feat: check project is trainable before build

Also fix an issue with translating a model configuration payload into ModelConfiguration object

src/ansys/simai/core/data/model_configuration.py

@@ -21,7 +21,7 @@
 # SOFTWARE.
 
 from dataclasses import asdict, dataclass, field
-from typing import TYPE_CHECKING, Any, Literal, Optional
+from typing import TYPE_CHECKING, Any, List, Literal, Optional
 
 from ansys.simai.core.errors import InvalidArguments, ProcessingError
 from ansys.simai.core.utils.misc import dict_get
@@ -223,7 +223,7 @@ class ModelConfiguration:
                     | *2_days*: < 2 days, default value.
 
                     | *7_days*: < 1 week
-        continuous: indicates if continuous learning is enabled. Default is False.
+        build_on_top: indicates if build_on_top learning is enabled. Default is False.
         input: the inputs of the model.
         output: the outputs of the model.
         global_coefficients: the Global Coefficients of the model.
@@ -275,20 +275,20 @@ class ModelConfiguration:
             new_conf = ModelConfiguration(
                 project=aero_dyn_project,
                 build_preset="debug",
-                continuous=False,
+                build_on_top=False,
                 input=model_input,
                 output=model_output,
                 global_coefficients=global_coefficients,
                 domain_of_analysis=doa,
                 pp_input=pp_input,
             )
 
-            # Launch a mode build with the new configuration
+            # Launch a model build with the new configuration
             new_model = simai.models.build(new_conf)
     """
 
     project: "Optional[Project]" = None
-    continuous: bool = False
+    build_on_top: bool = False
     input: ModelInput = field(default_factory=lambda: ModelInput())
     output: ModelOutput = field(default_factory=lambda: ModelOutput())
     domain_of_analysis: DomainOfAnalysis = field(default_factory=lambda: DomainOfAnalysis())
@@ -335,7 +335,7 @@ def __init__(
         project: "Project",
         boundary_conditions: Optional[dict[str, Any]] = None,
         build_preset: Optional[str] = "debug",
-        continuous: bool = False,
+        build_on_top: bool = False,
         fields: Optional[dict[str, Any]] = None,
         global_coefficients: Optional[list[GlobalCoefficientDefinition]] = None,
         simulation_volume: Optional[dict[str, Any]] = None,
@@ -358,7 +358,7 @@ def __init__(
         if boundary_conditions is not None and self.input.boundary_conditions is None:
             self.input.boundary_conditions = list(boundary_conditions.keys())
         self.build_preset = build_preset
-        self.continuous = continuous
+        self.build_on_top = build_on_top
         if fields is not None:
             if fields.get("surface_input"):
                 self.input.surface = [fd.get("name") for fd in fields["surface_input"]]
@@ -464,13 +464,24 @@ def _to_payload(self):
         return {
             "boundary_conditions": bcs,
             "build_preset": SupportedBuildPresets[self.build_preset],
-            "continuous": self.continuous,
+            "continuous": self.build_on_top,
             "fields": flds,
             "global_coefficients": gcs,
             "simulation_volume": simulation_volume,
         }
 
-    def compute_global_coefficient(self):
+    @classmethod
+    def _from_payload(cls, **kwargs) -> "ModelConfiguration":
+        # Retrieve SDK version of build preset from API version of build preset
+        build_preset = next(
+            (k for k, v in SupportedBuildPresets.items() if v == kwargs.get("build_preset")), None
+        )
+        kwargs["build_preset"] = build_preset
+        if build_on_top := kwargs.pop("continuous", None):
+            kwargs["build_on_top"] = build_on_top
+        return ModelConfiguration(**kwargs)
+
+    def compute_global_coefficient(self) -> List[float]:
         """Computes the results of the formula for all global coefficients with respect to the project's sample."""
 
