor(move(stream)));
+}
+
+ZlibDecompressor::ZlibDecompressor(MaybeOwned<Stream> stream)
+    : m_has_seen_header(false)
+    , m_stream(move(stream))
+{
+}
+
+ErrorOr<Bytes> ZlibDecompressor::read_some(Bytes bytes)
+{
+    if (!m_has_seen_header) {
+        auto header = TRY(m_stream->read_value<ZlibHeader>());
 
         if (header.compression_method != ZlibCompressionMethod::Deflate || header.compression_info > 7)
             return Error::from_string_literal("Non-DEFLATE compression inside Zlib is not supported");
@@ -28,20 +40,12 @@ ErrorOr<NonnullOwnPtr<ZlibDecompressor>> ZlibDecompressor::create(MaybeOwned<Str
         if (header.as_u16 % 31 != 0)
             return Error::from_string_literal("Zlib error correction code does not match");
 
-    auto bit_stream = make<LittleEndianInputBitStream>(move(stream));
+        auto bit_stream = make<LittleEndianInputBitStream>(move(m_stream));
         auto deflate_stream = TRY(Compress::DeflateDecompressor::construct(move(bit_stream)));
 
-    return adopt_nonnull_own_or_enomem(new (nothrow) ZlibDecompressor(header, move(deflate_stream)));
+        m_stream = move(deflate_stream);
+        m_has_seen_header = true;
     }
-
-ZlibDecompressor::ZlibDecompressor(ZlibHeader header, NonnullOwnPtr<Stream> stream)
-    : m_header(header)
-    , m_stream(move(stream))
-{
-}
-
-ErrorOr<Bytes> ZlibDecompressor::read_some(Bytes bytes)
-{
     return m_stream->read_some(bytes);
 }
 

Libraries/LibCompress/Zlib.h

@@ -55,10 +55,10 @@ public:
     virtual void close() override;
 
 private:
-    ZlibDecompressor(ZlibHeader, NonnullOwnPtr<Stream>);
+    ZlibDecompressor(MaybeOwned<Stream>);
 
-    ZlibHeader m_header;
-    NonnullOwnPtr<Stream> m_stream;
+    bool m_has_seen_header { false };
+    MaybeOwned<Stream> m_stream;
 };
 
 class ZlibCompressor : public Stream {

Libraries/LibWeb/Compression/DecompressionStream.cpp

@@ -32,8 +32,6 @@ WebIDL::ExceptionOr<GC::Ref<DecompressionStream>> DecompressionStream::construct
     auto decompressor = [&, input_stream = MaybeOwned<Stream> { *input_stream }]() mutable -> ErrorOr<Decompressor> {
         switch (format) {
         case Bindings::CompressionFormat::Deflate:
-            // FIXME: Our zlib decompressor assumes the initial data contains the zlib header. We don't have any data yet,
-            //        so this will always fail.
             return TRY(Compress::ZlibDecompressor::create(move(input_stream)));
         case Bindings::CompressionFormat::DeflateRaw:
             return TRY(Compress::DeflateDecompressor::construct(make<LittleEndianInputBitStream>(move(input_stream))));

Tests/LibCompress/TestZlib.cpp

@@ -27,6 +27,25 @@ TEST_CASE(zlib_decompress_simple)
     EXPECT(decompressed.bytes() == (ReadonlyBytes { uncompressed, sizeof(uncompressed) - 1 }));
 }
 
+TEST_CASE(zlib_decompress_stream)
+{
+    Array<u8, 40> const compressed {
+        0x78, 0x01, 0x01, 0x1D, 0x00, 0xE2, 0xFF, 0x54, 0x68, 0x69, 0x73, 0x20,
+        0x69, 0x73, 0x20, 0x61, 0x20, 0x73, 0x69, 0x6D, 0x70, 0x6C, 0x65, 0x20,
+        0x74, 0x65, 0x78, 0x74, 0x20, 0x66, 0x69, 0x6C, 0x65, 0x20, 0x3A, 0x29,
+        0x99, 0x5E, 0x09, 0xE8
+    };
+
+    u8 const uncompressed[] = "This is a simple text file :)";
+
+    auto stream = make<AllocatingMemoryStream>();
+    auto input = MaybeOwned<Stream> { *stream };
+    auto decompressor = TRY_OR_FAIL(Compress::ZlibDecompressor::create(move(input)));
+    TRY_OR_FAIL(stream->write_until_depleted(compressed));
+    auto decompressed = TRY_OR_FAIL(decompressor->read_until_eof());
+    EXPECT(decompressed.bytes() == (ReadonlyBytes { uncompressed, sizeof(uncompressed) - 1 }));
+}
+
 TEST_CASE(zlib_compress_simple)
 {
     // Note: This is just the output of our compression function from an arbitrary point in time.

Tests/LibWeb/Text/expected/Compression/DecompressionStream.txt

@@ -1,2 +1,3 @@
+format=deflate: Well hello friends!
 format=deflate-raw: Well hello friends!
 format=gzip: Well hello friends!

Tests/LibWeb/Text/input/Compression/DecompressionStream.html

@@ -2,8 +2,7 @@
 <script>
     asyncTest(async done => {
         const data = [
-            // FIXME: Enable this case when our DecompressionStream is functioning with the zlib format.
-            // { format: "deflate", text: "eJwLT83JUchIzcnJV0grykzNSylWBABGEQb1" },
+            { format: "deflate", text: "eJwLT83JUchIzcnJV0grykzNSylWBABGEQb1" },
             { format: "deflate-raw", text: "C0/NyVHISM3JyVdIK8pMzUspVgQA" },
             { format: "gzip", text: "H4sIAAAAAAADAwtPzclRyEjNyclXSCvKTM1LKVYEAHN0w4sTAAAA" },
         ];