Added message confidence levels

src/databricks/labs/pylint/airflow.py

@@ -2,6 +2,7 @@
 
 import astroid
 from pylint.checkers import BaseChecker
+from pylint.interfaces import INFERENCE
 
 
 class AirflowChecker(BaseChecker):
@@ -37,9 +38,11 @@ def visit_call(self, node: astroid.Call):
 
     def _check_new_cluster(self, key: str, new_cluster: Dict[str, Any], node: astroid.NodeNG):
         if "data_security_mode" not in new_cluster:
-            self.add_message("missing-data-security-mode", node=node, args=(key,))
+            self.add_message("missing-data-security-mode", node=node, args=(key,), confidence=INFERENCE)
         if "spark_version" in new_cluster and not self._is_supported(new_cluster["spark_version"]):
-            self.add_message("unsupported-runtime", node=node, args=(key, new_cluster["spark_version"]))
+            self.add_message(
+                "unsupported-runtime", node=node, args=(key, new_cluster["spark_version"]), confidence=INFERENCE
+            )
 
     @staticmethod
     def _is_supported(spark_version: str):

src/databricks/labs/pylint/dbutils.py

@@ -1,6 +1,7 @@
 # pylint checker for databricks dbutils
 import astroid
 from pylint.checkers import BaseChecker
+from pylint.interfaces import HIGH
 
 
 class DbutilsChecker(BaseChecker):
@@ -63,13 +64,15 @@ def visit_call(self, node: astroid.Call):
         # add message if dbutils.fs.cp() is used
         func_as_string = node.func.as_string()
         if func_as_string == "dbutils.fs.cp":
-            self.add_message("dbutils-fs-cp", node=node, args=(node.args[0].as_string(), node.args[1].as_string()))
+            self.add_message(
+                "dbutils-fs-cp", node=node, args=(node.args[0].as_string(), node.args[1].as_string()), confidence=HIGH
+            )
         # add message if dbutils.fs.head() is used
         elif func_as_string == "dbutils.fs.head":
-            self.add_message("dbutils-fs-head", node=node, args=(node.args[0].as_string(),))
+            self.add_message("dbutils-fs-head", node=node, args=(node.args[0].as_string(),), confidence=HIGH)
         # add message if dbutils.fs.ls("/tmp") is used
         elif func_as_string == "dbutils.fs.ls":
-            self.add_message("dbutils-fs-ls", node=node, args=(node.args[0].as_string(),))
+            self.add_message("dbutils-fs-ls", node=node, args=(node.args[0].as_string(),), confidence=HIGH)
         # add message if dbutils.fs.mount("s3a://%s" % aws_bucket_name, "/mnt/%s" % mount_name) is used
         elif func_as_string in {
             "dbutils.fs.mount",
@@ -78,43 +81,46 @@ def visit_call(self, node: astroid.Call):
             "dbutils.fs.updateMount",
             "dbutils.fs.refreshMounts",
         }:
-            self.add_message("dbutils-fs-mount", node=node)
+            self.add_message("dbutils-fs-mount", node=node, confidence=HIGH)
         # add message if dbutils.credentials.* is used
         elif func_as_string.startswith("dbutils.credentials."):
-            self.add_message("dbutils-credentials", node=node)
+            self.add_message("dbutils-credentials", node=node, confidence=HIGH)
         # add message if dbutils.notebook.run("My Other Notebook", 60) is used
         elif func_as_string == "dbutils.notebook.run":
             self.add_message(
-                "dbutils-notebook-run", node=node, args=(node.args[0].as_string(), node.args[1].as_string())
+                "dbutils-notebook-run",
+                node=node,
+                args=(node.args[0].as_string(), node.args[1].as_string()),
+                confidence=HIGH,
             )
         elif func_as_string.endswith("getDbutils"):
-            self.add_message("internal-api", node=node, args=(node.as_string(),))
+            self.add_message("internal-api", node=node, args=(node.as_string(),), confidence=HIGH)
         elif ".notebook().getContext()" in func_as_string:
-            self.add_message("internal-api", node=node, args=(node.as_string(),))
+            self.add_message("internal-api", node=node, args=(node.as_string(),), confidence=HIGH)
         elif ".notebook.entry_point" in func_as_string:
-            self.add_message("internal-api", node=node, args=(node.as_string(),))
+            self.add_message("internal-api", node=node, args=(node.as_string(),), confidence=HIGH)
         elif ".apiToken" in func_as_string:
-            self.add_message("internal-api", node=node, args=(node.as_string(),))
+            self.add_message("internal-api", node=node, args=(node.as_string(),), confidence=HIGH)
 
