nucliadb 6.3.1.post3584__py3-none-any.whl → 6.3.1.post3590__py3-none-any.whl

nucliadb/backups/create.py

@@ -30,7 +30,6 @@ from nucliadb.backups.const import (
 )
 from nucliadb.backups.models import BackupMetadata, CreateBackupRequest
 from nucliadb.backups.settings import settings
-from nucliadb.backups.utils import exists_in_storge
 from nucliadb.common import datamanagers
 from nucliadb.common.context import ApplicationContext
 from nucliadb.export_import.utils import (
@@ -261,7 +260,7 @@ async def delete_metadata(context: ApplicationContext, kbid: str, backup_id: str
 
 
 async def exists_cf(context: ApplicationContext, cf: resources_pb2.CloudFile) -> bool:
-    return await exists_in_storge(context.blob_storage, cf.bucket_name, cf.uri)
+    return await context.blob_storage.exists_object(bucket=cf.bucket_name, key=cf.uri)
 
 
 async def upload_to_bucket(context: ApplicationContext, bytes_iterator: AsyncIterator[bytes], key: str):

nucliadb/backups/utils.py

@@ -24,12 +24,7 @@ from nucliadb_utils.storages.storage import Storage
 
 
 async def exists_backup(storage: Storage, backup_id: str) -> bool:
-    return await exists_in_storge(
-        storage, settings.backups_bucket, StorageKeys.BACKUP_PREFIX.format(backup_id=backup_id)
+    # As the labels file is always created, we use it to check if the backup exists
+    return await storage.exists_object(
+        settings.backups_bucket, StorageKeys.LABELS.format(backup_id=backup_id)
     )
-
-
-async def exists_in_storge(storage: Storage, bucket: str, key: str) -> bool:
-    async for _ in storage.iterate_objects(bucket=bucket, prefix=key):
-        return True
-    return False

nucliadb/common/nidx.py

@@ -262,6 +262,7 @@ class NodeNidxAdapter:
         # Searcher methods
         self.Search = searcher_client.Search
         self.Suggest = searcher_client.Suggest
+        self.GraphSearch = searcher_client.GraphSearch
         self.Paragraphs = searcher_client.Paragraphs
         self.Documents = searcher_client.Documents
 

nucliadb/search/api/v1/__init__.py

@@ -17,16 +17,19 @@
 # You should have received a copy of the GNU Affero General Public License
 # along with this program. If not, see <http://www.gnu.org/licenses/>.
 #
-from . import ask  # noqa
-from . import catalog  # noqa
-from . import feedback  # noqa
-from . import find  # noqa
-from . import knowledgebox  # noqa
-from . import predict_proxy  # noqa
-from . import search  # noqa
-from . import suggest  # noqa
-from . import summarize  # noqa
-from .resource import ask as ask_resource  # noqa
-from .resource import search as search_resource  # noqa
-from .resource import ingestion_agents as ingestion_agents_resource  # noqa
-from .router import api  # noqa
+from . import (  # noqa: F401
+    ask,
+    catalog,
+    feedback,
+    find,
+    graph,
+    knowledgebox,
+    predict_proxy,
+    search,
+    suggest,
+    summarize,
+)
+from .resource import ask as ask_resource  # noqa: F401
+from .resource import ingestion_agents as ingestion_agents_resource  # noqa: F401
+from .resource import search as search_resource  # noqa: F401
+from .router import api  # noqa: F401

nucliadb/search/api/v1/catalog.py

@@ -36,7 +36,7 @@ from nucliadb.search.search import cache
 from nucliadb.search.search.exceptions import InvalidQueryError
 from nucliadb.search.search.merge import fetch_resources
 from nucliadb.search.search.pgcatalog import pgcatalog_search
-from nucliadb.search.search.query_parser.catalog import parse_catalog
+from nucliadb.search.search.query_parser.parsers import parse_catalog
 from nucliadb.search.search.utils import (
     maybe_log_request_payload,
 )

