emmet-core 0.87.0.dev3__tar.gz → 0.87.0.dev5__tar.gz

{emmet_core-0.87.0.dev3 → emmet_core-0.87.0.dev5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: emmet-core
-Version: 0.87.0.dev3
+Version: 0.87.0.dev5
 Summary: Core Emmet Library
 Author-email: The Materials Project <feedback@materialsproject.org>
 License: Modified BSD

{emmet_core-0.87.0.dev3 → emmet_core-0.87.0.dev5}/emmet/core/material.py

@@ -16,7 +16,7 @@ from emmet.core.types.pymatgen_types.structure_adapter import (
     MoleculeType,
     StructureType,
 )
-from emmet.core.types.typing import DateTimeType, IdentifierType
+from emmet.core.types.typing import DateTimeType, IdentifierType, MaterialIdentifierType
 
 if TYPE_CHECKING:
 
@@ -46,7 +46,7 @@ class BasePropertyMetadata(StructureMetadata, EmmetBaseModel):
     extended, not used directly
     """
 
-    material_id: IdentifierType | None = Field(
+    material_id: MaterialIdentifierType | None = Field(
         None,
         description="The Materials Project ID of the material, used as a universal reference across property documents."
         "This comes in the form: mp-******.",

{emmet_core-0.87.0.dev3 → emmet_core-0.87.0.dev5}/emmet/core/tasks.py

@@ -316,8 +316,7 @@ class CoreTaskDoc(StructureMetadata):
     )
     task_id: IdentifierType | None = Field(
         None,
-        description="The (task) ID of this calculation, used as a universal reference across property documents."
-        "This comes in the form: mp-******.",
+        description="The (task) ID of this calculation, used as a universal reference across property documents.",
     )
     task_type: TaskType | CalcType | None = Field(
         None, description="The type of calculation."

{emmet_core-0.87.0.dev3 → emmet_core-0.87.0.dev5}/emmet/core/thermo.py

@@ -38,7 +38,7 @@ class DecompositionProduct(BaseModel):
     )
     amount: float | None = Field(
         None,
-        description="The amount of the decomposed material by formula units this this material decomposes to.",
+        description="The amount of the decomposed material by formula units this material decomposes to.",
     )
 
 

{emmet_core-0.87.0.dev3 → emmet_core-0.87.0.dev5}/emmet/core/types/pymatgen_types/computed_entries_adapter.py

@@ -1,15 +1,17 @@
 from typing import Annotated, Any, TypeVar
 
 import orjson
-from pydantic import BeforeValidator, WrapSerializer
+from pydantic import BeforeValidator, TypeAdapter, WrapSerializer
 from pymatgen.entries.computed_entries import ComputedEntry, ComputedStructureEntry
 from typing_extensions import NotRequired, TypedDict
 
+from emmet.core.mpid_ext import ThermoID
 from emmet.core.types.pymatgen_types.element_adapter import ElementType
 from emmet.core.types.pymatgen_types.structure_adapter import (
-    TypedStructureDict,
+    StructureType,
     pop_empty_structure_keys,
 )
+from emmet.core.types.typing import IdentifierType, MaterialIdentifierType
 from emmet.core.vasp.calculation import PotcarSpec
 
 # TypedEnergyAdjustmentDict = TypedDict(
@@ -59,20 +61,20 @@ from emmet.core.vasp.calculation import PotcarSpec
 
 
 class TypedCEDataDict(TypedDict):
-    oxide_type: str
-    aspherical: bool
-    last_updated: str
-    task_id: str
-    material_id: str
-    oxidation_states: dict[ElementType, float]
-    license: str
-    run_type: str
+    oxide_type: NotRequired[str | None]
+    aspherical: NotRequired[bool | None]
+    last_updated: NotRequired[str | None]
+    task_id: NotRequired[IdentifierType | None]
+    material_id: NotRequired[MaterialIdentifierType | None]
+    oxidation_states: NotRequired[dict[ElementType, float] | None]
+    license: NotRequired[str | None]
+    run_type: NotRequired[str | None]
 
 
 class TypedCEParameterDict(TypedDict):
-    potcar_spec: list[PotcarSpec]
-    run_type: str
-    is_hubbard: bool
+    potcar_spec: NotRequired[list[PotcarSpec] | None]
+    run_type: NotRequired[str | None]
+    is_hubbard: NotRequired[bool | None]
     hubbards: NotRequired[dict[str, float] | None]  # type: ignore[type-arg]
 
