cognite-neat 0.123.23__py3-none-any.whl → 0.123.25__py3-none-any.whl

cognite/neat/_version.py

@@ -1,2 +1,2 @@
-__version__ = "0.123.23"
+__version__ = "0.123.25"
 __engine__ = "^2.0.4"

cognite/neat/core/_constants.py

@@ -250,6 +250,11 @@ def is_hierarchy_property(container: dm.ContainerId, property_: str) -> bool:
     )
 
 
+def cognite_prefixes() -> dict[str, Namespace]:
+    """Returns the Cognite prefixes and namespaces."""
+    return {space: Namespace(CDF_NAMESPACE[space] + "/") for space in COGNITE_SPACES}
+
+
 DMS_RESERVED_PROPERTIES = frozenset(
     {
         "createdTime",

cognite/neat/core/_data_model/importers/_rdf/_shared.py

@@ -1,17 +1,21 @@
-from typing import cast
+from typing import Any, Literal, cast
 from urllib.parse import quote
 
-from rdflib import BNode, Graph
+from rdflib import BNode, Graph, Namespace, URIRef
+from rdflib import Literal as RdfLiteral
 from rdflib.plugins.sparql import prepareQuery
 from rdflib.query import ResultRow
 
+from cognite.neat.core._constants import cognite_prefixes
+from cognite.neat.core._data_model.models.entities._constants import Unknown
+from cognite.neat.core._data_model.models.entities._single_value import ConceptEntity
 from cognite.neat.core._issues._base import IssueList
 from cognite.neat.core._issues.errors._general import NeatValueError
 from cognite.neat.core._issues.warnings._resources import (
     ResourceRedefinedWarning,
     ResourceRetrievalWarning,
 )
-from cognite.neat.core._utils.rdf_ import convert_rdflib_content
+from cognite.neat.core._utils.rdf_ import remove_namespace_from_uri, uri_to_entity_components
 
 
 def parse_concepts(
@@ -33,27 +37,31 @@ def parse_concepts(
     concepts: dict[str, dict] = {}
 
     query = prepareQuery(query.format(language=language), initNs={k: v for k, v in graph.namespaces()})
+    prefixes = cognite_prefixes()
 
     for raw in graph.query(query):
-        res: dict = convert_rdflib_content(cast(ResultRow, raw).asdict(), True)
+        res: dict = convert_rdflib_content(
+            cast(ResultRow, raw).asdict(), uri_handling="as-concept-entity", prefixes=prefixes
+        )
         res = {key: res.get(key, None) for key in parameters}
 
-        # Quote the concept id to ensure it is web-safe
-        res["concept"] = quote(res["concept"], safe="")
-
-        concept_id = res["concept"]
-
         # Safeguarding against incomplete semantic definitions
         if res["implements"] and isinstance(res["implements"], BNode):
             issue_list.append(
                 ResourceRetrievalWarning(
-                    concept_id,
+                    res["concept"],
                     "implements",
                     error=("Unable to determine concept that is being implemented"),
                 )
             )
             continue
 
+        # sanitize the concept and implements
+        res["concept"] = sanitize_entity(res["concept"])
+        res["implements"] = sanitize_entity(res["implements"]) if res["implements"] else None
+
+        concept_id = res["concept"]
+
         if concept_id not in concepts:
             concepts[concept_id] = res
         else:
@@ -72,7 +80,6 @@ def parse_concepts(
 
             handle_meta("concept", concepts, concept_id, res, "name", issue_list)
             handle_meta("concept", concepts, concept_id, res, "description", issue_list)
-
     if not concepts:
         issue_list.append(NeatValueError("Unable to parse concepts"))
 
@@ -98,9 +105,12 @@ def parse_properties(
     properties: dict[str, dict] = {}
 
     query = prepareQuery(query.format(language=language), initNs={k: v for k, v in graph.namespaces()})
+    prefixes = cognite_prefixes()
 
     for raw in graph.query(query):
-        res: dict = convert_rdflib_content(cast(ResultRow, raw).asdict(), True)
+        res: dict = convert_rdflib_content(
+            cast(ResultRow, raw).asdict(), uri_handling="as-concept-entity", prefixes=prefixes
+        )
         res = {key: res.get(key, None) for key in parameters}
 
         # Quote the concept id to ensure it is web-safe
@@ -129,9 +139,9 @@ def parse_properties(
             )
             continue
 
