graphiti-core 0.19.0rc2__py3-none-any.whl → 0.20.0__py3-none-any.whl

graphiti_core/utils/bulk_utils.py

@@ -14,6 +14,7 @@ See the License for the specific language governing permissions and
 limitations under the License.
 """
 
+import json
 import logging
 import typing
 from datetime import datetime
@@ -22,20 +23,21 @@ import numpy as np
 from pydantic import BaseModel, Field
 from typing_extensions import Any
 
-from graphiti_core.driver.driver import GraphDriver, GraphDriverSession
+from graphiti_core.driver.driver import GraphDriver, GraphDriverSession, GraphProvider
 from graphiti_core.edges import Edge, EntityEdge, EpisodicEdge, create_entity_edge_embeddings
 from graphiti_core.embedder import EmbedderClient
 from graphiti_core.graphiti_types import GraphitiClients
 from graphiti_core.helpers import normalize_l2, semaphore_gather
 from graphiti_core.models.edges.edge_db_queries import (
-    EPISODIC_EDGE_SAVE_BULK,
     get_entity_edge_save_bulk_query,
+    get_episodic_edge_save_bulk_query,
 )
 from graphiti_core.models.nodes.node_db_queries import (
     get_entity_node_save_bulk_query,
     get_episode_node_save_bulk_query,
 )
 from graphiti_core.nodes import EntityNode, EpisodeType, EpisodicNode, create_entity_node_embeddings
+from graphiti_core.utils.datetime_utils import convert_datetimes_to_strings
 from graphiti_core.utils.maintenance.edge_operations import (
     extract_edges,
     resolve_extracted_edge,
@@ -116,11 +118,16 @@ async def add_nodes_and_edges_bulk_tx(
     episodes = [dict(episode) for episode in episodic_nodes]
     for episode in episodes:
         episode['source'] = str(episode['source'].value)
-        episode['group_label'] = 'Episodic_' + episode['group_id'].replace('-', '')
-    nodes: list[dict[str, Any]] = []
+        episode.pop('labels', None)
+        if driver.provider == GraphProvider.NEO4J:
+            episode['group_label'] = 'Episodic_' + episode['group_id'].replace('-', '')
+
+    nodes = []
+
     for node in entity_nodes:
         if node.name_embedding is None:
             await node.generate_name_embedding(embedder)
+
         entity_data: dict[str, Any] = {
             'uuid': node.uuid,
             'name': node.name,
@@ -130,13 +137,19 @@ async def add_nodes_and_edges_bulk_tx(
             'created_at': node.created_at,
         }
 
-        entity_data.update(node.attributes or {})
-        entity_data['labels'] = list(
-            set(node.labels + ['Entity', 'Entity_' + node.group_id.replace('-', '')])
-        )
+        entity_data['labels'] = list(set(node.labels + ['Entity']))
+        if driver.provider == GraphProvider.KUZU:
+            attributes = convert_datetimes_to_strings(node.attributes) if node.attributes else {}
+            entity_data['attributes'] = json.dumps(attributes)
+        else:
+            entity_data.update(node.attributes or {})
+            entity_data['labels'] = list(
+                set(node.labels + ['Entity', 'Entity_' + node.group_id.replace('-', '')])
+            )
+
         nodes.append(entity_data)
 
-    edges: list[dict[str, Any]] = []
+    edges = []
     for edge in entity_edges:
         if edge.fact_embedding is None:
             await edge.generate_embedding(embedder)
@@ -155,17 +168,36 @@ async def add_nodes_and_edges_bulk_tx(
             'invalid_at': edge.invalid_at,
         }
 
-        edge_data.update(edge.attributes or {})
+        if driver.provider == GraphProvider.KUZU:
+            attributes = convert_datetimes_to_strings(edge.attributes) if edge.attributes else {}
+            edge_data['attributes'] = json.dumps(attributes)
+        else:
+            edge_data.update(edge.attributes or {})
+
         edges.append(edge_data)
 
