refactor: migrate from ConjunctiveGraph to Dataset for RDF quadstore handling

Author: arcangelo7

.gitignore

@@ -35,8 +35,4 @@ nosetests.xml
 
 # Virtuoso lock files
 tests/test_dataset_db/*
-tests/test_provenance_db/*
-
-.claude/
-CLAUDE.md
-icon_oc_positive_animated.html
+tests/test_provenance_db/*

docker-compose.dev.yaml

@@ -5,7 +5,7 @@ services:
       dockerfile: Dockerfile.dev
     ports:
       - "5000:5000"
-      - "6379:6379"  # Redis port for development
+      - "6381:6379"
     environment:
       # Application Configuration
       - FLASK_ENV=demo

heritrace/editor.py

@@ -5,7 +5,7 @@
 from heritrace.extensions import SPARQLWrapperWithRetry
 from rdflib import Graph, Literal, URIRef
 from rdflib_ocdm.counter_handler.counter_handler import CounterHandler
-from rdflib_ocdm.ocdm_graph import OCDMConjunctiveGraph, OCDMGraph
+from rdflib_ocdm.ocdm_graph import OCDMDataset, OCDMGraph
 from rdflib_ocdm.reader import Reader
 from rdflib_ocdm.storer import Storer
 from SPARQLWrapper import JSON
@@ -30,7 +30,7 @@ def __init__(
         self.c_time = self.to_posix_timestamp(c_time)
         self.dataset_is_quadstore = dataset_is_quadstore
         self.g_set = (
-            OCDMConjunctiveGraph(self.counter_handler)
+            OCDMDataset(self.counter_handler)
             if self.dataset_is_quadstore
             else OCDMGraph(self.counter_handler)
         )

heritrace/routes/entity.py

@@ -8,7 +8,7 @@
                    render_template, request, url_for)
 from flask_babel import gettext
 from flask_login import current_user, login_required
-from rdflib import RDF, XSD, ConjunctiveGraph, Graph, Literal, URIRef
+from rdflib import RDF, XSD, Dataset, Graph, Literal, URIRef
 from SPARQLWrapper import JSON
 from time_agnostic_library.agnostic_entity import AgnosticEntity
 
@@ -186,13 +186,19 @@ def about(subject):
             abort(404)
 
         if data_graph:
-            triples = list(data_graph.triples((None, None, None)))
-            subject_classes = [o for s, p, o in data_graph.triples((URIRef(subject), RDF.type, None))]
+            # For Dataset (quadstore), we need to use quads() instead of triples()
+            if isinstance(data_graph, Dataset):
+                # Convert quads to triples by extracting only s, p, o
+                triples = [(s, p, o) for s, p, o, g in data_graph.quads((None, None, None))]
+                subject_classes = [o for s, p, o, g in data_graph.quads((URIRef(subject), RDF.type, None))]
+                subject_triples = [(s, p, o) for s, p, o, g in data_graph.quads((URIRef(subject), None, None))]
+            else:
+                triples = list(data_graph.triples((None, None, None)))
+                subject_classes = [o for s, p, o in data_graph.triples((URIRef(subject), RDF.type, None))]
+                subject_triples = list(data_graph.triples((URIRef(subject), None, None)))
 
             highest_priority_class = get_highest_priority_class(subject_classes)
-            entity_shape = determine_shape_for_entity_triples(
-                list(data_graph.triples((URIRef(subject), None, None)))
-            )
+            entity_shape = determine_shape_for_entity_triples(subject_triples)
 
             (
                 can_be_added,
@@ -432,7 +438,7 @@ def create_nested_entity(
     editor: Editor, entity_uri, entity_data, graph_uri=None
 ):
     form_fields = get_form_fields()
-    
+
     editor.create(
         entity_uri,
         URIRef("http://www.w3.org/1999/02/22-rdf-syntax-ns#type"),
@@ -1388,7 +1394,7 @@ def restore_version(entity_uri, timestamp):
 
 
 def compute_graph_differences(
-    current_graph: Graph | ConjunctiveGraph, historical_graph: Graph | ConjunctiveGraph
+    current_graph: Graph | Dataset, historical_graph: Graph | Dataset
 ):
     if get_dataset_is_quadstore():
         current_data = set(current_graph.quads())

heritrace/utils/display_rules_utils.py

@@ -4,7 +4,7 @@
 
 from heritrace.extensions import (get_custom_filter, get_display_rules,
                                   get_form_fields, get_sparql)
-from rdflib import ConjunctiveGraph, Graph, Literal, URIRef
+from rdflib import Dataset, Graph, Literal, URIRef
 from rdflib.plugins.sparql.algebra import translateQuery
 from rdflib.plugins.sparql.parser import parseQuery
 from SPARQLWrapper import JSON
@@ -629,7 +629,7 @@ def process_ordering(
     grouped_triples,
     display_name,
     fetched_values_map,
-    historical_snapshot: ConjunctiveGraph | Graph | None = None,
+    historical_snapshot: Dataset | Graph | None = None,
 ):
     def get_ordered_sequence(order_results):
         order_map = {}

heritrace/utils/filters.py

@@ -10,7 +10,7 @@
 from flask_babel import format_datetime, gettext, lazy_gettext
 from heritrace.apis.orcid import format_orcid_attribution, is_orcid_url
 from heritrace.apis.zenodo import format_zenodo_source, is_zenodo_url
-from rdflib import ConjunctiveGraph, Graph
+from rdflib import Dataset, Graph
 from SPARQLWrapper import JSON
 
 
@@ -91,7 +91,7 @@ def human_readable_class(self, entity_key):
         return format_uri_as_readable(class_uri)
 
     def human_readable_entity(
-        self, uri: str, entity_key: tuple[str, str | None], graph: Graph | ConjunctiveGraph = None
+        self, uri: str, entity_key: tuple[str, str | None], graph: Graph | Dataset = None
     ) -> str:
         """Convert an entity URI to human-readable format using display rules.
         
