property kind and repr context

Author: valentin-gauthier-geosiris

energyml-utils/example/attic/property_kind_searcher.py

@@ -0,0 +1,38 @@
+import json
+from typing import List, Dict
+from energyml.resqml.v2_0_1.resqmlv2 import ResqmlPropertyKind
+from energyml.utils.epc_utils import __CACHE_PROP_KIND_DICT__, get_property_kind_by_title
+
+def title_inclusion(t0: str, t1: str) -> bool:
+    t0_words = set(t0.lower().split())
+    t1_words = set(t1.lower().split())
+    return t0_words.issubset(t1_words) or t1_words.issubset(t0_words)
+    
+
+def search_matching_title(title: List[str]) -> Dict[str, List[str]]:
+    potential_matches = {t: [] for t in title}
+    for pk in __CACHE_PROP_KIND_DICT__.values():
+        for t in title:
+            if title_inclusion(t, pk.citation.title):
+                potential_matches[t].append(pk.citation.title)
+    return potential_matches
+
+if __name__ == "__main__":
+    not_found = []
+    for p in ResqmlPropertyKind:
+        print(p)
+        pk = get_property_kind_by_title(p.value)
+        if pk is None:
+            not_found.append(p.value)
+        print(f"\t{pk.uuid}" if pk is not None else "\tNot found")
+        
+    
+    print("Not found tried to match:")
+    matches = search_matching_title(not_found)
+    for title, matched_titles in matches.items():
+        print(f"\t{title}: ")
+        for mt in matched_titles:
+            print(f"\t\t{mt}")
+            
+    with open("pk_not_found_matches.json", "w") as f:
+        json.dump(matches, f, indent=4)

energyml-utils/pyproject.toml

@@ -103,7 +103,7 @@ mypy = "^0.971"
 bandit = "^1.7.0"
 safety = "^1.10.0"
 memory-profiler = "^0.60.0"
-line-profiler = "^4.0.0"
+# line-profiler = "^4.0.0"
 
 [tool.coverage.run]
 branch = true

energyml-utils/src/energyml/utils/data/datasets_io.py

@@ -856,7 +856,7 @@ def open_file_no_cache(self, file_path: str, mode: str = "r") -> Optional[Any]:
             try:
                 return h5py.File(file_path, mode)  # type: ignore
             except Exception as e:
-                logging.debug(f"Failed to open HDF5 file {file_path}: {e}")
+                # logging.debug(f"Failed to open HDF5 file {file_path}: {e}")
                 return None
 
         def read_array(

energyml-utils/src/energyml/utils/data/representation_context.py

@@ -8,8 +8,8 @@
 
 from energyml.utils.uri import Uri
 from energyml.utils.storage_interface import EnergymlStorageInterface
-from energyml.utils.epc_utils import extract_uuid_and_version_from_obj_path
-from energyml.utils.introspection import get_obj_uri, get_obj_uuid, search_attribute_matching_name
+from energyml.utils.epc_utils import extract_uuid_and_version_from_obj_path, get_property_kind_by_title, get_property_kind_uuid_from_property_object
+from energyml.utils.introspection import get_obj_uri, get_obj_uuid, get_object_attribute, is_enum, search_attribute_matching_name
 from energyml.utils.data.helper import RgbaColor, ScalarRenderingInfo, read_graphical_rendering_info
 
 NO_KIND = "NO_KIND"
@@ -60,16 +60,17 @@ class RepresentationContext(BaseModel):
     crs: List[Any] = Field(default_factory=list)
     rels: List[Relationship] = Field(default_factory=list)
 
-    # Properties keyed by object uuid → property object
-    properties_by_kind: dict = Field(default_factory=dict)
-
     # Graphical information keyed by GraphicalInformationSet uri → list of entries
     graphical_info: dict = Field(default_factory=dict)
 
     time_series: list = Field(default_factory=list)
 
     def __init__(self, obj: Any, workspace: EnergymlStorageInterface, **data):
         super().__init__(obj=obj, workspace=workspace, uri=get_obj_uri(obj), **data)
+        
+        # Properties keyed by object uuid → property object
+        self._props = {}
+        self._props_by_kind = {}
         self.update()
 
     def update(self):
@@ -92,7 +93,8 @@ def collect_time_series(self):
 
     def _collect_properties(self, rels: List[Relationship]):
         # Collect related properties keyed by property uuid
-        self.properties_by_kind = {}
+        self._props = {}
+        self._props_by_kind = {}
         for r in self.rels:
             if "Property" in r.target:
                 uuid, version = extract_uuid_and_version_from_obj_path(r.target)
@@ -101,7 +103,26 @@ def _collect_properties(self, rels: List[Relationship]):
                     logging.warning(f"Property {r.target} not found in workspace")
                     continue
                 prop_uuid = getattr(prop, "uuid", NO_KIND)
-                self.properties_by_kind[prop_uuid] = prop
+                self._props[prop_uuid] = prop
+                prop_kind_uuid = get_property_kind_uuid_from_property_object(prop) or NO_KIND
+                if prop_kind_uuid not in self._props_by_kind:
+                    self._props_by_kind[prop_kind_uuid] = []
+                    
+                if prop_kind_uuid != NO_KIND:
+                    self._props_by_kind[prop_kind_uuid].append(prop)
+                else:
+                    # Search if a standard name is given (in resqml 201)
+                    title = get_object_attribute(prop, "propertyKind.kind")
+                    if title is not None:
+                        if is_enum(title):
+                            title = title.value
+                        pk = get_property_kind_by_title(title)
+                        if pk is not None:
+                            self._props_by_kind[pk.uuid].append(prop)
+                        else:
+                            self._props_by_kind[NO_KIND].append(prop)
+                    else:
+                        self._props_by_kind[NO_KIND].append(prop)
 
