RGBCube/serenity
1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-07-25 19:37:35 +00:00
Code Issues Activity Actions

LibGfx/TIFF: Support TIFF 5.0 LZW images

Browse source 
This makes us decode `quad-lzw.tif` from libtiff's test suite:
https://libtiff.gitlab.io/libtiff/images.html
This commit is contained in:
Lucas CHOLLET 2023-12-01 20:37:59 -05:00 • committed by Andreas Kling
parent 4c145fdb12
commit e6fc18d96a
1 changed files with 17 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

18
Userland/Libraries/LibGfx/ImageFormats/TIFFLoader.cpp
View file

@ -119,7 +119,23 @@ private:
u32 read_head {};
auto initializer = [&](u32 bytes) -> ErrorOr<void> {
decoded_bytes = TRY(Compress::LZWDecoder<BigEndianInputBitStream>::decode_all(TRY(m_stream->read_in_place<u8 const>(bytes)), 8, -1));
auto const encoded_bytes = TRY(m_stream->read_in_place<u8 const>(bytes));
if (encoded_bytes.is_empty())
return Error::from_string_literal("TIFFImageDecoderPlugin: Unable to read from empty LZW strip");
// Note: AFAIK, there are two common ways to use LZW compression:
// - With a LittleEndian stream and no Early-Change, this is used in the GIF format
// - With a BigEndian stream and an EarlyChange of 1, this is used in the PDF format
// The fun begins when they decided to change from the former to the latter when moving
// from TIFF 5.0 to 6.0, and without including a way for files to be identified.
// Fortunately, as the first byte of a LZW stream is a constant we can guess the endianess
// and deduce the version from it. The first code is 0x100 (9-bits).
if (encoded_bytes[0] == 0x00)
decoded_bytes = TRY(Compress::LZWDecoder<LittleEndianInputBitStream>::decode_all(encoded_bytes, 8, 0));
else
decoded_bytes = TRY(Compress::LZWDecoder<BigEndianInputBitStream>::decode_all(encoded_bytes, 8, -1));
read_head = 0;
return {};
};