Convert directory fbcode/kats to use the Ruff Formatter

Summary:
Converts the directory specified to use the Ruff formatter in pyfmt

ruff_dog

If this diff causes merge conflicts when rebasing, please run
`hg status -n -0 --change . -I '**/*.{py,pyi}' | xargs -0 arc pyfmt`
on your diff, and amend any changes before rebasing onto latest.
That should help reduce or eliminate any merge conflicts.

allow-large-files

Reviewed By: amyreese

Differential Revision: D66250543

fbshipit-source-id: 1b9ae94cb20f7a1b66468ec76dcf9a1a6735ad89

kats/detectors/bocpd.py

@@ -1201,7 +1201,6 @@ class _NormalKnownPrec(_PredictiveModel):
     _data_shape: Union[int, Tuple[int, int]]
 
     def __init__(self, data: TimeSeriesData, parameters: NormalKnownParameters) -> None:
-
         # \mu \sim N(\mu0, \frac{1}{\lambda0})
         # x \sim N(\mu,\frac{1}{\lambda})
 
@@ -1560,7 +1559,6 @@ def _sample_bayesian_linreg(
         b_n: float,
         num_samples: int,
     ) -> Tuple[npt.NDArray, npt.NDArray]:
-
         # this is to make sure the results are consistent
         # and tests don't break randomly
         seed_value = 100

kats/detectors/cusum_detection.py

@@ -780,7 +780,6 @@ def detector(self, **kwargs: Any) -> List[CUSUMChangePoint]:
         # multivariate detection. We keep using change_direction = "increase"
         # here to be consistent with the univariate detector.
         for change_direction in ["increase"]:
-
             change_meta = self._get_change_point(
                 ts,
                 max_iter=max_iter,
@@ -816,23 +815,25 @@ def _get_llr(
         sigma0: Optional[float],
         sigma1: Optional[float],
     ) -> float:
-
         mu_tilde = np.mean(ts, axis=0)
         sigma_pooled = np.cov(ts, rowvar=False)
-        llr = -2 * (
-            self._log_llr_multi(
-                ts[: (changepoint + 1)],
-                mu_tilde,
-                sigma_pooled,
-                mu0,
-                sigma0,  # pyre-fixme
-            )
-            - self._log_llr_multi(
-                ts[(changepoint + 1) :],
-                mu_tilde,
-                sigma_pooled,
-                mu1,
-                sigma1,  # pyre-fixme
+        llr = (
+            -2
+            * (
+                self._log_llr_multi(
+                    ts[: (changepoint + 1)],
+                    mu_tilde,
+                    sigma_pooled,
+                    mu0,
+                    sigma0,  # pyre-fixme
+                )
+                - self._log_llr_multi(
+                    ts[(changepoint + 1) :],
+                    mu_tilde,
+                    sigma_pooled,
+                    mu1,
+                    sigma1,  # pyre-fixme
+                )
             )
         )
         return llr
@@ -868,7 +869,6 @@ def _get_change_point(
         start_point: int,
         change_direction: str = "increase",
     ) -> CUSUMChangePointVal:
-
         # locate the change point using cusum method
         changepoint_func = np.argmin
         n = 0

kats/detectors/cusum_model.py

@@ -861,7 +861,6 @@ def fit_predict(
             change_tsd.extend(change_tsd_vec, validate=False)
 
         else:
-
             for start_time in pd.date_range(
                 anomaly_start_time,
                 min(

kats/detectors/detector_consts.py

@@ -518,7 +518,6 @@ class ConfidenceBand:
 
 
 class AnomalyResponse:
-
     key_mapping: List[str]
     num_series: int
 
@@ -625,7 +624,7 @@ def inplace_update(
         self._inplace_update_ts(self.scores, time, score)
         cb = self.confidence_band
         if cb is not None:
-            self._inplace_update_ts(cb.lower, time, ci_lower),
+            (self._inplace_update_ts(cb.lower, time, ci_lower),)
             self._inplace_update_ts(cb.upper, time, ci_upper)
 
         if self.predicted_ts is not None:

kats/detectors/distribution_distance_model.py

@@ -87,7 +87,6 @@ def _percentile_to_prob(
 
         # if all values in l2 are lager than l1
         if n_idx_at2 == 0:
-
             # use distance from l2[0] to l2[j] as base
             j = 1
             while j < len(l2_merge) and l2_merge[j] == l2_merge[0]:

kats/detectors/dtwcpd.py

@@ -112,7 +112,6 @@ def __init__(
         min_value: float = 1e-9,  # Controls when a value is considered to be zero
         match_against_same_time_series: bool = False,  # Whether to allow historical matches in the same time series.
     ) -> None:
-
         self.data = data
         self.sliding_window_size = sliding_window_size
         self.MIN_TS_VALUE = min_value
@@ -263,7 +262,6 @@ def LB_Keogh(s1: List[float], s2: List[float], w: int) -> float:
         # TODO: vectorize
         LB_sum = 0.0
         for ind, i in enumerate(s1):
-
             # Rolling min/max
             lower_bound = min(s2[(ind - w if ind - w >= 0 else 0) : (ind + w)])
             upper_bound = max(s2[(ind - w if ind - w >= 0 else 0) : (ind + w)])

