cognite-neat 0.78.5__tar.gz → 0.79.0__tar.gz

{cognite_neat-0.78.5 → cognite_neat-0.79.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: cognite-neat
-Version: 0.78.5
+Version: 0.79.0
 Summary: Knowledge graph transformation
 Home-page: https://cognite-neat.readthedocs-hosted.com/
 License: Apache-2.0

cognite_neat-0.79.0/cognite/neat/_version.py

@@ -0,0 +1 @@
+__version__ = "0.79.0"

{cognite_neat-0.78.5 → cognite_neat-0.79.0}/cognite/neat/graph/stores/_base.py

@@ -3,10 +3,12 @@ import sys
 import time
 from abc import ABC, abstractmethod
 from collections.abc import Iterable, Iterator
+from datetime import datetime
 from pathlib import Path
 from typing import Literal, TypeAlias, cast
 
 import pandas as pd
+import pytz
 from prometheus_client import Gauge, Summary
 from rdflib import RDF, Graph, Namespace, URIRef
 from rdflib.query import Result, ResultRow
@@ -16,6 +18,8 @@ from cognite.neat.graph.models import Triple
 from cognite.neat.graph.stores._rdf_to_graph import rdf_file_to_graph
 from cognite.neat.utils import remove_namespace
 
+from ._provenance import Change, Provenance
+
 if sys.version_info >= (3, 11):
     pass
 else:
@@ -49,6 +53,7 @@ class NeatGraphStoreBase(ABC):
         namespace: Namespace = DEFAULT_NAMESPACE,
         prefixes: dict = PREFIXES,
     ):
+        _start = datetime.now(pytz.utc)
         self.graph = graph or Graph()
         self.base_prefix: str = base_prefix
         self.namespace: Namespace = namespace
@@ -63,6 +68,16 @@ class NeatGraphStoreBase(ABC):
         self.internal_storage_dir_orig: Path | None = None
         self.storage_dirs_to_delete: list[Path] = []
         self.queries = _Queries(self)
+        self.provenance = Provenance(
+            [
+                Change.record(
+                    activity=f"{type(self).__name__}.__init__",
+                    start=_start,
+                    end=datetime.now(pytz.utc),
+                    description="Initialize graph store",
+                )
+            ]
+        )
 
     @abstractmethod
     def _set_graph(self) -> None:

cognite_neat-0.79.0/cognite/neat/graph/stores/_provenance.py

@@ -0,0 +1,99 @@
+# we will use prov-o to represent the provenance of the neat graph store
+# basically tracking changes that occur in the graph store
+# prov-o use concepts of Agent, Activity and Entity to represent provenance
+# where in case of neat we have:
+# Agent: triples extractors, graph enhancers, contextualizers, etc.
+# Activity: write triple, add connection, etc.
+# Entity: neat graph store
+
+
+import uuid
+from collections import UserList
+from collections.abc import Sequence
+from dataclasses import dataclass
+from datetime import datetime
+from typing import TypeVar
+
+from rdflib import PROV, RDF, Literal, URIRef
+
+from cognite.neat.constants import DEFAULT_NAMESPACE
+
+
+@dataclass(frozen=True)
+class Agent:
+    id_: URIRef = DEFAULT_NAMESPACE.agent
+    acted_on_behalf_of: str = "NEAT"
+
+    def as_triples(self):
+        return [
+            (self.id_, RDF.type, PROV[type(self).__name__]),
+            (self.id_, PROV.actedOnBehalfOf, self.acted_on_behalf_of),
+        ]
+
+
+@dataclass(frozen=True)
+class Activity:
+    was_associated_with: Agent
+    ended_at_time: datetime
+    started_at_time: datetime
+    used: str  # this would be set to for example Extractor, Enhancer, Contextualizer, etc.
+    id_: URIRef = DEFAULT_NAMESPACE[f"activity-{uuid.uuid4()}"]
+
+    def as_triples(self):
+        return [
+            (self.id_, RDF.type, PROV[type(self).__name__]),
+            (self.id_, PROV.wasAssociatedWith, self.was_associated_with.id_),
+            (self.id_, PROV.startedAtTime, Literal(self.started_at_time)),
+            (self.id_, PROV.endedAtTime, Literal(self.ended_at_time)),
+            (self.id_, PROV.used, self.used),
+        ]
+
+
+@dataclass(frozen=True)
+class Entity:
+    was_generated_by: Activity
+    was_attributed_to: Agent
+    id_: URIRef = DEFAULT_NAMESPACE["graph-store"]
+
+    def as_triples(self):
+        return [
+            (self.id_, RDF.type, PROV[type(self).__name__]),
+            (self.id_, PROV.wasGeneratedBy, self.was_generated_by.id_),
+            (self.id_, PROV.wasAttributedTo, self.was_attributed_to.id_),
+        ]
+
+
+@dataclass(frozen=True)
+class Change:
+    agent: Agent
+    activity: Activity
+    entity: Entity
+    description: str
+
+    def as_triples(self):
+        return self.agent.as_triples() + self.activity.as_triples() + self.entity.as_triples()
+
+    @classmethod
+    def record(cls, activity: str, start: datetime, end: datetime, description: str):
+        """User friendly method to record a change that occurred in the graph store."""
+        agent = Agent()
+        activity = Activity(used=activity, was_associated_with=agent, started_at_time=start, ended_at_time=end)
+        entity = Entity(was_generated_by=activity, was_attributed_to=agent)
+        return cls(agent, activity, entity, description)
+
+
+T_Change = TypeVar("T_Change", bound=Change)
+
+
+class Provenance(UserList[T_Change]):
+    def __init__(self, changes: Sequence[T_Change] | None = None):
+        super().__init__(changes or [])
+
+    def did_this_happen(self, this: str) -> bool:
+        return any(change.description == this for change in self)
+
+    def __delitem__(self, *args, **kwargs):
+        raise TypeError("Cannot delete change from provenance")
+
+    def __setitem__(self, *args, **kwargs):
+        raise TypeError("Cannot modify change from provenance")

{cognite_neat-0.78.5 → cognite_neat-0.79.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "cognite-neat"
-version = "0.78.5"
+version = "0.79.0"
 readme = "README.md"
 description = "Knowledge graph transformation"
 authors = [

cognite_neat-0.78.5/cognite/neat/_version.py

@@ -1 +0,0 @@
-__version__ = "0.78.5"