nucliadb/search/api/v1/graph.py

@@ -0,0 +1,130 @@
+# Copyright (C) 2021 Bosutech XXI S.L.
+#
+# nucliadb is offered under the AGPL v3.0 and as commercial software.
+# For commercial licensing, contact us at info@nuclia.com.
+#
+# AGPL:
+# This program is free software: you can redistribute it and/or modify
+# it under the terms of the GNU Affero General Public License as
+# published by the Free Software Foundation, either version 3 of the
+# License, or (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU Affero General Public License for more details.
+#
+# You should have received a copy of the GNU Affero General Public License
+# along with this program. If not, see <http://www.gnu.org/licenses/>.
+#
+from fastapi import Header, Request, Response
+from fastapi_versioning import version
+
+from nucliadb.search.api.v1.router import KB_PREFIX, api
+from nucliadb.search.requesters.utils import Method, node_query
+from nucliadb.search.search.graph_merge import (
+    build_graph_nodes_response,
+    build_graph_relations_response,
+    build_graph_response,
+)
+from nucliadb.search.search.query_parser.parsers import (
+    parse_graph_node_search,
+    parse_graph_relation_search,
+    parse_graph_search,
+)
+from nucliadb_models.graph.requests import (
+    GraphNodesSearchRequest,
+    GraphRelationsSearchRequest,
+    GraphSearchRequest,
+)
+from nucliadb_models.graph.responses import (
+    GraphNodesSearchResponse,
+    GraphRelationsSearchResponse,
+    GraphSearchResponse,
+)
+from nucliadb_models.resource import NucliaDBRoles
+from nucliadb_models.search import (
+    NucliaDBClientType,
+)
+from nucliadb_utils.authentication import requires
+
+
+@api.post(
+    f"/{KB_PREFIX}/{{kbid}}/graph",
+    status_code=200,
+    summary="Search Knowledge Box graph",
+    description="Search on the Knowledge Box graph and retrieve triplets of vertex-edge-vertex",
+    response_model_exclude_unset=True,
+    include_in_schema=False,
+    tags=["Search"],
+)
+@requires(NucliaDBRoles.READER)
+@version(1)
+async def graph_search_knowledgebox(
+    request: Request,
+    response: Response,
+    kbid: str,
+    item: GraphSearchRequest,
+    x_ndb_client: NucliaDBClientType = Header(NucliaDBClientType.API),
+    x_nucliadb_user: str = Header(""),
+    x_forwarded_for: str = Header(""),
+) -> GraphSearchResponse:
+    pb_query = parse_graph_search(item)
+
+    results, _, _ = await node_query(kbid, Method.GRAPH, pb_query)
+
+    return build_graph_response(results)
+
+
+@api.post(
+    f"/{KB_PREFIX}/{{kbid}}/graph/nodes",
+    status_code=200,
+    summary="Search Knowledge Box graph nodes",
+    description="Search on the Knowledge Box graph and retrieve nodes (vertices)",
+    response_model_exclude_unset=True,
+    include_in_schema=False,
+    tags=["Search"],
+)
+@requires(NucliaDBRoles.READER)
+@version(1)
+async def graph_nodes_search_knowledgebox(
+    request: Request,
+    response: Response,
+    kbid: str,
+    item: GraphNodesSearchRequest,
+    x_ndb_client: NucliaDBClientType = Header(NucliaDBClientType.API),
+    x_nucliadb_user: str = Header(""),
+    x_forwarded_for: str = Header(""),
+) -> GraphNodesSearchResponse:
+    pb_query = parse_graph_node_search(item)
+
+    results, _, _ = await node_query(kbid, Method.GRAPH, pb_query)
+
+    return build_graph_nodes_response(results)
+
+
+@api.post(
+    f"/{KB_PREFIX}/{{kbid}}/graph/relations",
+    status_code=200,
+    summary="Search Knowledge Box graph relations",
+    description="Search on the Knowledge Box graph and retrieve relations (edges)",
+    response_model_exclude_unset=True,
+    include_in_schema=False,
+    tags=["Search"],
+)
+@requires(NucliaDBRoles.READER)
+@version(1)
+async def graph_relations_search_knowledgebox(
+    request: Request,
+    response: Response,
+    kbid: str,
+    item: GraphRelationsSearchRequest,
+    x_ndb_client: NucliaDBClientType = Header(NucliaDBClientType.API),
+    x_nucliadb_user: str = Header(""),
+    x_forwarded_for: str = Header(""),
+) -> GraphRelationsSearchResponse:
+    pb_query = parse_graph_relation_search(item)
+
+    results, _, _ = await node_query(kbid, Method.GRAPH, pb_query)
+
+    return build_graph_relations_response(results)

nucliadb/search/requesters/utils.py

@@ -33,11 +33,14 @@ from nucliadb.common.cluster.exceptions import ShardsNotFound
 from nucliadb.common.cluster.utils import get_shard_manager
 from nucliadb.search import logger
 from nucliadb.search.search.shards import (
+    graph_search_shard,
     query_shard,
     suggest_shard,
 )
 from nucliadb.search.settings import settings
 from nucliadb_protos.nodereader_pb2 import (
+    GraphSearchRequest,
+    GraphSearchResponse,
     SearchRequest,
     SearchResponse,
     SuggestRequest,
@@ -50,19 +53,22 @@ from nucliadb_telemetry import errors
 class Method(Enum):
     SEARCH = auto()
     SUGGEST = auto()
+    GRAPH = auto()
 
 
 METHODS = {
     Method.SEARCH: query_shard,
     Method.SUGGEST: suggest_shard,
+    Method.GRAPH: graph_search_shard,
 }
 
-REQUEST_TYPE = Union[SuggestRequest, SearchRequest]
+REQUEST_TYPE = Union[SuggestRequest, SearchRequest, GraphSearchRequest]
 
 T = TypeVar(
     "T",
     SuggestResponse,
     SearchResponse,
+    GraphSearchResponse,
 )
 
 
@@ -84,6 +90,15 @@ async def node_query(
 ) -> tuple[list[SearchResponse], bool, list[tuple[AbstractIndexNode, str]]]: ...
 
 
+@overload
+async def node_query(
+    kbid: str,
+    method: Method,
+    pb_query: GraphSearchRequest,
+    timeout: Optional[float] = None,
+) -> tuple[list[GraphSearchResponse], bool, list[tuple[AbstractIndexNode, str]]]: ...
+
+
 async def node_query(
     kbid: str,
     method: Method,

nucliadb/search/search/find.py

@@ -40,7 +40,7 @@ from nucliadb.search.search.metrics import (
 )
 from nucliadb.search.search.query import QueryParser
 from nucliadb.search.search.query_parser.old_filters import OldFilterParams
-from nucliadb.search.search.query_parser.parser import parse_find
+from nucliadb.search.search.query_parser.parsers import parse_find
 from nucliadb.search.search.rank_fusion import (
     RankFusionAlgorithm,
     get_rank_fusion,

