refactor(AIR302): refactor regex usage

crates/ruff_linter/resources/test/fixtures/airflow/AIR302_class_attribute.py

@@ -0,0 +1,59 @@
+from airflow.datasets.manager import DatasetManager
+from airflow.lineage.hook import DatasetLineageInfo, HookLineageCollector
+from airflow.providers.amazon.auth_manager.aws_auth_manager import AwsAuthManager
+from airflow.providers.apache.beam.hooks import BeamHook, NotAir302HookError
+from airflow.providers.google.cloud.secrets.secret_manager import (
+    CloudSecretManagerBackend,
+)
+from airflow.providers.hashicorp.secrets.vault import NotAir302SecretError, VaultBackend
+from airflow.providers_manager import ProvidersManager
+from airflow.secrets.base_secrets import BaseSecretsBackend
+
+dm = DatasetManager()
+dm.register_dataset_change()
+dm.create_datasets()
+dm.notify_dataset_created()
+dm.notify_dataset_changed()
+dm.notify_dataset_alias_created()
+
+hlc = HookLineageCollector()
+hlc.create_dataset()
+hlc.add_input_dataset()
+hlc.add_output_dataset()
+hlc.collected_datasets()
+
+aam = AwsAuthManager()
+aam.is_authorized_dataset()
+
+pm = ProvidersManager()
+pm.initialize_providers_asset_uri_resources()
+pm.dataset_factories
+
+base_secret_backend = BaseSecretsBackend()
+base_secret_backend.get_conn_uri()
+base_secret_backend.get_connections()
+
+csm_backend = CloudSecretManagerBackend()
+csm_backend.get_conn_uri()
+csm_backend.get_connections()
+
+vault_backend = VaultBackend()
+vault_backend.get_conn_uri()
+vault_backend.get_connections()
+
+not_an_error = NotAir302SecretError()
+not_an_error.get_conn_uri()
+
+beam_hook = BeamHook()
+beam_hook.get_conn_uri()
+
+not_an_error = NotAir302HookError()
+not_an_error.get_conn_uri()
+
+provider_manager = ProvidersManager()
+provider_manager.dataset_factories
+provider_manager.dataset_uri_handlers
+provider_manager.dataset_to_openlineage_converters
+
+dl_info = DatasetLineageInfo()
+dl_info.dataset

crates/ruff_linter/resources/test/fixtures/airflow/AIR302_names.py

@@ -269,68 +269,3 @@
 
 # airflow.www.utils
 get_sensitive_variables_fields, should_hide_value_for_key
-
-
-# methods
-
-from airflow.datasets.manager import DatasetManager
-
-dm = DatasetManager()
-dm.register_dataset_change()
-dm.create_datasets()
-dm.notify_dataset_created()
-dm.notify_dataset_changed()
-dm.notify_dataset_alias_created()
-
-
-from airflow.lineage.hook import HookLineageCollector
-
-hlc = HookLineageCollector()
-hlc.create_dataset()
-hlc.add_input_dataset()
-hlc.add_output_dataset()
-hlc.collected_datasets()
-
-from airflow.providers.amazon.auth_manager.aws_auth_manager import AwsAuthManager
-
-aam = AwsAuthManager()
-aam.is_authorized_dataset()
-
-from airflow.providers_manager import ProvidersManager
-
-pm = ProvidersManager()
-pm.initialize_providers_asset_uri_resources()
-pm.dataset_factories
-
-
-from airflow.secrets.base_secrets import BaseSecretsBackend
-
-base_secret_backend = BaseSecretsBackend()
-base_secret_backend.get_conn_uri()
-base_secret_backend.get_connections()
-
-from airflow.providers.google.cloud.secrets.secret_manager import (
-    CloudSecretManagerBackend,
-)
-
-csm_backend = CloudSecretManagerBackend()
-csm_backend.get_conn_uri()
-csm_backend.get_connections()
-
-from airflow.providers.hashicorp.secrets.vault import VaultBackend
-
-vault_backend = VaultBackend()
-vault_backend.get_conn_uri()
-vault_backend.get_connections()
-
-from airflow.providers_manager import ProvidersManager
-
-provider_manager = ProvidersManager()
-provider_manager.dataset_factories
-provider_manager.dataset_uri_handlers
-provider_manager.dataset_to_openlineage_converters
-
-from airflow.lineage.hook import DatasetLineageInfo
-
-dl_info = DatasetLineageInfo()
-dl_info.dataset

