From 139e809282053f736cb039229e844b7478605c55 Mon Sep 17 00:00:00 2001
From: Sophie Herold <sophie@hemio.de>
Date: Mon, 25 Dec 2023 00:22:41 +0100
Subject: [PATCH] tiff: Support CMYK images (#2075)

---
 src/codecs/tiff.rs                            |  50 ++++++++++++++++--
 src/color.rs                                  |   6 ++-
 .../images/tiff/testsuite/hpredict_cmyk.tiff  | Bin 0 -> 2668 bytes
 3 files changed, 52 insertions(+), 4 deletions(-)
 create mode 100644 tests/images/tiff/testsuite/hpredict_cmyk.tiff

diff --git a/src/codecs/tiff.rs b/src/codecs/tiff.rs
index 7ee05e1047..6ad356230a 100644
--- a/src/codecs/tiff.rs
+++ b/src/codecs/tiff.rs
@@ -28,6 +28,7 @@ where
 {
     dimensions: (u32, u32),
     color_type: ColorType,
+    original_color_type: ExtendedColorType,
 
     // We only use an Option here so we can call with_limits on the decoder without moving.
     inner: Option<tiff::decoder::Decoder<R>>,
@@ -42,7 +43,7 @@ where
         let mut inner = tiff::decoder::Decoder::new(r).map_err(ImageError::from_tiff_decode)?;
 
         let dimensions = inner.dimensions().map_err(ImageError::from_tiff_decode)?;
-        let color_type = inner.colortype().map_err(ImageError::from_tiff_decode)?;
+        let tiff_color_type = inner.colortype().map_err(ImageError::from_tiff_decode)?;
         match inner.find_tag_unsigned_vec::<u16>(tiff::tags::Tag::SampleFormat) {
             Ok(Some(sample_formats)) => {
                 for format in sample_formats {
@@ -53,7 +54,7 @@ where
             Err(other) => return Err(ImageError::from_tiff_decode(other)),
         };
 
-        let color_type = match color_type {
+        let color_type = match tiff_color_type {
             tiff::ColorType::Gray(8) => ColorType::L8,
             tiff::ColorType::Gray(16) => ColorType::L16,
             tiff::ColorType::GrayA(8) => ColorType::La8,
@@ -62,6 +63,7 @@ where
             tiff::ColorType::RGB(16) => ColorType::Rgb16,
             tiff::ColorType::RGBA(8) => ColorType::Rgba8,
             tiff::ColorType::RGBA(16) => ColorType::Rgba16,
+            tiff::ColorType::CMYK(8) => ColorType::Rgb8,
 
             tiff::ColorType::Palette(n) | tiff::ColorType::Gray(n) => {
                 return Err(err_unknown_color_type(n))
@@ -74,12 +76,30 @@ where
             }
         };
 
+        let original_color_type = match tiff_color_type {
+            tiff::ColorType::CMYK(8) => ExtendedColorType::Cmyk8,
+            _ => color_type.into(),
+        };
+
         Ok(TiffDecoder {
             dimensions,
             color_type,
+            original_color_type,
             inner: Some(inner),
         })
     }
+
+    // The buffer can be larger for CMYK than the RGB output
+    fn total_bytes_buffer(&self) -> u64 {
+        let dimensions = self.dimensions();
+        let total_pixels = u64::from(dimensions.0) * u64::from(dimensions.1);
+        let bytes_per_pixel = if self.original_color_type == ExtendedColorType::Cmyk8 {
+            16
+        } else {
+            u64::from(self.color_type().bytes_per_pixel())
+        };
+        total_pixels.saturating_mul(bytes_per_pixel)
+    }
 }
 
 fn check_sample_format(sample_format: u16) -> Result<(), ImageError> {
@@ -176,6 +196,10 @@ impl<'a, R: 'a + Read + Seek> ImageDecoder<'a> for TiffDecoder<R> {
         self.color_type
     }
 
