pre-commit: pyupgrade (#658)

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

.pre-commit-config.yaml

@@ -34,6 +34,13 @@ repos:
       - id: check-added-large-files
       - id: detect-private-key
 
+  - repo: https://github.com/asottile/pyupgrade
+    rev: v2.23.0
+    hooks:
+      - id: pyupgrade
+        args: [--py36-plus]
+        name: Upgrade code
+
   - repo: https://github.com/PyCQA/isort
     rev: 5.9.3
     hooks:

docs/source/conf.py

@@ -140,7 +140,7 @@ def setup(app):
 # https://stackoverflow.com/questions/15889621/sphinx-how-to-exclude-imports-in-automodule
 def _package_list_from_file(pfile):
     assert os.path.isfile(pfile)
-    with open(pfile, "r") as fp:
+    with open(pfile) as fp:
         lines = fp.readlines()
     list_pkgs = []
     for ln in lines:

flash/core/data/data_source.py

@@ -684,9 +684,7 @@ def predict_load_data(data: SampleCollection, dataset: Optional[Any] = None) ->
     def _validate(self, data):
         label_type = data._get_label_field_type(self.label_field)
         if not issubclass(label_type, self.label_cls):
-            raise ValueError(
-                "Expected field '%s' to have type %s; found %s" % (self.label_field, self.label_cls, label_type)
-            )
+            raise ValueError(f"Expected field '{self.label_field}' to have type {self.label_cls}; found {label_type}")
 
     def _get_classes(self, data):
         classes = data.classes.get(self.label_field, None)

flash/core/data/transforms.py

@@ -31,7 +31,7 @@ class ApplyToKeys(nn.Sequential):
     """
 
     def __init__(self, keys: Union[str, Sequence[str]], *args):
-        super().__init__(*[convert_to_modules(arg) for arg in args])
+        super().__init__(*(convert_to_modules(arg) for arg in args))
         if isinstance(keys, str):
             keys = [keys]
         self.keys = keys
@@ -72,7 +72,7 @@ class KorniaParallelTransforms(nn.Sequential):
     """
 
     def __init__(self, *args):
-        super().__init__(*[convert_to_modules(arg) for arg in args])
+        super().__init__(*(convert_to_modules(arg) for arg in args))
 
     def forward(self, inputs: Any):
         result = list(inputs) if isinstance(inputs, Sequence) else [inputs]

flash/core/serve/component.py

@@ -94,12 +94,12 @@ def _validate_model_args(
         raise ValueError(f"Iterable args={args} must have length >= 1")
 
     if isinstance(args, (list, tuple)):
-        if not all((isinstance(x, _Servable_t) for x in args)):
+        if not all(isinstance(x, _Servable_t) for x in args):
             raise TypeError(f"One of arg in args={args} is not type {_Servable_t}")
     elif isinstance(args, dict):
-        if not all((isinstance(x, str) for x in args.keys())):
+        if not all(isinstance(x, str) for x in args.keys()):
             raise TypeError(f"One of keys in args={args.keys()} is not type {str}")
-        if not all((isinstance(x, _Servable_t) for x in args.values())):
+        if not all(isinstance(x, _Servable_t) for x in args.values()):
             raise TypeError(f"One of values in args={args} is not type {_Servable_t}")
     elif not isinstance(args, _Servable_t):
         raise TypeError(f"Args must be instance, list/tuple, or mapping of {_Servable_t}")

flash/core/serve/dag/rewrite.py

@@ -189,7 +189,7 @@ def _apply(self, sub_dict):
         return term
 
     def __str__(self):
-        return "RewriteRule({0}, {1}, {2})".format(self.lhs, self.rhs, self.vars)
+        return f"RewriteRule({self.lhs}, {self.rhs}, {self.vars})"
 
     def __repr__(self):
         return str(self)

flash/core/serve/dag/task.py

@@ -41,12 +41,10 @@ def preorder_traversal(task):
 
     for item in task:
         if istask(item):
-            for i in preorder_traversal(item):
-                yield i
+            yield from preorder_traversal(item)
         elif isinstance(item, list):
             yield list
-            for i in preorder_traversal(item):
-                yield i
+            yield from preorder_traversal(item)
         else:
             yield item
 
@@ -222,8 +220,7 @@ def flatten(seq, container=list):
     else:
         for item in seq:
             if isinstance(item, container):
-                for item2 in flatten(item, container=container):
-                    yield item2
+                yield from flatten(item, container=container)
             else:
                 yield item
 

flash/core/serve/types/label.py

@@ -32,7 +32,7 @@ def __post_init__(self):
                     "Must provide either classes as a list or " "path to a text file that contains classes"
                 )
             with Path(self.path).open(mode="r") as f:
-                self.classes = tuple([item.strip() for item in f.readlines()])
+                self.classes = tuple(item.strip() for item in f.readlines())
         if isinstance(self.classes, dict):
             self._reverse_map = {}
             for key, value in self.classes.items():