crates/ruff_linter/src/rules/airflow/mod.rs

@@ -16,6 +16,7 @@ mod tests {
     #[test_case(Rule::AirflowDagNoScheduleArgument, Path::new("AIR301.py"))]
     #[test_case(Rule::Airflow3Removal, Path::new("AIR302_args.py"))]
     #[test_case(Rule::Airflow3Removal, Path::new("AIR302_names.py"))]
+    #[test_case(Rule::Airflow3Removal, Path::new("AIR302_class_attribute.py"))]
     #[test_case(Rule::Airflow3MovedToProvider, Path::new("AIR303.py"))]
     fn rules(rule_code: Rule, path: &Path) -> Result<()> {
         let snapshot = format!("{}_{}", rule_code.noqa_code(), path.to_string_lossy());

crates/ruff_linter/src/rules/airflow/rules/removal_in_3.rs

@@ -4,16 +4,58 @@ use ruff_python_ast::{name::QualifiedName, Arguments, Expr, ExprAttribute, ExprC
 use ruff_python_semantic::analyze::typing;
 use ruff_python_semantic::Modules;
 use ruff_text_size::Ranged;
-use std::sync::LazyLock;
 
 use crate::checkers::ast::Checker;
-use regex::Regex;
 
-static SECRET_BACKEND_REGEX: LazyLock<Regex> =
-    LazyLock::new(|| Regex::new(r"airflow\..*secrets\.\w+\.\w+Backend").unwrap());
+fn is_airflow_secret_backend(qualname: &QualifiedName) -> bool {
+    match qualname.segments() {
+        ["airflow", "secrets", rest @ ..] => {
+            if let Some(last_element) = rest.last() {
+                last_element.ends_with("Backend")
+            } else {
+                false
+            }
+        }
+
+        ["airflow", "providers", rest @ ..] => {
+            // Ensure 'operators' exists somewhere in the middle
+            if let (Some(pos), Some(last_element)) =
+                (rest.iter().position(|&s| s == "secrets"), rest.last())
+            {
+                pos + 1 < rest.len() && last_element.ends_with("Backend") // Check that 'operators' is not the last element
+            } else {
+                false
+            }
+        }
 
-static AIRFLOW_HOOK_REGEX: LazyLock<Regex> =
-    LazyLock::new(|| Regex::new(r"airflow\..*hooks\.\w+\.\w+Hook").unwrap());
+        _ => false,
+    }
+}
+
+fn is_airflow_hook(qualname: &QualifiedName) -> bool {
+    match qualname.segments() {
+        ["airflow", "hooks", rest @ ..] => {
+            if let Some(last_element) = rest.last() {
+                last_element.ends_with("Hook")
+            } else {
+                false
+            }
+        }
+
+        ["airflow", "providers", rest @ ..] => {
+            // Ensure 'operators' exists somewhere in the middle
+            if let (Some(pos), Some(last_element)) =
+                (rest.iter().position(|&s| s == "hooks"), rest.last())
+            {
+                pos + 1 < rest.len() && last_element.ends_with("Hook")
+            } else {
+                false
+            }
+        }
+
+        _ => false,
+    }
+}
 
 #[derive(Debug, Eq, PartialEq)]
 enum Replacement {
@@ -172,14 +214,6 @@ fn removed_class_attribute(checker: &mut Checker, expr: &Expr) {
         return;
     };
 
-    // checker.diagnostics.push(Diagnostic::new(
-    //     Airflow3Removal {
-    //         deprecated: qualname.to_string(),
-    //         replacement: Replacement::Name("te"),
-    //     },
-    //     attr.range(),
-    // ));
-
     let replacement = match *qualname.segments() {
         ["airflow", "providers_manager", "ProvidersManager"] => match attr.as_str() {
             "dataset_factories" => Some(Replacement::Name("asset_factories")),
@@ -252,12 +286,12 @@ fn removed_method(checker: &mut Checker, expr: &Expr) {
             "iter_dataset_aliases" => Some(Replacement::Name("iter_asset_aliases")),
             &_ => None,
         },
-        _ if SECRET_BACKEND_REGEX.is_match(&qualname.segments().join(".")) => match attr.as_str() {
+        _ if is_airflow_secret_backend(&qualname) => match attr.as_str() {
             "get_conn_uri" => Some(Replacement::Name("get_conn_value")),
             "get_connections" => Some(Replacement::Name("get_connection")),
             &_ => None,
         },
-        _ if AIRFLOW_HOOK_REGEX.is_match(&qualname.segments().join(".")) => match attr.as_str() {
+        _ if is_airflow_hook(&qualname) => match attr.as_str() {
             "get_connections" => Some(Replacement::Name("get_connection")),
             &_ => None,
         },