diff --git a/src/backend/base/langflow/components/data/kb_ingest.py b/src/backend/base/langflow/components/data/kb_ingest.py index 3e70d18a8..1dd63088d 100644 --- a/src/backend/base/langflow/components/data/kb_ingest.py +++ b/src/backend/base/langflow/components/data/kb_ingest.py @@ -139,8 +139,8 @@ class KBIngestionComponent(Component): { "column_name": "text", "vectorize": True, - "identifier": False, - } + "identifier": True, + }, ], ), IntInput( @@ -402,16 +402,22 @@ class KBIngestionComponent(Component): # Convert each row to a Data object for _, row in df_source.iterrows(): - # Build content text from vectorized columns using list comprehension - content_parts = [str(row[col]) for col in content_cols if col in row and pd.notna(row[col])] + # Build content text from identifier columns using list comprehension + identifier_parts = [str(row[col]) for col in content_cols if col in row and pd.notna(row[col])] - page_content = " ".join(content_parts) + # Join all parts into a single string + page_content = " ".join(identifier_parts) # Build metadata from NON-vectorized columns only (simple key-value pairs) data_dict = { "text": page_content, # Main content for vectorization } + # Add identifier columns if they exist + if identifier_cols: + identifier_parts = [str(row[col]) for col in identifier_cols if col in row and pd.notna(row[col])] + page_content = " ".join(identifier_parts) + # Add metadata columns as simple key-value pairs for col in df_source.columns: if col not in content_cols and col in row and pd.notna(row[col]): diff --git a/src/backend/base/langflow/initial_setup/starter_projects/Knowledge Ingestion.json b/src/backend/base/langflow/initial_setup/starter_projects/Knowledge Ingestion.json index f58e152a8..0371c75ec 100644 --- a/src/backend/base/langflow/initial_setup/starter_projects/Knowledge Ingestion.json +++ b/src/backend/base/langflow/initial_setup/starter_projects/Knowledge Ingestion.json @@ -702,7 +702,7 @@ "last_updated": "2025-08-13T19:45:49.122Z", "legacy": false, "metadata": { - "code_hash": "0161c64fee9a", + "code_hash": "e1ebcd66ecbc", "module": "langflow.components.data.kb_ingest.KBIngestionComponent" }, "minimized": false, @@ -795,7 +795,7 @@ "show": true, "title_case": false, "type": "code", - "value": "from __future__ import annotations\n\nimport hashlib\nimport json\nimport re\nimport uuid\nfrom dataclasses import asdict, dataclass, field\nfrom datetime import datetime, timezone\nfrom pathlib import Path\nfrom typing import Any\n\nimport pandas as pd\nfrom cryptography.fernet import InvalidToken\nfrom langchain_chroma import Chroma\nfrom loguru import logger\n\nfrom langflow.base.models.openai_constants import OPENAI_EMBEDDING_MODEL_NAMES\nfrom langflow.custom import Component\nfrom langflow.io import BoolInput, DataFrameInput, DropdownInput, IntInput, Output, SecretStrInput, StrInput, TableInput\nfrom langflow.schema.data import Data\nfrom langflow.schema.dotdict import dotdict # noqa: TC001\nfrom langflow.schema.table import EditMode\nfrom langflow.services.auth.utils import decrypt_api_key, encrypt_api_key\nfrom langflow.services.deps import get_settings_service\n\nHUGGINGFACE_MODEL_NAMES = [\"sentence-transformers/all-MiniLM-L6-v2\", \"sentence-transformers/all-mpnet-base-v2\"]\nCOHERE_MODEL_NAMES = [\"embed-english-v3.0\", \"embed-multilingual-v3.0\"]\n\nsettings = get_settings_service().settings\nknowledge_directory = settings.knowledge_bases_dir\nif not knowledge_directory:\n msg = \"Knowledge bases directory is not set in the settings.