Disallow non-ASCII characters (outside of [0..127] range) in the interopID field.

Signed-off-by: cuneyt.ozdas <cuneyt.ozdas@autodesk.com>

Author: cozdas

src/OpenColorIO/ColorSpace.cpp

@@ -236,7 +236,7 @@ void ColorSpace::setInteropID(const char * interopID)
     if (!id.empty())
     {
         // Count the number of ':' characters in the string
-        // This is UTF-8 safe.
+        // and check for non-ASCII characters
         size_t colonCount = 0;
         size_t lastColonPos = std::string::npos;
 
@@ -247,6 +247,12 @@ void ColorSpace::setInteropID(const char * interopID)
                 colonCount++;
                 lastColonPos = i;
             }
+            else if (static_cast<unsigned char>(id[i]) >= 0x80)
+            {
+                std::ostringstream oss;
+                oss << "InteropID '" << id << "' is invalid: only ASCII characters [0x00..0x7F] are allowed.";
+                throw Exception(oss.str().c_str());
+            }
         }
 
         // Validate: only zero or one ':' character allowed
@@ -265,7 +271,7 @@ void ColorSpace::setInteropID(const char * interopID)
             throw Exception(oss.str().c_str());
         }
 
-        // TODO: If no namespace is given, verify the interopID against CIF list and warn if not found.
+        // TODO: Do we want to verify the interopID against the CIF list here?
     }
 
     getImpl()->m_interopID = id;

tests/cpu/ColorSpace_tests.cpp

@@ -1855,59 +1855,22 @@ OCIO_ADD_TEST(ColorSpace, interop_id)
 
     // Test invalid InteropID with multiple colons.
     const char * invalidMultipleColons = "name:space:cs_name";
-    OCIO_CHECK_THROW_WHAT(cs->setInteropID(invalidMultipleColons), 
-                          OCIO::Exception,
-                          "InteropID 'name:space:cs_name' is invalid: only zero or one ':' character is allowed.");
+    OCIO_CHECK_THROW_WHAT(cs->setInteropID(invalidMultipleColons), OCIO::Exception,
+                          "only zero or one ':' character is allowed");
 
     // Test invalid InteropID with colon at the end.
     const char * invalidColonAtEnd = "namespace:";
-    OCIO_CHECK_THROW_WHAT(cs->setInteropID(invalidColonAtEnd), 
-                          OCIO::Exception,
-                          "InteropID 'namespace:' is invalid: ':' character cannot be the last character.");
-
-    // Test invalid InteropID with three colons.
-    const char * invalidThreeColons = "a:b:c:d";
-    OCIO_CHECK_THROW_WHAT(cs->setInteropID(invalidThreeColons), 
-                          OCIO::Exception,
-                          "InteropID 'a:b:c:d' is invalid: only zero or one ':' character is allowed.");
-
-    // Test UTF-8 strings with valid single colon.
-    const char * utf8ValidColon = "標準:萬國碼";
-    OCIO_CHECK_NO_THROW(cs->setInteropID(utf8ValidColon));
-    OCIO_CHECK_EQUAL(std::string(cs->getInteropID()), utf8ValidColon);
-
-    // Test UTF-8 strings with invalid multiple colons.
-    const char * utf8InvalidMultipleColons = "標準:萬國:碼";
-    OCIO_CHECK_THROW_WHAT(cs->setInteropID(utf8InvalidMultipleColons), 
-                          OCIO::Exception,
-                          "only zero or one ':' character is allowed.");
-
-    // Test UTF-8 strings with invalid colon at end.
-    const char * utf8InvalidColonAtEnd = "標準萬國碼:";
-    OCIO_CHECK_THROW_WHAT(cs->setInteropID(utf8InvalidColonAtEnd), 
-                          OCIO::Exception,
-                          "':' character cannot be the last character.");
-
-    // Test UTF-8 strings without colon (should be valid).
-    const char * utf8NoColon = "標準萬國碼";
-    OCIO_CHECK_NO_THROW(cs->setInteropID(utf8NoColon));
-    OCIO_CHECK_EQUAL(std::string(cs->getInteropID()), utf8NoColon);
-
-    // Test edge case: single colon character.
-    const char * singleColon = ":";
-    OCIO_CHECK_THROW_WHAT(cs->setInteropID(singleColon), 
-                          OCIO::Exception,
-                          "':' character cannot be the last character.");
-
-    // Test edge case: colon at beginning (should be valid).
-    const char * colonAtBeginning = ":valid_name";
-    OCIO_CHECK_NO_THROW(cs->setInteropID(colonAtBeginning));
-    OCIO_CHECK_EQUAL(std::string(cs->getInteropID()), colonAtBeginning);
-
-    // Test valid InteropID with one colon (not at the end).
-    const char * validWithColon = "namespace:cs_name";
-    OCIO_CHECK_NO_THROW(cs->setInteropID(validWithColon));
-    OCIO_CHECK_EQUAL(std::string(cs->getInteropID()), validWithColon);
+    OCIO_CHECK_THROW_WHAT(cs->setInteropID(invalidColonAtEnd), OCIO::Exception,
+                          "':' character cannot be the last character");
+
+    // Test invalid InteropID with non-ASCII characters.
+    const char * invalidNonASCII1 = "café_scene";  // Contains é (0xC3 0xA9)
+    OCIO_CHECK_THROW_WHAT(cs->setInteropID(invalidNonASCII1), OCIO::Exception,
+                          "contains non-ASCII characters");
+
+    const char * invalidNonASCII2 = "space±_name";  // Contains ± (ANSI 0xB1)
+    OCIO_CHECK_THROW_WHAT(cs->setInteropID(invalidNonASCII2), OCIO::Exception,
+                          "contains non-ASCII characters");
 }
 
