Add support for reading the ICC profile

Co-authored-by: Rasmus Piorr <piorr@google.com>

Author: 1c3t3a

src/reader/decoder.rs

@@ -19,8 +19,9 @@ use weezl::{decode::Decoder as LzwDecoder, BitOrder, LzwError, LzwStatus};
 /// GIF palettes are RGB
 pub const PLTE_CHANNELS: usize = 3;
 /// Headers for supported extensions.
-const EXT_NAME_NETSCAPE: &[u8] = b"\x0bNETSCAPE2.0\x03";
+const EXT_NAME_NETSCAPE: &[u8] = b"\x0bNETSCAPE2.0\x01";
 const EXT_NAME_XMP: &[u8] = b"\x0bXMP DataXMP";
+const EXT_NAME_ICC: &[u8] = b"\x0bICCRGBG1012";
 
 /// An error returned in the case of the image not being formatted properly.
 #[derive(Debug)]
@@ -382,6 +383,8 @@ pub struct StreamingDecoder {
     header_end_reached: bool,
     /// XMP metadata bytes.
     xmp_metadata: Option<Vec<u8>>,
+    /// ICC profile bytes.
+    icc_profile: Option<Vec<u8>>,
 }
 
 /// One version number of the GIF standard.
@@ -396,6 +399,7 @@ pub enum Version {
 struct ExtensionData {
     id: AnyExtension,
     data: Vec<u8>,
+    sub_block_lens: Vec<u8>,
     is_block_end: bool,
 }
 
@@ -465,11 +469,13 @@ impl StreamingDecoder {
             ext: ExtensionData {
                 id: AnyExtension(0),
                 data: Vec::with_capacity(256), // 0xFF + 1 byte length
+                sub_block_lens: Vec::new(),
                 is_block_end: true,
             },
             current: None,
             header_end_reached: false,
             xmp_metadata: None,
+            icc_profile: None,
         }
     }
 
@@ -542,6 +548,12 @@ impl StreamingDecoder {
         self.xmp_metadata.as_deref()
     }
 
+    /// ICC profile stored in the image.
+    #[must_use]
+    pub fn icc_profile(&self) -> Option<&[u8]> {
+        self.icc_profile.as_deref()
+    }
+
     /// The version number of the GIF standard used in this image.
     ///
     /// We suppose a minimum of `V87a` compatibility. This value will be reported until we have
@@ -705,6 +717,7 @@ impl StreamingDecoder {
                     }
                     Some(Block::Extension) => {
                         self.ext.data.clear();
+                        self.ext.sub_block_lens.clear();
                         self.ext.id = AnyExtension(b);
                         if self.ext.id.into_known().is_none() {
                             if !self.allow_unknown_blocks {
@@ -766,33 +779,15 @@ impl StreamingDecoder {
                         }
                     }
                 } else {
-                    self.ext.data.push(b);
+                    self.ext.sub_block_lens.push(b);
                     self.ext.is_block_end = false;
                     goto!(ExtensionDataBlock(b as usize), emit Decoded::SubBlockFinished(self.ext.id))
                 }
             }
             ApplicationExtension => {
                 debug_assert_eq!(0, b);
-
-                // the parser removes sub-block lenghts, so app name and data are concatenated
-                if let Some(&[first, second, ..]) = self.ext.data.strip_prefix(EXT_NAME_NETSCAPE) {
-                    let repeat = u16::from(first) | u16::from(second) << 8;
-                    goto!(BlockEnd, emit Decoded::Repetitions(if repeat == 0 { Repeat::Infinite } else { Repeat::Finite(repeat) }))
-                } else if let Some(mut rest) = self.ext.data.strip_prefix(EXT_NAME_XMP) {
-                    // XMP adds a "ramp" of 257 bytes to the end of the metadata to let the "pascal-strings"
-                    // parser converge to the null byte. The ramp looks like "0x01, 0xff, .., 0x01, 0x00".
-                    // For convenience and to allow consumers to not be bothered with this implementation detail,
-                    // we cut the ramp.
-                    const RAMP_SIZE: usize = 257;
-                    if rest.len() >= RAMP_SIZE
-                        && rest.ends_with(&[0x03, 0x02, 0x01, 0x00])
-                        && rest[rest.len() - RAMP_SIZE..].starts_with(&[0x01, 0x0ff])
-                    {
-                        rest = &rest[..(rest.len() - RAMP_SIZE)];
-                    }
-
-                    self.xmp_metadata = Some(rest.to_vec());
-                    goto!(BlockEnd)
+                if let Some(decoded) = self.read_application_extension() {
+                    goto!(BlockEnd, emit decoded)
                 } else {
                     goto!(BlockEnd)
                 }
@@ -901,6 +896,44 @@ impl StreamingDecoder {
         Ok(())
     }
 