flash/core/serve/types/repeated.py

@@ -50,7 +50,7 @@ def __post_init__(self):
     def deserialize(self, *args: Dict) -> Tuple[Tensor, ...]:
         if (self.max_len is not None) and (len(args) > self.max_len):
             raise ValueError(f"len(arg)={len(args)} > self.max_len={self.max_len}")
-        return tuple((self.dtype.deserialize(**item) for item in args))
+        return tuple(self.dtype.deserialize(**item) for item in args)
 
     def packed_deserialize(self, args):
         """Arguments are positional arguments for deserialize, unlike other datatypes."""
@@ -59,4 +59,4 @@ def packed_deserialize(self, args):
     def serialize(self, args: Sequence) -> Tuple[Any, ...]:
         if (self.max_len is not None) and (len(args) > self.max_len):
             raise ValueError(f"len(arg)={len(args)} > self.max_len={self.max_len}")
-        return tuple((self.dtype.serialize(item) for item in args))
+        return tuple(self.dtype.serialize(item) for item in args)

flash/image/classification/backbones/resnet.py

@@ -62,7 +62,7 @@ def __init__(
         dilation: int = 1,
         norm_layer: Optional[Callable[..., nn.Module]] = None,
     ) -> None:
-        super(BasicBlock, self).__init__()
+        super().__init__()
         if norm_layer is None:
             norm_layer = nn.BatchNorm2d
         if groups != 1 or base_width != 64:
@@ -118,7 +118,7 @@ def __init__(
         dilation: int = 1,
         norm_layer: Optional[Callable[..., nn.Module]] = None,
     ) -> None:
-        super(Bottleneck, self).__init__()
+        super().__init__()
         if norm_layer is None:
             norm_layer = nn.BatchNorm2d
         width = int(planes * (base_width / 64.0)) * groups
@@ -171,7 +171,7 @@ def __init__(
         remove_first_maxpool: bool = False,
     ) -> None:
 
-        super(ResNet, self).__init__()
+        super().__init__()
 
         if norm_layer is None:
             norm_layer = nn.BatchNorm2d
@@ -320,7 +320,7 @@ def _resnet(
     model_weights = None
     if pretrained_flag:
         if "supervised" not in weights_paths:
-            raise KeyError("Supervised pretrained weights not available for {0}".format(model_name))
+            raise KeyError(f"Supervised pretrained weights not available for {model_name}")
 
         model_weights = load_state_dict_from_url(
             weights_paths["supervised"], map_location=torch.device("cpu") if device == -1 else torch.device(device)

flash/image/detection/serialization.py

@@ -87,7 +87,7 @@ def serialize(self, sample: Dict[str, Any]) -> Union[Detections, Dict[str, Any]]
             if self.threshold is not None and confidence < self.threshold:
                 continue
 
-            xmin, ymin, xmax, ymax = [c.tolist() for c in det["boxes"]]
+            xmin, ymin, xmax, ymax = (c.tolist() for c in det["boxes"])
             box = [
                 xmin / width,
                 ymin / height,

flash/image/segmentation/data.py

@@ -345,7 +345,7 @@ def from_data_source(
         if flash._IS_TESTING:
             data_fetcher.block_viz_window = True
 
-        dm = super(SemanticSegmentationData, cls).from_data_source(
+        dm = super().from_data_source(
             data_source=data_source,
             train_data=train_data,
             val_data=val_data,

flash/pointcloud/detection/open3d_ml/data_sources.py

@@ -55,7 +55,7 @@ def load_meta(self, root_dir, dataset: Optional[BaseAutoDataset]):
         if not exists(meta_file):
             raise MisconfigurationException(f"The {root_dir} should contain a `meta.yaml` file about the classes.")
 
-        with open(meta_file, "r") as f:
+        with open(meta_file) as f:
             self.meta = yaml.safe_load(f)
 
         if "label_to_names" not in self.meta:

flash/pointcloud/segmentation/open3d_ml/sequences_dataset.py

@@ -77,13 +77,13 @@ def load_meta(self, root_dir):
                 f"The {root_dir} should contain a `meta.yaml` file about the pointcloud sequences."
             )
 
-        with open(meta_file, "r") as f:
+        with open(meta_file) as f:
             self.meta = yaml.safe_load(f)
 
         self.label_to_names = self.get_label_to_names()
         self.num_classes = len(self.label_to_names)
 