-    await tx.run(get_episode_node_save_bulk_query(driver.provider), episodes=episodes)
-    entity_node_save_bulk = get_entity_node_save_bulk_query(driver.provider, nodes)
-    await tx.run(entity_node_save_bulk, nodes=nodes)
-    await tx.run(
-        EPISODIC_EDGE_SAVE_BULK, episodic_edges=[edge.model_dump() for edge in episodic_edges]
-    )
-    entity_edge_save_bulk = get_entity_edge_save_bulk_query(driver.provider)
-    await tx.run(entity_edge_save_bulk, entity_edges=edges)
+    if driver.provider == GraphProvider.KUZU:
+        # FIXME: Kuzu's UNWIND does not currently support STRUCT[] type properly, so we insert the data one by one instead for now.
+        episode_query = get_episode_node_save_bulk_query(driver.provider)
+        for episode in episodes:
+            await tx.run(episode_query, **episode)
+        entity_node_query = get_entity_node_save_bulk_query(driver.provider, nodes)
+        for node in nodes:
+            await tx.run(entity_node_query, **node)
+        entity_edge_query = get_entity_edge_save_bulk_query(driver.provider)
+        for edge in edges:
+            await tx.run(entity_edge_query, **edge)
+        episodic_edge_query = get_episodic_edge_save_bulk_query(driver.provider)
+        for edge in episodic_edges:
+            await tx.run(episodic_edge_query, **edge.model_dump())
+    else:
+        await tx.run(get_episode_node_save_bulk_query(driver.provider), episodes=episodes)
+        await tx.run(get_entity_node_save_bulk_query(driver.provider, nodes), nodes=nodes)
+        await tx.run(
+            get_episodic_edge_save_bulk_query(driver.provider),
+            episodic_edges=[edge.model_dump() for edge in episodic_edges],
+        )
+        await tx.run(get_entity_edge_save_bulk_query(driver.provider), entity_edges=edges)
 
 
 async def extract_nodes_and_edges_bulk(

graphiti_core/utils/datetime_utils.py

@@ -40,3 +40,16 @@ def ensure_utc(dt: datetime | None) -> datetime | None:
         return dt.astimezone(timezone.utc)
 
     return dt
+
+
+def convert_datetimes_to_strings(obj):
+    if isinstance(obj, dict):
+        return {k: convert_datetimes_to_strings(v) for k, v in obj.items()}
+    elif isinstance(obj, list):
+        return [convert_datetimes_to_strings(item) for item in obj]
+    elif isinstance(obj, tuple):
+        return tuple(convert_datetimes_to_strings(item) for item in obj)
+    elif isinstance(obj, datetime):
+        return obj.isoformat()
+    else:
+        return obj

graphiti_core/utils/maintenance/community_operations.py

@@ -4,11 +4,12 @@ from collections import defaultdict
 
 from pydantic import BaseModel
 
-from graphiti_core.driver.driver import GraphDriver
+from graphiti_core.driver.driver import GraphDriver, GraphProvider
 from graphiti_core.edges import CommunityEdge
 from graphiti_core.embedder import EmbedderClient
 from graphiti_core.helpers import semaphore_gather
 from graphiti_core.llm_client import LLMClient
+from graphiti_core.models.nodes.node_db_queries import COMMUNITY_NODE_RETURN
 from graphiti_core.nodes import CommunityNode, EntityNode, get_community_node_from_record
 from graphiti_core.prompts import prompt_library
 from graphiti_core.prompts.summarize_nodes import Summary, SummaryDescription
@@ -33,11 +34,11 @@ async def get_community_clusters(
     if group_ids is None:
         group_id_values, _, _ = await driver.execute_query(
             """
-        MATCH (n:Entity)
-        WHERE n.group_id IS NOT NULL
-        RETURN
-            collect(DISTINCT n.group_id) AS group_ids
-        """,
+            MATCH (n:Entity)
+            WHERE n.group_id IS NOT NULL
+            RETURN
+                collect(DISTINCT n.group_id) AS group_ids
+            """
         )
 