+    fn read_application_extension(&mut self) -> Option<Decoded> {
+        if let Some(&[first, second]) = self.ext.data.strip_prefix(EXT_NAME_NETSCAPE) {
+            let repeat = u16::from(first) | u16::from(second) << 8;
+            return Some(Decoded::Repetitions(if repeat == 0 {
+                Repeat::Infinite
+            } else {
+                Repeat::Finite(repeat)
+            }));
+        } else if let Some(mut rest) = self.ext.data.strip_prefix(EXT_NAME_XMP) {
+            // XMP is not written as a valid "pascal-string", so we need to stitch together
+            // the text from our collected sublock-lengths.
+            let mut xmp_metadata = Vec::new();
+            for len in self.ext.sub_block_lens.iter() {
+                xmp_metadata.push(*len);
+                let (sub_block, tail) = rest.split_at(*len as usize);
+                xmp_metadata.extend_from_slice(sub_block);
+                rest = tail;
+            }
+
+            // XMP adds a "ramp" of 257 bytes to the end of the metadata to let the "pascal-strings"
+            // parser converge to the null byte. The ramp looks like "0x01, 0xff, .., 0x01, 0x00".
+            // For convenience and to allow consumers to not be bothered with this implementation detail,
+            // we cut the ramp.
+            const RAMP_SIZE: usize = 257;
+            if xmp_metadata.len() >= RAMP_SIZE
+                && xmp_metadata.ends_with(&[0x03, 0x02, 0x01, 0x00])
+                && xmp_metadata[xmp_metadata.len() - RAMP_SIZE..].starts_with(&[0x01, 0x0ff])
+            {
+                xmp_metadata.truncate(xmp_metadata.len() - RAMP_SIZE);
+            }
+
+            self.xmp_metadata = Some(xmp_metadata);
+        } else if let Some(rest) = self.ext.data.strip_prefix(EXT_NAME_ICC) {
+            self.icc_profile = Some(rest.to_vec());
+        }
+        None
+    }
+
     fn add_frame(&mut self) {
         if self.current.is_none() {
             self.current = Some(Frame::default());

src/reader/mod.rs

@@ -479,6 +479,12 @@ where
         self.decoder.decoder.xmp_metadata()
     }
 
+    /// ICC profile stored in the image.
+    #[inline]
+    pub fn icc_profile(&self) -> Option<&[u8]> {
+        self.decoder.decoder.icc_profile()
+    }
+
     /// Abort decoding and recover the `io::Read` instance
     pub fn into_inner(self) -> io::BufReader<R> {
         self.decoder.into_inner()

tests/decode.rs

@@ -273,6 +273,15 @@ fn xmp_metadata() {
     assert_eq!(decoder.xmp_metadata(), Some(EXPECTED_METADATA.as_bytes()))
 }
 
+#[test]
+fn icc_profile() {
+    let image: &[u8] = include_bytes!("samples/beacon_icc.gif");
+    let icc_profile: &[u8] = include_bytes!("samples/profile.icc");
+    let decoder = DecodeOptions::new().read_info(image).unwrap();
+
+    assert_eq!(decoder.icc_profile(), Some(icc_profile))
+}
+
 #[test]
 fn check_last_extension_returns() {
     let mut buf: &[u8] = include_bytes!("samples/beacon_xmp.gif");
@@ -296,5 +305,5 @@ fn check_last_extension_returns() {
         }
     }
 
-    assert_eq!(decoder.last_ext().1.len(), 3129);
+    assert_eq!(decoder.last_ext().1.len(), 3048);
 }

tests/results.txt

@@ -1,5 +1,6 @@
 tests/samples/alpha_gif_a.gif: 3871893825
 tests/samples/anim-gr.gif: 291646878
+tests/samples/beacon_icc.gif: 2462153529
 tests/samples/beacon_xmp.gif: 2462153529
 tests/samples/beacon.gif: 2462153529
 tests/samples/interlaced.gif: 2115495372