+    fn original_color_type(&self) -> ExtendedColorType {
+        self.original_color_type
+    }
+
     fn icc_profile(&mut self) -> Option<Vec<u8>> {
         if let Some(decoder) = &mut self.inner {
             decoder.get_tag_u8_vec(tiff::tags::Tag::Unknown(34675)).ok()
@@ -191,7 +215,7 @@ impl<'a, R: 'a + Read + Seek> ImageDecoder<'a> for TiffDecoder<R> {
         limits.check_dimensions(width, height)?;
 
         let max_alloc = limits.max_alloc.unwrap_or(u64::MAX);
-        let max_intermediate_alloc = max_alloc.saturating_sub(self.total_bytes());
+        let max_intermediate_alloc = max_alloc.saturating_sub(self.total_bytes_buffer());
 
         let mut tiff_limits: tiff::decoder::Limits = Default::default();
         tiff_limits.decoding_buffer_size =
@@ -234,6 +258,14 @@ impl<'a, R: 'a + Read + Seek> ImageDecoder<'a> for TiffDecoder<R> {
             .read_image()
             .map_err(ImageError::from_tiff_decode)?
         {
+            tiff::decoder::DecodingResult::U8(v)
+                if self.original_color_type == ExtendedColorType::Cmyk8 =>
+            {
+                let mut out_cur = Cursor::new(buf);
+                for cmyk in v.chunks_exact(4) {
+                    out_cur.write_all(&cmyk_to_rgb(cmyk))?;
+                }
+            }
             tiff::decoder::DecodingResult::U8(v) => {
                 buf.copy_from_slice(&v);
             }
@@ -274,6 +306,18 @@ pub struct TiffEncoder<W> {
     w: W,
 }
 
+fn cmyk_to_rgb(cmyk: &[u8]) -> [u8; 3] {
+    let c = cmyk[0] as f32;
+    let m = cmyk[1] as f32;
+    let y = cmyk[2] as f32;
+    let kf = 1. - cmyk[3] as f32 / 255.;
+    [
+        ((255. - c) * kf) as u8,
+        ((255. - m) * kf) as u8,
+        ((255. - y) * kf) as u8,
+    ]
+}
+
 // Utility to simplify and deduplicate error handling during 16-bit encoding.
 fn u8_slice_as_u16(buf: &[u8]) -> ImageResult<&[u16]> {
     bytemuck::try_cast_slice(buf).map_err(|err| {
diff --git a/src/color.rs b/src/color.rs
index 5ec6a4677b..47f7fee3e4 100644
--- a/src/color.rs
+++ b/src/color.rs
@@ -142,6 +142,9 @@ pub enum ExtendedColorType {
     /// Pixel is 32-bit float RGBA
     Rgba32F,
 