         group_ids = group_id_values[0]['group_ids'] if group_id_values else []
@@ -46,14 +47,21 @@ async def get_community_clusters(
         projection: dict[str, list[Neighbor]] = {}
         nodes = await EntityNode.get_by_group_ids(driver, [group_id])
         for node in nodes:
-            records, _, _ = await driver.execute_query(
+            match_query = """
+                MATCH (n:Entity {group_id: $group_id, uuid: $uuid})-[e:RELATES_TO]-(m: Entity {group_id: $group_id})
+            """
+            if driver.provider == GraphProvider.KUZU:
+                match_query = """
+                MATCH (n:Entity {group_id: $group_id, uuid: $uuid})-[:RELATES_TO]-(e:RelatesToNode_)-[:RELATES_TO]-(m: Entity {group_id: $group_id})
                 """
-            MATCH (n:Entity {group_id: $group_id, uuid: $uuid})-[r:RELATES_TO]-(m: Entity {group_id: $group_id})
-            WITH count(r) AS count, m.uuid AS uuid
-            RETURN
-                uuid,
-                count
-            """,
+            records, _, _ = await driver.execute_query(
+                match_query
+                + """
+                WITH count(e) AS count, m.uuid AS uuid
+                RETURN
+                    uuid,
+                    count
+                """,
                 uuid=node.uuid,
                 group_id=group_id,
             )
@@ -235,9 +243,9 @@ async def build_communities(
 async def remove_communities(driver: GraphDriver):
     await driver.execute_query(
         """
-    MATCH (c:Community)
-    DETACH DELETE c
-    """,
+        MATCH (c:Community)
+        DETACH DELETE c
+        """
     )
 
 
@@ -247,14 +255,10 @@ async def determine_entity_community(
     # Check if the node is already part of a community
     records, _, _ = await driver.execute_query(
         """
-    MATCH (c:Community)-[:HAS_MEMBER]->(n:Entity {uuid: $entity_uuid})
-    RETURN
-        c.uuid AS uuid,
-        c.name AS name,
-        c.group_id AS group_id,
-        c.created_at AS created_at,
-        c.summary AS summary
-    """,
+        MATCH (c:Community)-[:HAS_MEMBER]->(n:Entity {uuid: $entity_uuid})
+        RETURN
+        """
+        + COMMUNITY_NODE_RETURN,
         entity_uuid=entity.uuid,
     )
 
@@ -262,16 +266,19 @@ async def determine_entity_community(
         return get_community_node_from_record(records[0]), False
 
     # If the node has no community, add it to the mode community of surrounding entities
+    match_query = """
+        MATCH (c:Community)-[:HAS_MEMBER]->(m:Entity)-[:RELATES_TO]-(n:Entity {uuid: $entity_uuid})
+    """
+    if driver.provider == GraphProvider.KUZU:
+        match_query = """
+            MATCH (c:Community)-[:HAS_MEMBER]->(m:Entity)-[:RELATES_TO]-(e:RelatesToNode_)-[:RELATES_TO]-(n:Entity {uuid: $entity_uuid})
+        """
     records, _, _ = await driver.execute_query(
+        match_query
+        + """
+        RETURN
         """
-    MATCH (c:Community)-[:HAS_MEMBER]->(m:Entity)-[:RELATES_TO]-(n:Entity {uuid: $entity_uuid})
-    RETURN
-        c.uuid AS uuid,
-        c.name AS name,
-        c.group_id AS group_id,
-        c.created_at AS created_at,
-        c.summary AS summary
-    """,
+        + COMMUNITY_NODE_RETURN,
         entity_uuid=entity.uuid,
     )
 

graphiti_core/utils/maintenance/edge_operations.py

@@ -531,17 +531,28 @@ async def filter_existing_duplicate_of_edges(
             routing_='r',
         )
     else:
-        query: LiteralString = """
-            UNWIND $duplicate_node_uuids AS duplicate_tuple
-            MATCH (n:Entity {uuid: duplicate_tuple[0]})-[r:RELATES_TO {name: 'IS_DUPLICATE_OF'}]->(m:Entity {uuid: duplicate_tuple[1]})
-            RETURN DISTINCT
-                n.uuid AS source_uuid,
-                m.uuid AS target_uuid
-        """
+        if driver.provider == GraphProvider.KUZU:
+            query = """
+                UNWIND $duplicate_node_uuids AS duplicate
+                MATCH (n:Entity {uuid: duplicate.src})-[:RELATES_TO]->(e:RelatesToNode_ {name: 'IS_DUPLICATE_OF'})-[:RELATES_TO]->(m:Entity {uuid: duplicate.dst})
+                RETURN DISTINCT
+                    n.uuid AS source_uuid,
+                    m.uuid AS target_uuid
+            """
+            duplicate_node_uuids = [{'src': src, 'dst': dst} for src, dst in duplicate_nodes_map]
+        else:
+            query: LiteralString = """
+                UNWIND $duplicate_node_uuids AS duplicate_tuple
+                MATCH (n:Entity {uuid: duplicate_tuple[0]})-[r:RELATES_TO {name: 'IS_DUPLICATE_OF'}]->(m:Entity {uuid: duplicate_tuple[1]})
+                RETURN DISTINCT
+                    n.uuid AS source_uuid,
+                    m.uuid AS target_uuid
+            """
+            duplicate_node_uuids = list(duplicate_nodes_map.keys())
 