-        # Quote the concept and value_type to ensure they are web-safe
-        res["concept"] = quote(res["concept"], safe="") if res["concept"] else "#N/A"
-        res["value_type"] = quote(res["value_type"], safe="") if res["value_type"] else "#N/A"
+        # Quote the concept and value_type if they exist if not signal neat that they are not available
+        res["concept"] = sanitize_entity(res["concept"]) if res["concept"] else str(Unknown)
+        res["value_type"] = sanitize_entity(res["value_type"]) if res["value_type"] else str(Unknown)
 
         id_ = f"{res['concept']}.{res['property_']}"
 
@@ -173,14 +183,73 @@ def handle_meta(
     if not resources[resource_id][feature] and res[feature]:
         resources[resource_id][feature] = res[feature]
 
-    # RAISE warning only if the feature is being redefined
-    elif resources[resource_id][feature] and res[feature]:
+    current_value = resources[resource_id][feature]
+    new_value = res[feature]
+
+    if not current_value and new_value:
+        resources[resource_id][feature] = new_value
+    elif current_value and new_value and current_value != new_value:
         issue_list.append(
             ResourceRedefinedWarning(
                 identifier=resource_id,
                 resource_type=resource_type,
                 feature=feature,
-                current_value=resources[resource_id][feature],
-                new_value=res[feature],
+                current_value=current_value,
+                new_value=new_value,
             )
         )
+
+
+def convert_rdflib_content(
+    content: RdfLiteral | URIRef | dict | list,
+    uri_handling: Literal["skip", "remove-namespace", "as-concept-entity"] = "skip",
+    prefixes: dict[str, Namespace] | None = None,
+) -> Any:
+    """Converts rdflib content to a more Python-friendly format.
+
+    Args:
+        content: The content to convert, can be a RdfLiteral, URIRef, dict, or list.
+        uri_handling: How to handle URIs. Options are:
+            - "skip": Leave URIs as is.
+            - "remove-namespace": Remove the namespace from URIs.
+            - "short-form": Convert URIs to a short form using prefixes.
+
+    """
+    if isinstance(content, RdfLiteral):
+        return content.toPython()
+    elif isinstance(content, URIRef):
+        if uri_handling == "remove-namespace":
+            return remove_namespace_from_uri(content)
+        elif uri_handling == "as-concept-entity":
+            if components := uri_to_entity_components(content, prefixes or {}):
+                return ConceptEntity(prefix=components[0], suffix=components[3], version=components[2])
+            # fallback to "remove-namespace"
+            else:
+                return convert_rdflib_content(content, uri_handling="remove-namespace", prefixes=prefixes)
+        else:
+            return content.toPython()
+    elif isinstance(content, dict):
+        return {key: convert_rdflib_content(value, uri_handling, prefixes) for key, value in content.items()}
+    elif isinstance(content, list):
+        return [convert_rdflib_content(item, uri_handling, prefixes) for item in content]
+    else:
+        return content
+
+
+def sanitize_entity(entity: str | ConceptEntity, safe: str = "") -> str:
+    """Sanitize an entity to ensure it yields entity form that will pass downstream validation.
+
+    Args:
+        entity: The entity to sanitize. Can be a string or a ConceptEntity.
+        safe: Characters that should not be quoted during sanitization.
+
+    Returns:
+        A web-safe string representation of the entity
+    """
+    if isinstance(entity, str):
+        return quote(entity, safe=safe)
+    # if it already we dont need to quote it so we return its string representation
+    elif isinstance(entity, ConceptEntity):
+        return str(entity)
+    else:
+        raise ValueError(f"Invalid entity type: {type(entity)}. Expected str, ConceptEntity.")

cognite/neat/core/_utils/rdf_.py

@@ -147,19 +147,6 @@ def as_neat_compliant_uri(uri: URIRef) -> URIRef:
     return URIRef(f"{namespace}{compliant_uri}")
 
 
-def convert_rdflib_content(content: RdfLiteral | URIRef | dict | list, remove_namespace: bool = False) -> Any:
-    if isinstance(content, RdfLiteral):
-        return content.toPython()
-    elif isinstance(content, URIRef):
-        return remove_namespace_from_uri(content) if remove_namespace else content.toPython()
-    elif isinstance(content, dict):
-        return {key: convert_rdflib_content(value, remove_namespace) for key, value in content.items()}
-    elif isinstance(content, list):
-        return [convert_rdflib_content(item, remove_namespace) for item in content]
-    else:
-        return content
-
-
 def uri_to_short_form(URI: URIRef, prefixes: dict[str, Namespace]) -> str | URIRef:
     """Returns the short form of a URI if its namespace is present in the prefixes dict,
     otherwise returns the URI itself
@@ -179,6 +166,43 @@ def uri_to_short_form(URI: URIRef, prefixes: dict[str, Namespace]) -> str | URIR
     return min(uris, key=len)
 
