Merge branch 'main' into 540-classification-arg

src/fmu/dataio/case.py

@@ -124,10 +124,11 @@ def generate_metadata(self) -> dict:
                 model=global_configuration.Model.model_validate(
                     self.config["model"],
                 ),
-                case=internal.CaseMetadata(
+                case=meta.FMUCase(
                     name=self.casename,
-                    uuid=str(self._case_uuid()),
+                    uuid=self._case_uuid(),
                     user=meta.User(id=self.caseuser),
+                    description=None,
                 ),
             ),
             tracklog=_metadata.generate_meta_tracklog(),

src/fmu/dataio/dataio.py

@@ -32,10 +32,9 @@
 )
 from .aggregation import AggregatedData
 from .case import InitializeCase
-from .datastructure._internal.internal import (
-    AllowedContent,
-)
+from .datastructure._internal.internal import AllowedContent
 from .datastructure.configuration import global_configuration
+from .datastructure.meta import meta
 from .providers._fmu import FmuProvider, get_fmu_context_from_environment
 
 # always show PendingDeprecationWarnings
@@ -433,7 +432,7 @@ class ExportData:
     unit: str = ""
     verbosity: str = "DEPRECATED"  # remove in version 2
     vertical_domain: dict = field(default_factory=dict)
-    workflow: str = ""
+    workflow: Optional[str | dict[str, str]] = None
     table_index: Optional[list] = None
 
     # some keys that are modified version of input, prepended with _use
@@ -456,37 +455,23 @@ class ExportData:
     def __post_init__(self) -> None:
         logger.info("Running __post_init__ ExportData")
 
+        self._show_deprecations_or_notimplemented()
+
         self._fmurun = get_fmu_context_from_environment() != FmuContext.NON_FMU
 
         # set defaults for mutable keys
         self.vertical_domain = {"depth": "msl"}
 
         self._validate_content_key()
         self._validate_and_establish_fmucontext()
-        self._validate_access_ssdl_key()
-
-        # parse and validate the config
-        logger.debug("Global config given %s", prettyprint_dict(self.config))
-
-        # NB overwriting self.config. Ugly, but needed (?)
-        # Input argument is called "config", don't want to change that.
-        # Class variable self.config is used a lot, don't want to change it.
-        self.config = global_configuration.parse(self.config)
-        logger.debug("Global config after parsing is %s", prettyprint_dict(self.config))
-        # TODO: Consider if we should wipe self.config at this point, to avoid any
-        # (misuse) of it downstream. Since we are no longer carrying it and updating it,
-        # it will potentially NOT contain the values we are actually using. So there is
-        # a risk keeping it around.
-
-        # keeping this block here, while we still require the "_config_is_valid"
+        self._validate_workflow_key()
 
         self._config_is_valid = global_configuration.is_valid(self.config)
         if self._config_is_valid:
             # TODO: This needs refinement: _config_is_valid should be removed
             self.config = global_configuration.roundtrip(self.config)
 
         self._establish_pwd_rootpath()
-        self._show_deprecations_or_notimplemented()
         logger.info("Ran __post_init__")
 
     def _validate_access_ssdl_key(self) -> None:
@@ -582,6 +567,23 @@ def _show_deprecations_or_notimplemented(self) -> None:
                 "manner instead.",
                 UserWarning,
             )
+        if isinstance(self.workflow, dict):
+            warn(
+                "The 'workflow' argument should be given as a string. "
+                "Support for dictionary will be deprecated.",
+                FutureWarning,
+            )
+
+    def _validate_workflow_key(self) -> None:
+        if self.workflow:
+            if isinstance(self.workflow, str):
+                workflow = meta.Workflow(reference=self.workflow)
+            elif isinstance(self.workflow, dict):
+                workflow = meta.Workflow.model_validate(self.workflow)
+            else:
+                raise TypeError("'workflow' should be string.")
+
+            self.workflow = workflow.model_dump(mode="json", exclude_none=True)
 
