Skip to content

LLM-Driven Extraction of Unstructured Data — Built for API Deployments & ETL Pipeline Workflows

feat: Added x2text / ocr defaults to tool settings schema API (#121)

Chandrasekharan M68b818bdb372
diff --git a/backend/adapter_processor/adapter_processor.py b/backend/adapter_processor/adapter_processor.py
index 9a711f34f..159952a14 100644
--- a/backend/adapter_processor/adapter_processor.py
+++ b/backend/adapter_processor/adapter_processor.py
@@ -99,18 +99,18 @@ class AdapterProcessor:
adapter_metadata.pop(AdapterKeys.ADAPTER_TYPE)
== AdapterKeys.X2TEXT
):
- adapter_metadata[X2TextConstants.X2TEXT_HOST] = (
- settings.X2TEXT_HOST
- )
- adapter_metadata[X2TextConstants.X2TEXT_PORT] = (
- settings.X2TEXT_PORT
- )
+ adapter_metadata[
+ X2TextConstants.X2TEXT_HOST
+ ] = settings.X2TEXT_HOST
+ adapter_metadata[
+ X2TextConstants.X2TEXT_PORT
+ ] = settings.X2TEXT_PORT
platform_key = (
PlatformAuthenticationService.get_active_platform_key()
)
- adapter_metadata[X2TextConstants.PLATFORM_SERVICE_API_KEY] = (
- str(platform_key.key)
- )
+ adapter_metadata[
+ X2TextConstants.PLATFORM_SERVICE_API_KEY
+ ] = str(platform_key.key)
adapter_instance = adapter_class(adapter_metadata)
test_result: bool = adapter_instance.test_connection()
@@ -272,7 +272,10 @@ class AdapterProcessor:
return adapters
except ObjectDoesNotExist as e:
logger.error(f"No default adapters found: {e}")
- raise InternalServiceError("No default adapters found")
+ raise InternalServiceError(
+ "No default adapters found, "
+ "configure them through Platform Settings"
+ )
except Exception as e:
logger.error(f"Error occurred while fetching default adapters: {e}")
raise InternalServiceError("Error fetching default adapters")
diff --git a/backend/tool_instance/tool_instance_helper.py b/backend/tool_instance/tool_instance_helper.py
index 102460276..c0d07fab2 100644
--- a/backend/tool_instance/tool_instance_helper.py
+++ b/backend/tool_instance/tool_instance_helper.py
@@ -179,6 +179,7 @@ class ToolInstanceHelper:
adapter_type=AdapterTypes.OCR,
)
+ # TODO: Review if adding this metadata is still required
@staticmethod
def get_altered_metadata(
tool_instance: ToolInstance,
diff --git a/unstract/tool-registry/src/unstract/tool_registry/constants.py b/unstract/tool-registry/src/unstract/tool_registry/constants.py
index db03fa067..334fc7442 100644
--- a/unstract/tool-registry/src/unstract/tool_registry/constants.py
+++ b/unstract/tool-registry/src/unstract/tool_registry/constants.py
@@ -65,6 +65,9 @@ class AdapterPropertyKey:
ADAPTER_ID_KEY = "adapterIdKey"
ADAPTER_ID = "adapterId"
ADAPTER_TYPE = "adapterType"
- LLM_ADAPTER_ID = "llmAdapterId"
- EMBEDDING_ADAPTER_ID = "embeddingAdapterId"
- VECTOR_DB_ADAPTER_ID = "vectorDbAdapterId"
+ # TODO: Define defaults in SDK and use within tools
+ DEFAULT_LLM_ADAPTER_ID = "llmAdapterId"
+ DEFAULT_EMBEDDING_ADAPTER_ID = "embeddingAdapterId"
+ DEFAULT_VECTOR_DB_ADAPTER_ID = "vectorDbAdapterId"
+ DEFAULT_X2TEXT_ADAPTER_ID = "x2TextAdapterId"
+ DEFAULT_OCR_ADAPTER_ID = "ocrAdapterId"
diff --git a/unstract/tool-registry/src/unstract/tool_registry/tool_utils.py b/unstract/tool-registry/src/unstract/tool_registry/tool_utils.py
index c8b98586d..3dcb33f04 100644
--- a/unstract/tool-registry/src/unstract/tool_registry/tool_utils.py
+++ b/unstract/tool-registry/src/unstract/tool_registry/tool_utils.py
@@ -254,28 +254,30 @@ class ToolUtils:
models=language_models,
adapter_type=AdapterTypes.LLM,
schema=schema,
- default_adapter_id=AdapterPropertyKey.LLM_ADAPTER_ID,
+ default_adapter_id=AdapterPropertyKey.DEFAULT_LLM_ADAPTER_ID,
)
ToolUtils.process_adapter_models(
models=embeddings,
adapter_type=AdapterTypes.EMBEDDING,
schema=schema,
- default_adapter_id=AdapterPropertyKey.EMBEDDING_ADAPTER_ID,
+ default_adapter_id=AdapterPropertyKey.DEFAULT_EMBEDDING_ADAPTER_ID,
)
ToolUtils.process_adapter_models(
models=vector_stores,
adapter_type=AdapterTypes.VECTOR_DB,
schema=schema,
- default_adapter_id=AdapterPropertyKey.VECTOR_DB_ADAPTER_ID,
+ default_adapter_id=AdapterPropertyKey.DEFAULT_VECTOR_DB_ADAPTER_ID,
)
ToolUtils.process_adapter_models(
models=text_extractors,
adapter_type=AdapterTypes.X2TEXT,
schema=schema,
+ default_adapter_id=AdapterPropertyKey.DEFAULT_X2TEXT_ADAPTER_ID,
)
ToolUtils.process_adapter_models(
models=ocrs,
adapter_type=AdapterTypes.OCR,
schema=schema,
+ default_adapter_id=AdapterPropertyKey.DEFAULT_OCR_ADAPTER_ID,
)
return schema