 
@@ -83,14 +85,14 @@ TypedComputedEntryDict = TypedDict(
         "@class": str,
         "energy": float,
         "composition": dict[ElementType, float],
-        "entry_id": str,
+        "entry_id": ThermoID,
         "correction": float,
         # "energy_adjustments": list[
         #     TypedCompositionEnergyAdjustmentDict
         #     | TypedEnergyAdjustmentDict
         #     | TypedTemperatureEnergyAdjustmentDict
         # ],
-        "energy_adjustments": str,
+        "energy_adjustments": NotRequired[str | None],
         "parameters": TypedCEParameterDict,
         "data": TypedCEDataDict,
     },
@@ -98,7 +100,7 @@ TypedComputedEntryDict = TypedDict(
 
 
 class TypedComputedStructureEntryDict(TypedComputedEntryDict):
-    structure: TypedStructureDict
+    structure: StructureType
 
 
 ComputedEntryTypeVar = TypeVar(
@@ -116,6 +118,9 @@ ComputedStructureEntryTypeVar = TypeVar(
 
 
 def entry_serializer(entry, nxt, info) -> dict[str, Any]:
+    # need to beat pmg serialization to get correct id serialization
+    entry.data = TypeAdapter(TypedCEDataDict).dump_python(entry.data)
+
     default_serialized_object = nxt(entry.as_dict(), info)
 
     format = info.context.get("format") if info.context else None
@@ -142,11 +147,18 @@ def entry_deserializer(entry: dict[str, Any] | ComputedEntry | ComputedStructure
         match entry["@class"]:
             case "ComputedEntry":
                 entry_cls = ComputedEntry
+                entry_type = TypedComputedEntryDict
             case "ComputedStructureEntry":
                 entry_cls = ComputedStructureEntry
+                entry_type = TypedComputedStructureEntryDict
                 entry = pop_cse_empty_keys(entry)
 
-        if isinstance(entry["energy_adjustments"], str):
+        # must be before 'energy_adjustments' deserialization
+        entry = TypeAdapter(entry_type).validate_python(entry)
+
+        if "energy_adjustments" in entry and isinstance(
+            entry["energy_adjustments"], str
+        ):
             entry["energy_adjustments"] = orjson.loads(entry["energy_adjustments"])
 
         return entry_cls.from_dict(entry)  # type: ignore[arg-type]

{emmet_core-0.87.0.dev3 → emmet_core-0.87.0.dev5}/emmet/core/types/pymatgen_types/phase_diagram_adapter.py

@@ -101,9 +101,15 @@ def phase_diagram_serializer(phase_diagram, nxt, info) -> dict[str, Any]:
             entry.as_dict() for entry in default_serialized_object["computed_data"][key]
         ]
 
+    # ndarray -> list[list[int]]
     for simplex in default_serialized_object["computed_data"]["simplexes"]:
         simplex["coords"] = simplex["coords"].tolist()
 
+    # ndarray -> list[list[float]]
+    default_serialized_object["computed_data"]["qhull_data"] = (
+        default_serialized_object["computed_data"]["qhull_data"].tolist()
+    )
+
     format = info.context.get("format") if info.context else None
     if format == "arrow":
         phase_diagram_serde(

{emmet_core-0.87.0.dev3 → emmet_core-0.87.0.dev5}/emmet/core/types/typing.py

@@ -11,8 +11,9 @@ from __future__ import annotations
 
 import os
 from datetime import datetime
+from functools import partial
 from pathlib import Path
-from typing import TYPE_CHECKING, Annotated, Union, Any
+from typing import TYPE_CHECKING, Annotated, Any, Union
 
 import orjson
 from pydantic import BeforeValidator, Field, PlainSerializer, WrapSerializer
@@ -52,22 +53,41 @@ for why this separate class is necesary instead of `DateTimeType | None`
 """
 
 
-def _fault_tolerant_id_serde(val: Any, serialize: bool = False) -> Any:
+def _fault_tolerant_id_serde(
+    val: Any,
+    legacy: bool = False,
+    serialize: bool = False,
+    **kwargs,
+) -> Any:
     """Needed for the API and safe de-/serialization behavior."""
     try:
-        alpha_id = AlphaID(val)
+        alpha_id = AlphaID(val, **kwargs)
         if serialize:
             return str(alpha_id)
-        return alpha_id.formatted
+        return alpha_id.formatted if legacy else alpha_id
     except Exception:
         return val
 