 OCIO_ADD_TEST(ColorSpace, interop_id_serialization)

tests/python/ColorSpaceTest.py

@@ -680,65 +680,24 @@ def test_interop_id(self):
         self.colorspace.setInteropID(valid_with_colon)
         self.assertEqual(self.colorspace.getInteropID(), valid_with_colon)
 
-        # Test valid InteropID with colon in the middle.
-        valid_colon_middle = 'namespace:colorspace_name'
-        self.colorspace.setInteropID(valid_colon_middle)
-        self.assertEqual(self.colorspace.getInteropID(), valid_colon_middle)
-
         # Test invalid InteropID with multiple colons.
         with self.assertRaises(Exception) as context:
             self.colorspace.setInteropID('name:space:cs_name')
         self.assertIn("only zero or one ':' character is allowed", str(context.exception))
 
         # Test invalid InteropID with colon at the end.
         with self.assertRaises(Exception) as context:
-            self.colorspace.setInteropID('lin_ap1_scene:')
+            self.colorspace.setInteropID('namespace:')
         self.assertIn("':' character cannot be the last character", str(context.exception))
 
-        # Test invalid InteropID with three colons.
-        with self.assertRaises(Exception) as context:
-            self.colorspace.setInteropID('a:b:c:d')
-        self.assertIn("only zero or one ':' character is allowed", str(context.exception))
-
-        # Test UTF-8 strings with valid single colon.
-        utf8_valid_colon = '標準:萬國碼'
-        self.colorspace.setInteropID(utf8_valid_colon)
-        self.assertEqual(self.colorspace.getInteropID(), utf8_valid_colon)
-
-        # Test UTF-8 strings with invalid multiple colons.
-        with self.assertRaises(Exception) as context:
-            self.colorspace.setInteropID('標準:萬國:碼')
-        self.assertIn("only zero or one ':' character is allowed", str(context.exception))
-
-        # Test UTF-8 strings with invalid colon at end.
+        # Test invalid InteropID with non-ASCII characters.
         with self.assertRaises(Exception) as context:
-            self.colorspace.setInteropID('標準萬國碼:')
-        self.assertIn("':' character cannot be the last character", str(context.exception))
+            self.colorspace.setInteropID('café_scene')  # Contains é (UTF-8)
+        self.assertIn("contains non-ASCII characters", str(context.exception))
 
-        # Test UTF-8 strings without colon (should be valid).
-        utf8_no_colon = '標準萬國碼'
-        self.colorspace.setInteropID(utf8_no_colon)
-        self.assertEqual(self.colorspace.getInteropID(), utf8_no_colon)
-
-        # Test edge case: single colon character.
         with self.assertRaises(Exception) as context:
-            self.colorspace.setInteropID(':')
-        self.assertIn("':' character cannot be the last character", str(context.exception))
-
-        # Test edge case: colon at beginning (should be valid).
-        colon_at_beginning = ':valid_name'
-        self.colorspace.setInteropID(colon_at_beginning)
-        self.assertEqual(self.colorspace.getInteropID(), colon_at_beginning)
-
-        # Test with Unicode characters that might look like colons but aren't ASCII colons.
-        unicode_similar = 'test：name'  # This uses a full-width colon (U+FF1A), not ASCII colon
-        self.colorspace.setInteropID(unicode_similar)
-        self.assertEqual(self.colorspace.getInteropID(), unicode_similar)
-
-        # Test mixed ASCII and Unicode with valid single ASCII colon.
-        mixed_valid = 'ñamespace:色空間'
-        self.colorspace.setInteropID(mixed_valid)
-        self.assertEqual(self.colorspace.getInteropID(), mixed_valid)
+            self.colorspace.setInteropID('space±_name')  # Contains ± (ANSI 0xB1)
+        self.assertIn("contains non-ASCII characters", str(context.exception))
 
     def test_amf_transform_ids(self):
         """
@@ -900,7 +859,7 @@ def test_processor_to_known_colorspace(self):
   - !<ColorSpace>
     name: Linear ITU-R BT.709
     description: A linear Rec.709 space with an unusual spelling.
-    interop_id: "mycompany:lin_rec709_scene"
+    interop_id: "mycompany:led_wall_1"
     isdata: false
     from_scene_reference: !<GroupTransform>
       name: AP0 to Linear Rec.709 (sRGB)