Fix dataset versioning

Co-authored-by: Gustavo Viera López <[email protected]>

pactus/dataset/dataset.py

@@ -264,7 +264,7 @@ class Dataset(Data):
         Dataset version.
     """
 
-    _last_tag = None
+    _last_tag: Union[str, None] = None
 
     def __init__(
         self,
@@ -277,7 +277,7 @@ def __init__(
         self.version = version
         self.trajs = trajs
         self.labels = labels
-        super().__init__(trajs, labels)
+        super().__init__(trajs, labels, dataset_name=name)
 
     def __len__(self):
         return len(self.trajs)
@@ -286,13 +286,12 @@ def __len__(self):
     def _from_json(name: str, data: dict) -> Dataset:
         assert "trajs" in data, "trajs not found in dataset"
         assert "labels" in data, "labels not found in dataset"
-        assert "version" in data, "version not found in dataset"
         trajs = [JSONSerializer.from_json(traj) for traj in data["trajs"]]
         return Dataset(
             name=name,
             trajs=trajs,
             labels=data["labels"],
-            version=data["version"],
+            version=data.get("version", 0),
         )
 
     @staticmethod
@@ -315,7 +314,7 @@ def _from_url(name: str, force: bool = False) -> Dataset:
         if not force and yupi_data_file.exists():
             with open(yupi_data_file, "r", encoding="utf-8") as yupi_fd:
                 data = json.load(yupi_fd)
-                if "trajs" in data and "labels" in data and "version" in data:
+                if "trajs" in data and "labels" in data:
                     return Dataset._from_json(name, data)
             logging.warning("Invalid dataset file, downloading again")
 

pactus/models/evaluation_comparison.py

@@ -45,7 +45,7 @@ def __init__(self, evals: List[Evaluation]):
 
         self.evals_by_dataset: Dict[str, List[Evaluation]] = {}
         for evaluation in self.evals:
-            ds_name = evaluation.dataset.name
+            ds_name = evaluation.dataset_name
             self.evals_by_dataset[ds_name] = self.evals_by_dataset.get(ds_name, []) + [
                 evaluation
             ]

pactus/models/lstm_model.py

@@ -5,7 +5,6 @@
 import numpy as np
 from sklearn.preprocessing import LabelEncoder
 from tensorflow import keras
-from tensorflow.keras import layers
 from yupi import Trajectory
 
 from pactus import Dataset
@@ -65,25 +64,27 @@ def _get_model(self, input_shape, n_classes):
         max_len, traj_dim = input_shape
         model = keras.Sequential()
         model.add(
-            layers.Masking(
-                mask_value=self.masking_value, input_shape=(max_len, traj_dim)
+            keras.layers.Masking(
+                mask_value=self.masking_value,
+                input_shape=(max_len, traj_dim),
             )
         )
         for units_val in self.units:
             model.add(
-                layers.LSTM(
+                keras.layers.LSTM(
                     units_val,
                     input_shape=(max_len, traj_dim),
                     return_sequences=True,
                 )
             )
         model.add(
-            layers.Bidirectional(
-                layers.LSTM(32, input_shape=(max_len, traj_dim)), merge_mode="ave"
+            keras.layers.Bidirectional(
+                keras.layers.LSTM(32, input_shape=(max_len, traj_dim)),
+                merge_mode="ave",
             )
         )
-        model.add(layers.Dense(15, activation="relu"))
-        model.add(layers.Dense(n_classes, activation="softmax"))
+        model.add(keras.layers.Dense(15, activation="relu"))
+        model.add(keras.layers.Dense(n_classes, activation="softmax"))
         model.compile(**self.compile_args)
         return model
 

pyproject.toml

@@ -70,6 +70,9 @@ push = false
 "pactus/__init__.py" = [
     '__version__ = "{version}"',
 ]
+"docs/source/conf.py" = [
+    'release = "{version}"',
+]
 
 [build-system]
 requires      = ["setuptools>=61.0.0",
@@ -78,6 +81,9 @@ build-backend = "setuptools.build_meta"
 
 [tool.mypy]
 python_version = "3.8"
+exclude = [
+    "docs",
+]
 
 [[tool.mypy.overrides]]
 module = [