feat: reduce the size of literal identifiers

nada_dsl/ast_util.py

@@ -10,6 +10,7 @@
 
 OPERATION_ID_COUNTER = 0
 
+
 def next_operation_id() -> int:
     """Returns the next value of the operation id counter."""
     global OPERATION_ID_COUNTER
@@ -55,6 +56,9 @@ def to_mir(self):
 # The key is the operation identifier, the value the operation
 AST_OPERATIONS: Dict[int, ASTOperation] = SortedDict()
 
+# Map of literal hashes to index
+LITERALS: Dict[str, int] = {}
+
 
 @dataclass
 class BinaryASTOperation(ASTOperation):
@@ -167,7 +171,7 @@ class LiteralASTOperation(ASTOperation):
 
     name: str
     value: object
-    literal_name: str
+    literal_index: str
 
     def __init__(
         self,
@@ -184,16 +188,23 @@ def __init__(
         self.source_ref = source_ref
         # Generate a unique name depending on the value and type
         # to prevent duplicating literals in the bytecode.
-        self.literal_name = hashlib.md5(
+        literal_name = hashlib.md5(
             (str(self.value) + str(self.ty)).encode("UTF-8")
         ).hexdigest()
+
+        # Replace the literal name by an index so it is shorter
+        if literal_name not in LITERALS:
+            LITERALS[literal_name] = len(LITERALS)
+
+        self.literal_index = str(LITERALS[literal_name])
+
         super().__init__(id=self.id, source_ref=self.source_ref, ty=self.ty)
 
     def to_mir(self):
         return {
             "LiteralReference": {
                 "id": self.id,
-                "refers_to": self.literal_name,
+                "refers_to": self.literal_index,
                 "type": self.ty,
                 "source_ref_index": self.source_ref.to_index(),
             }

nada_dsl/compiler_frontend.py

@@ -305,7 +305,7 @@ def process_operation(
         processed_operation = ProcessOperationOutput(operation.to_mir(), None)
     elif isinstance(operation, LiteralASTOperation):
 
-        LITERALS[operation.literal_name] = (str(operation.value), operation.ty)
+        LITERALS[operation.literal_index] = (str(operation.value), operation.ty)
         processed_operation = ProcessOperationOutput(operation.to_mir(), None)
     elif isinstance(
         operation, (MapASTOperation, ReduceASTOperation, NadaFunctionCallASTOperation)

nada_dsl/compiler_frontend_test.py

@@ -577,7 +577,6 @@ def test_binary_operator_integer_publicinteger(operator, name, ty):
     right_ast = AST_OPERATIONS[inner["right"]]
     assert isinstance(left_ast, LiteralASTOperation)
     assert left_ast.value == -3
-    assert len(left_ast.literal_name) == 32
     assert isinstance(right_ast, InputASTOperation)
     assert right_ast.name == "right"
     assert inner["type"] == to_type(ty)