nucliadb/search/search/graph_merge.py

@@ -0,0 +1,90 @@
+# Copyright (C) 2021 Bosutech XXI S.L.
+#
+# nucliadb is offered under the AGPL v3.0 and as commercial software.
+# For commercial licensing, contact us at info@nuclia.com.
+#
+# AGPL:
+# This program is free software: you can redistribute it and/or modify
+# it under the terms of the GNU Affero General Public License as
+# published by the Free Software Foundation, either version 3 of the
+# License, or (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU Affero General Public License for more details.
+#
+# You should have received a copy of the GNU Affero General Public License
+# along with this program. If not, see <http://www.gnu.org/licenses/>.
+#
+
+
+from nucliadb.common.models_utils.from_proto import RelationNodeTypePbMap
+from nucliadb_models.graph import responses as graph_responses
+from nucliadb_models.graph.responses import (
+    GraphNodesSearchResponse,
+    GraphRelationsSearchResponse,
+    GraphSearchResponse,
+)
+from nucliadb_protos import nodereader_pb2
+
+
+def build_graph_response(results: list[nodereader_pb2.GraphSearchResponse]) -> GraphSearchResponse:
+    paths = []
+    for shard_results in results:
+        for pb_path in shard_results.graph:
+            source = shard_results.nodes[pb_path.source]
+            relation = shard_results.relations[pb_path.relation]
+            destination = shard_results.nodes[pb_path.destination]
+
+            path = graph_responses.GraphPath(
+                source=graph_responses.GraphNode(
+                    value=source.value,
+                    type=RelationNodeTypePbMap[source.ntype],
+                    group=source.subtype,
+                ),
+                relation=graph_responses.GraphRelation(
+                    label=relation.label,
+                ),
+                destination=graph_responses.GraphNode(
+                    value=destination.value,
+                    type=RelationNodeTypePbMap[destination.ntype],
+                    group=destination.subtype,
+                ),
+            )
+            paths.append(path)
+
+    response = GraphSearchResponse(paths=paths)
+    return response
+
+
+def build_graph_nodes_response(
+    results: list[nodereader_pb2.GraphSearchResponse],
+) -> GraphNodesSearchResponse:
+    nodes = []
+    for shard_results in results:
+        for node in shard_results.nodes:
+            nodes.append(
+                graph_responses.GraphNode(
+                    value=node.value,
+                    type=RelationNodeTypePbMap[node.ntype],
+                    group=node.subtype,
+                )
+            )
+    response = GraphNodesSearchResponse(nodes=nodes)
+    return response
+
+
+def build_graph_relations_response(
+    results: list[nodereader_pb2.GraphSearchResponse],
+) -> GraphRelationsSearchResponse:
+    relations = []
+    for shard_results in results:
+        for relation in shard_results.relations:
+            relations.append(
+                graph_responses.GraphRelation(
+                    label=relation.label,
+                )
+            )
+    response = GraphRelationsSearchResponse(relations=relations)
+    return response

nucliadb/search/search/pgcatalog.py

@@ -111,7 +111,7 @@ def _convert_date_filter(date: CatalogExpression.Date, filter_params: dict[str,
         raise ValueError(f"Invalid date operator")
 
 
-def _prepare_query(catalog_query: CatalogQuery):
+def _prepare_query_filters(catalog_query: CatalogQuery) -> tuple[str, dict[str, Any]]:
     filter_sql = ["kbid = %(kbid)s"]
     filter_params: dict[str, Any] = {"kbid": catalog_query.kbid}
 
@@ -127,7 +127,17 @@ def _prepare_query(catalog_query: CatalogQuery):
     if catalog_query.filters:
         filter_sql.append(_convert_filter(catalog_query.filters, filter_params))
 
-    order_sql = ""
+    return (
+        f"SELECT * FROM catalog WHERE {' AND '.join(filter_sql)}",
+        filter_params,
+    )
+
+
+def _prepare_query(catalog_query: CatalogQuery) -> tuple[str, dict[str, Any]]:
+    # Base query with all the filters
+    query, filter_params = _prepare_query_filters(catalog_query)
+
+    # Sort
     if catalog_query.sort:
         if catalog_query.sort.field == SortField.CREATED:
             order_field = "created_at"
@@ -144,12 +154,15 @@ def _prepare_query(catalog_query: CatalogQuery):
         else:
             order_dir = "DESC"
 
-        order_sql = f" ORDER BY {order_field} {order_dir}"
+        query += f" ORDER BY {order_field} {order_dir}"
 
-    return (
-        f"SELECT * FROM catalog WHERE {' AND '.join(filter_sql)}{order_sql}",
-        filter_params,
-    )
+    # Pagination
+    offset = catalog_query.page_size * catalog_query.page_number
+    query += f" LIMIT %(page_size)s OFFSET %(offset)s"
+    filter_params["page_size"] = catalog_query.page_size
+    filter_params["offset"] = offset
+
+    return query, filter_params
 
 
 def _pg_driver() -> PGDriver:
@@ -159,7 +172,7 @@ def _pg_driver() -> PGDriver:
 @observer.wrap({"op": "search"})
 async def pgcatalog_search(catalog_query: CatalogQuery) -> Resources:
     # Prepare SQL query
-    query, query_params = _prepare_query(catalog_query)
+    query, query_params = _prepare_query_filters(catalog_query)
 
     async with _pg_driver()._get_connection() as conn, conn.cursor(row_factory=dict_row) as cur:
         facets = {}
@@ -210,15 +223,8 @@ async def pgcatalog_search(catalog_query: CatalogQuery) -> Resources:
 
         # Query
         with observer({"op": "query"}):
-            offset = catalog_query.page_size * catalog_query.page_number
-            await cur.execute(
-                f"{query} LIMIT %(page_size)s OFFSET %(offset)s",
-                {
-                    **query_params,
-                    "page_size": catalog_query.page_size,
-                    "offset": offset,
-                },
-            )
+            query, query_params = _prepare_query(catalog_query)
+            await cur.execute(query, query_params)
             data = await cur.fetchall()
 
     return Resources(
@@ -237,6 +243,6 @@ async def pgcatalog_search(catalog_query: CatalogQuery) -> Resources:
         total=total,
         page_number=catalog_query.page_number,
         page_size=catalog_query.page_size,
-        next_page=(offset + len(data) < total),
+        next_page=(catalog_query.page_size * catalog_query.page_number + len(data) < total),
         min_score=0,
     )

nucliadb/search/search/query_parser/models.py

@@ -28,6 +28,7 @@ from pydantic import (
 )
 
 from nucliadb_models import search as search_models
+from nucliadb_protos import nodereader_pb2
 
