disambiguate non-minecraft identifier namespaces in codegen

6 files changed, 37 insertions, 25 deletions

diff --git a/codegen/lib/code/data_components.py b/codegen/lib/code/data_components.py
index abbe2eb6..e6de6201 100644
--- a/codegen/lib/code/data_components.py
+++ b/codegen/lib/code/data_components.py
@@ -1,7 +1,7 @@
 from typing import Any, Optional
 import lib.code.utils
 import lib.extract
-import lib.utils
+from lib.utils import identifier_to_path, to_camel_case, to_snake_case
 
 
 DATA_COMPONENTS_DIR = "azalea-inventory/src/components/mod.rs"
@@ -54,7 +54,7 @@ def get_expected_variants(version_id: str):
         registry["entries"].items(), key=lambda x: x[1]["protocol_id"]
     )
     for variant_name, _variant in registry_entries:
-        variant_struct_name = lib.utils.to_camel_case(variant_name.split(":")[-1])
+        variant_struct_name = to_camel_case(identifier_to_path(variant_name))
         expected_variants.append(variant_struct_name)
 
     return expected_variants
@@ -198,10 +198,10 @@ use crate::{
     components_to_item_defaults = {}
 
     for item_resource_id, data in items.items():
-        item_resource_id = item_resource_id.split(":")[1]
+        item_resource_id = identifier_to_path(item_resource_id)
         components = data["components"]
         for component_resource_id, component_value in components.items():
-            component_resource_id = component_resource_id.split(":")[1]
+            component_resource_id = identifier_to_path(component_resource_id)
             if component_resource_id not in components_to_item_defaults:
                 components_to_item_defaults[component_resource_id] = {}
             components_to_item_defaults[component_resource_id][item_resource_id] = (
@@ -212,7 +212,7 @@ use crate::{
     item_resource_id_to_protocol_id = {}
     item_resource_ids = [None] * len(registries["minecraft:item"]["entries"])
     for item_resource_id, item_data in registries["minecraft:item"]["entries"].items():
-        item_resource_id = item_resource_id.split(":")[-1]
+        item_resource_id = identifier_to_path(item_resource_id)
         item_protocol_id = item_data["protocol_id"]
         item_resource_id_to_protocol_id[item_resource_id] = item_protocol_id
         item_resource_ids[item_protocol_id] = item_resource_id
@@ -249,7 +249,7 @@ use crate::{
         # manual implementations
         if isinstance(python_value, dict) and len(python_value) > 0:
             if target_rust_type == "ConsumeEffect":
-                variant = lib.utils.to_camel_case(python_value["type"].split(":")[-1])
+                variant = to_camel_case(python_value["type"].split(":")[-1])
                 type_with_variant = f"ConsumeEffect::{variant}"
                 details_without_type = python_value.copy()
                 del details_without_type["type"]
@@ -329,7 +329,7 @@ use crate::{
 
             if entity_id and entity_id.startswith("minecraft:"):
                 entity_name = entity_id[10:]  # Remove "minecraft:" prefix
-                entity_name_camel = lib.utils.to_camel_case(entity_name)
+                entity_name_camel = to_camel_case(entity_name)
                 return f"EntityKind::{entity_name_camel}"
             raise ValueError(f"Unknown or missing EntityKind: {python_value}")
         elif target_rust_type == "NbtCompound":
@@ -435,7 +435,7 @@ use crate::{
                 return "DamageType::Registry(azalea_registry::data::DamageKind::new_raw(0))"
             else:
                 # enum variant
-                return f"{target_rust_type}::{lib.utils.to_camel_case(python_value.split(':')[-1])}"
+                return f"{target_rust_type}::{to_camel_case(identifier_to_path(python_value))}"
         if isinstance(python_value, list):
             # convert Vec<Thing> into Thing
             main_vec = "vec!["
@@ -463,7 +463,7 @@ use crate::{
             for v in python_value:
                 # handle tags correctly
                 if isinstance(v, str) and v.startswith("#minecraft:"):
-                    tag_name = lib.utils.to_snake_case(v.split(":")[-1]).upper()
+                    tag_name = to_snake_case(identifier_to_path(v)).upper()
                     if inner_type == "EntityKind":
                         tag_module = "entities"
                     elif inner_type == "ItemKind":
@@ -494,7 +494,7 @@ use crate::{
         return str(python_value)
 
