Feature/schema tests are more unique

CHANGELOG.md

@@ -26,6 +26,7 @@ Contributors:
 - Ensure that schema test macros are properly processed ([#3229](https://github.com/fishtown-analytics/dbt/issues/3229), [#3272](https://github.com/fishtown-analytics/dbt/pull/3272))
 - Use absolute path for profiles directory instead of a path relative to the project directory. Note: If a user supplies a relative path to the profiles directory, the value of `args.profiles_dir` will still be absolute. ([#3133](https://github.com/fishtown-analytics/dbt/issues/3133))
 - Fix FQN selector unable to find models whose name contains dots ([#3246](https://github.com/fishtown-analytics/dbt/issues/3246))
+- Fix unique_id generation for generic tests so tests with the same FQN but different configuration will run. ([#3254](https://github.com/fishtown-analytics/dbt/issues/3254)) 
 
 ### Features
 - Support commit hashes in dbt deps package revision ([#3268](https://github.com/fishtown-analytics/dbt/issues/3268), [#3270](https://github.com/fishtown-analytics/dbt/pull/3270))

core/dbt/exceptions.py

@@ -772,6 +772,10 @@ def raise_duplicate_resource_name(node_1, node_2):
     elif node_1.resource_type == NodeType.Documentation:
         get_func = 'doc("{}")'.format(duped_name)
     elif node_1.resource_type == NodeType.Test and 'schema' in node_1.tags:
+        # TODO: This causes duplicate schema test errors to be ignored!!
+        # This PR should ensure this never gets called when the tests are actually different,
+        # but removing this breaks 50+ tests that currently have the exact same schema
+        # test defined more than once. :/
         return
     else:
         get_func = '"{}"'.format(duped_name)

core/dbt/parser/base.py

@@ -2,7 +2,7 @@
 import itertools
 import os
 from typing import (
-    List, Dict, Any, Generic, TypeVar
+    List, Dict, Any, Generic, Optional, TypeVar
 )
 
 from dbt.dataclass_schema import ValidationError
@@ -55,11 +55,25 @@ def parse_file(self, block: FileBlock) -> None:
     def resource_type(self) -> NodeType:
         pass
 
-    def generate_unique_id(self, resource_name: str) -> str:
-        """Returns a unique identifier for a resource"""
-        return "{}.{}.{}".format(self.resource_type,
-                                 self.project.project_name,
-                                 resource_name)
+    def generate_unique_id(
+        self,
+        resource_name: str,
+        hash: Optional[str] = None
+    ) -> str:
+        """Returns a unique identifier for a resource
+        An optional hash may be passed in to ensure uniqueness for edge cases"""
+
+        return '.'.join(
+            filter(
+                None,
+                [
+                    self.resource_type,
+                    self.project.project_name,
+                    resource_name,
+                    hash
+                ]
+            )
+        )
 
 
 class Parser(BaseParser[FinalValue], Generic[FinalValue]):

core/dbt/parser/docs.py

@@ -1,4 +1,4 @@
-from typing import Iterable
+from typing import Iterable, Optional
 
 import re
 
@@ -23,7 +23,7 @@ def resource_type(self) -> NodeType:
     def get_compiled_path(cls, block: FileBlock):
         return block.path.relative_path
 
-    def generate_unique_id(self, resource_name: str) -> str:
+    def generate_unique_id(self, resource_name: str, _: Optional[str] = None) -> str:
         # because docs are in their own graph namespace, node type doesn't
         # need to be part of the unique ID.
         return '{}.{}'.format(self.project.project_name, resource_name)

core/dbt/parser/schemas.py

@@ -2,8 +2,9 @@
 import os
 
 from abc import ABCMeta, abstractmethod
+from hashlib import md5
 from typing import (
-    Iterable, Dict, Any, Union, List, Optional, Generic, TypeVar, Type
+    Iterable, Dict, Any, Union, List, Optional, Generic, TypeVar, Type, Iterator
 )
 
 from dbt.dataclass_schema import ValidationError, dbtClassMixin
@@ -351,6 +352,30 @@ def create_test_node(
         column_name: Optional[str],
     ) -> ParsedSchemaTestNode:
 
+        # TODO: (PR input needed) Should this be applied to all tests?
+        # In theory we already check non-schema tests for "unique" name uniqueness in
+        # core/dbt/contracts/graph/manifest.py @ _check_duplicates()
+
+        HASH_LENGTH = 10
+
+        if 'schema' in tags:
+            # TODO: (PR input needed) is this a complete set of uniquie-ifying data?
+            hash_data: Iterator = filter(
+                None,
+                [
+                    name,
+                    test_metadata.get('kwargs').get('model'),  # type: ignore
+                    column_name
+                ])
+            hash_string = ''.join(hash_data).encode('utf-8')
+            test_hash = md5(hash_string).hexdigest()[-HASH_LENGTH:]
+            unique_id = self.generate_unique_id(
+                name,
+                test_hash
+            )
+        else:
+            unique_id = self.generate_unique_id(name)
+
         dct = {
             'alias': name,
             'schema': self.default_schema,
@@ -364,7 +389,7 @@ def create_test_node(
             'original_file_path': target.original_file_path,
             'package_name': self.project.project_name,
             'raw_sql': raw_sql,
-            'unique_id': self.generate_unique_id(name),
+            'unique_id': unique_id,
             'config': self.config_dict(config),
             'test_metadata': test_metadata,
             'column_name': column_name,