 ### Retrieval
 
@@ -101,3 +102,11 @@ class CatalogQuery(BaseModel):
     faceted: list[str]
     page_size: int
     page_number: int
+
+
+### Graph
+
+
+# Right now, we don't need a more generic model for graph queries, we can
+# directly use the protobuffer directly
+GraphRetrieval = nodereader_pb2.GraphSearchRequest

nucliadb/search/search/query_parser/parsers/__init__.py

@@ -0,0 +1,23 @@
+# Copyright (C) 2021 Bosutech XXI S.L.
+#
+# nucliadb is offered under the AGPL v3.0 and as commercial software.
+# For commercial licensing, contact us at info@nuclia.com.
+#
+# AGPL:
+# This program is free software: you can redistribute it and/or modify
+# it under the terms of the GNU Affero General Public License as
+# published by the Free Software Foundation, either version 3 of the
+# License, or (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU Affero General Public License for more details.
+#
+# You should have received a copy of the GNU Affero General Public License
+# along with this program. If not, see <http://www.gnu.org/licenses/>.
+#
+
+from .catalog import parse_catalog  # noqa: F401
+from .find import parse_find  # noqa: F401
+from .graph import parse_graph_node_search, parse_graph_relation_search, parse_graph_search  # noqa: F401

nucliadb/search/search/query_parser/{catalog.py → parsers/catalog.py}

@@ -18,10 +18,10 @@
 # along with this program. If not, see <http://www.gnu.org/licenses/>.
 #
 
-
 from nucliadb.common import datamanagers
 from nucliadb.search.search.exceptions import InvalidQueryError
 from nucliadb.search.search.filters import translate_label
+from nucliadb.search.search.query_parser.filter_expression import FacetFilterTypes, facet_from_filter
 from nucliadb.search.search.query_parser.models import (
     CatalogExpression,
     CatalogQuery,
@@ -44,8 +44,6 @@ from nucliadb_models.search import (
     SortOrder,
 )
 
-from .filter_expression import FacetFilterTypes, facet_from_filter
-
 
 async def parse_catalog(kbid: str, item: search_models.CatalogRequest) -> CatalogQuery:
     has_old_filters = (

nucliadb/search/search/query_parser/{parser.py → parsers/find.py}

@@ -18,7 +18,6 @@
 # along with this program. If not, see <http://www.gnu.org/licenses/>.
 #
 