+    /// Pixel is 8-bit CMYK
+    Cmyk8,
+
     /// Pixel is of unknown color type with the specified bits per pixel. This can apply to pixels
     /// which are associated with an external palette. In that case, the pixel value is an index
     /// into the palette.
@@ -180,7 +183,8 @@ impl ExtendedColorType {
             | ExtendedColorType::Rgba8
             | ExtendedColorType::Rgba16
             | ExtendedColorType::Rgba32F
-            | ExtendedColorType::Bgra8 => 4,
+            | ExtendedColorType::Bgra8
+            | ExtendedColorType::Cmyk8 => 4,
         }
     }
 }
diff --git a/tests/images/tiff/testsuite/hpredict_cmyk.tiff b/tests/images/tiff/testsuite/hpredict_cmyk.tiff
new file mode 100644
index 0000000000000000000000000000000000000000..412072e9d35fbc5dbd324e719cebee41f1f00af8
GIT binary patch
literal 2668
zcmYk82T)UK7lywSl8{gm2#5hf0O?54P=bJ=hzSzFf<Fp2qy=Qbf)r5`Kx9Dziu4wm
zL{U&_0n5@9WJMuC6c-dBG${sjRZw^Tu<On+cfNU_bMAT1IrGijnM<b|0<1IuEI{)3
zidU9JCORJR<ojEy%O^Yg?X-4NKtND%Qa^9c2}NQejkF|<SJA*+X$vqj*id#rJ*S;Y
z0+=7~q*CD;i!Assz-6{MU_;c=4Ce|5>Ze>Kgw(bwA;6DHIaVVo`?(FS4X4xh^8`0`
zRmDApekcK2?b5@{w@?&|xT$rI$~r_#+b?3Dr|1>%;BC#zOFFX{Xe-O~Yr$Wp@~p0L
zF4x;>RKxxZ{1_J$oD+sNvWQ;*?3p(>;|s~PHmQ}pl@QAHDzZyG($sk@GF!doc_hC(
z!rUTQaZV{5*g@*1zUo;OJ#S4S5Vv*h0%^fh-}3^yf$ZDX`)cxPm<JOeifzufs=-Wu
zi1}{!cS-pOA`Yha{M;$&1u4C*0&?C3#F;W+!yK>Ym&n=@=8JDprul~H&c>f@lgTDo
z_O}OYZ*_dYJFbwXZmWohT%XTy3bGGdqET~78m5QBvie$PFX_$!<nOfE@|``R><#FO
zR9v@s{H=_^U4Pn}8z)Zov59Ui@H5uaxEVm&-YO>|nhI0KulY%zhdrKo_d}z8N=TdE
zvT?)dG`-f8vG{lzd4cU=4#n896ldwSMxWG{?MMUovV>x4Vw*WYR<DTI28&BDio)wS
z7|U(A7nXwSpd-t-Pq+x%!kpY=$Q&vO&27z>LGQJ=f$=6aWQ98qb}L1&|F;*4jfyu|
z97R%)h0U(J&12zb0mtD(5I}Zz1I(kd)V`8iIt-#u%?LfD{_>|7^%@KR=*Sw-A{#~G
zw#&wln&RP@1Jh=k<KFNr#0f=*t23h5gJgF$k!uT6-m$mzGo%^$N}4eyT~r6VBrBse
z-+Np%!XZzMco1p3bHRHS($!mzG~|bt4`@!o<dF6y1t!3P9r`d;T|Q*C`T+Z#aeZUy
z0<;-93NInnKgLyZ5BpeBGsV`UeIo#2pQiOYrxEzOk33B~X40@%px#UA=EIZCTob~0
z`V4+px0y?#8g)G5WITQGg+FOOF|WckcTa6Qg|taf$K76nXR9yT1bSUwsdEaivHY>a
zX?uY_q&JkI``exoTK_A7>aT|b6C#84nS-RZE7;J)Ul%*k6xEfu(}JLM4cbhvi(W@F