-        with open(meta_file, "r") as f:
+        with open(meta_file) as f:
             self.meta = yaml.safe_load(f)
 
         remap_dict_val = self.meta["learning_map"]
@@ -169,7 +169,7 @@ def get_attr(self, idx):
         pc_path = self.path_list[idx]
         dir, file = split(pc_path)
         _, seq = split(split(dir)[0])
-        name = "{}_{}".format(seq, file[:-4])
+        name = f"{seq}_{file[:-4]}"
 
         pc_path = str(pc_path)
         attr = {"idx": idx, "name": name, "path": pc_path, "split": self.split}

flash/setup_tools.py

@@ -20,7 +20,7 @@
 
 
 def _load_requirements(path_dir: str, file_name: str = "requirements.txt", comment_chars: str = "#@") -> List[str]:
-    with open(os.path.join(path_dir, file_name), "r") as file:
+    with open(os.path.join(path_dir, file_name)) as file:
         lines = [ln.strip() for ln in file.readlines()]
     reqs = []
     for ln in lines:

flash/text/seq2seq/core/metrics.py

@@ -217,7 +217,7 @@ def aggregate(self):
             # Percentiles are returned as (interval, measure).
             percentiles = self._bootstrap_resample(score_matrix)
             # Extract the three intervals (low, mid, high).
-            intervals = tuple((Score(*percentiles[j, :]) for j in range(3)))
+            intervals = tuple(Score(*percentiles[j, :]) for j in range(3))
             result[score_type] = AggregateScore(low=intervals[0], mid=intervals[1], high=intervals[2])
         return result
 

flash/video/classification/data.py

@@ -54,7 +54,7 @@
 _PYTORCHVIDEO_DATA = Dict[str, Union[str, torch.Tensor, int, float, List]]
 
 
-class BaseVideoClassification(object):
+class BaseVideoClassification:
     def __init__(
         self,
         clip_sampler: "ClipSampler",

tests/core/serve/test_dag/test_order.py

@@ -20,14 +20,14 @@ def f(*args):
 
 def test_ordering_keeps_groups_together(abcde):
     a, b, c, d, e = abcde
-    d = dict(((a, i), (f,)) for i in range(4))
+    d = {(a, i): (f,) for i in range(4)}
     d.update({(b, 0): (f, (a, 0), (a, 1)), (b, 1): (f, (a, 2), (a, 3))})
     o = order(d)
 
     assert abs(o[(a, 0)] - o[(a, 1)]) == 1
     assert abs(o[(a, 2)] - o[(a, 3)]) == 1
 
-    d = dict(((a, i), (f,)) for i in range(4))
+    d = {(a, i): (f,) for i in range(4)}
     d.update({(b, 0): (f, (a, 0), (a, 2)), (b, 1): (f, (a, 1), (a, 3))})
     o = order(d)
 
@@ -220,7 +220,7 @@ def test_prefer_deep(abcde):
 
 
 def test_stacklimit(abcde):
-    dsk = dict(("x%s" % (i + 1), (inc, "x%s" % i)) for i in range(10000))
+    dsk = {"x%s" % (i + 1): (inc, "x%s" % i) for i in range(10000)}
     dependencies, dependents = get_deps(dsk)
     ndependencies(dependencies, dependents)
 
@@ -280,7 +280,7 @@ def test_run_smaller_sections(abcde):
     Prefer to run acb first because then we can get that out of the way
     """
     a, b, c, d, e = abcde
-    aa, bb, cc, dd = [x * 2 for x in [a, b, c, d]]
+    aa, bb, cc, dd = (x * 2 for x in [a, b, c, d])
 
     expected = [a, c, b, e, d, cc, bb, aa, dd]
 
@@ -325,9 +325,9 @@ def test_local_parents_of_reduction(abcde):
     Prefer to finish a1 stack before proceeding to b2
     """
     a, b, c, d, e = abcde
-    a1, a2, a3 = [a + i for i in "123"]
-    b1, b2, b3 = [b + i for i in "123"]
-    c1, c2, c3 = [c + i for i in "123"]
+    a1, a2, a3 = (a + i for i in "123")
+    b1, b2, b3 = (b + i for i in "123")
+    c1, c2, c3 = (c + i for i in "123")
 
     expected = [a3, a2, a1, b3, b2, b1, c3, c2, c1]
 
@@ -368,8 +368,8 @@ def test_nearest_neighbor(abcde):
     This is difficult because all groups are connected.
     """
     a, b, c, _, _ = abcde
-    a1, a2, a3, a4, a5, a6, a7, a8, a9 = [a + i for i in "123456789"]
-    b1, b2, b3, b4 = [b + i for i in "1234"]
+    a1, a2, a3, a4, a5, a6, a7, a8, a9 = (a + i for i in "123456789")
+    b1, b2, b3, b4 = (b + i for i in "1234")
 
     dsk = {
         b1: (f,),

tests/examples/utils.py

@@ -21,7 +21,7 @@ def call_script(
     args: Optional[List[str]] = None,
     timeout: Optional[int] = 60 * 10,
 ) -> Tuple[int, str, str]:
-    with open(filepath, "r") as original:
+    with open(filepath) as original:
         data = original.read()
 
     with open(filepath, "w") as modified:

tests/image/test_backbones.py

@@ -67,7 +67,7 @@ def test_pretrained_weights_registry(backbone, pretrained, expected_num_features
     ],
 )
 def test_wide_resnets(backbone, pretrained):
-    with pytest.raises(KeyError, match="Supervised pretrained weights not available for {0}".format(backbone)):
+    with pytest.raises(KeyError, match=f"Supervised pretrained weights not available for {backbone}"):
         IMAGE_CLASSIFIER_BACKBONES.get(backbone)(pretrained=pretrained)