feat(jpeg): Support encoding/decoding arbitrary metadata as comments

This is needed to port Blender's current JPEG IO code to using OIIO,
but is also a useful feature to have in general.

For reading, the code tries to parse comments as colon-separated key-value
pairs and sets metadata accordingly. For writing, this needs to be explicitly
enabled by setting jpeg:com_attributes to 1 in order to avoid accidentally
bloating files for existing applications.

Signed-off-by: Lukas Stockner <[email protected]>

src/cmake/testing.cmake

@@ -144,7 +144,7 @@ macro (oiio_add_all_tests)
                     oiiotool-demosaic
                     diff
                     dither dup-channels
-                    jpeg-corrupt
+                    jpeg-corrupt jpeg-metadata
                     maketx oiiotool-maketx
                     misnamed-file
                     missingcolor

src/doc/builtinplugins.rst

@@ -1037,6 +1037,10 @@ anywhere near the acceptance of the original JPEG/JFIF format.
        reader/writer, and you should assume that nearly everything described
        Appendix :ref:`chap-stdmetadata` is properly translated when using
        JPEG files.
+   * - *other*
+     -
+     - Extra attributes will be read from comment blocks in the JPEG file,
+       and can optionally be written if ``jpeg:com_attributes`` is enabled.
 
 **Configuration settings for JPEG input**
 
@@ -1084,6 +1088,10 @@ control aspects of the writing itself:
    * - ``jpeg:progressive``
      - int
      - If nonzero, will write a progressive JPEG file.
+   * - ``jpeg:com_attributes``
+     - int
+     - If nonzero, extra attributes will be written into the file as comment
+       blocks.
 
 
 **Custom I/O Overrides**

src/jpeg.imageio/jpeginput.cpp

@@ -284,10 +284,24 @@ JpgInput::open(const std::string& name, ImageSpec& newspec)
                    && !strcmp((const char*)m->data, "Photoshop 3.0"))
             jpeg_decode_iptc((unsigned char*)m->data);
         else if (m->marker == JPEG_COM) {
+            std::string data((const char*)m->data, m->data_length);
             if (!m_spec.find_attribute("ImageDescription", TypeDesc::STRING))
-                m_spec.attribute("ImageDescription",
-                                 std::string((const char*)m->data,
-                                             m->data_length));
+                m_spec.attribute("ImageDescription", data);
+            // Additional string metadata can be stored in JPEG files as
+            // comment markers in the form "key:value".
+            // Since the key might also commonly contain a colon, the
+            // heuristic to separate them here is that if there are multiple
+            // colons, the first one belongs to the key, the second one is the
+            // separator, and any further ones belong to the value.
+            auto separator = data.find(':');
+            if (separator != std::string::npos && separator > 0) {
+                auto second_separator = data.find(':', separator + 1);
+                if (second_separator != std::string::npos)
+                    separator = second_separator;
+                std::string key = data.substr(0, separator);
+                if (!m_spec.find_attribute(key, TypeDesc::STRING))
+                    m_spec.attribute(key, data.substr(separator + 1));
+            }
         }
     }
 

src/jpeg.imageio/jpegoutput.cpp

@@ -4,6 +4,7 @@
 
 #include <cassert>
 #include <cstdio>
+#include <set>
 #include <vector>
 
 #include <OpenImageIO/filesystem.h>
@@ -117,6 +118,15 @@ OIIO_PLUGIN_EXPORTS_END
 
 
 
+static std::set<std::string> metadata_include { "oiio:ConstantColor",
+                                                "oiio:AverageColor",
+                                                "oiio:SHA-1" };
+static std::set<std::string> metadata_exclude {
+    "XResolution",    "YResolution", "PixelAspectRatio",
+    "ResolutionUnit", "Orientation", "ImageDescription",
+    "Compression"
+};
+
 bool
 JpgOutput::open(const std::string& name, const ImageSpec& newspec,
                 OpenMode mode)
@@ -229,6 +239,36 @@ JpgOutput::open(const std::string& name, const ImageSpec& newspec,
                           comment.size() + 1);
     }
 
+    // Write other metadata as JPEG comments if requested
+    if (m_spec.get_int_attribute("jpeg:com_attributes")) {
+        for (const auto& p : m_spec.extra_attribs) {
+            std::string name = p.name().string();
+            auto colon       = name.find(':');
+            if (metadata_include.count(name)) {
+                // Allow explicitly included metadata
+            } else if (metadata_exclude.count(name))
+                continue;  // Suppress metadata that is processed separately
+            else if (Strutil::istarts_with(name, "ICCProfile"))
+                continue;  // Suppress ICC profile, gets written separately
+            else if (colon != ustring::npos) {
+                auto prefix = p.name().substr(0, colon);
+                if (Strutil::iequals(prefix, "oiio"))
+                    continue;  // Suppress internal metadata
+                else if (Strutil::iequals(prefix, "exif")
+                         || Strutil::iequals(prefix, "GPS")
+                         || Strutil::iequals(prefix, "XMP"))
+                    continue;  // Suppress EXIF metadata, gets written separately
+                else if (Strutil::iequals(prefix, "iptc"))
+                    continue;  // Suppress IPTC metadata
+                else if (is_imageio_format_name(prefix))
+                    continue;  // Suppress format-specific metadata
+            }
+            auto data = p.name().string() + ":" + p.get_string();
+            jpeg_write_marker(&m_cinfo, JPEG_COM, (JOCTET*)data.c_str(),
+                              data.size());
+        }
+    }
+
     if (Strutil::iequals(m_spec.get_string_attribute("oiio:ColorSpace"), "sRGB"))
         m_spec.attribute("Exif:ColorSpace", 1);
 