     def _validate_content_key(self) -> None:
         """Validate the given 'content' input."""
@@ -643,6 +645,7 @@ def _update_check_settings(self, newsettings: dict) -> None:
 
         self._show_deprecations_or_notimplemented()
         self._validate_content_key()
+        self._validate_workflow_key()
         self._validate_and_establish_fmucontext()
 
     def _establish_pwd_rootpath(self) -> None:
@@ -719,6 +722,7 @@ def _check_process_object(self, obj: types.Inferrable) -> None:
 
     def _get_fmu_provider(self) -> FmuProvider:
         assert isinstance(self.fmu_context, FmuContext)
+        assert isinstance(self.workflow, dict) or self.workflow is None
         return FmuProvider(
             model=self.config.get("model"),
             fmu_context=self.fmu_context,

src/fmu/dataio/datastructure/_internal/internal.py

@@ -152,15 +152,9 @@ class JsonSchemaMetadata(BaseModel, populate_by_name=True):
     source: str = Field(default=SOURCE)
 
 
-class CaseMetadata(BaseModel):
-    name: str
-    uuid: str
-    user: meta.User
-
-
 class FMUModel(BaseModel):
     model: GlobalConfigurationModel
-    case: CaseMetadata
+    case: meta.FMUCase
 
 
 class PreprocessedInfo(BaseModel):

src/fmu/dataio/datastructure/configuration/global_configuration.py

@@ -52,6 +52,7 @@ class Model(BaseModel):
 
     name: str
     revision: str
+    description: Optional[List[str]] = Field(default=None)
 
 
 class Ssdl(BaseModel):

src/fmu/dataio/providers/_fmu.py

@@ -97,7 +97,7 @@ class FmuProvider:
     include_ertjobs: bool = True
     casepath_proposed: str | Path = ""
     forced_realization: Optional[int] = None
-    workflow: str | dict = ""
+    workflow: Optional[dict[str, str]] = None
 
     # private properties for this class
     _runpath: Path | str = field(default="", init=False)
@@ -364,29 +364,8 @@ def _generate_ert_metadata(self) -> None:
         meta = self._metadata  # shortform
 
         meta["model"] = self.model
-
         meta["context"] = {"stage": self.fmu_context.name.lower()}
-
-        if self.workflow:
-            if isinstance(self.workflow, str):
-                meta["workflow"] = {"reference": self.workflow}
-            elif isinstance(self.workflow, dict):
-                if "reference" not in self.workflow:
-                    raise ValueError(
-                        "When workflow is given as a dict, the 'reference' "
-                        "key must be included and be a string"
-                    )
-                warn(
-                    "The 'workflow' argument should be given as a string. "
-                    "Support for dictionary input is scheduled for deprecation.",
-                    PendingDeprecationWarning,
-                )
-
-                meta["workflow"] = {"reference": self.workflow["reference"]}
-
-            else:
-                raise TypeError("'workflow' should be string.")
-
+        meta["workflow"] = self.workflow
         case_uuid = "not_present"  # TODO! not allow missing case metadata?
         if self._case_metadata and "fmu" in self._case_metadata:
             meta["case"] = deepcopy(self._case_metadata["fmu"]["case"])

tests/test_units/test_fmuprovider_class.py

@@ -5,6 +5,7 @@
 from pathlib import Path
 
 import fmu.dataio as dataio
+import pydantic
 import pytest
 
 # from conftest import pretend_ert_env_run1
@@ -14,6 +15,7 @@
 logger = logging.getLogger(__name__)
 
 FOLDERTREE = "/scratch/myfield/case/realization-13/iter-2/"
+WORKFLOW = {"reference": "some_work_flow"}
 
 GLOBAL_CONFIG_MODEL = {"name": "Model2", "revision": "22.1.0"}
 
@@ -45,7 +47,7 @@ def test_fmuprovider_no_provider():
         casepath_proposed="",
         include_ertjobs=False,
         forced_realization=None,
-        workflow="some work flow",
+        workflow=WORKFLOW,
     )
     assert myfmu.get_provider() is None
 