         records, _, _ = await driver.execute_query(
             query,
-            duplicate_node_uuids=list(duplicate_nodes_map.keys()),
+            duplicate_node_uuids=duplicate_node_uuids,
             routing_='r',
         )
 

graphiti_core/utils/maintenance/graph_data_operations.py

@@ -35,7 +35,8 @@ logger = logging.getLogger(__name__)
 
 async def build_indices_and_constraints(driver: GraphDriver, delete_existing: bool = False):
     if driver.provider == GraphProvider.NEPTUNE:
-        return  # Neptune does not need indexes built
+        await driver.create_aoss_indices()  # pyright: ignore[reportAttributeAccessIssue]
+        return
     if delete_existing:
         records, _, _ = await driver.execute_query(
             """
@@ -52,10 +53,29 @@ async def build_indices_and_constraints(driver: GraphDriver, delete_existing: bo
                 for name in index_names
             ]
         )
+
     range_indices: list[LiteralString] = get_range_indices(driver.provider)
 
     fulltext_indices: list[LiteralString] = get_fulltext_indices(driver.provider)
 
+    if driver.provider == GraphProvider.KUZU:
+        # Skip creating fulltext indices if they already exist. Need to do this manually
+        # until Kuzu supports `IF NOT EXISTS` for indices.
+        result, _, _ = await driver.execute_query('CALL SHOW_INDEXES() RETURN *;')
+        if len(result) > 0:
+            fulltext_indices = []
+
+        # Only load the `fts` extension if it's not already loaded, otherwise throw an error.
+        result, _, _ = await driver.execute_query('CALL SHOW_LOADED_EXTENSIONS() RETURN *;')
+        if len(result) == 0:
+            fulltext_indices.insert(
+                0,
+                """
+                INSTALL fts;
+                LOAD fts;
+                """,
+            )
+
     index_queries: list[LiteralString] = range_indices + fulltext_indices
 
     await semaphore_gather(
@@ -75,10 +95,19 @@ async def clear_data(driver: GraphDriver, group_ids: list[str] | None = None):
             await tx.run('MATCH (n) DETACH DELETE n')
 
         async def delete_group_ids(tx):
-            await tx.run(
-                'MATCH (n) WHERE (n:Entity OR n:Episodic OR n:Community) AND n.group_id IN $group_ids DETACH DELETE n',
-                group_ids=group_ids,
-            )
+            labels = ['Entity', 'Episodic', 'Community']
+            if driver.provider == GraphProvider.KUZU:
+                labels.append('RelatesToNode_')
+
+            for label in labels:
+                await tx.run(
+                    f"""
+                    MATCH (n:{label})
+                    WHERE n.group_id IN $group_ids
+                    DETACH DELETE n
+                    """,
+                    group_ids=group_ids,
+                )
 
         if group_ids is None:
             await session.execute_write(delete_all)
@@ -107,18 +136,23 @@ async def retrieve_episodes(
     Returns:
         list[EpisodicNode]: A list of EpisodicNode objects representing the retrieved episodes.
     """
-    group_id_filter: LiteralString = (
-        '\nAND e.group_id IN $group_ids' if group_ids and len(group_ids) > 0 else ''
-    )
-    source_filter: LiteralString = '\nAND e.source = $source' if source is not None else ''
+
+    query_params: dict = {}
+    query_filter = ''
+    if group_ids and len(group_ids) > 0:
+        query_filter += '\nAND e.group_id IN $group_ids'
+        query_params['group_ids'] = group_ids
+
+    if source is not None:
+        query_filter += '\nAND e.source = $source'
+        query_params['source'] = source.name
 
