feat(pyrefly): add pyrefly type checker integration

mkoura · mkoura · commit a46918424b25 · 2025-05-28T10:02:32.000+02:00
- Add pyrefly as a dev dependency in pyproject.toml and poetry.lock.
- Configure pyrefly with project-specific settings.
- Integrate pyrefly into pre-commit hooks for Python files.
- Add type ignore comments for pyrefly in various modules to suppress
  specific errors and improve compatibility.
- Refactor some type annotations and variable initializations for better
  type checking.
- Remove unused method from dbsync_service_manager.Table.
diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
@@ -34,6 +34,13 @@ repos:
     additional_dependencies: [sphinx, toml]
 - repo: local
   hooks:
+  - id: pyrefly
+    name: pyrefly
+    entry: pyrefly
+    args: [ check, --remove-unused-ignores ]
+    pass_filenames: false
+    language: system
+    types: [python]
   - id: mypy
     name: mypy
     entry: mypy
diff --git a/cardano_node_tests/cardano_cli_coverage.py b/cardano_node_tests/cardano_cli_coverage.py
@@ -105,9 +105,11 @@ def merge_coverage(dict_a: dict, dict_b: dict) -> dict:
     mergeable = (list, set, tuple)
     addable = (int, float)
     for key, value in dict_b.items():
+        # pyrefly: ignore  # invalid-argument
         if key in dict_a and isinstance(value, mergeable) and isinstance(dict_a[key], mergeable):
             new_list = set(dict_a[key]).union(value)
             dict_a[key] = sorted(new_list)
+        # pyrefly: ignore  # invalid-argument
         elif key in dict_a and isinstance(value, addable) and isinstance(dict_a[key], addable):
             dict_a[key] += value
         # Skipped arguments and commands are not in the available commands dict
@@ -126,6 +128,7 @@ def cli(cli_args: tp.Iterable[str]) -> str:
     assert not isinstance(cli_args, str), "`cli_args` must be sequence of strings"
     with subprocess.Popen(list(cli_args), stdout=subprocess.PIPE, stderr=subprocess.PIPE) as p:
         __, stderr = p.communicate()
+    # pyrefly: ignore  # missing-attribute
     return stderr.decode()
 
 
diff --git a/cardano_node_tests/pytest_plugins/xdist_scheduler.py b/cardano_node_tests/pytest_plugins/xdist_scheduler.py
@@ -104,6 +104,7 @@ def _assign_work_unit(self, node: workermanage.WorkerController) -> None:
         scope, work_unit = None, None
 
         # Check if there are any long-running tests already pending
+        # pyrefly: ignore  # bad-argument-type
         long_pending = self._is_long_pending(assigned_to_node)
 
         if long_pending:
diff --git a/cardano_node_tests/tests/test_dbsync.py b/cardano_node_tests/tests/test_dbsync.py
@@ -197,8 +197,8 @@ def test_blocks(self, cluster: clusterlib.ClusterLib):  # noqa: C901
                     f"Expected: value > {prev_rec.epoch_slot_no} vs Returned: {rec.epoch_slot_no}"
                 )
 
-            if rec.block_no is None or (
-                prev_rec.block_no and rec.block_no != prev_rec.block_no + 1
+            if (rec.block_no is None and prev_rec.block_no is not None) or (
+                prev_rec.block_no is not None and rec.block_no != prev_rec.block_no + 1
             ):
                 errors.append(
                     "'block_no' value is different than expected; "
@@ -211,6 +211,7 @@ def test_blocks(self, cluster: clusterlib.ClusterLib):  # noqa: C901
                     f"Expected: {prev_rec.id} vs Returned: {rec.previous_id}"
                 )
 