@@ -83,7 +85,7 @@ def test_fmuprovider_ert_provider_guess_casemeta_path(fmurun):
             casepath_proposed="",  # if casepath is undef, try deduce from, _ERT_RUNPATH
             include_ertjobs=False,
             forced_realization=None,
-            workflow="some work flow",
+            workflow=WORKFLOW,
         )
 
     assert myfmu.get_provider() == "ERT"
@@ -106,7 +108,7 @@ def test_fmuprovider_ert_provider_missing_parameter_txt(
             model=GLOBAL_CONFIG_MODEL,
             fmu_context=FmuContext.REALIZATION,
             include_ertjobs=True,
-            workflow="some work flow",
+            workflow=WORKFLOW,
         )
     assert myfmu._case_name == "ertrun1"
     assert myfmu._real_name == "realization-0"
@@ -121,7 +123,7 @@ def test_fmuprovider_arbitrary_iter_name(fmurun_w_casemetadata_pred):
         model=GLOBAL_CONFIG_MODEL,
         fmu_context=FmuContext.REALIZATION,
         include_ertjobs=True,
-        workflow="some work flow",
+        workflow=WORKFLOW,
     )
     assert myfmu._case_name == "ertrun1"
     assert myfmu._real_name == "realization-0"
@@ -169,7 +171,7 @@ def test_fmuprovider_prehook_case(tmp_path, globalconfig2, fmurun_prehook):
         model=GLOBAL_CONFIG_MODEL,
         fmu_context=FmuContext.CASE,
         include_ertjobs=False,
-        workflow="some work flow",
+        workflow=WORKFLOW,
         casepath_proposed=caseroot,
     )
 
@@ -257,7 +259,7 @@ def test_fmuprovider_no_restart_env(monkeypatch, fmurun_w_casemetadata, fmurun_p
     assert "restart_from" not in fmu_restart._metadata["iteration"]
 
 
-def test_fmuprovider_workflow_reference(fmurun_w_casemetadata):
+def test_fmuprovider_workflow_reference(fmurun_w_casemetadata, edataobj1):
     """Testing the handling of workflow reference input.
 
     Metadata definitions of fmu.workflow is that it is a dictionary with 'reference'
@@ -274,20 +276,25 @@ def test_fmuprovider_workflow_reference(fmurun_w_casemetadata):
     os.chdir(fmurun_w_casemetadata)
 
     # workflow input is a string
-    myfmu = FmuProvider(workflow="workflow as string")
+    edata = dataio.ExportData(workflow="workflow as string")
+    # check that the conversion to dict works
+    assert edata.workflow == {"reference": "workflow as string"}
+    myfmu = FmuProvider(workflow=edata.workflow)
     assert "workflow" in myfmu._metadata
-    assert myfmu._metadata["workflow"] == {"reference": "workflow as string"}
+    assert myfmu._metadata["workflow"] == edata.workflow
 
     # workflow input is a correct dict
-    with pytest.warns(PendingDeprecationWarning, match="The 'workflow' argument"):
-        myfmu = FmuProvider(workflow={"reference": "workflow as dict"})
+    with pytest.warns(FutureWarning, match="The 'workflow' argument"):
+        edata = dataio.ExportData(workflow={"reference": "workflow as dict"})
+    assert edata.workflow == {"reference": "workflow as dict"}
+    myfmu = FmuProvider(workflow=edata.workflow)
     assert "workflow" in myfmu._metadata
-    assert myfmu._metadata["workflow"] == {"reference": "workflow as dict"}
+    assert myfmu._metadata["workflow"] == edata.workflow
 
     # workflow input is non-correct dict
-    with pytest.raises(ValueError):
-        myfmu = FmuProvider(workflow={"wrong": "workflow as dict"})
+    with pytest.raises(pydantic.ValidationError):
+        dataio.ExportData(workflow={"wrong": "workflow as dict"})
 
     # workflow input is other types - shall fail
     with pytest.raises(TypeError):
-        myfmu = FmuProvider(workflow=123.4)
+        dataio.ExportData(workflow=123.4)