 
+def uri_to_entity_components(uri: URIRef, prefixes: dict[str, Namespace]) -> tuple[str, str, str, str] | None:
+    """Converts a URI to its components: space, data_model_id, version, and entity_id.
+    Args:
+        uri: URI to be converted
+        prefixes: dict of prefixes
+
+    Returns:
+        tuple of space, data_model_id, version, and entity_id if found,
+        otherwise None
+
+    !!! note "URI Format"
+        The URI is expected to be in the form of `.../<space>/<data_model_id>/<version>/<entity_id>` to
+        be able to extract the components correctly.
+
+        An example of a valid entity URI is:
+
+        `https://cognitedata.com/cdf_cdm/CogniteCore/v1/CogniteAsset` , where:
+
+        - space is `cdf_cdm`
+        - data_model_id is `CogniteCore`
+        - version is `v1`
+        - entity_id is `CogniteAsset`
+
+        to be able to parse the URI correctly, the prefixes dict must have
+        the corresponding prefix registered:
+        {'cdf_cdm': Namespace('https://cognitedata.com/cdf_cdm/CogniteCore/v1/')}
+
+        for this method to return the correct components.
+    """
+    for prefix, namespace in prefixes.items():
+        if uri.startswith(namespace):
+            remainder = str(uri)[len(str(namespace)) :]
+            if (components := remainder.split("/")) and len(components) == 3 and all(components):
+                return prefix, components[0], components[1], components[2]
+    return None
+
+
 def _traverse(hierarchy: dict, graph: dict, names: list[str]) -> dict:
     """traverse the graph and return the hierarchy"""
     for name in names:

{cognite_neat-0.123.23.dist-info → cognite_neat-0.123.25.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cognite-neat
-Version: 0.123.23
+Version: 0.123.25
 Summary: Knowledge graph transformation
 Project-URL: Documentation, https://cognite-neat.readthedocs-hosted.com/
 Project-URL: Homepage, https://cognite-neat.readthedocs-hosted.com/