     for component_resource_id, item_defaults in components_to_item_defaults.items():
-        component_struct_name = lib.utils.to_camel_case(component_resource_id)
+        component_struct_name = to_camel_case(component_resource_id)
         component_struct_fields = enum_and_struct_fields[component_struct_name]
 
         if len(component_struct_fields) == 1 and isinstance(
@@ -608,7 +608,7 @@ use crate::{
                 for item_resource_id, value in item_defaults.items():
                     if value == most_common_default_value:
                         continue
-                    item_variant_name = lib.utils.to_camel_case(item_resource_id)
+                    item_variant_name = to_camel_case(item_resource_id)
                     code.append(
                         f"            ItemKind::{item_variant_name} => {value},"
                     )
@@ -629,7 +629,7 @@ use crate::{
             code.append("    fn default_for_item(item: ItemKind) -> Option<Self> {")
             code.append("        let value = match item {")
             for item_resource_id, value in item_defaults.items():
-                item_variant_name = lib.utils.to_camel_case(item_resource_id)
+                item_variant_name = to_camel_case(item_resource_id)
                 code.append(f"            ItemKind::{item_variant_name} => {value},")
             code.append("            _ => return None,")
             code.append("        };")
diff --git a/codegen/lib/code/inventory.py b/codegen/lib/code/inventory.py
index 066319e0..7acb269d 100644
--- a/codegen/lib/code/inventory.py
+++ b/codegen/lib/code/inventory.py
@@ -1,5 +1,5 @@
 from typing import Any
-from lib.utils import to_camel_case, get_dir_location
+from lib.utils import identifier_to_path, to_camel_case, get_dir_location
 
 # The directory where declare_menus! {} is done
 inventory_menus_dir = get_dir_location("../azalea-inventory/src/lib.rs")
@@ -128,4 +128,4 @@ def update_menus(initial_menu_entries: dict[str, Any]):
 
 
 def menu_name_to_enum_name(menu_name: str) -> str:
-    return to_camel_case(menu_name.split(":")[-1])
+    return to_camel_case(identifier_to_path(menu_name))
diff --git a/codegen/lib/code/packet.py b/codegen/lib/code/packet.py
index 50d01ce2..9ce5c137 100644
--- a/codegen/lib/code/packet.py
+++ b/codegen/lib/code/packet.py
@@ -1,4 +1,4 @@
-from lib.utils import to_snake_case, to_camel_case, get_dir_location
+from lib.utils import identifier_to_path, to_snake_case, to_camel_case, get_dir_location
 from lib.code.utils import burger_type_to_rust_type, write_packet_file
 from lib.mappings import Mappings
 from typing import Optional
@@ -120,7 +120,7 @@ def packet_direction_report_to_packet_names(report):
     name_to_id = {}
     for identifier, packet in report.items():
         packet_id = packet["protocol_id"]
-        name_to_id[identifier.split(":")[-1]] = packet_id
+        name_to_id[identifier_to_path(identifier)] = packet_id
 
     names_sorted = [name for name in sorted(name_to_id, key=lambda x: name_to_id[x])]
     return names_sorted
diff --git a/codegen/lib/code/registry.py b/codegen/lib/code/registry.py
index 8957a8fe..01462e1f 100644
--- a/codegen/lib/code/registry.py
+++ b/codegen/lib/code/registry.py
@@ -1,4 +1,4 @@
-from lib.utils import get_dir_location, to_camel_case
+from lib.utils import get_dir_location, identifier_to_path, to_camel_case
 
