nucliadb 6.7.2.post4862__py3-none-any.whl → 6.9.2.post5282__py3-none-any.whl

migrations/0016_upgrade_to_paragraphs_v2.py

@@ -55,7 +55,7 @@ async def migrate_kb(context: ExecutionContext, kbid: str) -> None:
 
 
 # async def has_old_paragraphs_index(context: ExecutionContext, kbid: str) -> bool:
-#     async with context.kv_driver.transaction(read_only=True) as txn:
+#     async with context.kv_driver.ro_transaction() as txn:
 #         shards_object = await datamanagers.cluster.get_kb_shards(txn, kbid=kbid, for_update=False)
 #         if not shards_object:
 #             raise ShardsObjectNotFound()

migrations/0017_multiple_writable_shards.py

@@ -44,7 +44,7 @@ async def migrate_kb(context: ExecutionContext, kbid: str) -> None:
 
     # No longer relevant with nidx
 
-    # async with context.kv_driver.transaction() as txn:
+    # async with context.kv_driver.rw_transaction() as txn:
     #     shards = await datamanagers.cluster.get_kb_shards(txn, kbid=kbid, for_update=True)
     #     if shards is None:
     #         logger.error("KB without shards", extra={"kbid": kbid})

migrations/0018_purge_orphan_kbslugs.py

@@ -35,7 +35,7 @@ logger = logging.getLogger(__name__)
 
 
 async def migrate(context: ExecutionContext) -> None:
-    async with context.kv_driver.transaction() as txn:
+    async with context.kv_driver.rw_transaction() as txn:
         async for key in txn.keys(KB_SLUGS_BASE):
             slug = key.replace(KB_SLUGS_BASE, "")
             value = await txn.get(key, for_update=False)

migrations/0019_upgrade_to_paragraphs_v3.py

@@ -55,7 +55,7 @@ async def migrate_kb(context: ExecutionContext, kbid: str) -> None:
 
 
 # async def has_old_paragraphs_index(context: ExecutionContext, kbid: str) -> bool:
-#     async with context.kv_driver.transaction(read_only=True) as txn:
+#     async with context.kv_driver.ro_transaction() as txn:
 #         shards_object = await datamanagers.cluster.get_kb_shards(txn, kbid=kbid)
 #         if not shards_object:
 #             raise ShardsObjectNotFound()

migrations/0021_overwrite_vectorsets_key.py

@@ -38,7 +38,7 @@ async def migrate(context: ExecutionContext) -> None: ...
 
 
 async def migrate_kb(context: ExecutionContext, kbid: str) -> None:
-    async with context.kv_driver.transaction() as txn:
+    async with context.kv_driver.rw_transaction() as txn:
         logger.info(f"Overwriting vectorsets key", extra={"kbid": kbid})
         await datamanagers.vectorsets.initialize(txn, kbid=kbid)
         await txn.commit()

migrations/0023_backfill_pg_catalog.py

@@ -28,9 +28,10 @@ import logging
 from typing import cast
 
 from nucliadb.common import datamanagers
+from nucliadb.common.catalog import catalog_update, get_catalog
+from nucliadb.common.catalog.pg import PGCatalog
 from nucliadb.common.maindb.pg import PGDriver, PGTransaction
 from nucliadb.ingest.orm.index_message import get_resource_index_message
-from nucliadb.ingest.orm.processor.pgcatalog import pgcatalog_update
 from nucliadb.migrator.context import ExecutionContext
 
 logger = logging.getLogger(__name__)
@@ -43,8 +44,11 @@ async def migrate_kb(context: ExecutionContext, kbid: str) -> None:
     if not isinstance(context.kv_driver, PGDriver):
         return
 
+    if not isinstance(get_catalog(), PGCatalog):
+        return
+
     BATCH_SIZE = 100
-    async with context.kv_driver.transaction() as txn:
+    async with context.kv_driver.rw_transaction() as txn:
         txn = cast(PGTransaction, txn)
         continue_sql = ""
         while True:
@@ -75,7 +79,7 @@ async def migrate_kb(context: ExecutionContext, kbid: str) -> None:
                         continue
 
                     index_message = await get_resource_index_message(resource, reindex=False)
-                    await pgcatalog_update(txn, kbid, resource, index_message)
+                    await catalog_update(txn, kbid, resource, index_message)
 
                 await txn.commit()
                 continue_sql = f"AND key > '/kbs/{kbid}/r/{rid}'"

migrations/0025_assign_models_to_kbs_v2.py

@@ -47,7 +47,7 @@ async def migrate(context: ExecutionContext) -> None: ...
 