{cognite_neat-0.123.23.dist-info → cognite_neat-0.123.25.dist-info}/RECORD

@@ -1,9 +1,9 @@
 cognite/neat/__init__.py,sha256=12StS1dzH9_MElqxGvLWrNsxCJl9Hv8A2a9D0E5OD_U,193
-cognite/neat/_version.py,sha256=0ecHvh-Qq-gIDd9cT8TLmFo99So1sxYnycVfCgckPL8,47
+cognite/neat/_version.py,sha256=r7KKTj1I5Kj3gInGNGwe1UoX8sdMd4-Z1v_2uzagy1s,47
 cognite/neat/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 cognite/neat/core/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 cognite/neat/core/_config.py,sha256=WT1BS8uADcFvGoUYOOfwFOVq_VBl472TisdoA3wLick,280
-cognite/neat/core/_constants.py,sha256=Cc7p1QbpbzPSjlf-PB_Ow2pKRhWQL-K8VJma96HzU3A,8863
+cognite/neat/core/_constants.py,sha256=wIpGOzZAKS2vhshXR1K51cbcsgq2TlvZaZ3zw91CU9I,9054
 cognite/neat/core/_shared.py,sha256=Ov59SWYboRRsncB_5V1ZC_BAoACfNLjo80vqE5Ru6wo,2325
 cognite/neat/core/_client/__init__.py,sha256=RQ7MwL8mwGqGHokRzsPqO3XStDzmI4-c12_gz1UPJ74,177
 cognite/neat/core/_client/_api_client.py,sha256=CqgG4kEArI9jiKAh82YrRZv_SzeMKA5guIZOvDg2R5I,860
@@ -45,7 +45,7 @@ cognite/neat/core/_data_model/importers/_rdf/__init__.py,sha256=1yOjV2PKCxwH7uCT
 cognite/neat/core/_data_model/importers/_rdf/_base.py,sha256=XTdKqN4nf0d2Vvk56LShrDoyCQCg9MXvKwxXT0uMnGg,6016
 cognite/neat/core/_data_model/importers/_rdf/_inference2rdata_model.py,sha256=PCgM9-qGSLlupN7tYCFLHjivgICtMiahNry1ub8JCYk,28934
 cognite/neat/core/_data_model/importers/_rdf/_owl2data_model.py,sha256=l_89N1LewZVjSttcJkFAJj63JW8wI-THKJYjAxcNiDg,3093
-cognite/neat/core/_data_model/importers/_rdf/_shared.py,sha256=qibq94dFoy_g1ED5nIFYwHC_LYI4jI67HFVcMy7abrM,6665
+cognite/neat/core/_data_model/importers/_rdf/_shared.py,sha256=Cni9yaIKnKh6tHngxhxgseVveB7ZV_doA92fUbMd8gY,9774
 cognite/neat/core/_data_model/models/__init__.py,sha256=hmF7MDR1XmpLxYdMkOEuPuHUqOQKE4AgsuUqdc-ySSQ,1249
 cognite/neat/core/_data_model/models/_base_unverified.py,sha256=1Wfbp-tJaEF6hd1bFdp2FhTgPkInf-1ZokuEoVJRPxQ,6842
 cognite/neat/core/_data_model/models/_base_verified.py,sha256=nzPrlj7ZvYull_Fdh2zeDXz98hux-eQOdTGy9jhUtYA,15127
@@ -149,7 +149,7 @@ cognite/neat/core/_utils/auxiliary.py,sha256=FpeVlscVue4QFaUM3jt7LLiAMkiec6IvpYr
 cognite/neat/core/_utils/collection_.py,sha256=zVrSmm4045pjw6Pt6y4VPTIJ4dXdMJPyOV70LdFyDBM,2376
 cognite/neat/core/_utils/graph_transformations_report.py,sha256=ORVH7lw357TPOq4elU5lH46Qx6GCLVrSj-1nX6Ggk1U,1235
 cognite/neat/core/_utils/io_.py,sha256=D2Mg8sOxfBoDg3fC0jBzaxO3vkXmr0QvZSgYIv6xRkM,386
-cognite/neat/core/_utils/rdf_.py,sha256=8AALp8H_nXEDSBo6jZ1idyT_x3K4PJT5ZyBEyxPmgxI,10403
+cognite/neat/core/_utils/rdf_.py,sha256=2O4yaOO2af6NBBCFcKm5YMQRxOhjspqS6N3p5T3sV5g,11223
 cognite/neat/core/_utils/spreadsheet.py,sha256=VdjcCh339cKu9UwxJkYrmmnrbLguD71tdFmyd3MlIZA,5951
 cognite/neat/core/_utils/tarjan.py,sha256=IZvwaIITryGVNbo9Bv5EA9_sW3DyfUNAe7uYyPOCL0g,1357
 cognite/neat/core/_utils/text.py,sha256=ON4ihfscFJkQqQ-Rj46XXtf-9tAobwXbbfa3wuekSu4,8519
@@ -195,7 +195,7 @@ cognite/neat/session/engine/__init__.py,sha256=D3MxUorEs6-NtgoICqtZ8PISQrjrr4dvc
 cognite/neat/session/engine/_import.py,sha256=1QxA2_EK613lXYAHKQbZyw2yjo5P9XuiX4Z6_6-WMNQ,169
 cognite/neat/session/engine/_interface.py,sha256=3W-cYr493c_mW3P5O6MKN1xEQg3cA7NHR_ev3zdF9Vk,533
 cognite/neat/session/engine/_load.py,sha256=g52uYakQM03VqHt_RDHtpHso1-mFFifH5M4T2ScuH8A,5198
-cognite_neat-0.123.23.dist-info/METADATA,sha256=f78SKr8-LucvAchjKK1niKDLVAWwdadAm5x4dOdl_5s,9172
-cognite_neat-0.123.23.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-cognite_neat-0.123.23.dist-info/licenses/LICENSE,sha256=W8VmvFia4WHa3Gqxq1Ygrq85McUNqIGDVgtdvzT-XqA,11351
-cognite_neat-0.123.23.dist-info/RECORD,,
+cognite_neat-0.123.25.dist-info/METADATA,sha256=UATxdmkTTf9TwT0BwXNWjPVRT1LKrWejsWN5S5DLFcM,9172
+cognite_neat-0.123.25.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+cognite_neat-0.123.25.dist-info/licenses/LICENSE,sha256=W8VmvFia4WHa3Gqxq1Ygrq85McUNqIGDVgtdvzT-XqA,11351
+cognite_neat-0.123.25.dist-info/RECORD,,