kats/detectors/meta_learning/metalearning_detection_model.py

@@ -64,7 +64,6 @@ def metadata_detect_preprocessor(
     rawdata: pd.DataFrame,
     params_to_scale_down: Set[str] = PARAMS_TO_SCALE_DOWN,
 ) -> List[Dict[str, Any]]:
-
     rawdata["features"] = rawdata["features"].map(change_str_to_dict)
     rawdata["features"] = rawdata["features"].map(change_dtype)
     rawdata["hpt_res"] = rawdata["hpt_res"].map(change_str_to_dict)

kats/detectors/multivariate_detector.py

@@ -9,6 +9,7 @@
 """
 This module implements the multivariate Outlier Detection algorithm as a Detector Model.
 """
+
 import json
 from typing import Any, Optional
 

kats/detectors/outlier_detector.py

@@ -9,6 +9,7 @@
 """
 This module implements the univariate Outlier Detection algorithm as a Detector Model.
 """
+
 import json
 from typing import Any, Optional
 

kats/detectors/prophet_detector.py

@@ -59,7 +59,6 @@ class SilentStdoutStderr(object):
     stdout, stderr = sys.__stdout__.fileno(), sys.__stderr__.fileno()  # type: ignore
 
     def __enter__(self) -> None:
-
         # pyre-fixme typing # type: ignore
         self.devnull = os.open(os.devnull, os.O_RDWR)
         # pyre-fixme typing
@@ -206,9 +205,8 @@ def seasonalities_to_dict(
         List[SeasonalityTypes],
         List[str],
         Dict[SeasonalityTypes, Union[bool, str]],
-    ]
+    ],
 ) -> Dict[SeasonalityTypes, Union[bool, str]]:
-
     if isinstance(seasonalities, SeasonalityTypes):
         seasonalities = {seasonalities: True}
     elif isinstance(seasonalities, list):
@@ -599,9 +597,9 @@ def predict(
             )
             if holidays_df is not None:
                 scores_ts = pd.Series(list(scores.value), index=data.time)
-                scores_ts.loc[
-                    scores_ts.index.floor("d").isin(holidays_df)
-                ] *= self.holiday_multiplier
+                scores_ts.loc[scores_ts.index.floor("d").isin(holidays_df)] *= (
+                    self.holiday_multiplier
+                )
                 scores = TimeSeriesData(
                     time=pd.Series(scores_ts.index), value=scores_ts
                 )

kats/detectors/residual_translation.py

@@ -14,6 +14,7 @@
 asymmetric). This module “learns” the distribution of the residual (using kernel
 density estimation), and outputs a false-alarm probability based on it.
 """
+
 from __future__ import annotations
 
 from typing import Optional

kats/detectors/stat_sig_detector.py

@@ -101,7 +101,6 @@ def __init__(
         anomaly_scores_only: bool = False,
         min_perc_change: float = 0.0,
     ) -> None:
-
         if serialized_model:
             model_dict = json.loads(serialized_model)
             self.n_test: int = model_dict["n_test"]
@@ -999,7 +998,6 @@ def __init__(
         use_corrected_scores: bool = False,
         min_perc_change: float = 0.0,
     ) -> None:
-
         StatSigDetectorModel.__init__(
             self,
             n_control=n_control,
@@ -1180,7 +1178,6 @@ def fit_predict(
         return self.response.get_last_n(self.last_N)
 
     def _init_response(self, data: TimeSeriesData) -> None:
-
         zeros_df = pd.DataFrame(
             {
                 **{"time": data.time},

kats/models/bayesian_var.py

@@ -6,14 +6,14 @@
 # pyre-strict
 
 """
- Bayesian estimation of Vector Autoregressive Model using
- Minnesota prior on the coefficient matrix. This version is
- useful for regularization when they are too many coefficients
- to be estimated.
-
- Implementation inspired by the following two articles/papers:
-    https://www.mathworks.com/help/econ/normalbvarm.html#mw_4a1ab118-9ef3-4380-8c5a-12b848254117
-    http://apps.eui.eu/Personal/Canova/Articles/ch10.pdf (page 5)
+Bayesian estimation of Vector Autoregressive Model using
+Minnesota prior on the coefficient matrix. This version is
+useful for regularization when they are too many coefficients
+to be estimated.
+
+Implementation inspired by the following two articles/papers:
+   https://www.mathworks.com/help/econ/normalbvarm.html#mw_4a1ab118-9ef3-4380-8c5a-12b848254117
+   http://apps.eui.eu/Personal/Canova/Articles/ch10.pdf (page 5)
 """
 
 import logging
@@ -208,11 +208,12 @@ def fit(self) -> None:
             ]  # shape: [m * (m * p + r + 1)] x 1
 