 
 async def migrate_kb(context: ExecutionContext, kbid: str) -> None:
-    async with context.kv_driver.transaction(read_only=True) as txn:
+    async with context.kv_driver.ro_transaction() as txn:
         vectorsets_count = len([vs async for vs in datamanagers.vectorsets.iter(txn, kbid=kbid)])
     if vectorsets_count > 0:
         logger.info("Skipping KB with vectorsets already populated", extra={"kbid": kbid})
@@ -65,7 +65,7 @@ async def migrate_kb(context: ExecutionContext, kbid: str) -> None:
     learning_matryoshka_dimensions = learning_model_metadata.matryoshka_dimensions
     learning_normalize_vectors = len(learning_matryoshka_dimensions) > 0
 
-    async with context.kv_driver.transaction(read_only=True) as txn:
+    async with context.kv_driver.ro_transaction() as txn:
         semantic_model = await datamanagers.kb.get_model_metadata(txn, kbid=kbid)
 
         maindb_similarity = semantic_model.similarity_function
@@ -103,7 +103,7 @@ async def migrate_kb(context: ExecutionContext, kbid: str) -> None:
         matryoshka_dimensions=maindb_matryoshka_dimensions,
     )
 
-    async with context.kv_driver.transaction() as txn:
+    async with context.kv_driver.rw_transaction() as txn:
         # Populate KB vectorsets with data from learning. We are skipping KBs
         # with this key already set, so we can set here safely
         await datamanagers.vectorsets.set(txn, kbid=kbid, config=default_vectorset)

migrations/0027_rollover_texts3.py

@@ -49,7 +49,7 @@ async def maybe_fix_vector_dimensions(context: ExecutionContext, kbid: str) -> N
         logger.warning(f"KB has no learning config", extra={"kbid": kbid})
         return
 
-    async with context.kv_driver.transaction() as txn:
+    async with context.kv_driver.rw_transaction() as txn:
         vectorsets = [vs async for vs in datamanagers.vectorsets.iter(txn, kbid=kbid)]
         if len(vectorsets) != 1:
             # If multiple vectorsets, they are new shards created correctly, we can safely skip it

migrations/0028_extracted_vectors_reference.py

@@ -39,7 +39,7 @@ async def migrate_kb(context: ExecutionContext, kbid: str) -> None:
     async with datamanagers.with_rw_transaction() as txn:
         vectorsets = [vs async for (_vid, vs) in datamanagers.vectorsets.iter(txn, kbid=kbid)]
 
-        if len(vectorsets) == 0:  # pragma: nocover
+        if len(vectorsets) == 0:  # pragma: no cover
             # should never happen, everyone should have at least one
             logger.warning(f"KB has no vectorsets!", extra={"kbid": kbid})
             return

migrations/0029_backfill_field_status.py

@@ -45,7 +45,7 @@ async def migrate_kb(context: ExecutionContext, kbid: str) -> None: ...
 
 async def do_batch(context: ExecutionContext, start: str) -> Optional[str]:
     logger.info(f"Running batch from {start}")
-    async with context.kv_driver.transaction(read_only=False) as txn:
+    async with context.kv_driver.rw_transaction() as txn:
         async with txn.connection.cursor() as cur:  # type: ignore
             # Retrieve a batch of fields
             await cur.execute(

migrations/0032_remove_old_relations.py

@@ -47,7 +47,7 @@ async def migrate_kb(context: ExecutionContext, kbid: str) -> None:
 
 async def do_batch(context: ExecutionContext, start: str) -> Optional[str]:
     logger.info(f"Running batch from {start}")
-    async with context.kv_driver.transaction(read_only=False) as txn:
+    async with context.kv_driver.rw_transaction() as txn:
         async with txn.connection.cursor() as cur:  # type: ignore
             # Retrieve a batch of fields
             await cur.execute(

migrations/0036_backfill_catalog_slug.py

@@ -37,7 +37,7 @@ async def migrate(context: ExecutionContext) -> None:
     driver = cast(PGDriver, context.kv_driver)
 
     BATCH_SIZE = 10_000
-    async with driver.transaction() as txn:
+    async with driver.rw_transaction() as txn:
         txn = cast(PGTransaction, txn)
         start_key = ""
         while True:

migrations/0037_backfill_catalog_facets.py

@@ -37,7 +37,7 @@ async def migrate(context: ExecutionContext) -> None:
     driver = cast(PGDriver, context.kv_driver)
 