 
-IdentifierType: TypeAlias = Annotated[
-    Union[MPID, AlphaID],
-    BeforeValidator(_fault_tolerant_id_serde),
-    PlainSerializer(lambda x: _fault_tolerant_id_serde(x, serialize=True)),
-]
+_id_base_metadata = (BeforeValidator(_fault_tolerant_id_serde),)
+
+
+def _make_id_type(render_order, **kwargs) -> Annotated:
+    match render_order:
+        case 0:
+            _order = Union[AlphaID, MPID]
+        case 1:
+            _order = Union[MPID, AlphaID]
+
+    return Annotated[
+        _order,
+        BeforeValidator(partial(_fault_tolerant_id_serde, **kwargs)),
+        PlainSerializer(partial(_fault_tolerant_id_serde, serialize=True, **kwargs)),
+    ]
+
+
+IdentifierType: TypeAlias = _make_id_type(0, padlen=8)
+MaterialIdentifierType: TypeAlias = _make_id_type(1, legacy=True, prefix="mp", padlen=8)
 """MPID / AlphaID serde."""
 
 

{emmet_core-0.87.0.dev3 → emmet_core-0.87.0.dev5}/emmet/core/vasp/material.py

@@ -5,11 +5,12 @@ from typing import Mapping
 from pydantic import BaseModel, Field
 from pymatgen.analysis.structure_analyzer import SpacegroupAnalyzer, oxide_type
 from pymatgen.analysis.structure_matcher import StructureMatcher
-from pymatgen.entries.computed_entries import ComputedStructureEntry
 
 from emmet.core.base import EmmetMeta
 from emmet.core.material import MaterialsDoc as CoreMaterialsDoc
 from emmet.core.material import PropertyOrigin
+from emmet.core.mpid import AlphaID
+from emmet.core.mpid_ext import ThermoID
 from emmet.core.settings import EmmetSettings
 from emmet.core.tasks import TaskDoc
 from emmet.core.types.pymatgen_types.computed_entries_adapter import (
@@ -105,7 +106,7 @@ class MaterialsDoc(CoreMaterialsDoc):
         if use_statics:
             possible_mat_ids += [task.task_id for task in statics]
 
-        material_id = min(possible_mat_ids)
+        material_id = AlphaID(min(possible_mat_ids), prefix="mp")
 
         # Always prefer a static over a structure opt
         structure_task_quality_scores = {"Structure Optimization": 1, "Static": 2}
@@ -207,10 +208,12 @@ class MaterialsDoc(CoreMaterialsDoc):
 
             if relevant_calcs:
                 best_task_doc = relevant_calcs[0]
-                entry = ComputedStructureEntry(
-                    composition=best_task_doc.output.structure.composition,
-                    correction=0.0,
-                    data={
+                entry = {
+                    "@class": "ComputedStructureEntry",
+                    "@module": "pymatgen.entries.computed_entries",
+                    "composition": best_task_doc.output.structure.composition,
+                    "correction": 0.0,
+                    "data": {
                         "aspherical": best_task_doc.input.parameters.get(
                             "LASPH", False
                         ),
@@ -219,9 +222,9 @@ class MaterialsDoc(CoreMaterialsDoc):
                         "material_id": material_id,
                         "task_id": best_task_doc.task_id,
                     },
-                    energy=best_task_doc.output.energy,
-                    entry_id="{}-{}".format(material_id, rt.value),
-                    parameters={
+                    "energy": best_task_doc.output.energy,
+                    "entry_id": ThermoID(identifier=material_id, suffix=rt),
+                    "parameters": {
                         "hubbards": best_task_doc.input.hubbards,
                         "is_hubbard": best_task_doc.input.is_hubbard,
                         "potcar_spec": (
@@ -231,8 +234,8 @@ class MaterialsDoc(CoreMaterialsDoc):
                         ),
                         "run_type": str(best_task_doc.run_type),
                     },
-                    structure=best_task_doc.output.structure,
-                )
+                    "structure": best_task_doc.output.structure,
+                }
                 entries[rt] = entry
 
         if not any(

{emmet_core-0.87.0.dev3 → emmet_core-0.87.0.dev5}/emmet_core.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: emmet-core
-Version: 0.87.0.dev3
+Version: 0.87.0.dev5
 Summary: Core Emmet Library
 Author-email: The Materials Project <feedback@materialsproject.org>
 License: Modified BSD