         if self.project is None:

src/ansys/simai/core/data/models.py

@@ -23,6 +23,7 @@
 
 from ansys.simai.core.data.base import ComputableDataModel, Directory
 from ansys.simai.core.data.model_configuration import ModelConfiguration
+from ansys.simai.core.errors import InvalidArguments
 
 
 class Model(ComputableDataModel):
@@ -39,7 +40,7 @@ def project_id(self) -> str:
     @property
     def configuration(self) -> ModelConfiguration:
         """Build configuration of a model."""
-        return ModelConfiguration(
+        return ModelConfiguration._from_payload(
             project=self._client.projects.get(self.fields["project_id"]),
             **self.fields["configuration"],
         )
@@ -99,6 +100,12 @@ def build(
                 new_model = simai.models.build(build_conf)
 
         """
+        if not configuration.project:
+            raise InvalidArguments("The model configuration does not have a project set")
+
+        is_trainable = configuration.project.is_trainable()
+        if not is_trainable:
+            raise InvalidArguments(f"Cannot train model because: {is_trainable.reason}")
 
         return self._model_from(
             self._client._api.launch_build(

src/ansys/simai/core/data/projects.py

@@ -27,6 +27,7 @@
 from ansys.simai.core.data.model_configuration import ModelConfiguration
 from ansys.simai.core.data.types import Identifiable, get_id_from_identifiable
 from ansys.simai.core.errors import InvalidArguments, ProcessingError
+from ansys.simai.core.utils.numerical import cast_values_to_float
 
 if TYPE_CHECKING:
     from ansys.simai.core.data.global_coefficients_requests import (
@@ -124,17 +125,19 @@ def sample(self, new_sample: Identifiable["TrainingData"]):
     @property
     def last_model_configuration(self) -> ModelConfiguration:
         """Last :class:`configuration <ansys.simai.core.data.model_configuration.ModelConfiguration>` used for model training in this project."""
-        return ModelConfiguration(project=self, **self.fields.get("last_model_configuration"))
-
-    def delete(self) -> None:
-        """Delete the project."""
-        self._client._api.delete_project(self.id)
+        return ModelConfiguration._from_payload(
+            project=self, **self.fields.get("last_model_configuration")
+        )
 
-    def is_trainable(self) -> bool:
+    def is_trainable(self) -> IsTrainableInfo:
         """Check if the project meets the prerequisites to be trained."""
         tt = self._client._api.is_project_trainable(self.id)
         return IsTrainableInfo(**tt)
 
+    def delete(self) -> None:
+        """Delete the project."""
+        self._client._api.delete_project(self.id)
+
     def get_variables(self) -> Optional[Dict[str, List[str]]]:
         """Get the available variables for the model's input/output."""
         if not self.sample:
@@ -177,7 +180,7 @@ def verify_gc_formula(
 
     def compute_gc_formula(
         self, gc_formula: str, bc: list[str] = None, surface_variables: list[str] = None
-    ):
+    ) -> float:
         """Compute the result of a global coefficient formula according to the project sample."""
 
         if not self.sample:
@@ -203,7 +206,7 @@ def compute_gc_formula(
         gc_compute.run()
         gc_compute.wait()
 
-        return gc_compute.result if gc_compute.is_ready else None
+        return cast_values_to_float(gc_compute.result["value"])
 
     def cancel_build(self):
         """Cancels a build if there is one pending."""

tests/test_models.py

@@ -226,12 +226,18 @@ def test_build_with_last_config(simai_client):
         json=raw_project,
         status=200,
     )
+    responses.add(
+        responses.GET,
+        f"https://test.test/projects/{MODEL_RAW['project_id']}/trainable",
+        json={"is_trainable": True},
+        status=200,
+    )
 
     project: Project = simai_client._project_directory._model_from(raw_project)
 
     project.verify_gc_formula = Mock()
 
-    in_model_conf = ModelConfiguration(project=project, **MODEL_CONF_RAW)
+    in_model_conf = ModelConfiguration._from_payload(project=project, **MODEL_CONF_RAW)
 
     launched_model: Model = simai_client.models.build(in_model_conf)
 