-
 from pydantic import ValidationError
 
 from nucliadb.search.search.query_parser.exceptions import InternalParserError

nucliadb/search/search/query_parser/parsers/graph.py

@@ -0,0 +1,177 @@
+# Copyright (C) 2021 Bosutech XXI S.L.
+#
+# nucliadb is offered under the AGPL v3.0 and as commercial software.
+# For commercial licensing, contact us at info@nuclia.com.
+#
+# AGPL:
+# This program is free software: you can redistribute it and/or modify
+# it under the terms of the GNU Affero General Public License as
+# published by the Free Software Foundation, either version 3 of the
+# License, or (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU Affero General Public License for more details.
+#
+# You should have received a copy of the GNU Affero General Public License
+# along with this program. If not, see <http://www.gnu.org/licenses/>.
+#
+
+
+from nucliadb.common.models_utils.from_proto import RelationNodeTypeMap
+from nucliadb.search.search.query_parser.models import GraphRetrieval
+from nucliadb_models.graph import requests as graph_requests
+from nucliadb_protos import nodereader_pb2
+
+
+def parse_graph_search(item: graph_requests.GraphSearchRequest) -> GraphRetrieval:
+    pb = nodereader_pb2.GraphSearchRequest()
+    pb.query.path.CopyFrom(_parse_path_query(item.query))
+    pb.top_k = item.top_k
+    return pb
+
+
+def parse_graph_node_search(item: graph_requests.GraphNodesSearchRequest) -> GraphRetrieval:
+    pb = nodereader_pb2.GraphSearchRequest()
+    pb.query.path.CopyFrom(_parse_node_query(item.query))
+    pb.top_k = item.top_k
+    return pb
+
+
+def parse_graph_relation_search(item: graph_requests.GraphRelationsSearchRequest) -> GraphRetrieval:
+    pb = nodereader_pb2.GraphSearchRequest()
+    pb.query.path.CopyFrom(_parse_relation_query(item.query))
+    pb.top_k = item.top_k
+    return pb
+
+
+def _parse_path_query(expr: graph_requests.GraphPathQuery) -> nodereader_pb2.GraphQuery.PathQuery:
+    pb = nodereader_pb2.GraphQuery.PathQuery()
+
+    if isinstance(expr, graph_requests.And):
+        for op in expr.operands:
+            pb.bool_and.operands.append(_parse_path_query(op))
+
+    elif isinstance(expr, graph_requests.Or):
+        for op in expr.operands:
+            pb.bool_or.operands.append(_parse_path_query(op))
+
+    elif isinstance(expr, graph_requests.Not):
+        pb.bool_not.CopyFrom(_parse_path_query(expr.operand))
+
+    elif isinstance(expr, graph_requests.GraphPath):
+        if expr.source is not None:
+            _set_node_to_pb(expr.source, pb.path.source)
+
+        if expr.destination is not None:
+            _set_node_to_pb(expr.destination, pb.path.destination)
+
+        if expr.relation is not None:
+            relation = expr.relation
+            if relation.label is not None:
+                pb.path.relation.value = relation.label
+
+        pb.path.undirected = expr.undirected
+
+    elif isinstance(expr, graph_requests.SourceNode):
+        _set_node_to_pb(expr, pb.path.source)
+
+    elif isinstance(expr, graph_requests.DestinationNode):
+        _set_node_to_pb(expr, pb.path.destination)
+
+    elif isinstance(expr, graph_requests.AnyNode):
+        _set_node_to_pb(expr, pb.path.source)
+        pb.path.undirected = True
+
+    elif isinstance(expr, graph_requests.Relation):
+        if expr.label is not None:
+            pb.path.relation.value = expr.label
+
+    else:  # pragma: nocover
+        # This is a trick so mypy generates an error if this branch can be reached,
+        # that is, if we are missing some ifs
+        _a: int = "a"
+
+    return pb
+
+
+def _parse_node_query(expr: graph_requests.GraphNodesQuery) -> nodereader_pb2.GraphQuery.PathQuery:
+    pb = nodereader_pb2.GraphQuery.PathQuery()
+
+    if isinstance(expr, graph_requests.And):
+        for op in expr.operands:
+            pb.bool_and.operands.append(_parse_node_query(op))
+
+    elif isinstance(expr, graph_requests.Or):
+        for op in expr.operands:
+            pb.bool_or.operands.append(_parse_node_query(op))
+
+    elif isinstance(expr, graph_requests.Not):
+        pb.bool_not.CopyFrom(_parse_node_query(expr.operand))
+
+    elif isinstance(expr, graph_requests.SourceNode):
+        _set_node_to_pb(expr, pb.path.source)
+
+    elif isinstance(expr, graph_requests.DestinationNode):
+        _set_node_to_pb(expr, pb.path.destination)
+
+    elif isinstance(expr, graph_requests.AnyNode):
+        _set_node_to_pb(expr, pb.path.source)
+        pb.path.undirected = True
+
+    else:  # pragma: nocover
+        # This is a trick so mypy generates an error if this branch can be reached,
+        # that is, if we are missing some ifs
+        _a: int = "a"
+
+    return pb
+
+
+def _parse_relation_query(
+    expr: graph_requests.GraphRelationsQuery,
+) -> nodereader_pb2.GraphQuery.PathQuery:
+    pb = nodereader_pb2.GraphQuery.PathQuery()
+
+    if isinstance(expr, graph_requests.And):
+        for op in expr.operands:
+            pb.bool_and.operands.append(_parse_relation_query(op))
+
+    elif isinstance(expr, graph_requests.Or):
+        for op in expr.operands:
+            pb.bool_or.operands.append(_parse_relation_query(op))
+
+    elif isinstance(expr, graph_requests.Not):
+        pb.bool_not.CopyFrom(_parse_relation_query(expr.operand))
+
+    elif isinstance(expr, graph_requests.Relation):
+        if expr.label is not None:
+            pb.path.relation.value = expr.label
+
+    else:  # pragma: nocover
+        # This is a trick so mypy generates an error if this branch can be reached,
+        # that is, if we are missing some ifs
+        _a: int = "a"
+
+    return pb
+
+
+def _set_node_to_pb(node: graph_requests.GraphNode, pb: nodereader_pb2.GraphQuery.Node):
+    if node.value is not None:
+        pb.value = node.value
+        if node.match == graph_requests.NodeMatchKind.EXACT:
+            pb.match_kind = nodereader_pb2.GraphQuery.Node.MatchKind.EXACT
+
+        elif node.match == graph_requests.NodeMatchKind.FUZZY:
+            pb.match_kind = nodereader_pb2.GraphQuery.Node.MatchKind.FUZZY
+
+        else:  # pragma: nocover
+            # This is a trick so mypy generates an error if this branch can be reached,
+            # that is, if we are missing some ifs
+            _a: int = "a"
+
+    if node.type is not None:
+        pb.node_type = RelationNodeTypeMap[node.type]
+
+    if node.group is not None:
+        pb.node_subtype = node.group

nucliadb/search/search/shards.py

@@ -26,6 +26,8 @@ from grpc.aio import AioRpcError
 from nucliadb.common.cluster.base import AbstractIndexNode
 from nucliadb_protos.nodereader_pb2 import (
     GetShardRequest,
+    GraphSearchRequest,
+    GraphSearchResponse,
     SearchRequest,
     SearchResponse,
     SuggestRequest,
@@ -79,3 +81,16 @@ async def suggest_shard(node: AbstractIndexNode, shard: str, query: SuggestReque
     req.shard = shard
     with node_observer({"type": "suggest", "node_id": node.id}):
         return await node.reader.Suggest(req)  # type: ignore
+
+
+@backoff.on_exception(
+    backoff.expo, Exception, jitter=None, factor=0.1, max_tries=3, giveup=should_giveup
+)
+async def graph_search_shard(
+    node: AbstractIndexNode, shard: str, query: GraphSearchRequest
+) -> GraphSearchResponse:
+    req = GraphSearchRequest()
+    req.CopyFrom(query)
+    req.shard = shard
+    with node_observer({"type": "graph_search", "node_id": node.id}):
+        return await node.reader.GraphSearch(req)  # type: ignore

