🚀 simplify Sample and move to pydantic v2 conventions

Author: casenave

CHANGELOG.md

@@ -18,6 +18,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Changed
 
+- (sample) **Breaking**: remove the `path` constructor argument from `Sample`; use `Sample.load_from_dir(path)` or `sample.load(path)` for explicit disk loading.
 - (infos) add an infos.py file with a pydantic Info class at the same level as ProblemDefinition.
 - (cgns_backend) remove constant/variable mechanism in reading and writing paths for the cgns_backend.
 - (docs) update logos and migrate to Zensical.

docs/source/concepts/sample.md

@@ -93,6 +93,10 @@ sample.save_to_dir("sample_000000000", overwrite=True)
 sample = Sample.load_from_dir("sample_000000000")
 ```
 
+`Sample()` creates an empty in-memory sample. Loading from disk is explicit:
+use `Sample.load_from_dir(path)` to create a loaded sample, or instantiate an
+empty sample and call `sample.load(path)`.
+
 The CGNS files can be open in Paraview for advanced visualization and post-treatment.
 
 For large samples, `save_to_dir(..., memory_safe=True)` writes CGNS files through

docs/source/upgrade_guide.md

@@ -165,6 +165,7 @@ the [reader API](api/storage/reader.md), and the [backend API](api/storage/backe
 | `Dataset()` + `add_sample` / `add_samples` / `from_list_of_samples` | `save_to_disk(sample_constructor=..., ids=...)` |
 | `Dataset.save_to_dir(path)` / `add_to_dir` | `save_to_disk(path, sample_constructor=..., ids=...)` |
 | `Dataset.load(path)` | `init_from_disk(path)` → `(datasetdict, converterdict)` |
+| `Sample(path=sample_dir)` | `Sample.load_from_dir(sample_dir)` or `sample = Sample(); sample.load(sample_dir)` |
 | `dataset[i]` / `get_samples()` | `converter.to_plaid(dataset, i)` |
 | `len(dataset)` / `get_number_of_samples()` | `len(dataset)` (per-split backend object) |
 | `dataset.set_infos(...)` / `get_infos()` | pass [`Infos`](concepts/infos.md) to `save_to_disk(infos=...)`; read back with `Infos.from_path(path)` |

examples/containers/sample_example.py

@@ -559,10 +559,10 @@
 sample.save_to_dir(sample_save_fname)
 
 # %% [markdown]
-# #### Load a Sample from a directory via initialization
+# #### Load a Sample from a directory via the Sample class
 
 # %%
-new_sample = Sample(path=sample_save_fname)
+new_sample = Sample.load_from_dir(sample_save_fname)
 
 print(sample.summarize())
 

src/plaid/containers/managers/default_manager.py

@@ -83,7 +83,7 @@ def set_default_time(self, time: float) -> None:
             .. code-block:: python
 
                 from plaid import Sample
-                sample = Sample("path_to_plaid_sample")
+                sample = Sample.load_from_dir("path_to_plaid_sample")
                 print(sample)
                 >>> Sample(2 scalars, 1 timestamp, 5 fields)
                 print(sample.show_tree(0.5))
@@ -122,7 +122,7 @@ def set_default_base(self, base: str, time: Optional[float] = None) -> None:
             .. code-block:: python
 
                 from plaid import Sample
-                sample = Sample("path_to_plaid_sample")
+                sample = Sample.load_from_dir("path_to_plaid_sample")
                 print(sample)
                 >>> Sample(2 scalars, 1 timestamp, 5 fields)
                 print(sample.get_physical_dim("BaseA", 0.5))
@@ -173,7 +173,7 @@ def set_default_zone_base(
             .. code-block:: python
 
                 from plaid import Sample
-                sample = Sample("path_to_plaid_sample")
+                sample = Sample.load_from_dir("path_to_plaid_sample")
                 print(sample)
                 >>> Sample(2 scalars, 1 timestamp, 5 fields)
                 print(sample.get_zone_type("ZoneX", "BaseA", 0.5))

src/plaid/containers/sample.py

@@ -20,15 +20,15 @@
 import subprocess
 from copy import deepcopy
 from pathlib import Path
-from typing import Any, Optional, Union
+from typing import Optional, Union
 
 import CGNS.MAP as CGM
 import CGNS.PAT.cgnskeywords as CGK
 import CGNS.PAT.cgnslib as CGL
 import CGNS.PAT.cgnsutils as CGU
 import numpy as np
 from CGNS.PAT.cgnsutils import __CHILDREN__, __NAME__
-from pydantic import BaseModel, ConfigDict, model_validator
+from pydantic import BaseModel, model_validator
 from pydantic import Field as PydanticField
 
 from ..constants import (
@@ -48,35 +48,24 @@
 CGNS_WORKER = Path(__file__).parent.parent / "utils" / "cgns_worker.py"
 
 
-class Sample(BaseModel):
+class Sample(BaseModel, arbitrary_types_allowed=True, extra="forbid"):
     """Represents a single sample. It contains data and information related to a single observation or measurement within a dataset.
 
-    By default, the sample is empty but:
-        - You can provide a path to a folder containing the sample data, and it will be loaded during initialization.
+    By default, the sample is empty. Use :meth:`Sample.load_from_dir` or