     def visit_const(self, node: astroid.Const):
         value = node.value
         if not isinstance(value, str):
             return
         # add a message if string matches dapi[0-9a-f]{32}, dkea[0-9a-f]{32}, or dosa[0-9a-f]{32}
         if value.startswith("dapi") or value.startswith("dkea") or value.startswith("dosa"):
-            self.add_message("pat-token-leaked", node=node)
+            self.add_message("pat-token-leaked", node=node, confidence=HIGH)
 
     def visit_import(self, node: astroid.Import):
         # add a message if dbruntime is imported
         for name_tuple in node.names:
             real_name, _ = name_tuple
             if real_name.startswith("dbruntime"):
-                self.add_message("internal-api", node=node, args=(node.as_string(),))
+                self.add_message("internal-api", node=node, args=(node.as_string(),), confidence=HIGH)
 
     def visit_importfrom(self, node: astroid.ImportFrom):
         # add a message if dbruntime is imported
         if node.modname.startswith("dbruntime"):
-            self.add_message("internal-api", node=node, args=(node.as_string(),))
+            self.add_message("internal-api", node=node, args=(node.as_string(),), confidence=HIGH)
 
 
 def register(linter):

src/databricks/labs/pylint/legacy.py

@@ -2,6 +2,7 @@
 
 import astroid
 from pylint.checkers import BaseChecker
+from pylint.interfaces import HIGH, INFERENCE
 
 
 class LegacyChecker(BaseChecker):
@@ -68,26 +69,26 @@ def visit_import(self, node: astroid.Import):
         # add message if databricks_cli is imported
         for name, _ in node.names:
             if name.startswith("databricks_cli"):
-                self.add_message("legacy-cli", node=node)
+                self.add_message("legacy-cli", node=node, confidence=HIGH)
             # very coarse check for UC incompatibility
             for needle in self.UC_INCOMPATIBLE_BRUTE_FORCE:
                 if needle in name:
-                    self.add_message("incompatible-with-uc", node=node, args=(node.as_string(),))
+                    self.add_message("incompatible-with-uc", node=node, args=(node.as_string(),), confidence=INFERENCE)
 
     def visit_importfrom(self, node: astroid.ImportFrom):
         if node.modname.startswith("databricks_cli"):
             self.add_message("legacy-cli", node=node)
         # very coarse check for UC incompatibility
         for needle in self.UC_INCOMPATIBLE_BRUTE_FORCE:
             if needle in node.modname:
-                self.add_message("incompatible-with-uc", node=node, args=(node.as_string(),))
+                self.add_message("incompatible-with-uc", node=node, args=(node.as_string(),), confidence=INFERENCE)
 
     def visit_call(self, node: astroid.Call):
         func_as_string = node.func.as_string()
         # very coarse check for UC incompatibility
         for needle in self.UC_INCOMPATIBLE_BRUTE_FORCE:
             if needle in func_as_string:
-                self.add_message("incompatible-with-uc", node=node, args=(node.as_string(),))
+                self.add_message("incompatible-with-uc", node=node, args=(node.as_string(),), confidence=INFERENCE)
 