+            # pyrefly: ignore  # unknown
             prev_rec = rec
 
         if errors:
diff --git a/cardano_node_tests/tests/test_kes.py b/cardano_node_tests/tests/test_kes.py
@@ -290,7 +290,9 @@ def _check_kes_period_info(
             ignore_file_id=worker_id,
         )
         # Search for expected errors only in log file corresponding to pool with expired KES
-        expected_errors = [(f"{expire_node_name}.stdout", err) for err in expected_err_regexes]
+        expected_errors: list[tuple[str, str]] = [
+            (f"{expire_node_name}.stdout", err) for err in expected_err_regexes
+        ]
 
         with logfiles.expect_errors(expected_errors, worker_id=worker_id):
             LOGGER.info(
diff --git a/cardano_node_tests/tests/test_reconnect.py b/cardano_node_tests/tests/test_reconnect.py
@@ -65,9 +65,9 @@ def node_query_utxo(
         try:
             os.environ["CARDANO_NODE_SOCKET_PATH"] = str(new_socket)
             utxos = cluster_obj.g_query.get_utxo(address=address, tx_raw_output=tx_raw_output)
-            return utxos
         finally:
             os.environ["CARDANO_NODE_SOCKET_PATH"] = orig_socket
+        return utxos
 
     def node_get_tip(
         self,
@@ -82,9 +82,9 @@ def node_get_tip(
         try:
             os.environ["CARDANO_NODE_SOCKET_PATH"] = str(new_socket)
             tip = cluster_obj.g_query.get_tip()
-            return tip
         finally:
             os.environ["CARDANO_NODE_SOCKET_PATH"] = orig_socket
+        return tip
 
     def node_submit_tx(
         self,
@@ -112,9 +112,9 @@ def node_submit_tx(
                 tx_files=tx_files,
                 verify_tx=False,
             )
-            return tx_raw_output
         finally:
             os.environ["CARDANO_NODE_SOCKET_PATH"] = orig_socket
+        return tx_raw_output
 
     def get_prometheus_metrics(self, port: int) -> requests.Response:
         response = http_client.get_session().get(f"http://localhost:{port}/metrics", timeout=10)
diff --git a/cardano_node_tests/tests/test_rollback.py b/cardano_node_tests/tests/test_rollback.py
@@ -128,9 +128,9 @@ def node_query_utxo(
         try:
             os.environ["CARDANO_NODE_SOCKET_PATH"] = str(new_socket)
             utxos = cluster_obj.g_query.get_utxo(address=address, tx_raw_output=tx_raw_output)
-            return utxos
         finally:
             os.environ["CARDANO_NODE_SOCKET_PATH"] = orig_socket
+        return utxos
 
     def node_submit_tx(
         self,
@@ -157,9 +157,9 @@ def node_submit_tx(
                 txouts=txouts,
                 tx_files=tx_files,
             )
-            return tx_raw_output
         finally:
             os.environ["CARDANO_NODE_SOCKET_PATH"] = orig_socket
+        return tx_raw_output
 
     def node_wait_for_block(
         self,
@@ -174,9 +174,10 @@ def node_wait_for_block(
 
         try:
             os.environ["CARDANO_NODE_SOCKET_PATH"] = str(new_socket)
-            return cluster_obj.wait_for_block(block=block_no)
+            new_block = cluster_obj.wait_for_block(block=block_no)
         finally:
             os.environ["CARDANO_NODE_SOCKET_PATH"] = orig_socket
+        return new_block
 
     @allure.link(helpers.get_vcs_link())
     # There's a submission delay of 60 sec. Therefore on testnet with low `securityParam`,
diff --git a/cardano_node_tests/tests/test_scripts.py b/cardano_node_tests/tests/test_scripts.py
@@ -2144,6 +2144,7 @@ def test_incremental_signing(
                     tx_name=f"{temp_template}_from{idx}_r{r}",
                 )
 
+        assert tx_signed is not None  # for pyrefly
         submit_utils.submit_tx(
             submit_method=submit_method,
             cluster_obj=cluster,
diff --git a/cardano_node_tests/tests/test_tx_basic.py b/cardano_node_tests/tests/test_tx_basic.py
@@ -1701,6 +1701,7 @@ def test_incremental_signing(
                 )
 