zN-HjIzdr8cI~LF>f3kFk2eytgbZyd()+4PlwikBU3%|Wob(=3@Q1ba`o|Pc}N3ePP
z`esC*afw=MXvXRN7n?F^%NJ`u3-XVC)Aj#)_PPK)0Kim;^V0|vT<%Zlv3DjLdWjZ&
zfKP`jl0W!2IMmS*s?Z7Y9M|hgyRkze%1f_fUkM8`C|P0YIBWotIeGO2avBmXUO$=X
z2X}i${hCyPI&a;fB+mj+KQ0u(Jx-h$T=Fy_p$!1CR)MhrukGo{XKi*5kQ>^k?Hs{>
zc=r?orOyUf9+34r=kxMtk{1Vu>a<`$vgd4kGgP3;W#LkoSlMLet;g%U&nl_#Mgf*G
z<Gd83;XXu3Q*ahO8gEj<i84$!QPb#tV>@a6<0eMm)3#_g0!_EnEDtxfqbD$<!o&a;
z47Qp93THX|)U5Tu{i|ByiREyP=*{6Ln||po714Fa3tN@<DQTnDLlqaP^N)CUEc~xm
zWTuk^F4}me{CdLPi(&W5pB$W}g6zm6-T++&CnPIQAwPDqWh1~m6^sG7QDDCp4Um#a
zmAzf1%0;I1=IarO-UZy3SJI2s&c=wdrhBmv@)J=pz_Duxt`sY6nM34@{4_95E1?M~
zR~P!?FS)#-Uf2?bB7s90q`TB2nk~uc7=Y$SP^*NoY(sd#^RYJ!!5A(g<xYoZ*_*_X
zW%9<!ysqpoI@_?;wPcgna2FIg-B1h)9*Ow*(~p@tI@G{~1NkyZ$ADCUab$QQfD7QS
z+a6J?8iv8<z(Ihdf}o5RD;7eKdt=*f!;0a4K0T>|3UhbdNR6lR2VNY^_>F9Q&LBU=
z%r|gxU*=@tgB*fex{@F&$R8h3d!+sW06mS+#V9BXSfa4bBsZQa)(AR02LSBz958k9
zI-;Ve2Nb^-+&-oCE^V`nV<_daIm%Jo<D-OWg@o<PbPAdt9;(n6m*psYq%$-fv@gNq
zITzxs^;CdrHJo`9mv~b(^f~}v*(R;~sXUnpE{VmqiJYN&wa+#0z!wkS!{=c0ezf?H
zSlMQ})MbSD--|k(D)Y_Cii!K<s+8~iGd!(mP*X}h1a-3NPA4~M(VZ|e7oeBn2oTHA
zrbfbpVXiYZwZJmWXko%ANtmKg`fB4uueqk04)xa-duBJo1m9o-nE64m-qFPeG0&sj
zM>cSGvtIKA=q;`hv)UbnRdOeLtdhkq4LZnarp<ZAzH)XZ(SIXEPK?W#e7^4c>E_ob
z;;I$yT;D<2l{9Fa9UDkVF>VfkrG0c-w;4jH-+h)y6fI=Y^cQW*?=NwLJrtUF($Xn;
zDNWd8;Q`bE>=GN=L;!jQ*8(5s%Ia_-uJyzw_sLvteyF?e3$JHp%4RpRKMmJqzu(l@
zEl6eh#Z*<}s_yeMI9h&a+jM=gr4xUN)8TIuX<uzy9x)`RIQUfWMf+Ca+e4ub&e!I!
zuN4L}?DPGs6Xz+{O!emx*R);EGYG!qZh0p6_}j`}-9tBx_6*5yhUU`r#bz7f@e5&&
zVTc)``nA4rRn6NWFFPiH@~efQPTiWzpC)uCq%^`CV719(IDLJ$#ZrY7&X98I7SV9L
zwvK;LagMxRK{&1vx?(4T?)l@p??cYg$`&`&tZ6y7m9uNbddKu1Chz-&EcC&|UkEoU
zT`FTWl_EmC1UUD}3T}RZd!ons>lQBzj4B0Hjz^;*)6;6_^YMS>xSkM_9vT>BK?J)m
zmU<GmQKJhSdiEeKnT9YfFDCYzJx$5bp}HSYov2#7OQURO@QP4;@cVsCUW`N4y`e8-
zsduH11SAs`JKRHMTtDtatBGzNzw*%0!R3z8`T$S%VLILsQbW}Q)wUnCa5rx=k=++6
z5B$iA`EolLqI+CR?y^Vk^55HMejK`a0gF`v00ILL002pHAZr*ZVJS&Y3oZFXe&$zW
z#2QBb<W)Zo0{^EEO2P_j@=9y!m47lxQnT*AdT9U%5UGE>3X)l>Kbat5wKaY8e==0!
zHIVSCN;a1K=lr|>YHTm@uBxxZyH&yn2^&k;ZB3tDYw9Jh0ZnpN_y12(;J@d>>`qCZ
p1v>uZNfbbPpSrCo<X>fhpD{GO{m<1r){4MgRjAQL$w<k4{vVM>4@Cd~

literal 0
HcmV?d00001