     def visit_const(self, node: astroid.Const):
         # very coarse check for UC incompatibility
@@ -96,7 +97,7 @@ def visit_const(self, node: astroid.Const):
             return
         for needle in self.UC_INCOMPATIBLE_BRUTE_FORCE:
             if needle in value:
-                self.add_message("incompatible-with-uc", node=node, args=(node.as_string(),))
+                self.add_message("incompatible-with-uc", node=node, args=(node.as_string(),), confidence=INFERENCE)
 
 
 def register(linter):

src/databricks/labs/pylint/notebooks.py

@@ -1,5 +1,6 @@
 import astroid
 from pylint.checkers import BaseRawFileChecker
+from pylint.interfaces import CONTROL_FLOW, HIGH
 
 
 class NotebookChecker(BaseRawFileChecker):
@@ -55,11 +56,11 @@ def process_module(self, node: astroid.Module):
                 if line == b"# COMMAND ----------\n":
                     cells += 1
                 if cells > self.linter.config.max_cells and not too_many_cells_raised:
-                    self.add_message("notebooks-too-many-cells", line=lineno + 1)
+                    self.add_message("notebooks-too-many-cells", line=lineno + 1, confidence=CONTROL_FLOW)
                     too_many_cells_raised = True
                     continue
                 if line.startswith(b"# MAGIC %run"):
-                    self.add_message("notebooks-percent-run", line=lineno)
+                    self.add_message("notebooks-percent-run", line=lineno, confidence=HIGH)
 
 
 def register(linter):

src/databricks/labs/pylint/spark.py

@@ -1,5 +1,6 @@
 import astroid
 from pylint.checkers import BaseChecker
+from pylint.interfaces import CONTROL_FLOW, INFERENCE
 
 
 class SparkChecker(BaseChecker):
@@ -33,19 +34,23 @@ def visit_name(self, node: astroid.Name):
         while in_node and not isinstance(in_node, astroid.FunctionDef):
             in_node = in_node.parent
         if not in_node:
-            self.add_message("spark-outside-function", node=node)
+            self.add_message("spark-outside-function", node=node, confidence=CONTROL_FLOW)
             return
         has_spark_arg = False
         for arg in in_node.args.arguments:
             if arg.name == "spark":
                 has_spark_arg = True
                 break
         if not has_spark_arg:
-            self.add_message("no-spark-argument-in-function", node=in_node, args=(in_node.name,))
+            self.add_message(
+                "no-spark-argument-in-function", node=in_node, args=(in_node.name,), confidence=CONTROL_FLOW
+            )
 
     def visit_attribute(self, node: astroid.Attribute):
         if node.attrname == "show" and isinstance(node.expr, astroid.Call):
-            self.add_message("use-display-instead-of-show", node=node, args=(node.expr.as_string(),))
+            self.add_message(
+                "use-display-instead-of-show", node=node, args=(node.expr.as_string(),), confidence=INFERENCE
+            )
 
 
 def register(linter):

tests/samples/m/many_cells.txt

@@ -1 +1 @@
-notebooks-too-many-cells:303:0:None:None::Notebooks should not have more than 75 cells:UNDEFINED
+notebooks-too-many-cells:303:0:None:None::Notebooks should not have more than 75 cells:CONTROL_FLOW

tests/samples/p/percent_run.txt

@@ -1,3 +1,3 @@
-notebooks-percent-run:7:0:None:None::Using %run is not allowed:UNDEFINED
-spark-outside-function:15:5:15:10::Using spark outside the function is leading to untestable code:UNDEFINED
-spark-outside-function:24:5:24:10::Using spark outside the function is leading to untestable code:UNDEFINED
+notebooks-percent-run:7:0:None:None::Using %run is not allowed:HIGH
+spark-outside-function:15:5:15:10::Using spark outside the function is leading to untestable code:CONTROL_FLOW
+spark-outside-function:24:5:24:10::Using spark outside the function is leading to untestable code:CONTROL_FLOW