make getting shapes faster (#149)

* supposed to make getting shapes faster

* why was this reversed

* forgot to run codegen

* don't panic when getting the shape for invalid block ids

---------

Co-authored-by: mat <git@matdoes.dev>

2 files changed, 28 insertions, 18 deletions

diff --git a/codegen/lib/code/shapes.py b/codegen/lib/code/shapes.py
index 6906328c..bedab647 100755
--- a/codegen/lib/code/shapes.py
+++ b/codegen/lib/code/shapes.py
@@ -73,9 +73,11 @@ def generate_block_shapes_code(blocks: dict, shapes: dict, block_states_report,
         generated_shape_code += generate_code_for_shape(shape_id, shape)
 
 
-    # 1..100 | 200..300 => &SHAPE1,
-    generated_match_inner_code = ''
-    shape_ids_to_block_state_ids = {}
+    # static SHAPES_MAP: [&Lazy<VoxelShape>; 26644] = [&SHAPE0, &SHAPE1, &SHAPE1, ...]
+    empty_shapes = []
+    full_shapes = []
+
+    block_state_ids_to_shape_ids = []
     for block_id, shape_ids in blocks.items():
         if isinstance(shape_ids, int):
             shape_ids = [shape_ids]
@@ -84,19 +86,24 @@ def generate_block_shapes_code(blocks: dict, shapes: dict, block_states_report,
         for possible_state, shape_id in zip(block_report_data['states'], shape_ids):
             block_state_id = possible_state['id']
 
-            if shape_id not in shape_ids_to_block_state_ids:
-                shape_ids_to_block_state_ids[shape_id] = []
-            shape_ids_to_block_state_ids[shape_id].append(block_state_id)
+            if shape_id == 0 :
+                empty_shapes.append(block_state_id)
+            elif shape_id == 1 :
+                full_shapes.append(block_state_id)
+
+            block_state_ids_to_shape_ids.append((block_state_id, shape_id))
+
+
+    generated_map_code = f'static SHAPES_MAP: [&Lazy<VoxelShape>; {len(block_state_ids_to_shape_ids)}] = ['
 
-    empty_shape_match_code = convert_ints_to_rust_ranges(shape_ids_to_block_state_ids[0])
-    block_shape_match_code = convert_ints_to_rust_ranges(shape_ids_to_block_state_ids[1])
+    block_state_ids_to_shape_ids = sorted(block_state_ids_to_shape_ids, key=lambda x: x[0])
 
-    # shape 1 is the most common so we have a _ => &SHAPE1 at the end
-    del shape_ids_to_block_state_ids[1]
+    empty_shape_match_code = convert_ints_to_rust_ranges(empty_shapes)
+    block_shape_match_code = convert_ints_to_rust_ranges(full_shapes)
 
-    for shape_id, block_state_ids in shape_ids_to_block_state_ids.items():
-        generated_match_inner_code += f'{convert_ints_to_rust_ranges(block_state_ids)} => &SHAPE{shape_id},\n'
-    generated_match_inner_code += '_ => &SHAPE1'
+    for block_state_id, shape_id in block_state_ids_to_shape_ids:
+        generated_map_code += f'&SHAPE{shape_id},\n'
+    generated_map_code += '];'
 
     if empty_shape_match_code == '':
         print('Error: shape 0 was not found')
@@ -126,11 +133,10 @@ pub trait BlockWithShape {{
 
 {generated_shape_code}
 
+
 impl BlockWithShape for BlockState {{
     fn shape(&self) -> &'static VoxelShape {{
-        match self.id {{
-            {generated_match_inner_code}
-        }}
+        SHAPES_MAP.get(self.id as usize).unwrap_or(&&SHAPE1)
     }}
 
     fn is_shape_empty(&self) -> bool {{
@@ -141,8 +147,12 @@ impl BlockWithShape for BlockState {{
         matches!(self.id, {block_shape_match_code})
     }}
 }}
+
+{generated_map_code}
 '''
 
+    
+
 
 def generate_code_for_shape(shape_id: str, parts: list[list[float]]):
     def make_arguments(part: list[float]):
diff --git a/codegen/lib/extract.py b/codegen/lib/extract.py
index fa75c49e..d3bd46d3 100755
--- a/codegen/lib/extract.py
+++ b/codegen/lib/extract.py
@@ -236,9 +236,9 @@ def get_pixlyzer_data(version_id: str, category: str):
         assert pom_xml_dependencies != ''
         pom_xml = open(f'{pixlyzer_dir}/pom.xml', 'r').read()
         pom_xml = re.sub(
-            '<dependencies>.*?</dependencies>', f'<dependencies>{pom_xml_dependencies}</dependencies>', pom_xml, flags=re.DOTALL)
+            r'<dependencies>.*?</dependencies>', f'<dependencies>{pom_xml_dependencies}</dependencies>', pom_xml, flags=re.DOTALL)
         pom_xml = re.sub(
-            '<minecraft\.version>.*?</minecraft\.version>', f'<minecraft.version>{version_id}</minecraft.version>', pom_xml, flags=re.DOTALL)
+            r'<minecraft\.version>.*?</minecraft\.version>', f'<minecraft.version>{version_id}</minecraft.version>', pom_xml, flags=re.DOTALL)
         open(f'{pixlyzer_dir}/pom.xml', 'w').write(pom_xml)
 
         # compile