     BATCH_SIZE = 1_000
-    async with driver.transaction() as txn:
+    async with driver.rw_transaction() as txn:
         txn = cast(PGTransaction, txn)
         start_kbid = "00000000000000000000000000000000"
         start_rid = "00000000000000000000000000000000"

migrations/0038_backfill_catalog_field_labels.py

@@ -28,9 +28,10 @@ import logging
 from typing import cast
 
 from nucliadb.common import datamanagers
+from nucliadb.common.catalog import catalog_update, get_catalog
+from nucliadb.common.catalog.pg import PGCatalog
 from nucliadb.common.maindb.pg import PGDriver, PGTransaction
 from nucliadb.ingest.orm.index_message import get_resource_index_message
-from nucliadb.ingest.orm.processor.pgcatalog import pgcatalog_update
 from nucliadb.migrator.context import ExecutionContext
 from nucliadb_protos import resources_pb2
 
@@ -44,8 +45,11 @@ async def migrate_kb(context: ExecutionContext, kbid: str) -> None:
     if not isinstance(context.kv_driver, PGDriver):
         return
 
+    if not isinstance(get_catalog(), PGCatalog):
+        return
+
     BATCH_SIZE = 100
-    async with context.kv_driver.transaction() as txn:
+    async with context.kv_driver.rw_transaction() as txn:
         txn = cast(PGTransaction, txn)
         start = ""
         while True:
@@ -84,7 +88,7 @@ async def migrate_kb(context: ExecutionContext, kbid: str) -> None:
                         continue
 
                     index_message = await get_resource_index_message(resource, reindex=False)
-                    await pgcatalog_update(txn, kbid, resource, index_message)
+                    await catalog_update(txn, kbid, resource, index_message)
 
                 if to_index:
                     await txn.commit()

migrations/0039_backfill_converation_splits_metadata.py

@@ -0,0 +1,106 @@
+# 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/>.
+#
+
+"""Migration #39
+
+Backfill splits metadata on conversation fields
+
+"""
+
+import logging
+from typing import cast
+
+from nucliadb.common.maindb.driver import Transaction
+from nucliadb.common.maindb.pg import PGTransaction
+from nucliadb.ingest.fields.conversation import (
+    CONVERSATION_SPLITS_METADATA,
+    Conversation,
+)
+from nucliadb.ingest.orm.knowledgebox import KnowledgeBox as KnowledgeBoxORM
+from nucliadb.migrator.context import ExecutionContext
+from nucliadb_protos import resources_pb2
+from nucliadb_protos.resources_pb2 import SplitMetadata, SplitsMetadata
+from nucliadb_utils.storages.storage import Storage
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(context: ExecutionContext) -> None: ...
+
+
+async def migrate_kb(context: ExecutionContext, kbid: str) -> None:
+    BATCH_SIZE = 100
+    start = ""
+    while True:
+        to_fix: list[tuple[str, str]] = []
+        async with context.kv_driver.rw_transaction() as txn:
+            txn = cast(PGTransaction, txn)
+            async with txn.connection.cursor() as cur:
+                # Retrieve a bunch of conversation fields
+                await cur.execute(
+                    """
+                    SELECT key FROM resources
+                    WHERE key ~ ('^/kbs/' || %s || '/r/[^/]*/f/c/[^/]*$')
+                    AND key > %s
+                    ORDER BY key
+                    LIMIT %s""",
+                    (kbid, start, BATCH_SIZE),
+                )
+                rows = await cur.fetchall()
+                if len(rows) == 0:
+                    return
+                for row in rows:
+                    key = row[0]
+                    start = key
+                    rid = key.split("/")[4]
+                    field_id = key.split("/")[7]
+                    to_fix.append((rid, field_id))
+
+        for rid, field_id in to_fix:
+            async with context.kv_driver.rw_transaction() as txn2:
+                splits_metadata = await build_splits_metadata(
+                    txn2, context.blob_storage, kbid, rid, field_id
+                )
+                splits_metadata_key = CONVERSATION_SPLITS_METADATA.format(
+                    kbid=kbid, uuid=rid, type="c", field=field_id
+                )
+                await txn2.set(splits_metadata_key, splits_metadata.SerializeToString())
+                await txn2.commit()
+
+
+async def build_splits_metadata(
+    txn: Transaction, storage: Storage, kbid: str, rid: str, field_id: str
+) -> SplitsMetadata:
+    splits_metadata = SplitsMetadata()
+    kb_orm = KnowledgeBoxORM(txn, storage, kbid)
+    resource_obj = await kb_orm.get(rid)
+    if resource_obj is None:
+        return splits_metadata
+    field_obj: Conversation = await resource_obj.get_field(
+        field_id, resources_pb2.FieldType.CONVERSATION, load=False
+    )
+    conv_metadata = await field_obj.get_metadata()
+    for i in range(1, conv_metadata.pages + 1):
+        page = await field_obj.get_value(page=i)
+        if page is None:
+            continue
+        for message in page.messages:
+            splits_metadata.metadata.setdefault(message.ident, SplitMetadata())
+    return splits_metadata