\"\n raise ValueError(msg)\nKNOWLEDGE_BASES_ROOT_PATH = Path(knowledge_directory).expanduser()\n\n\nclass KBIngestionComponent(Component):\n \"\"\"Create or append to Langflow Knowledge from a DataFrame.\"\"\"\n\n # ------ UI metadata ---------------------------------------------------\n display_name = \"Knowledge Ingestion\"\n description = \"Create or update knowledge in Langflow.\"\n icon = \"database\"\n name = \"KBIngestion\"\n\n @dataclass\n class NewKnowledgeBaseInput:\n functionality: str = \"create\"\n fields: dict[str, dict] = field(\n default_factory=lambda: {\n \"data\": {\n \"node\": {\n \"name\": \"create_knowledge_base\",\n \"description\": \"Create new knowledge in Langflow.\",\n \"display_name\": \"Create new knowledge\",\n \"field_order\": [\"01_new_kb_name\", \"02_embedding_model\", \"03_api_key\"],\n \"template\": {\n \"01_new_kb_name\": StrInput(\n name=\"new_kb_name\",\n display_name=\"Knowledge Name\",\n info=\"Name of the new knowledge to create.\",\n required=True,\n ),\n \"02_embedding_model\": DropdownInput(\n name=\"embedding_model\",\n display_name=\"Model Name\",\n info=\"Select the embedding model to use for this knowledge base.\",\n required=True,\n options=OPENAI_EMBEDDING_MODEL_NAMES + HUGGINGFACE_MODEL_NAMES + COHERE_MODEL_NAMES,\n options_metadata=[{\"icon\": \"OpenAI\"} for _ in OPENAI_EMBEDDING_MODEL_NAMES]\n + [{\"icon\": \"HuggingFace\"} for _ in HUGGINGFACE_MODEL_NAMES]\n + [{\"icon\": \"Cohere\"} for _ in COHERE_MODEL_NAMES],\n ),\n \"03_api_key\": SecretStrInput(\n name=\"api_key\",\n display_name=\"API Key\",\n info=\"Provider API key for embedding model\",\n required=True,\n load_from_db=True,\n ),\n },\n },\n }\n }\n )\n\n # ------ Inputs --------------------------------------------------------\n inputs = [\n DropdownInput(\n name=\"knowledge_base\",\n display_name=\"Knowledge\",\n info=\"Select the knowledge to load data from.\",\n required=True,\n options=[\n str(d.name) for d in KNOWLEDGE_BASES_ROOT_PATH.iterdir() if not d.name.startswith(\".\") and d.is_dir()\n ]\n if KNOWLEDGE_BASES_ROOT_PATH.exists()\n else [],\n refresh_button=True,\n dialog_inputs=asdict(NewKnowledgeBaseInput()),\n ),\n DataFrameInput(\n name=\"input_df\",\n display_name=\"Data\",\n info=\"Table with all original columns (already chunked / processed).\",\n required=True,\n ),\n TableInput(\n name=\"column_config\",\n display_name=\"Column Configuration\",\n info=\"Configure column behavior for the knowledge base.\",\n required=True,\n table_schema=[\n {\n \"name\": \"column_name\",\n \"display_name\": \"Column Name\",\n \"type\": \"str\",\n \"description\": \"Name of the column in the source DataFrame\",\n \"edit_mode\": EditMode.INLINE,\n },\n {\n \"name\": \"vectorize\",\n \"display_name\": \"Vectorize\",\n \"type\": \"boolean\",\n \"description\": \"Create embeddings for this column\",\n \"default\": False,\n \"edit_mode\": EditMode.INLINE,\n },\n {\n \"name\": \"identifier\",\n \"display_name\": \"Identifier\",\n \"type\": \"boolean\",\n \"description\": \"Use this column as unique identifier\",\n \"default\": False,\n \"edit_mode\": EditMode.INLINE,\n },\n ],\n value=[\n {\n \"column_name\": \"text\",\n \"vectorize\": True,\n \"identifier\": False,\n }\n ],\n ),\n IntInput(\n name=\"chunk_size\",\n display_name=\"Chunk Size\",\n info=\"Batch size for processing embeddings\",\n advanced=True,\n value=1000,\n ),\n SecretStrInput(\n name=\"api_key\",\n display_name=\"Embedding Provider API Key\",\n info=\"API key for the embedding provider to generate embeddings.