         # It is not possible to submit Tx with missing required skey
+        assert tx_signed is not None  # for pyrefly
         with pytest.raises((clusterlib.CLIError, submit_api.SubmitApiError)) as excinfo:
             submit_utils.submit_tx(
                 submit_method=submit_method,
diff --git a/cardano_node_tests/tests/tests_conway/test_drep.py b/cardano_node_tests/tests/tests_conway/test_drep.py
@@ -438,6 +438,8 @@ def test_register_and_retire_drep(
                 if use_build_cmd and submit_method == "cli":
 
                     def _query_func():
+                        if not drep_data:
+                            return
                         dbsync_utils.check_off_chain_drep_registration(
                             drep_data=drep_data, metadata=drep_metadata_content
                         )
@@ -587,9 +589,10 @@ def _retire_drep() -> None:
         assert drep_data and drep_data.voting_anchor_id
 
         def _query_func():
-            dbsync_utils.check_off_chain_vote_fetch_error(
-                voting_anchor_id=drep_data.voting_anchor_id
-            )
+            if drep_data and drep_data.voting_anchor_id is not None:
+                dbsync_utils.check_off_chain_vote_fetch_error(
+                    voting_anchor_id=drep_data.voting_anchor_id
+                )
 
         reqc.db021.start(url=helpers.get_vcs_link())
         dbsync_utils.retry_query(query_func=_query_func, timeout=360)
diff --git a/cardano_node_tests/tests/tests_conway/test_pparam_update.py b/cardano_node_tests/tests/tests_conway/test_pparam_update.py
@@ -682,6 +682,7 @@ def _propose_pparams_update(
             anchor_data = governance_utils.get_default_anchor_data()
             # Increment count for a submitted proposal
             nonlocal submitted_proposal_count
+            # pyrefly: ignore  # unknown-name
             submitted_proposal_count += 1
             return conway_common.propose_pparams_update(
                 cluster_obj=cluster,
diff --git a/cardano_node_tests/utils/artifacts.py b/cardano_node_tests/utils/artifacts.py
@@ -20,6 +20,7 @@
 def save_cli_coverage(cluster_obj: clusterlib.ClusterLib, pytest_config: Config) -> pl.Path | None:
     """Save CLI coverage info."""
     cli_coverage_dir = pytest_config.getoption(CLI_COVERAGE_ARG)
+    # pyrefly: ignore  # missing-attribute
     if not (cli_coverage_dir and hasattr(cluster_obj, "cli_coverage") and cluster_obj.cli_coverage):
         return None
 
diff --git a/cardano_node_tests/utils/cluster_nodes.py b/cardano_node_tests/utils/cluster_nodes.py
@@ -67,7 +67,11 @@ class ClusterType:
 
     def __init__(self) -> None:
         self.type = "unknown"
-        self.cluster_scripts = cluster_scripts.ScriptsTypes()
+        self.cluster_scripts: (
+            cluster_scripts.ScriptsTypes
+            | cluster_scripts.TestnetScripts
+            | cluster_scripts.LocalScripts
+        ) = cluster_scripts.ScriptsTypes()
 
     @property
     def testnet_type(self) -> str:
@@ -204,9 +208,7 @@ class TestnetCluster(ClusterType):
     def __init__(self) -> None:
         super().__init__()
         self.type = ClusterType.TESTNET
-        self.cluster_scripts: cluster_scripts.ScriptsTypes | cluster_scripts.TestnetScripts = (
-            cluster_scripts.TestnetScripts()
-        )
+        self.cluster_scripts = cluster_scripts.TestnetScripts()
 
         # Cached values
         self._testnet_type = ""
diff --git a/cardano_node_tests/utils/clusterlib_utils.py b/cardano_node_tests/utils/clusterlib_utils.py
@@ -629,6 +629,7 @@ def mint_or_burn_witness(
         )
 