migrations/0040_migrate_search_configurations.py

@@ -0,0 +1,79 @@
+# 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/>.
+#
+
+"""Migration #40
+
+Replaces deprecated and removed generative models from search configurations
+
+"""
+
+import logging
+from typing import cast
+
+from nucliadb.common import datamanagers
+from nucliadb.migrator.context import ExecutionContext
+from nucliadb_models.configuration import SearchConfiguration
+from nucliadb_models.search import AskRequest, FindRequest
+
+logger = logging.getLogger(__name__)
+
+REPLACEMENTS = {
+    "claude-3-5-small": "claude-4-5-sonnet",
+    "gcp-claude-3-5-sonnet-v2": "gcp-claude-4-5-sonnet",
+}
+
+
+async def migrate(context: ExecutionContext) -> None: ...
+
+
+async def migrate_kb(context: ExecutionContext, kbid: str) -> None:
+    affected = await get_affected_search_configurations(kbid)
+    if not affected:
+        return
+
+    async with datamanagers.with_rw_transaction() as txn:
+        for name, config in affected.items():
+            logger.info(
+                "Migrating search config for kb",
+                extra={
+                    "kbid": kbid,
+                    "search_config": name,
+                    "generative_model": config.config.generative_model,  # type: ignore
+                },
+            )
+            config.config.generative_model = REPLACEMENTS[config.config.generative_model]  # type: ignore
+            await datamanagers.search_configurations.set(txn, kbid=kbid, name=name, config=config)
+        await txn.commit()
+
+
+async def get_affected_search_configurations(kbid: str) -> dict[str, SearchConfiguration]:
+    result: dict[str, SearchConfiguration] = {}
+    async with datamanagers.with_ro_transaction() as txn:
+        search_configs = await datamanagers.search_configurations.list(txn, kbid=kbid)
+        for name, config in search_configs.items():
+            if config.kind == "find":
+                find_config = cast(FindRequest, config.config)
+                if find_config.generative_model in REPLACEMENTS:
+                    result[name] = config
+            elif config.kind == "ask":
+                ask_config = cast(AskRequest, config.config)
+                if ask_config.generative_model in REPLACEMENTS:
+                    result[name] = config
+    return result

migrations/pg/0010_shards_index.py

@@ -0,0 +1,34 @@
+# 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.maindb.pg import PGTransaction
+
+
+async def migrate(txn: PGTransaction) -> None:
+    # Concurrent index must be created outside of a transaction but psycopg automatically
+    # creates transactions. We temporarily disable this for building indexes.
+    await txn.connection.commit()
+    try:
+        await txn.connection.set_autocommit(True)
+        await txn.connection.execute(
+            "CREATE INDEX CONCURRENTLY ON resources (key, value) WHERE key ~ '/kbs/[^/]*/r/[^/]*/shard$';"
+        )
+    finally:
+        await txn.connection.set_autocommit(False)

nucliadb/backups/create.py

@@ -269,7 +269,7 @@ async def backup_search_configurations(context: ApplicationContext, kbid: str, b
 async def get_metadata(
     context: ApplicationContext, kbid: str, backup_id: str
 ) -> Optional[BackupMetadata]:
-    async with context.kv_driver.transaction(read_only=True) as txn:
+    async with context.kv_driver.ro_transaction() as txn:
         metadata_raw = await txn.get(MaindbKeys.METADATA.format(kbid=kbid, backup_id=backup_id))
         if metadata_raw is None:
             return None
@@ -277,7 +277,7 @@ async def get_metadata(
 
 
 async def set_metadata(context: ApplicationContext, kbid: str, backup_id: str, metadata: BackupMetadata):
-    async with context.kv_driver.transaction() as txn:
+    async with context.kv_driver.rw_transaction() as txn:
         await txn.set(
             MaindbKeys.METADATA.format(kbid=kbid, backup_id=backup_id),
             metadata.model_dump_json().encode(),
@@ -286,7 +286,7 @@ async def set_metadata(context: ApplicationContext, kbid: str, backup_id: str, m
 
 
 async def delete_metadata(context: ApplicationContext, kbid: str, backup_id: str):
-    async with context.kv_driver.transaction() as txn:
+    async with context.kv_driver.rw_transaction() as txn:
         await txn.delete(MaindbKeys.METADATA.format(kbid=kbid, backup_id=backup_id))
         await txn.commit()
 