@@ -123,7 +123,7 @@ def human_readable_entity(
         return uri
 
     def get_fetch_uri_display(
-        self, uri: str, rule: dict, graph: Graph | ConjunctiveGraph = None
+        self, uri: str, rule: dict, graph: Graph | Dataset = None
     ) -> str | None:
         """Get a display value for an entity URI using fetchUriDisplay rules.
         

heritrace/utils/sparql_utils.py

@@ -3,7 +3,7 @@
 from concurrent.futures import ProcessPoolExecutor, as_completed
 from typing import List
 
-from rdflib import RDF, ConjunctiveGraph, Graph, Literal, URIRef
+from rdflib import RDF, Dataset, Graph, Literal, URIRef
 from rdflib.plugins.sparql.algebra import translateUpdate
 from rdflib.plugins.sparql.parser import parseUpdate
 from SPARQLWrapper import JSON
@@ -482,7 +482,7 @@ def get_catalog_data(
     }
 
 
-def fetch_data_graph_for_subject(subject: str) -> Graph | ConjunctiveGraph:
+def fetch_data_graph_for_subject(subject: str) -> Graph | Dataset:
     """
     Fetch all triples/quads associated with a subject from the dataset.
     Handles both triplestore and quadstore cases appropriately.
@@ -491,9 +491,9 @@ def fetch_data_graph_for_subject(subject: str) -> Graph | ConjunctiveGraph:
         subject (str): The URI of the subject to fetch data for
 
     Returns:
-        Graph|ConjunctiveGraph: A graph containing all triples/quads for the subject
+        Graph|Dataset: A graph containing all triples/quads for the subject
     """
-    g = ConjunctiveGraph() if get_dataset_is_quadstore() else Graph()
+    g = Dataset() if get_dataset_is_quadstore() else Graph()
     sparql = get_sparql()
 
     if is_virtuoso():
@@ -593,17 +593,17 @@ def extract_quads(quads):
 
 def fetch_current_state_with_related_entities(
     provenance: dict,
-) -> Graph | ConjunctiveGraph:
+) -> Graph | Dataset:
     """
     Fetch the current state of an entity and all its related entities known from provenance.
 
     Args:
         provenance (dict): Dictionary containing provenance metadata for main entity and related entities
 
     Returns:
-        ConjunctiveGraph: A graph containing the current state of all entities
+        Dataset: A graph containing the current state of all entities
     """
-    combined_graph = ConjunctiveGraph() if get_dataset_is_quadstore() else Graph()
+    combined_graph = Dataset() if get_dataset_is_quadstore() else Graph()
 
     # Fetch state for all entities mentioned in provenance
     for entity_uri in provenance.keys():
@@ -746,7 +746,7 @@ def process_deleted_entity(result: dict, sortable_properties: list) -> dict | No
         return None
 
     last_valid_time = convert_to_datetime(last_valid_snapshot_time, stringify=True)
-    last_valid_state: ConjunctiveGraph = state[entity_uri][last_valid_time]
+    last_valid_state: Dataset = state[entity_uri][last_valid_time]
 
     entity_types = [
         str(o)

poetry.lock

@@ -12,7 +12,7 @@ python = "^3.10,<3.14"
 flask = "^2.3.3"
 sparqlwrapper = "^2.0.0"
 pyyaml = "^6.0.1"
-flask-babel = "^3.1.0"
+flask-babel = "^4.0.0"
 validators = "0.20.0"
 requests = "^2.31.0"
 flask-wtf = "^1.1.1"
@@ -21,7 +21,7 @@ setuptools = "^68.2.2"
 docker = "^7.1.0"
 requests-oauthlib = "^2.0.0"
 flask-login = "^0.6.3"
-rdflib-ocdm = "^1.0.4"
+rdflib-ocdm = "^1.0.7"
 pyopenssl = "^24.3.0"
 time-agnostic-library = "^5.0.5"
 

pyproject.toml

@@ -13,7 +13,7 @@
 # Import the editor module
 from heritrace.editor import Editor
 from rdflib import RDF, XSD, Literal, URIRef
-from rdflib_ocdm.ocdm_graph import OCDMConjunctiveGraph, OCDMGraph
+from rdflib_ocdm.ocdm_graph import OCDMDataset, OCDMGraph
 from tests.test_config import TestConfig
 
 
@@ -63,7 +63,7 @@ def reset_editor_after_test(editor: Editor):
     yield
     # Reset editor state
     editor.g_set = (
-        OCDMConjunctiveGraph(editor.counter_handler)
+        OCDMDataset(editor.counter_handler)
         if editor.dataset_is_quadstore
         else OCDMGraph(editor.counter_handler)
     )