     def _collect_crs(self):
         # Collect related CRS objects referenced by the representation
@@ -137,7 +158,17 @@ def get_default_color(self) -> ScalarRenderingInfo:
 
     def get_property(self, property_uuid: str) -> Optional[Any]:
         """Return the property object with the given uuid, or None."""
-        return self.properties_by_kind.get(property_uuid)
+        return self._props.get(property_uuid)
+    
+    @property
+    def properties(self) -> Dict[str, Any]:
+        """Return a dict of all properties keyed by property uuid."""
+        return self._props
+    
+    @property
+    def properties_by_kind(self) -> Dict[str, List[Any]]:
+        """Return a dict of properties grouped by property kind uuid."""
+        return self._props_by_kind
 
     def get_properties_time_series(self, property_uuid: str) -> Dict[str, List[Any]]:
         """
@@ -220,9 +251,10 @@ def dump(self) -> str:
 
         lines.append("")
         lines.append(f"  Properties ({len(self.properties_by_kind)}):")
-        for uuid, prop in self.properties_by_kind.items():
-            kind = getattr(prop, "property_kind", "?")
-            lines.append(f"    - {type(prop).__name__}  uuid={uuid}  kind={kind}")
+        for kind, props in self.properties_by_kind.items():
+            lines.append(f"    - Kind {kind} ({len(props)} properties)")
+            for prop in props:
+                lines.append(f"      - {type(prop).__name__}  uuid={get_obj_uuid(prop)}")
 
         lines.append("")
         lines.append(f"  Graphical info ({len(self.graphical_info)} set(s)):")
@@ -272,21 +304,21 @@ def dump(self) -> str:
             print(f"    axis_order={info.projected_axis_order}")
 
     # Detail: property arrays (truncated)
-    if repr_ctx.properties_by_kind:
+    if repr_ctx._props:
         from energyml.utils.data.mesh import read_property
 
         print("\nProperty arrays (first 10 values):")
-        for uuid, prop in repr_ctx.properties_by_kind.items():
+        for uuid, prop in repr_ctx._props.items():
             try:
                 arr = read_property(prop, workspace)
                 print(f"  {type(prop).__name__} [{uuid}]: shape={getattr(arr, 'shape', len(arr))}  sample={arr[:10]}")
             except Exception as exc:
                 print(f"  {type(prop).__name__} [{uuid}]: ERROR reading — {exc}")
 
     # print property time series values
-    if repr_ctx.properties_by_kind:
+    if repr_ctx._props:
         print("\nProperty time series values:")
-        for uuid, prop in repr_ctx.properties_by_kind.items():
+        for uuid, prop in repr_ctx._props.items():
             try:
                 ts_values = repr_ctx.get_properties_time_series(uuid)
                 if ts_values:

energyml-utils/src/energyml/utils/epc_utils.py

@@ -550,6 +550,29 @@ def get_property_kind_by_uuid(uuid: str) -> Optional[Any]:
             logging.error(f"Failed to parse propertykind dict {e}")
     return __CACHE_PROP_KIND_DICT__.get(uuid, None)
 
+def get_property_kind_by_title(title: str) -> Optional[Any]:
+    """
+    Get a property kind by its title.
+    :param title: the title of the property kind
+    :return: the property kind or None if not found
+    """
+    
+    # TODO: Make a mapping between energyml.resqml.v2_0_1.resqmlv2.ResqmlPropertyKind enum and the real pk titles.
+    # Because some values from ResqmlPropertyKind are not the real pk title
+    
+    if len(__CACHE_PROP_KIND_DICT__) == 0:
+        # update the cache to check if it is a
+        try:
+            update_prop_kind_dict_cache()
+        except FileNotFoundError as e:
+            logging.error(f"Failed to parse propertykind dict {e}")
+    title_reshaped = title.replace(" ", "_").lower()
+    for prop in __CACHE_PROP_KIND_DICT__.values():
+        pk_title_reshaped = prop.citation.title.replace(" ", "_").lower() if prop.citation and prop.citation.title else ""
+        if pk_title_reshaped == title_reshaped:
+            return prop
+    return None
+
 
 def get_property_kind_and_parents(uuids: list) -> Dict[str, Any]:
     """Get PropertyKind objects and their parents from a list of UUIDs.
@@ -575,6 +598,13 @@ def get_property_kind_and_parents(uuids: list) -> Dict[str, Any]:
     return dict_props
 
 
+def get_property_kind_uuid_from_property_object(prop: Any) -> Optional[Any]:
+    prop_kind_uuid = get_object_attribute(prop, "propertyKind.*.uuid") # resqml 201
+    if prop_kind_uuid is None:
+        prop_kind_uuid = get_object_attribute(prop, "propertyKind.uuid") # resqml 22
+    return prop_kind_uuid
+
+
 #     ____  ____  ____     ______                 __  _
 #    / __ \/ __ \/ __ \   / ____/_______  ____ _/ /_(_)___  ____
 #   / / / / / / / /_/ /  / /   / ___/ _ \/ __ `/ __/ / __ \/ __ \