nucliadb/backups/restore.py

@@ -103,7 +103,7 @@ async def restore_resources(context: ApplicationContext, kbid: str, backup_id: s
 
 async def get_last_restored(context: ApplicationContext, kbid: str, backup_id: str) -> Optional[str]:
     key = MaindbKeys.LAST_RESTORED.format(kbid=kbid, backup_id=backup_id)
-    async with context.kv_driver.transaction(read_only=True) as txn:
+    async with context.kv_driver.ro_transaction() as txn:
         raw = await txn.get(key)
         if raw is None:
             return None
@@ -112,14 +112,14 @@ async def get_last_restored(context: ApplicationContext, kbid: str, backup_id: s
 
 async def set_last_restored(context: ApplicationContext, kbid: str, backup_id: str, resource_id: str):
     key = MaindbKeys.LAST_RESTORED.format(kbid=kbid, backup_id=backup_id)
-    async with context.kv_driver.transaction() as txn:
+    async with context.kv_driver.rw_transaction() as txn:
         await txn.set(key, resource_id.encode())
         await txn.commit()
 
 
 async def delete_last_restored(context: ApplicationContext, kbid: str, backup_id: str):
     key = MaindbKeys.LAST_RESTORED.format(kbid=kbid, backup_id=backup_id)
-    async with context.kv_driver.transaction() as txn:
+    async with context.kv_driver.rw_transaction() as txn:
         await txn.delete(key)
         await txn.commit()
 

nucliadb/common/cache.py

@@ -90,7 +90,7 @@ class ResourceCache(Cache[[str, str], ResourceORM]):
         @alru_cache(maxsize=cache_size)
         async def _get_resource(kbid: str, rid: str) -> Optional[ResourceORM]:
             storage = await get_storage()
-            async with get_driver().transaction(read_only=True) as txn:
+            async with get_driver().ro_transaction() as txn:
                 kb = KnowledgeBoxORM(txn, storage, kbid)
                 return await kb.get(rid)
 

nucliadb/common/catalog/__init__.py

@@ -0,0 +1,79 @@
+# 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/>.
+#
+# 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 nidx_protos.noderesources_pb2 import Resource as IndexMessage
+
+from nucliadb.common.catalog.dummy import DummyCatalog
+from nucliadb.common.catalog.interface import Catalog, CatalogQuery
+from nucliadb.common.catalog.pg import PGCatalog
+from nucliadb.common.catalog.utils import build_catalog_resource_data
+from nucliadb.common.maindb.driver import Transaction
+from nucliadb.ingest.orm.resource import Resource
+from nucliadb.ingest.settings import CatalogConfig, settings
+from nucliadb_models.search import CatalogFacetsRequest, Resources
+from nucliadb_utils.exceptions import ConfigurationError
+
+
+def get_catalog() -> Catalog:
+    if settings.catalog == CatalogConfig.UNSET:
+        return DummyCatalog()
+    elif settings.catalog == CatalogConfig.PG:
+        return PGCatalog()
+    else:
+        raise ConfigurationError(f"Unknown catalog configuration: {settings.catalog}")
+
+
+async def catalog_update(txn: Transaction, kbid: str, resource: Resource, index_message: IndexMessage):
+    catalog = get_catalog()
+    resource_data = build_catalog_resource_data(resource, index_message)
+    await catalog.update(txn, kbid, resource.uuid, resource_data)
+
+
+async def catalog_delete(txn: Transaction, kbid: str, rid: str):
+    catalog = get_catalog()
+    await catalog.delete(txn, kbid, rid)
+
+
+async def catalog_search(query: CatalogQuery) -> Resources:
+    catalog = get_catalog()
+    return await catalog.search(query)
+
+
+async def catalog_facets(kbid: str, request: CatalogFacetsRequest) -> dict[str, int]:
+    catalog = get_catalog()
+    return await catalog.facets(kbid, request)

nucliadb/common/catalog/dummy.py

@@ -0,0 +1,36 @@
+# 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.catalog.interface import Catalog, CatalogQuery, CatalogResourceData
+from nucliadb.common.maindb.driver import Transaction
+from nucliadb_models.search import CatalogFacetsRequest, Resources
+
+
+class DummyCatalog(Catalog):
+    async def update(self, txn: Transaction, kbid: str, rid: str, data: CatalogResourceData):
+        return
+
+    async def delete(self, txn: Transaction, kbid: str, rid: str):
+        return
+
+    async def search(self, query: CatalogQuery) -> Resources:
+        return Resources(results=[], min_score=0.0)
+
+    async def facets(self, kbid: str, request: CatalogFacetsRequest) -> dict[str, int]:
+        return {}