     query: LiteralString = (
         """
-                MATCH (e:Episodic)
-                WHERE e.valid_at <= $reference_time
-                """
-        + group_id_filter
-        + source_filter
+            MATCH (e:Episodic)
+            WHERE e.valid_at <= $reference_time
+            """
+        + query_filter
         + """
         RETURN
         """
@@ -135,9 +169,8 @@ async def retrieve_episodes(
     result, _, _ = await driver.execute_query(
         query,
         reference_time=reference_time,
-        source=source.name if source is not None else None,
         num_episodes=last_n,
-        group_ids=group_ids,
+        **query_params,
     )
 
     episodes = [get_episodic_node_from_record(record) for record in result]
@@ -147,41 +180,39 @@ async def retrieve_episodes(
 async def build_dynamic_indexes(driver: GraphDriver, group_id: str):
     # Make sure indices exist for this group_id in Neo4j
     if driver.provider == GraphProvider.NEO4J:
-        await semaphore_gather(
-            driver.execute_query(
-                """CREATE FULLTEXT INDEX $episode_content IF NOT EXISTS
-FOR (e:"""
-                + 'Episodic_'
-                + group_id.replace('-', '')
-                + """) ON EACH [e.content, e.source, e.source_description, e.group_id]""",
-                episode_content='episode_content_' + group_id.replace('-', ''),
-            ),
-            driver.execute_query(
-                """CREATE FULLTEXT INDEX $node_name_and_summary IF NOT EXISTS FOR (n:"""
-                + 'Entity_'
-                + group_id.replace('-', '')
-                + """) ON EACH [n.name, n.summary, n.group_id]""",
-                node_name_and_summary='node_name_and_summary_' + group_id.replace('-', ''),
-            ),
-            driver.execute_query(
-                """CREATE FULLTEXT INDEX $community_name IF NOT EXISTS
-                                                         FOR (n:"""
-                + 'Community_'
-                + group_id.replace('-', '')
-                + """) ON EACH [n.name, n.group_id]""",
-                community_name='Community_' + group_id.replace('-', ''),
-            ),
-            driver.execute_query(
-                """CREATE VECTOR INDEX $group_entity_vector IF NOT EXISTS
-                                                        FOR (n:"""
-                + 'Entity_'
-                + group_id.replace('-', '')
-                + """)
+        await driver.execute_query(
+            """CREATE FULLTEXT INDEX $episode_content IF NOT EXISTS 
+            FOR (e:"""
+            + 'Episodic_'
+            + group_id.replace('-', '')
+            + """) ON EACH [e.content, e.source, e.source_description, e.group_id]""",
+            episode_content='episode_content_' + group_id.replace('-', ''),
+        )
+        await driver.execute_query(
+            """CREATE FULLTEXT INDEX $node_name_and_summary IF NOT EXISTS FOR (n:"""
+            + 'Entity_'
+            + group_id.replace('-', '')
+            + """) ON EACH [n.name, n.summary, n.group_id]""",
+            node_name_and_summary='node_name_and_summary_' + group_id.replace('-', ''),
+        )
+        await driver.execute_query(
+            """CREATE FULLTEXT INDEX $community_name IF NOT EXISTS
+                                                     FOR (n:"""
+            + 'Community_'
+            + group_id.replace('-', '')
+            + """) ON EACH [n.name, n.group_id]""",
+            community_name='Community_' + group_id.replace('-', ''),
+        )
+        await driver.execute_query(
+            """CREATE VECTOR INDEX $group_entity_vector IF NOT EXISTS
+                                                    FOR (n:"""
+            + 'Entity_'
+            + group_id.replace('-', '')
+            + """)
                                ON n.embedding
                                OPTIONS { indexConfig: {
                                 `vector.dimensions`: 1024,
                                 `vector.similarity_function`: 'cosine'
                                }}""",
-                group_entity_vector='group_entity_vector_' + group_id.replace('-', ''),
-            ),
+            group_entity_vector='group_entity_vector_' + group_id.replace('-', ''),
         )