Fix: use natural string sorting for attribute names

1 files changed, 16 insertions, 5 deletions

diff --git a/source/blender/editors/space_node/node_draw.cc b/source/blender/editors/space_node/node_draw.cc
index 47d35ced371..90ab798bafc 100644
--- a/source/blender/editors/space_node/node_draw.cc
+++ b/source/blender/editors/space_node/node_draw.cc
@@ -1703,14 +1703,25 @@ static char *named_attribute_tooltip(bContext *UNUSED(C), void *argN, const char
 
   std::stringstream ss;
   ss << TIP_("Accessed attribute names:\n");
-  Vector<std::pair<StringRefNull, NamedAttributeUsage>> sorted_used_attribute;
+
+  struct NameWithUsage {
+    StringRefNull name;
+    NamedAttributeUsage usage;
+  };
+
+  Vector<NameWithUsage> sorted_used_attribute;
   for (auto &&item : arg.usage_by_attribute.items()) {
     sorted_used_attribute.append({item.key, item.value});
   }
-  std::sort(sorted_used_attribute.begin(), sorted_used_attribute.end());
-  for (const std::pair<StringRefNull, NamedAttributeUsage> &attribute : sorted_used_attribute) {
-    const StringRefNull name = attribute.first;
-    const NamedAttributeUsage usage = attribute.second;
+  std::sort(sorted_used_attribute.begin(),
+            sorted_used_attribute.end(),
+            [](const NameWithUsage &a, const NameWithUsage &b) {
+              return BLI_strcasecmp_natural(a.name.c_str(), b.name.c_str()) <= 0;
+            });
+
+  for (const NameWithUsage &attribute : sorted_used_attribute) {
+    const StringRefNull name = attribute.name;
+    const NamedAttributeUsage usage = attribute.usage;
     ss << "  \u2022 \"" << name << "\": ";
     Vector<std::string> usages;
     if ((usage & NamedAttributeUsage::Read) != NamedAttributeUsage::None) {