Make parsing of JPEG comments more robust, handle ImageDescription sm…

…arter

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

src/include/OpenImageIO/imageio.h

@@ -2902,6 +2902,13 @@ OIIO_API std::string geterror(bool clear = true);
 ///    When nonzero, use the new "OpenEXR core C library" when available,
 ///    for OpenEXR >= 3.1. This is experimental, and currently defaults to 0.
 ///
+/// - `int jpeg:com_attributes`
+///
+///    When nonzero, try to parse JPEG comment blocks as key-value attributes,
+///    and only set ImageDescription if the parsing fails. Otherwise, always
+///    set ImageDescription to the first comment block. Default is 1.
+///
+///
 /// - `int limits:channels` (1024)
 ///
 ///    When nonzero, the maximum number of color channels in an image. Image

src/include/imageio_pvt.h

@@ -42,6 +42,7 @@ extern OIIO_UTIL_API int oiio_print_debug;
 extern OIIO_UTIL_API int oiio_print_uncaught_errors;
 extern int oiio_log_times;
 extern int openexr_core;
+extern int jpeg_com_attributes;
 extern int limit_channels;
 extern int limit_imagesize_MB;
 extern int imagebuf_print_uncaught_errors;

src/jpeg.imageio/jpeginput.cpp

@@ -9,6 +9,7 @@
 #include <OpenImageIO/filesystem.h>
 #include <OpenImageIO/fmath.h>
 #include <OpenImageIO/imageio.h>
+#include <OpenImageIO/strutil.h>
 #include <OpenImageIO/tiffutils.h>
 
 #include "jpeg_pvt.h"
@@ -288,23 +289,41 @@ JpgInput::open(const std::string& name, ImageSpec& newspec)
             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", 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.
+            // comment markers in the form "key:value" or "ident:key:value".
+            // If the string contains a single colon, we assume key:value.
+            // If there's multiple, we try splitting as ident:key:value and
+            // check if ident and key are reasonable (in particular, whether
+            // ident is a C-style identifier and key is not surrounded by
+            // whitespace). If ident passes but key doesn't, assume key: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));
+            if (OIIO::get_int_attribute("jpeg:com_attributes")
+                && (separator != std::string::npos && separator > 0)) {
+                std::string left  = data.substr(0, separator);
+                std::string right = data.substr(separator + 1);
+                separator         = right.find(':');
+                if (separator != std::string::npos && separator > 0) {
+                    std::string mid   = right.substr(0, separator);
+                    std::string value = right.substr(separator + 1);
+                    if (Strutil::string_is_identifier(left)
+                        && (mid == Strutil::trimmed_whitespace(mid))) {
+                        // Valid parsing: left is ident, mid is key
+                        std::string attribute = left + ":" + mid;
+                        if (!m_spec.find_attribute(attribute, TypeDesc::STRING))
+                            m_spec.attribute(attribute, value);
+                        continue;
+                    }
+                }
+                if (left == Strutil::trimmed_whitespace(left)) {
+                    // Valid parsing: left is key, right is value
+                    if (!m_spec.find_attribute(left, TypeDesc::STRING))
+                        m_spec.attribute(left, right);
+                    continue;
+                }
             }
+            // If we made it this far, treat the comment as a description
+            if (!m_spec.find_attribute("ImageDescription", TypeDesc::STRING))
+                m_spec.attribute("ImageDescription", data);
         }
     }
 

src/libOpenImageIO/imageio.cpp

@@ -48,6 +48,7 @@ atomic_int oiio_try_all_readers(1);
 #endif
 // Should we use "Exr core C library"?
 int openexr_core(OIIO_OPENEXR_CORE_DEFAULT);
+int jpeg_com_attributes(1);
 int tiff_half(0);
 int tiff_multithread(1);
 int dds_bc5normal(0);
@@ -366,6 +367,10 @@ attribute(string_view name, TypeDesc type, const void* val)
         openexr_core = *(const int*)val;
         return true;
     }
+    if (name == "jpeg:com_attributes" && type == TypeInt) {
+        jpeg_com_attributes = *(const int*)val;
+        return true;
+    }
     if (name == "tiff:half" && type == TypeInt) {
         tiff_half = *(const int*)val;
         return true;
@@ -520,6 +525,10 @@ getattribute(string_view name, TypeDesc type, void* val)
         *(int*)val = openexr_core;
         return true;
     }
+    if (name == "jpeg:com_attributes" && type == TypeInt) {
+        *(int*)val = jpeg_com_attributes;
+        return true;
+    }
     if (name == "tiff:half" && type == TypeInt) {
         *(int*)val = tiff_half;
         return true;

testsuite/jpeg-metadata/ref/out.txt

@@ -2,7 +2,6 @@ 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>"
@@ -18,19 +17,15 @@ 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>"
@@ -46,7 +41,6 @@ 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>"
@@ -57,6 +51,64 @@ with-attribs.jpg     :  640 x  480, 3 channel, uint8 jpeg
     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
+    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-and-desc.jpg"
+PASS
+Reading with-attribs-and-desc.jpg
+with-attribs-and-desc.jpg :  640 x  480, 3 channel, uint8 jpeg
+    SHA-1: 329B449C07E6649023504E2C8E5130B41985CF7F
+    channel list: R, G, B
+    ImageDescription: "A photo"
+    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: "A photo"
+    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
+    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-colon-desc.jpg"
+PASS
+Reading with-colon-desc.jpg
+with-colon-desc.jpg  :  640 x  480, 3 channel, uint8 jpeg
+    SHA-1: 329B449C07E6649023504E2C8E5130B41985CF7F
+    channel list: R, G, B
+    ImageDescription: "Example:Text"
+    Exif:ColorSpace: 1
+    Exif:ExifVersion: "0230"
+    Exif:FlashPixVersion: "0100"
+    IPTC:Caption: "Example:Text"
     jpeg:subsampling: "4:2:0"
     oiio:ColorSpace: "sRGB"

testsuite/jpeg-metadata/run.py

@@ -21,3 +21,18 @@
                        output_filename="with-attribs.jpg",
                        extraargs="--attrib:type=int jpeg:com_attributes 1")
 command += info_command ("with-attribs.jpg", safematch=True)
+
+# Check that JPEG comments that don't match an attribute will be read as ImageDescription.
+command += rw_command ("src", "blender-render.jpg", use_oiiotool=1,
+                       output_filename="with-attribs-and-desc.jpg",
+                       extraargs="--attrib:type=int jpeg:com_attributes 1 "
+                       "--attrib:type=string ImageDescription 'A photo'")
+command += info_command ("with-attribs-and-desc.jpg", safematch=True)
+
+# Check that JPEG comments that would match an attribute will be read as ImageDescription
+# if jpeg:com_attributes is 0.
+command += rw_command ("src", "blender-render.jpg", use_oiiotool=1,
+                       output_filename="with-colon-desc.jpg",
+                       extraargs="--attrib:type=string ImageDescription 'Example:Text'")
+command += info_command ("with-colon-desc.jpg", safematch=True,
+                         extraargs="--oiioattrib:type=int jpeg:com_attributes 0")