nuclia 4.9.1__tar.gz → 4.9.2__tar.gz

{nuclia-4.9.1 → nuclia-4.9.2}/.gitignore

@@ -4,6 +4,7 @@ __pycache__
 build
 dist
 tmp
+_tmp
 data
 venv
 .venv

{nuclia-4.9.1 → nuclia-4.9.2}/CHANGELOG.md

@@ -1,6 +1,12 @@
 # Changelog
 
 
+## 4.9.2 (2025-06-12)
+
+
+- Support extract strategies
+
+
 ## 4.9.1 (2025-06-10)
 
 

{nuclia-4.9.1 → nuclia-4.9.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nuclia
-Version: 4.9.1
+Version: 4.9.2
 Summary: Nuclia Python SDK
 Author-email: Nuclia <info@nuclia.com>
 License-Expression: MIT

nuclia-4.9.2/VERSION

@@ -0,0 +1 @@
+4.9.2

{nuclia-4.9.1 → nuclia-4.9.2}/docs/04-upload.md

@@ -250,3 +250,43 @@ kb.update_graph(slug="graph1", overrides=True, graph=[
     }
 ])
 ```
+
+## Use extract strategies
+
+Extract strategies allow you to perform specific processing at ingestion time. It can be useful to handle complex tables extraction, or to support unusual layouts.
+
+### Manage extract strategies
+
+CLI:
+
+```bash
+nuclia kb extract_strategies add --config='{"name":"strategy1","vllm_config":{}}'
+nuclia kb extract_strategies list
+nuclia kb extract_strategies delete --id=1361c0c7-918a-4a7f-b44b-ba37437619fb
+```
+
+SDK:
+
+```python
+from nuclia import sdk
+extract_strategies = sdk.NucliaKB().extract_strategies
+print(extract_strategies.list())
+id = extract_strategies.add(config={"name": "strategy1", "vllm_config": {}})
+extract_strategies.delete(id=id)
+```
+
+### Use extract strategies
+
+CLI:
+
+```bash
+nuclia kb upload file --path=FILE_PATH --extract_strategy=1361c0c7-918a-4a7f-b44b-ba37437619fb
+```
+
+SDK:
+
+```python
+from nuclia import sdk
+upload = sdk.NucliaUpload()
+upload.file(path=FILE_PATH, extract_strategy="1361c0c7-918a-4a7f-b44b-ba37437619fb")
+```

{nuclia-4.9.1 → nuclia-4.9.2}/nuclia/lib/kb.py

@@ -23,6 +23,7 @@ from nuclia_models.events.activity_logs import (  # type: ignore
 )
 from nuclia_models.events.remi import RemiQuery
 from nuclia_models.worker.tasks import TaskStartKB
+from nuclia_models.config.proto import ExtractConfig
 from nuclia.exceptions import RateLimitError
 from nuclia.lib.utils import handle_http_sync_errors, handle_http_async_errors
 from datetime import datetime
@@ -58,6 +59,8 @@ STOP_TASK = "/task/{task_id}/stop"
 DELETE_TASK = "/task/{task_id}"
 GET_TASK = "/task/{task_id}/inspect"
 RESTART_TASK = "/task/{task_id}/restart"
+EXTRACT_STRATEGIES = "/extract_strategies"
+DELETE_EXTRACT_STRATEGY = "/extract_strategies/strategy/{id}"
 
 DOWNLOAD_FORMAT_HEADERS = {
     DownloadFormat.CSV: "text/csv",
@@ -487,6 +490,37 @@ class NucliaDBClient(BaseNucliaDBClient):
         handle_http_sync_errors(response)
         return response
 
+    def list_extract_strategies(self) -> httpx.Response:
+        if self.reader_session is None:
+            raise Exception("KB not configured")
+
+        response: httpx.Response = self.reader_session.get(
+            f"{self.url}{EXTRACT_STRATEGIES}"
+        )
+        handle_http_sync_errors(response)
+        return response
+
+    def add_extract_strategy(self, config: ExtractConfig) -> httpx.Response:
+        if self.writer_session is None:
+            raise Exception("KB not configured")
+
+        response: httpx.Response = self.writer_session.post(
+            f"{self.url}{EXTRACT_STRATEGIES}",
+            json=config.model_dump(mode="json", exclude_unset=True),
+        )
+        handle_http_sync_errors(response)
+        return response
+
+    def delete_extract_strategy(self, strategy_id: str) -> httpx.Response:
+        if self.writer_session is None:
+            raise Exception("KB not configured")
+
+        response: httpx.Response = self.writer_session.delete(
+            f"{self.url}{DELETE_EXTRACT_STRATEGY.format(id=strategy_id)}",
+        )
+        handle_http_sync_errors(response)
+        return response
+
 
 class AsyncNucliaDBClient(BaseNucliaDBClient):
     reader_session: Optional[httpx.AsyncClient] = None
@@ -813,3 +847,34 @@ class AsyncNucliaDBClient(BaseNucliaDBClient):
         )
         await handle_http_async_errors(response)
         return response
+
+    async def list_extract_strategies(self) -> httpx.Response:
+        if self.reader_session is None:
+            raise Exception("KB not configured")
+
+        response: httpx.Response = await self.reader_session.get(
+            f"{self.url}{EXTRACT_STRATEGIES}"
+        )
+        await handle_http_async_errors(response)
+        return response
+
+    async def add_extract_strategy(self, config: ExtractConfig) -> httpx.Response:
+        if self.writer_session is None:
+            raise Exception("KB not configured")
+
+        response: httpx.Response = await self.writer_session.post(
+            f"{self.url}{EXTRACT_STRATEGIES}",
+            json=config.model_dump(mode="json", exclude_unset=True),
+        )
+        await handle_http_async_errors(response)
+        return response
+
+    async def delete_extract_strategy(self, strategy_id: str) -> httpx.Response:
+        if self.writer_session is None:
+            raise Exception("KB not configured")
+
+        response: httpx.Response = await self.writer_session.delete(
+            f"{self.url}{DELETE_EXTRACT_STRATEGY.format(id=strategy_id)}",
+        )
+        await handle_http_async_errors(response)
+        return response

nuclia-4.9.2/nuclia/sdk/extract_strategy.py

@@ -0,0 +1,98 @@
+from nuclia.data import get_auth, get_async_auth
+from nuclia.decorators import kb, pretty
+from nuclia.lib.kb import NucliaDBClient, AsyncNucliaDBClient
+from nuclia.sdk.auth import NucliaAuth, AsyncNucliaAuth
+from nuclia_models.config.proto import ExtractConfig
+from typing import Dict
+
+
+class NucliaExtractStrategy:
+    @property
+    def _auth(self) -> NucliaAuth:
+        auth = get_auth()
+        return auth
+
+    @kb
+    @pretty
+    def list(self, *args, **kwargs) -> Dict[str, ExtractConfig]:
+        """
+        List extract strategies
+        """
+        ndb: NucliaDBClient = kwargs["ndb"]
+        response = ndb.list_extract_strategies()
+        return response.json()
+
+    @kb
+    def add(
+        self,
+        *args,
+        config: ExtractConfig,
+        **kwargs,
+    ) -> str:
+        """
+        Add extract strategy
+
+        :param config: strategy configuration
+        """
+        if isinstance(config, dict):
+            config = ExtractConfig.model_validate(config)
+
+        ndb: NucliaDBClient = kwargs["ndb"]
+        response = ndb.add_extract_strategy(config=config)
+        return response.json()
+
+    @kb
+    def delete(self, *args, id: str, **kwargs):
+        """
+        Delete extract strategy
+
+        :param id: ID of the strategy to delete
+        """
+        ndb: NucliaDBClient = kwargs["ndb"]
+        ndb.delete_extract_strategy(strategy_id=id)
+
+
+class AsyncNucliaExtractStrategy:
+    @property
+    def _auth(self) -> AsyncNucliaAuth:
+        auth = get_async_auth()
+        return auth
+
+    @kb
+    @pretty