{nucliadb-6.3.1.post3584.dist-info → nucliadb-6.3.1.post3590.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: nucliadb
-Version: 6.3.1.post3584
+Version: 6.3.1.post3590
 Summary: NucliaDB
 Author-email: Nuclia <nucliadb@nuclia.com>
 License: AGPL
@@ -20,11 +20,11 @@ Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3 :: Only
 Requires-Python: <4,>=3.9
 Description-Content-Type: text/markdown
-Requires-Dist: nucliadb-telemetry[all]>=6.3.1.post3584
-Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.3.1.post3584
-Requires-Dist: nucliadb-protos>=6.3.1.post3584
-Requires-Dist: nucliadb-models>=6.3.1.post3584
-Requires-Dist: nidx-protos>=6.3.1.post3584
+Requires-Dist: nucliadb-telemetry[all]>=6.3.1.post3590
+Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.3.1.post3590
+Requires-Dist: nucliadb-protos>=6.3.1.post3590
+Requires-Dist: nucliadb-models>=6.3.1.post3590
+Requires-Dist: nidx-protos>=6.3.1.post3590
 Requires-Dist: nucliadb-admin-assets>=1.0.0.post1224
 Requires-Dist: nuclia-models>=0.24.2
 Requires-Dist: uvicorn

{nucliadb-6.3.1.post3584.dist-info → nucliadb-6.3.1.post3590.dist-info}/RECORD

@@ -41,19 +41,19 @@ nucliadb/openapi.py,sha256=wDiw0dVEvTpJvbatkJ0JZLkKm9RItZT5PWRHjqRfqTA,2272
 nucliadb/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nucliadb/backups/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn4bIXMa20ks,835
 nucliadb/backups/const.py,sha256=9vPAhLxQO_gNAjSdPxWuv3V66s9WcdpjOQ89CZlfmuk,1894
-nucliadb/backups/create.py,sha256=mvirguMbtxNgSDGG81l0kkgHWJSZPk4GFyra9nAkBZM,11275
+nucliadb/backups/create.py,sha256=w6KpjkTOpIoPVrtOZrxB99oMdWp6jYkyIJ1-qo_k4Mw,11230
 nucliadb/backups/delete.py,sha256=1rnBhVUGYYZJXSZUrrgYMDZ5NyswEWkIA-G-crRCyHk,2404
 nucliadb/backups/models.py,sha256=-hITU4Mv6AxePu12toBu_fjpEv6vVGcwNVxV22O9jQA,1273
 nucliadb/backups/restore.py,sha256=wepEgv4vBN5yeiZU-f17PbuFV4xT4_SVKplNr8xSJrE,10001
 nucliadb/backups/settings.py,sha256=SyzsInj1BRbBI0atg5IXWbMbOZ_eVg4eSQ3IcnUhCxQ,1357
 nucliadb/backups/tasks.py,sha256=4_kOVJ2yCwMvDEpzJgTuTt75TNlpq5woyw9sTAcaSkw,4194
-nucliadb/backups/utils.py,sha256=b1hi0gEp90tNrWHejNVoUgRpa4D6uKGhbACq0yeLkJY,1375
+nucliadb/backups/utils.py,sha256=_Vogjqcru5oqNZM-bZ0q7Ju79Bv1PD-LVFEa7Z-Q13I,1261
 nucliadb/common/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn4bIXMa20ks,835
 nucliadb/common/constants.py,sha256=QpigxJh_CtD85Evy0PtV5cVq6x0U_f9xfIcXz1ymkUg,869
 nucliadb/common/counters.py,sha256=8lOi3A2HeLDDlcNaS2QT1SfD3350VPBjiY3FkmHH1V8,977
 nucliadb/common/ids.py,sha256=4QjoIofes_vtKj2HsFWZf8VVIVWXxdkYtLpx1n618Us,8239
 nucliadb/common/locking.py,sha256=RL0CabZVPzxHZyUjYeUyLvsJTm7W3J9o4fEgsY_ufNc,5896
-nucliadb/common/nidx.py,sha256=_LoU8D4afEtlW0c3vGUCoatDZvMr0-2l_GtIGap7VxA,10185
+nucliadb/common/nidx.py,sha256=Xuh7qT3fTDiyBlXe40gXtk3URE0EoJHoi0BsSA5fNZU,10240
 nucliadb/common/cluster/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn4bIXMa20ks,835
 nucliadb/common/cluster/base.py,sha256=kklDqyvsubNX0W494ttl9f3E58lGaX6AXqAd8XX8ZHE,5522
 nucliadb/common/cluster/exceptions.py,sha256=t7v_l93t44l2tQpdQXgO_w-c4YZRcaayOz1A2i0w4RQ,1258
@@ -197,11 +197,12 @@ nucliadb/search/run.py,sha256=aFb-CXRi_C8YMpP_ivNj8KW1BYhADj88y8K9Lr_nUPI,1402
 nucliadb/search/settings.py,sha256=vem3EcyYlTPSim0kEK-xe-erF4BZg0CT_LAb8ZRQAE8,1684
 nucliadb/search/utilities.py,sha256=9SsRDw0rJVXVoLBfF7rBb6q080h-thZc7u8uRcTiBeY,1037
 nucliadb/search/api/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn4bIXMa20ks,835
-nucliadb/search/api/v1/__init__.py,sha256=8w6VhZ5rbzX1xLSXr336d2IE-O0dQiv-ba6UYdRKnHA,1325