 BUILTIN_REGISTRIES_DIR = get_dir_location("../azalea-registry/src/builtin.rs")
 DATA_REGISTRIES_DIR = get_dir_location("../azalea-registry/src/data.rs")
@@ -16,7 +16,7 @@ def generate_builtin_registries(registries: dict):
         #     Stone => "minecraft:stone"
         # });
 
-        registry_name = registry_name.split(":")[1]
+        registry_name = identifier_to_path(registry_name)
         registry_enum_name = registry_name_to_enum_name(registry_name)
 
         existing_registry_enum_names.add(registry_enum_name)
@@ -27,8 +27,7 @@ def generate_builtin_registries(registries: dict):
             registry["entries"].items(), key=lambda x: x[1]["protocol_id"]
         )
         for variant_name, _variant in registry_entries:
-            # strip out the "minecraft:" prefix
-            variant_name = variant_name.split(":")[-1]
+            variant_name = identifier_to_path(variant_name)
             variant_struct_name = to_camel_case(variant_name)
             registry_code.append(f'\t{variant_struct_name} => "{variant_name}",')
         registry_code.append("}")
@@ -117,7 +116,7 @@ def generate_data_registries(data_registries: dict):
         registry_code.append(f"enum {registry_enum_name}Key {{")
         registry_entries.sort()
         for variant_name in registry_entries:
-            variant_struct_name = to_camel_case(variant_name.split(":")[-1])
+            variant_struct_name = to_camel_case(identifier_to_path(variant_name))
             registry_code.append(f'    {variant_struct_name} => "{variant_name}",')
         registry_code.append("}")
 
@@ -146,7 +145,7 @@ def generate_data_registries(data_registries: dict):
 
 
 def registry_name_to_enum_name(registry_name: str) -> str:
-    registry_name = registry_name.split(":")[-1]
+    registry_name = identifier_to_path(registry_name)
 
     if registry_name == "block_type":
         # avoid conflicting with BlockKind
diff --git a/codegen/lib/code/tags.py b/codegen/lib/code/tags.py
index 651b6892..b6e225ac 100644
--- a/codegen/lib/code/tags.py
+++ b/codegen/lib/code/tags.py
@@ -1,5 +1,5 @@
 from lib.code.registry import registry_name_to_enum_name
-from lib.utils import to_snake_case, upper_first_letter, get_dir_location, to_camel_case
+from lib.utils import identifier_to_namespace, identifier_to_path, to_snake_case, upper_first_letter, get_dir_location, to_camel_case
 
 TAGS_DIR = get_dir_location("../azalea-registry/src/tags")
 
@@ -19,14 +19,15 @@ use crate::{{builtin::{struct_name}, tags::RegistryTag}};
 
     protocol_ids = {}
     for k, v in registries["minecraft:" + registry_name]["entries"].items():
-        protocol_ids[k.split(":")[1]] = v["protocol_id"]
+        protocol_ids[identifier_to_path(k)] = v["protocol_id"]
 
     for tag_name, tag in sorted(tags.items(), key=lambda x: x[0]):
         entries = []
         queue = tag["values"].copy()
         while queue != []:
             ident = queue.pop(0)
-            namespace, entry_name = ident.split(":")
+            namespace = identifier_to_namespace(ident)
+            entry_name = identifier_to_path(ident)
             if namespace[0] == "#":
                 queue += tags[entry_name]["values"]
                 continue
diff --git a/codegen/lib/utils.py b/codegen/lib/utils.py
index 8d1e8756..52c7206c 100644
--- a/codegen/lib/utils.py
+++ b/codegen/lib/utils.py
@@ -66,3 +66,15 @@ def group_packets(packets: list[PacketIdentifier]):
 
 def get_dir_location(name: str):
     return os.path.join(os.path.dirname(os.path.dirname(__file__)), name)
+
+def identifier_to_namespace(ident: str):
+    return ident.split(":")[0]
+def identifier_to_path(ident: str):
+    if ":" not in ident:
+        return ident
+
+    namespace, path = ident.lstrip("#").split(":")
+    if namespace in {"minecraft", "brigadier"}:
+        return path
+    # support for mods
+    return f"{namespace}_{path}"