     # Submit signed TX
+    assert tx_witnessed_file is not None  # for pyrefly
     submit_utils.submit_tx(
         submit_method=submit_method,
         cluster_obj=cluster_obj,
@@ -729,6 +730,7 @@ def mint_or_burn_sign(
         )
 
     # Submit signed transaction
+    assert out_file_signed is not None  # for pyrefly
     submit_utils.submit_tx(
         submit_method=submit_method,
         cluster_obj=cluster_obj,
@@ -880,7 +882,9 @@ def _get_ledger_state_cmd(
     # Record cli coverage
     if hasattr(cluster_obj, "cli_coverage"):
         custom_clusterlib.record_cli_coverage(
-            cli_args=cardano_cli_args, coverage_dict=cluster_obj.cli_coverage
+            cli_args=cardano_cli_args,
+            # pyrefly: ignore  # missing-attribute
+            coverage_dict=cluster_obj.cli_coverage,
         )
 
     return ledger_state_cmd
diff --git a/cardano_node_tests/utils/dbsync_check_tx.py b/cardano_node_tests/utils/dbsync_check_tx.py
@@ -367,12 +367,14 @@ def check_tx_collaterals(
         and not (tx_raw_output.total_collateral_amount or tx_raw_output.return_collateral_txouts)
     ):
         protocol_params = cluster_obj.g_query.get_protocol_params()
+        # pyrefly: ignore  # no-matching-overload, bad-argument-type
         tx_collaterals_amount = clusterlib.calculate_utxos_balance(utxos=list(tx_collaterals))
         tx_collateral_output_amount = int(
             tx_collaterals_amount
             - tx_raw_output.fee * protocol_params["collateralPercentage"] / 100
         )
         db_collateral_output_amount = clusterlib.calculate_utxos_balance(
+            # pyrefly: ignore  # no-matching-overload, bad-argument-type
             utxos=list(response.collateral_outputs)
         )
 
diff --git a/cardano_node_tests/utils/dbsync_service_manager.py b/cardano_node_tests/utils/dbsync_service_manager.py
@@ -98,11 +98,6 @@ class Table(enum.StrEnum):
     VOTING_ANCHOR = "voting_anchor"
     VOTING_PROCEDURE = "voting_procedure"
 
-    @classmethod
-    def get_all_tables(cls) -> list[str]:
-        """Return all table names as a list of strings."""
-        return [table.value for table in cls]
-
 
 class Column:
     class Tx(enum.StrEnum):
@@ -395,7 +390,7 @@ def update_config(self, config: dict | DBSyncConfigBuilder | None) -> pl.Path:
         LOGGER.debug(f"Effective config: {config}")
 
         with open(config_file, encoding="utf-8") as fp_in:
-            current_dbsync_config = yaml.safe_load(fp_in) or {}
+            current_dbsync_config: dict = yaml.safe_load(fp_in) or {}
 
         current_dbsync_config["insert_options"] = config
 
diff --git a/cardano_node_tests/utils/helpers.py b/cardano_node_tests/utils/helpers.py
@@ -100,11 +100,14 @@ def run_command(
     stdout, stderr = p.communicate()
 
     if not ignore_fail and p.returncode != 0:
+        # pyrefly: ignore  # missing-attribute
         err_dec = stderr.decode()
+        # pyrefly: ignore  # missing-attribute
         err_dec = err_dec or stdout.decode()
         msg = f"An error occurred while running `{cmd_str}`: {err_dec}"
         raise AssertionError(msg)
 