+nucliadb/search/api/v1/__init__.py,sha256=DH16OYnw9jQ38OpKlmdXeoq2j40ZPXZRtGvClKOkMhw,1239
 nucliadb/search/api/v1/ask.py,sha256=F2dR3-swb3Xz8MfZPYL3G65KY2R_mgef4YVBbu8kLi4,4352
-nucliadb/search/api/v1/catalog.py,sha256=IyBkq99CND1kFelsedJiQaSD8GOPGBJpcKuw1Z6jB9E,7709
+nucliadb/search/api/v1/catalog.py,sha256=MRiuawjKoWXdd-tptw8r8HyBhZ1JgdUYpau__69oij0,7709
 nucliadb/search/api/v1/feedback.py,sha256=kNLc4dHz2SXHzV0PwC1WiRAwY88fDptPcP-kO0q-FrQ,2620
 nucliadb/search/api/v1/find.py,sha256=l2dRg0eYngq52vyn9_z9iK7bdO7ufHQDnJWBZgMVrqY,9628
+nucliadb/search/api/v1/graph.py,sha256=5APs0-W-jNQfH-mRLrdRkk6B_mnsJxLb68t0NE_KZUk,4238
 nucliadb/search/api/v1/knowledgebox.py,sha256=rWhx3PYWryingu19qwwFDbVvVYynq5Ky23FSlzmTutQ,8721
 nucliadb/search/api/v1/predict_proxy.py,sha256=QrGzo0hKjtmyGZ6pjlJHYAh4hxwVUIOTcVcerRCw7eE,3047
 nucliadb/search/api/v1/router.py,sha256=mtT07rBZcVfpa49doaw9b1tj3sdi3qLH0gn9Io6NYM0,988
@@ -215,27 +216,28 @@ nucliadb/search/api/v1/resource/ingestion_agents.py,sha256=fqqRCd8Wc9GciS5P98lcn
 nucliadb/search/api/v1/resource/search.py,sha256=s_si6iilqhmopEQ5GG5c7C_4QV3X8QneQyS5zP0d22I,5228
 nucliadb/search/api/v1/resource/utils.py,sha256=-NjZqAQtFEXKpIh8ui5S26ItnJ5rzmmG0BHxGSS9QPw,1141
 nucliadb/search/requesters/__init__.py,sha256=itSI7dtTwFP55YMX4iK7JzdMHS5CQVUiB1XzQu4UBh8,833
-nucliadb/search/requesters/utils.py,sha256=qL81UVPNgBftUMLpcxIYVr7ILsMqpKCo-9SY2EvAaXw,6681
+nucliadb/search/requesters/utils.py,sha256=BIJjfjJt3qClA6XRtqA7E5am5Re29CaY_R6QOuTG-24,7082
 nucliadb/search/search/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn4bIXMa20ks,835
 nucliadb/search/search/cache.py,sha256=n9vkN6Y6Xnr2RBJyoH0WzjzGTJOMfKekU9tfPTWWCPc,6810
 nucliadb/search/search/cut.py,sha256=ytY0_GY7ocNjfxTb4aosxEp4ZfhQNDP--JkhEMGD298,1153
 nucliadb/search/search/exceptions.py,sha256=klGLgAGGrXcSGix_W6418ZBMqDchAIGjN77ofkOScEI,1039
 nucliadb/search/search/fetch.py,sha256=XJHIFnZmXM_8Kb37lb4lg1GYG7cZ1plT-qAIb_QziX4,6184
 nucliadb/search/search/filters.py,sha256=1MkHlJjAQqoRCj7e5cEzK2HvBxGLE17I_omsjiklbtw,6476
-nucliadb/search/search/find.py,sha256=jQZOqu8VeX8k3ELV8bLK4TwUUjGrvmubouxvO1IvJV0,10236
+nucliadb/search/search/find.py,sha256=p7Odx3hjPwTsHEhQrVYsR_ufq0meMmCnQV0YihfOmlg,10237
 nucliadb/search/search/find_merge.py,sha256=3FnzKFEnVemg6FO_6zveulbAU7klvsiPEBvLrpBBMg8,17450
+nucliadb/search/search/graph_merge.py,sha256=7LzzU09HCFu6wm3LfgJ3Lm6Uuh_qPDmM2tWZS-fHmw8,3250
 nucliadb/search/search/graph_strategy.py,sha256=gisL2GpbSIa_SucyOwEt7TWdqURyAQqxvD_-PkXQct8,32339
 nucliadb/search/search/hydrator.py,sha256=-R37gCrGxkyaiHQalnTWHNG_FCx11Zucd7qA1vQCxuw,6985
 nucliadb/search/search/ingestion_agents.py,sha256=NeJr4EEX-bvFFMGvXOOwLv8uU7NuQ-ntJnnrhnKfMzY,3174
 nucliadb/search/search/merge.py,sha256=aUn6f5XnwWzUFhVC6uBqHE8NKdlfgw_xcTo57rS23U8,22950
 nucliadb/search/search/metrics.py,sha256=GGGtXHLhK79_ESV277xkBVjcaMURXHCxYG0EdGamUd8,2886
 nucliadb/search/search/paragraphs.py,sha256=pNAEiYqJGGUVcEf7xf-PFMVqz0PX4Qb-WNG-_zPGN2o,7799
-nucliadb/search/search/pgcatalog.py,sha256=V1NYLEUSXHpWmgcPIo1HS2riK_HDXSi-uykJjSoOOrE,9033
+nucliadb/search/search/pgcatalog.py,sha256=s_J98fsX_RuFXwpejpkGqG-tD9ELuzz4YQ6U3ew5h2g,9313
 nucliadb/search/search/predict_proxy.py,sha256=IFI3v_ODz2_UU1XZnyaD391fE7-2C0npSmj_HmDvzS4,3123
 nucliadb/search/search/query.py,sha256=CCz58Cs_VKxbwGO6VdhOzl9HXwKny2hfH2llJVR9emM,29859
 nucliadb/search/search/rank_fusion.py,sha256=tRGo_KlsFsVx1CQEy1iqQ6f0T1Dq1kf0axDXHuuzvvM,6946
 nucliadb/search/search/rerankers.py,sha256=3vep4EOVNeDJGsMdx-1g6Ar4ZGJG3IHym3HkxnbwtAQ,7321
