[CT-1591] Don't parse empty Python files

.changes/unreleased/Fixes-20221213-113915.yaml

@@ -0,0 +1,6 @@
+kind: Fixes
+body: '[CT-1591] Don''t parse empty Python files'
+time: 2022-12-13T11:39:15.818464-08:00
+custom:
+  Author: aranke
+  Issue: "6345"

core/dbt/parser/models.py

@@ -31,7 +31,6 @@
 from dbt.dataclass_schema import ValidationError
 from dbt.exceptions import ParsingException, validator_error_message, UndefinedMacroException
 
-
 dbt_function_key_words = set(["ref", "source", "config", "get"])
 dbt_function_full_names = set(["dbt.ref", "dbt.source", "dbt.config", "dbt.config.get"])
 
@@ -196,42 +195,47 @@ def get_compiled_path(cls, block: FileBlock):
         return block.path.relative_path
 
     def parse_python_model(self, node, config, context):
+        config_keys_used = []
+        config_keys_defaults = []
+
         try:
             tree = ast.parse(node.raw_code, filename=node.original_file_path)
         except SyntaxError as exc:
             msg = validator_error_message(exc)
             raise ParsingException(f"{msg}\n{exc.text}", node=node) from exc
 
-        # We are doing a validator and a parser because visit_FunctionDef in parser
-        # would actually make the parser not doing the visit_Calls any more
-        dbtValidator = PythonValidationVisitor()
-        dbtValidator.visit(tree)
-        dbtValidator.check_error(node)
+        # Only parse if AST tree has instructions in body
+        if tree.body:
+            # We are doing a validator and a parser because visit_FunctionDef in parser
+            # would actually make the parser not doing the visit_Calls any more
+            dbt_validator = PythonValidationVisitor()
+            dbt_validator.visit(tree)
+            dbt_validator.check_error(node)
+
+            dbt_parser = PythonParseVisitor(node)
+            dbt_parser.visit(tree)
+
+            for (func, args, kwargs) in dbt_parser.dbt_function_calls:
+                if func == "get":
+                    num_args = len(args)
+                    if num_args == 0:
+                        raise ParsingException(
+                            "dbt.config.get() requires at least one argument",
+                            node=node,
+                        )
+                    if num_args > 2:
+                        raise ParsingException(
+                            f"dbt.config.get() takes at most 2 arguments ({num_args} given)",
+                            node=node,
+                        )
+                    key = args[0]
+                    default_value = args[1] if num_args == 2 else None
+                    config_keys_used.append(key)
+                    config_keys_defaults.append(default_value)
+                    continue
+
+                context[func](*args, **kwargs)
 
-        dbtParser = PythonParseVisitor(node)
-        dbtParser.visit(tree)
-        config_keys_used = []
-        config_keys_defaults = []
-        for (func, args, kwargs) in dbtParser.dbt_function_calls:
-            if func == "get":
-                num_args = len(args)
-                if num_args == 0:
-                    raise ParsingException(
-                        "dbt.config.get() requires at least one argument",
-                        node=node,
-                    )
-                if num_args > 2:
-                    raise ParsingException(
-                        f"dbt.config.get() takes at most 2 arguments ({num_args} given)",
-                        node=node,
-                    )
-                key = args[0]
-                default_value = args[1] if num_args == 2 else None
-                config_keys_used.append(key)
-                config_keys_defaults.append(default_value)
-                continue
-
-            context[func](*args, **kwargs)
         if config_keys_used:
             # this is being used in macro build_config_dict
             context["config"](

test/unit/test_parser.py

@@ -571,6 +571,8 @@ def model1(dbt, session):
     return dbt.ref("some_model")
 """
 
+python_model_empty_file = """    """
+
 python_model_multiple_returns = """
 def model(dbt, session):
     dbt.config(materialized='table')
@@ -749,6 +751,11 @@ def test_python_model_incorrect_function_name(self):
         with self.assertRaises(ParsingException):
             self.parser.parse_file(block)
 
+    def test_python_model_empty_file(self):
+        block = self.file_block_for(python_model_empty_file, "nested/py_model.py")
+        self.parser.manifest.files[block.file.file_id] = block.file
+        self.assertIsNone(self.parser.parse_file(block))
+
     def test_python_model_multiple_returns(self):
         block = self.file_block_for(python_model_multiple_returns, 'nested/py_model.py')
         self.parser.manifest.files[block.file.file_id] = block.file
@@ -786,6 +793,7 @@ def test_python_model_custom_materialization(self):
         node = list(self.parser.manifest.nodes.values())[0]
         self.assertEqual(node.get_materialization(), "view")
 
+
 class StaticModelParserTest(BaseParserTest):
     def setUp(self):
         super().setUp()