testsuite/jpeg-metadata/ref/out.txt

@@ -0,0 +1,62 @@
+Reading src/blender-render.jpg
+src/blender-render.jpg :  640 x  480, 3 channel, uint8 jpeg
+    SHA-1: A60D05FC42FDEE2FC8907531E3641C17D7C1E3AB
+    channel list: R, G, B
+    ImageDescription: "Blender:File:<untitled>"
+    Blender:Camera: "Camera"
+    Blender:Date: "2024/09/17 15:50:17"
+    Blender:File: "<untitled>"
+    Blender:Frame: "001"
+    Blender:RenderTime: "00:03.49"
+    Blender:Scene: "Scene"
+    Blender:Time: "00:00:00:01"
+    jpeg:subsampling: "4:2:0"
+    oiio:ColorSpace: "sRGB"
+Comparing "src/blender-render.jpg" and "no-attribs.jpg"
+PASS
+Reading no-attribs.jpg
+no-attribs.jpg       :  640 x  480, 3 channel, uint8 jpeg
+    SHA-1: 329B449C07E6649023504E2C8E5130B41985CF7F
+    channel list: R, G, B
+    ImageDescription: "Blender:File:<untitled>"
+    Blender:File: "<untitled>"
+    Exif:ColorSpace: 1
+    Exif:ExifVersion: "0230"
+    Exif:FlashPixVersion: "0100"
+    IPTC:Caption: "Blender:File:<untitled>"
+    jpeg:subsampling: "4:2:0"
+    oiio:ColorSpace: "sRGB"
+Reading src/blender-render.jpg
+src/blender-render.jpg :  640 x  480, 3 channel, uint8 jpeg
+    SHA-1: A60D05FC42FDEE2FC8907531E3641C17D7C1E3AB
+    channel list: R, G, B
+    ImageDescription: "Blender:File:<untitled>"
+    Blender:Camera: "Camera"
+    Blender:Date: "2024/09/17 15:50:17"
+    Blender:File: "<untitled>"
+    Blender:Frame: "001"
+    Blender:RenderTime: "00:03.49"
+    Blender:Scene: "Scene"
+    Blender:Time: "00:00:00:01"
+    jpeg:subsampling: "4:2:0"
+    oiio:ColorSpace: "sRGB"
+Comparing "src/blender-render.jpg" and "with-attribs.jpg"
+PASS
+Reading with-attribs.jpg
+with-attribs.jpg     :  640 x  480, 3 channel, uint8 jpeg
+    SHA-1: 329B449C07E6649023504E2C8E5130B41985CF7F
+    channel list: R, G, B
+    ImageDescription: "Blender:File:<untitled>"
+    Blender:Camera: "Camera"
+    Blender:Date: "2024/09/17 15:50:17"
+    Blender:File: "<untitled>"
+    Blender:Frame: "001"
+    Blender:RenderTime: "00:03.49"
+    Blender:Scene: "Scene"
+    Blender:Time: "00:00:00:01"
+    Exif:ColorSpace: 1
+    Exif:ExifVersion: "0230"
+    Exif:FlashPixVersion: "0100"
+    IPTC:Caption: "Blender:File:<untitled>"
+    jpeg:subsampling: "4:2:0"
+    oiio:ColorSpace: "sRGB"

testsuite/jpeg-metadata/run.py

@@ -0,0 +1,23 @@
+#!/usr/bin/env python
+
+# Copyright Contributors to the OpenImageIO project.
+# SPDX-License-Identifier: Apache-2.0
+# https://github.com/AcademySoftwareFoundation/OpenImageIO
+
+
+redirect = ' >> out.txt 2>&1 '
+
+# This file was rendered and saved in Blender, and therefore contains metadata
+# in the form of comments.
+
+# Check if the comments are correctly decoded as attributes, and that writing
+# to a new JPEG does not include them by default.
+command += rw_command ("src", "blender-render.jpg", use_oiiotool=1,
+                       output_filename="no-attribs.jpg")
+command += info_command ("no-attribs.jpg", safematch=True)
+
+# Check that, when jpeg:com_attributes is set, the attributes are preserved.
+command += rw_command ("src", "blender-render.jpg", use_oiiotool=1,
+                       output_filename="with-attribs.jpg",
+                       extraargs="--attrib:type=int jpeg:com_attributes 1")
+command += info_command ("with-attribs.jpg", safematch=True)