+    :meth:`Sample.load` to load sample data from disk.
 
     Note:
         Mesh/field/global operations are directly implemented on ``Sample`` via
         inheritance from internal feature operations.
     """
 
-    # Pydantic configuration
-    # TODO(FB) check why arbitrary_types_allowed is needed, and if it can be removed
-    model_config = ConfigDict(
-        arbitrary_types_allowed=True, revalidate_instances="always", extra="forbid"
-    )
-
     # Attributes
-    path: Optional[Union[str, Path]] = PydanticField(
-        None,
-        description="Path to the folder containing the sample data. If provided, the sample will be loaded from this path during initialization. Defaults to None.",
-    )
-
     data: dict[float, CGNSTree] = PydanticField(
         default_factory=dict,
         description="A dictionary mapping time steps to CGNS trees.",
     )
 
-    defaults: DefaultManager = PydanticField(
+    defaults: Optional[DefaultManager] = PydanticField(
         default=None,
         exclude=True,
         repr=False,
@@ -90,13 +79,6 @@ def initialize_defaults(self) -> "Sample":
             self.defaults = DefaultManager(self)
         return self
 
-    def model_post_init(self, _context: Any) -> None:
-        """Run post-initialization hooks (e.g. load sample from path)."""
-        # Load if path is provided
-        if self.path is not None:
-            path = Path(self.path)
-            self.load(path)
-
     def copy(self) -> Self:  # pyright: ignore[reportIncompatibleMethodOverride]
         """Create a deep copy of the current `Sample` instance.
 

src/plaid/infos.py

@@ -7,18 +7,11 @@
 from typing import Any, Union
 
 import yaml
-from pydantic import BaseModel, ConfigDict, Field, ValidationError
-from pydantic.dataclasses import dataclass
+from pydantic import BaseModel, Field, ValidationError
 
 logger = logging.getLogger(__name__)
 
-_PD_CONFIG = ConfigDict(
-    extra="forbid", str_strip_whitespace=True, validate_assignment=True
-)
-
-
-@dataclass(config=_PD_CONFIG)
-class DataProduction:
+class DataProduction(BaseModel, revalidate_instances="always", str_strip_whitespace=True, validate_assignment=True, extra="forbid"):
     """Dataset production context metadata."""
 
     owner: str | None = None
@@ -43,10 +36,10 @@ class DataProduction:
 )
 
 
-class Infos(BaseModel):
+class Infos(BaseModel, revalidate_instances="always", str_strip_whitespace=True, validate_assignment=True, extra="forbid"):
     """Structured representation of a PLAID dataset ``infos`` payload."""
 
-    model_config = _PD_CONFIG
+    # model_config = _PD_CONFIG
 
     owner: str
     license: str

src/plaid/problem_definition.py

@@ -15,7 +15,7 @@
 from typing import Any, Literal, Sequence, Union
 
 import yaml
-from pydantic import BaseModel, ConfigDict, field_validator
+from pydantic import BaseModel, field_validator
 
 # %% Globals
 
@@ -26,12 +26,12 @@ def _normalize_list(v):
     return sorted(map(str, v))
 
 
-class ProblemDefinition(BaseModel):
+class ProblemDefinition(BaseModel, revalidate_instances="always", str_strip_whitespace=True, validate_assignment=True, extra="forbid"):
     """Defines the input and output features for a machine learning problem."""
 
-    model_config = ConfigDict(
-        revalidate_instances="always", validate_assignment=True, extra="forbid"
-    )
+    # model_config = ConfigDict(
+    #     revalidate_instances="always", validate_assignment=True, extra="forbid"
+    # )
 
     input_features: list[str]
     output_features: list[str]

src/plaid/storage/cgns/reader.py

@@ -75,7 +75,7 @@ def __getitem__(self, idx: int) -> Sample:
             Sample: A PLAID Sample object.
         """
         assert idx in self.ids
-        return Sample(path=self.path / f"sample_{idx:09d}")
+        return Sample.load_from_dir(self.path / f"sample_{idx:09d}")
 
     def __len__(self) -> int:
         """Get the number of samples in the dataset.
@@ -122,12 +122,9 @@ def sample_generator(
                 allow_patterns=[f"data/{split}/sample_{idx:09d}/"],
                 local_dir=temp_folder,
             )
-            sample = Sample(
-                path=Path(temp_folder) / "data" / f"{split}" / f"sample_{idx:09d}"
+            sample = Sample.load_from_dir(
+                Path(temp_folder) / "data" / f"{split}" / f"sample_{idx:09d}"
             )
-            # Sample data are eagerly loaded in memory during initialization;
-            # clear the transient on-disk path before leaving the temp dir.
-            sample.path = None
         yield sample
 
 

src/plaid/storage/common/bridge.py

@@ -208,7 +208,7 @@ def to_plaid_sample(
     for time, flat_tree in sample_dict.items():
         sample_data[time] = unflatten_cgns_tree(flat_tree, cgns_types)
 
-    return Sample(path=None, data=sample_data)
+    return Sample(data=sample_data)
 
 
 def plaid_to_sample_dict(