-nucliadb/search/search/shards.py,sha256=JSRSrHgHcF4sXyuZZoJdMfK0v_LHpoSRf1lCr5-K5ko,2742
+nucliadb/search/search/shards.py,sha256=OEtN1p9WX_cMX8t-myaafpmFAPTpUEOutR7z1sDuNcY,3242
 nucliadb/search/search/summarize.py,sha256=ksmYPubEQvAQgfPdZHfzB_rR19B2ci4IYZ6jLdHxZo8,4996
 nucliadb/search/search/utils.py,sha256=iF2tbBA56gRMJH1TlE2hMrqeXqjoeOPt4KgRdp2m9Ek,3313
 nucliadb/search/search/chat/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn4bIXMa20ks,835
@@ -245,13 +247,15 @@ nucliadb/search/search/chat/images.py,sha256=PA8VWxT5_HUGfW1ULhKTK46UBsVyINtWWqE
 nucliadb/search/search/chat/prompt.py,sha256=Jnja-Ss7skgnnDY8BymVfdeYsFPnIQFL8tEvcRXTKUE,47356
 nucliadb/search/search/chat/query.py,sha256=0IoeW-JNaRBe2d9C3bXNfkYpzmsN_IIg3U4Vqb8eOEk,16485
 nucliadb/search/search/query_parser/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn4bIXMa20ks,835
-nucliadb/search/search/query_parser/catalog.py,sha256=PtH5nb6UTzH8l7Lmdd1RgLVFsn9CN5M5-JkVq9YeR4k,7116
 nucliadb/search/search/query_parser/exceptions.py,sha256=szAOXUZ27oNY-OSa9t2hQ5HHkQQC0EX1FZz_LluJHJE,1224
 nucliadb/search/search/query_parser/fetcher.py,sha256=jhr__J0KmAzjdsTTadWQmD9qf6lZvqlKAfZdYjZH_UY,15742
 nucliadb/search/search/query_parser/filter_expression.py,sha256=rws5vsKTofX2iMUK4yvjmLZFxtcbWbyhIcwen4j0rQg,6578
-nucliadb/search/search/query_parser/models.py,sha256=VHDuyJlU2OLZN1usrQX53TZbPmWhzMeVYY0BiYNFzak,2464
+nucliadb/search/search/query_parser/models.py,sha256=7czH-jHskl9axEnZV5XnQY8cyN_fs14Cpivzd5aaSxc,2686
 nucliadb/search/search/query_parser/old_filters.py,sha256=-zbfN-RsXoj_DRjh3Lfp-wShwFXgkISawzVptVzja-A,9071
-nucliadb/search/search/query_parser/parser.py,sha256=9TwkSNna3s-lCQIqBoSJzm6YbXdu8VIHJUan8M4ysfE,4667
+nucliadb/search/search/query_parser/parsers/__init__.py,sha256=ySCNSdbesLXGZyR88919njulA6UE10_3PhqMG_Yj1o4,1034
+nucliadb/search/search/query_parser/parsers/catalog.py,sha256=XdBiTweGTQkj8m_V_i2xbwp7P5pPO8K1Tud692XKhMw,7149
+nucliadb/search/search/query_parser/parsers/find.py,sha256=q3wH_i0DGceeKckYEH3c5MqM5EvRiMCL7r-6nCAId9Q,4666
+nucliadb/search/search/query_parser/parsers/graph.py,sha256=4cR-Y_ZUX-QNYhL6DV2tanmBlSb0ducf4J3hb9VXLh8,6261
 nucliadb/standalone/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn4bIXMa20ks,835
 nucliadb/standalone/api_router.py,sha256=hgq9FXpihzgjHkwcVGfGCSwyXy67fqXTfLFHuINzIi0,5567
 nucliadb/standalone/app.py,sha256=mAApNK_iVsQgJyd-mtwCeZq5csSimwnXmlQGH9a70pE,5586
@@ -347,8 +351,8 @@ nucliadb/writer/tus/local.py,sha256=7jYa_w9b-N90jWgN2sQKkNcomqn6JMVBOVeDOVYJHto,
 nucliadb/writer/tus/s3.py,sha256=vF0NkFTXiXhXq3bCVXXVV-ED38ECVoUeeYViP8uMqcU,8357
 nucliadb/writer/tus/storage.py,sha256=ToqwjoYnjI4oIcwzkhha_MPxi-k4Jk3Lt55zRwaC1SM,2903
 nucliadb/writer/tus/utils.py,sha256=MSdVbRsRSZVdkaum69_0wku7X3p5wlZf4nr6E0GMKbw,2556
-nucliadb-6.3.1.post3584.dist-info/METADATA,sha256=1LFsE32EtmIRlKK9EAN1-kK6P5pfLq3pIRewpO0k2Ug,4291
-nucliadb-6.3.1.post3584.dist-info/WHEEL,sha256=52BFRY2Up02UkjOa29eZOS2VxUrpPORXg1pkohGGUS8,91
-nucliadb-6.3.1.post3584.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
-nucliadb-6.3.1.post3584.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
-nucliadb-6.3.1.post3584.dist-info/RECORD,,
+nucliadb-6.3.1.post3590.dist-info/METADATA,sha256=cnyGLiXObJPzQkmY-I7Ilyq2TvPG9l48K1TrnPhqreQ,4291
+nucliadb-6.3.1.post3590.dist-info/WHEEL,sha256=52BFRY2Up02UkjOa29eZOS2VxUrpPORXg1pkohGGUS8,91
+nucliadb-6.3.1.post3590.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
+nucliadb-6.3.1.post3590.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
+nucliadb-6.3.1.post3590.dist-info/RECORD,,