+    async def list(self, *args, **kwargs) -> Dict[str, ExtractConfig]:
+        """
+        List extract strategies
+        """
+        ndb: AsyncNucliaDBClient = kwargs["ndb"]
+        response = await ndb.list_extract_strategies()
+        return response.json()
+
+    @kb
+    async def add(
+        self,
+        *args,
+        config: ExtractConfig,
+        **kwargs,
+    ) -> str:
+        """
+        Add extract strategy
+
+        :param config: strategy configuration
+        """
+        if isinstance(config, dict):
+            config = ExtractConfig.model_validate(config)
+
+        ndb: AsyncNucliaDBClient = kwargs["ndb"]
+        response = await ndb.add_extract_strategy(config=config)
+        return response.json()
+
+    @kb
+    async def delete(self, *args, id: str, **kwargs):
+        """
+        Delete extract strategy
+
+        :param id: ID of the strategy to delete
+        """
+        ndb: AsyncNucliaDBClient = kwargs["ndb"]
+        await ndb.delete_extract_strategy(strategy_id=id)

{nuclia-4.9.1 → nuclia-4.9.2}/nuclia/sdk/kb.py

@@ -30,6 +30,10 @@ from nuclia.sdk.remi import NucliaRemi, AsyncNucliaRemi
 from nuclia.sdk.resource import AsyncNucliaResource, NucliaResource
 from nuclia.sdk.search import AsyncNucliaSearch, NucliaSearch
 from nuclia.sdk.task import NucliaTask, AsyncNucliaTask
+from nuclia.sdk.extract_strategy import (
+    NucliaExtractStrategy,
+    AsyncNucliaExtractStrategy,
+)
 from nuclia.sdk.upload import AsyncNucliaUpload, NucliaUpload
 
 
@@ -48,6 +52,7 @@ class NucliaKB:
         self.logs = NucliaLogs()
         self.task = NucliaTask()
         self.remi = NucliaRemi()
+        self.extract_strategies = NucliaExtractStrategy()
 
     @kb
     def list(
@@ -459,6 +464,7 @@ class AsyncNucliaKB:
         self.logs = AsyncNucliaLogs()
         self.task = AsyncNucliaTask()
         self.remi = AsyncNucliaRemi()
+        self.extract_strategies = AsyncNucliaExtractStrategy()
 
     @kb
     async def list(self, **kwargs) -> ResourceList:

nuclia-4.9.2/nuclia/tests/test_kb/test_extract_strategies.py

@@ -0,0 +1,16 @@
+from nuclia.sdk.kb import NucliaKB
+
+
+def test_extract_strategies(testing_config):
+    nkb = NucliaKB()
+    # preventive clean up
+    for id in nkb.extract_strategies.list().keys():
+        nkb.extract_strategies.delete(id=id)
+
+    # tests
+    nkb.extract_strategies.add(config={"name": "strategy1", "vllm_config": {}})
+    all = nkb.extract_strategies.list()
+    assert len(all.keys()) == 1
+    nkb.extract_strategies.delete(id=list(all.keys())[0])
+    all = nkb.extract_strategies.list()
+    assert len(all.keys()) == 0

{nuclia-4.9.1 → nuclia-4.9.2}/nuclia.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nuclia
-Version: 4.9.1
+Version: 4.9.2
 Summary: Nuclia Python SDK
 Author-email: Nuclia <info@nuclia.com>
 License-Expression: MIT

{nuclia-4.9.1 → nuclia-4.9.2}/nuclia.egg-info/SOURCES.txt

@@ -49,6 +49,7 @@ nuclia/sdk/agent.py
 nuclia/sdk/auth.py
 nuclia/sdk/backup.py
 nuclia/sdk/export_import.py
+nuclia/sdk/extract_strategy.py
 nuclia/sdk/kb.py
 nuclia/sdk/kbs.py
 nuclia/sdk/logger.py
@@ -70,6 +71,7 @@ nuclia/tests/assets/conversation.json
 nuclia/tests/test_kb/test_backup.py
 nuclia/tests/test_kb/test_conversation.py
 nuclia/tests/test_kb/test_export_import.py
+nuclia/tests/test_kb/test_extract_strategies.py
 nuclia/tests/test_kb/test_graph.py
 nuclia/tests/test_kb/test_labels.py
 nuclia/tests/test_kb/test_logs.py

nuclia-4.9.1/VERSION

@@ -1 +0,0 @@
-4.9.1