+    # pyrefly: ignore  # bad-return
     return stdout
 
 
@@ -293,7 +296,9 @@ def flatten(
             first = next(remainder)
         except StopIteration:
             break
+        # pyrefly: ignore  # invalid-argument
         if isinstance(first, ltypes) and not isinstance(first, (str, bytes)):
+            # pyrefly: ignore  # bad-argument-type
             remainder = itertools.chain(first, remainder)
         else:
             yield first
diff --git a/cardano_node_tests/utils/requirements.py b/cardano_node_tests/utils/requirements.py
@@ -100,17 +100,15 @@ def collect_executed_req(base_dir: pl.Path) -> dict:
             req_rec = json.load(in_fp)
 
         group_name = req_rec["group"]
-        group_collected = collected.get(group_name)
+        group_collected: dict = collected.get(group_name) or {}
         if not group_collected:
-            group_collected = {}
             collected[group_name] = group_collected
 
         req_id = req_rec["id"]
-        id_collected = group_collected.get(req_id)
+        id_collected: dict = group_collected.get(req_id) or {}
         if id_collected and id_collected["status"] == Statuses.success.name:
             continue
         if not id_collected:
-            id_collected = {}
             group_collected[req_id] = id_collected
         id_collected["status"] = req_rec["status"]
         id_collected["url"] = req_rec.get("url") or ""
@@ -126,8 +124,12 @@ def merge_reqs(*reqs: dict[str, dict]) -> dict:
             merged_group = merged.get(gname) or {}
             for req_id, req_data in greqs.items():
                 merged_rec = merged_group.get(req_id) or {}
-                merged_status_val = Statuses[merged_rec.get("status") or "uncovered"].value
+                merged_status_key = merged_rec.get("status") or "uncovered"
+                # pyrefly: ignore  # bad-specialization, not-a-type
+                merged_status_val = Statuses[merged_status_key].value
+                # pyrefly: ignore  # bad-specialization
                 req_status_val = Statuses[req_data["status"]].value
+                # pyrefly: ignore  # missing-attribute
                 if not merged_rec or req_status_val < merged_status_val:
                     merged_group[req_id] = req_data
             merged[gname] = merged_group
@@ -160,6 +162,7 @@ def get_mapped_req(mapping: pl.Path, executed_req: dict) -> dict:  # noqa: C901
             dependencies_failures = []
 
             for p_req in dependencies:
+                # pyrefly: ignore  # no-matching-overload, bad-argument-type
                 p_status = executed_group.get(p_req, {}).get("status")
 
                 if not url:
diff --git a/cardano_node_tests/utils/versions.py b/cardano_node_tests/utils/versions.py
@@ -40,7 +40,7 @@ def __init__(self) -> None:
         self.command_era_name = configuration.COMMAND_ERA
 
         if self.command_era_name and self.command_era_name in self.MAP.values():
-            self.transaction_era_name = self.command_era_name
+            self.transaction_era_name: str = self.command_era_name
         else:
             self.transaction_era_name = self.MAP[self.DEFAULT_TX_ERA]
 
diff --git a/poetry.lock b/poetry.lock
diff --git a/pyproject.toml b/pyproject.toml

Original file line number	Diff line number	Diff line change
`@@ -2144,6 +2144,7 @@ def test_incremental_signing(`
`2144`	`2144`	`tx_name=f"{temp_template}_from{idx}_r{r}",`
`2145`	`2145`	`)`
`2146`	`2146`
	`2147`	`+ assert tx_signed is not None # for pyrefly`
`2147`	`2148`	`submit_utils.submit_tx(`
`2148`	`2149`	`submit_method=submit_method,`
`2149`	`2150`	`cluster_obj=cluster,`
Original file line number	Diff line number	Diff line change
`@@ -1701,6 +1701,7 @@ def test_incremental_signing(`
`1701`	`1701`	`)`
`1702`	`1702`
`1703`	`1703`	`# It is not possible to submit Tx with missing required skey`
	`1704`	`+ assert tx_signed is not None # for pyrefly`
`1704`	`1705`	`with pytest.raises((clusterlib.CLIError, submit_api.SubmitApiError)) as excinfo:`
`1705`	`1706`	`submit_utils.submit_tx(`
`1706`	`1707`	`submit_method=submit_method,`