@@ -261,6 +267,12 @@ def test_build_with_new_config(simai_client):
         json=raw_project,
         status=200,
     )
+    responses.add(
+        responses.GET,
+        f"https://test.test/projects/{MODEL_RAW['project_id']}/trainable",
+        json={"is_trainable": True},
+        status=200,
+    )
 
     project: Project = simai_client._project_directory._model_from(raw_project)
 
@@ -280,7 +292,7 @@ def test_build_with_new_config(simai_client):
     new_conf = ModelConfiguration(
         project=project,
         build_preset="debug",
-        continuous=False,
+        build_on_top=False,
         input=model_input,
         output=model_output,
         global_coefficients=global_coefficients,
@@ -317,7 +329,7 @@ def test_set_doa(simai_client):
 
     project.verify_gc_formula = Mock()
 
-    model_conf = ModelConfiguration(project=project, **MODEL_CONF_RAW)
+    model_conf = ModelConfiguration._from_payload(project=project, **MODEL_CONF_RAW)
 
     new_height = {"position": "relative_to_center", "value": 0.5, "length": 15.2}
 
@@ -344,7 +356,7 @@ def test_get_doa(simai_client):
 
     project.verify_gc_formula = Mock()
 
-    model_conf = ModelConfiguration(project=project, **MODEL_CONF_RAW)
+    model_conf = ModelConfiguration._from_payload(project=project, **MODEL_CONF_RAW)
 
     doa_length_raw = MODEL_CONF_RAW.get("simulation_volume").get("X")
 
@@ -436,7 +448,7 @@ def test_exception_compute_global_coefficient(simai_client):
 
     project.verify_gc_formula = Mock()
 
-    model_conf = ModelConfiguration(project=project, **MODEL_CONF_RAW)
+    model_conf = ModelConfiguration._from_payload(project=project, **MODEL_CONF_RAW)
 
     model_conf.project = None
 
@@ -449,7 +461,7 @@ def test_exception_setting_global_coefficient():
     THEN an error is raise."""
 
     with pytest.raises(ProcessingError):
-        ModelConfiguration(project=None, **MODEL_CONF_RAW)
+        ModelConfiguration._from_payload(project=None, **MODEL_CONF_RAW)
 
 
 def test_sse_event_handler(simai_client, model_factory):
@@ -508,7 +520,7 @@ def test_throw_error_when_volume_is_missing_from_sample(simai_client):
     model_output = ModelOutput(surface=[], volume=["Velocity_0"])
     global_coefficients = []
 
-    model_conf = ModelConfiguration(
+    model_conf = ModelConfiguration._from_payload(
         project=project,
         build_preset="debug",
         continuous=False,
@@ -539,6 +551,12 @@ def test_post_process_input(simai_client):
         json=raw_project,
         status=200,
     )
+    responses.add(
+        responses.GET,
+        f"https://test.test/projects/{MODEL_RAW['project_id']}/trainable",
+        json={"is_trainable": True},
+        status=200,
+    )
 
     project: Project = simai_client._project_directory._model_from(raw_project)
     project.verify_gc_formula = Mock()
@@ -559,7 +577,7 @@ def test_post_process_input(simai_client):
     model_request = deepcopy(MODEL_RAW)
     model_request["configuration"] = model_conf_dict
 
-    config_with_pp_input = ModelConfiguration(
+    config_with_pp_input = ModelConfiguration._from_payload(
         project=project,
         **model_conf_dict,
         pp_input=pp_input,
@@ -606,6 +624,12 @@ def test_failed_build_with_resolution(simai_client):
         json=raw_project,
         status=200,
     )
+    responses.add(
+        responses.GET,
+        f"https://test.test/projects/{MODEL_RAW['project_id']}/trainable",
+        json={"is_trainable": True},
+        status=200,
+    )
 
     project: Project = simai_client._project_directory._model_from(raw_project)
 
@@ -619,7 +643,7 @@ def test_failed_build_with_resolution(simai_client):
         height=hght,
     )
 
-    new_conf = ModelConfiguration(
+    new_conf = ModelConfiguration._from_payload(
         project=project,
         build_preset="debug",
         domain_of_analysis=doa,