             assert (
-                num_mu,
-                num_mu,
-            ) == z_sum_term.shape, (
-                f"Expected {(num_mu, num_mu)}, got {z_sum_term.shape}"
-            )
+                (
+                    num_mu,
+                    num_mu,
+                )
+                == z_sum_term.shape
+            ), f"Expected {(num_mu, num_mu)}, got {z_sum_term.shape}"
             assert (
                 num_mu,
             ) == y_sum_term.shape, f"Expected {(num_mu,)}, got {y_sum_term.shape}"
@@ -258,11 +259,12 @@ def _construct_Zt(self, X: npt.NDArray, Y: npt.NDArray, t: int) -> npt.NDArray:
         Z_t = block_diag(*([z] * self.m))
 
         assert (
-            self.m,
-            self.num_mu_coefficients,
-        ) == Z_t.shape, (
-            f"Expected {(self.m, self.num_mu_coefficients)}, got {Z_t.shape}"
-        )
+            (
+                self.m,
+                self.num_mu_coefficients,
+            )
+            == Z_t.shape
+        ), f"Expected {(self.m, self.num_mu_coefficients)}, got {Z_t.shape}"
 
         return Z_t  # shape: m x [m * (m * p + m + 1)]
 

kats/models/ensemble/median_ensemble.py

@@ -59,8 +59,9 @@ def predict(self, steps: int, **kwargs) -> pd.DataFrame:
         """
 
         logging.debug(
-            "Call predict() with parameters. "
-            "steps:{steps}, kwargs:{kwargs}".format(steps=steps, kwargs=kwargs)
+            "Call predict() with parameters. " "steps:{steps}, kwargs:{kwargs}".format(
+                steps=steps, kwargs=kwargs
+            )
         )
         # Keep freq in the parameters passed to _predict_all()
         self.freq = freq = kwargs.get("freq", "D")

kats/models/ensemble/weighted_avg_ensemble.py

@@ -12,6 +12,7 @@
 back testing results, i.e., model with better performance should have higher
 weight.
 """
+
 import logging
 import sys
 from multiprocessing import cpu_count, Pool

kats/models/globalmodel/backtester.py

@@ -69,7 +69,6 @@ def __init__(
         earliest_timestamp: Union[str, pd.Timestamp, None] = None,
         max_core: Optional[int] = None,
     ) -> None:
-
         if not isinstance(gmparam, GMParam):
             msg = f"gmparam should be GMParam object but receives {type(gmparam)}."
             logging.error(msg)

kats/models/globalmodel/data_processor.py

@@ -207,7 +207,6 @@ def __init__(
         ] = None,
         mode: str = "train",
     ) -> None:
-
         if not isinstance(params, GMParam):
             msg = f"params should be a GMParam object but receives {type(params)}."
             logging.error(msg)

kats/models/globalmodel/ensemble.py

@@ -60,7 +60,6 @@ def __init__(
         multi: bool = False,
         max_core: Optional[int] = None,
     ) -> None:
-
         if not isinstance(gmparam, GMParam):
             msg = f"gmparam should be GMParam object but receives {type(gmparam)}."
             logging.error(msg)
@@ -210,7 +209,6 @@ def train(
         split_data = split(self.splits, self.overlap, train_TSs, valid_TSs)
         # multi processing
         if self.multi:
-
             t0 = time.time()
             rds = np.random.randint(1, int(10000 * self.model_num), self.model_num)
             model_params = [

kats/models/globalmodel/model.py

@@ -55,7 +55,6 @@ class GMModel:
 
     # pyre-fixme[3]: Return type must be annotated.
     def __init__(self, params: GMParam):
-
         if not isinstance(params, GMParam):
             msg = f"params should be a GMParam object but receives {type(params)}."
             logging.error(msg)
@@ -321,7 +320,6 @@ def _process(
         x_t = xi_t / anchor_level
 
         if period > 1:
-
             input_season = torch.cat(seasonality[idx - input_window : idx], dim=1)
             x_t = x_t / input_season
 
@@ -544,7 +542,6 @@ def _format_fcst(
             actual = np.column_stack(fcst_store["actual"])[:, :steps]
 
         for i, idx in enumerate(ids):
-
             df = pd.DataFrame(
                 fcst[i].transpose()[:steps,],
                 columns=cols,
@@ -607,7 +604,6 @@ def predict(
         self._set_nn_status("test")
         fcst_collects = {}
         for i in range(m):
-
             self._reset_nn_states()
 
             ids = dl.get_batch(batch_size)
@@ -1083,7 +1079,6 @@ def _single_pass_s2s(
         prev_idx = 0
 
         while cur_step < total_step_num:
-
             cur_idx = batch.indices[cur_step]
             is_valid = cur_idx >= first_valid_idx
 
@@ -1125,7 +1120,6 @@ def _single_pass_s2s(
             self._valid_tensor(tmp_encode, "encoder_output")
 
             if training_mode or is_valid:
-
                 # pyre-fixme[29]: `Union[Module, Tensor]` is not a function.
                 encoder.prepare_decoder(decoder)
                 encoder_step = (

kats/models/globalmodel/serialize.py

@@ -77,7 +77,6 @@ def global_model_to_json(gme: Union[GMModel, GMEnsemble]) -> str:
 
 
 def load_global_model_from_json(json_str: str) -> Union[GMModel, GMEnsemble]:
-
     param_dict = json.loads(json_str)
 
     # string for GMEnsemble