\",\n advanced=True,\n required=False,\n ),\n BoolInput(\n name=\"allow_duplicates\",\n display_name=\"Allow Duplicates\",\n info=\"Allow duplicate rows in the knowledge base\",\n advanced=True,\n value=False,\n ),\n ]\n\n # ------ Outputs -------------------------------------------------------\n outputs = [Output(display_name=\"DataFrame\", name=\"dataframe\", method=\"build_kb_info\")]\n\n # ------ Internal helpers ---------------------------------------------\n def _get_kb_root(self) -> Path:\n \"\"\"Return the root directory for knowledge bases.\"\"\"\n return KNOWLEDGE_BASES_ROOT_PATH\n\n def _validate_column_config(self, df_source: pd.DataFrame) -> list[dict[str, Any]]:\n \"\"\"Validate column configuration using Structured Output patterns.\"\"\"\n if not self.column_config:\n msg = \"Column configuration cannot be empty\"\n raise ValueError(msg)\n\n # Convert table input to list of dicts (similar to Structured Output)\n config_list = self.column_config if isinstance(self.column_config, list) else []\n\n # Validate column names exist in DataFrame\n df_columns = set(df_source.columns)\n for config in config_list:\n col_name = config.get(\"column_name\")\n if col_name not in df_columns:\n msg = f\"Column '{col_name}' not found in DataFrame. Available columns: {sorted(df_columns)}\"\n raise ValueError(msg)\n\n return config_list\n\n def _get_embedding_provider(self, embedding_model: str) -> str:\n \"\"\"Get embedding provider by matching model name to lists.\"\"\"\n if embedding_model in OPENAI_EMBEDDING_MODEL_NAMES:\n return \"OpenAI\"\n if embedding_model in HUGGINGFACE_MODEL_NAMES:\n return \"HuggingFace\"\n if embedding_model in COHERE_MODEL_NAMES:\n return \"Cohere\"\n return \"Custom\"\n\n def _build_embeddings(self, embedding_model: str, api_key: str):\n \"\"\"Build embedding model using provider patterns.\"\"\"\n # Get provider by matching model name to lists\n provider = self._get_embedding_provider(embedding_model)\n\n # Validate provider and model\n if provider == \"OpenAI\":\n from langchain_openai import OpenAIEmbeddings\n\n if not api_key:\n msg = \"OpenAI API key is required when using OpenAI provider\"\n raise ValueError(msg)\n return OpenAIEmbeddings(\n model=embedding_model,\n api_key=api_key,\n chunk_size=self.chunk_size,\n )\n if provider == \"HuggingFace\":\n from langchain_huggingface import HuggingFaceEmbeddings\n\n return HuggingFaceEmbeddings(\n model=embedding_model,\n )\n if provider == \"Cohere\":\n from langchain_cohere import CohereEmbeddings\n\n if not api_key:\n msg = \"Cohere API key is required when using Cohere provider\"\n raise ValueError(msg)\n return CohereEmbeddings(\n model=embedding_model,\n cohere_api_key=api_key,\n )\n if provider == \"Custom\":\n # For custom embedding models, we would need additional configuration\n msg = \"Custom embedding models not yet supported\"\n raise NotImplementedError(msg)\n msg = f\"Unknown provider: {provider}\"\n raise ValueError(msg)\n\n def _build_embedding_metadata(self, embedding_model, api_key) -> dict[str, Any]:\n \"\"\"Build embedding model metadata.\"\"\"\n # Get provider by matching model name to lists\n embedding_provider = self._get_embedding_provider(embedding_model)\n\n api_key_to_save = None\n if api_key and hasattr(api_key, \"get_secret_value\"):\n api_key_to_save = api_key.get_secret_value()\n elif isinstance(api_key, str):\n api_key_to_save = api_key\n\n encrypted_api_key = None\n if api_key_to_save:\n settings_service = get_settings_service()\n try:\n encrypted_api_key = encrypt_api_key(api_key_to_save, settings_service=settings_service)\n except (TypeError, ValueError) as e:\n self.log(f\"Could not encrypt API key: {e}\")\n logger.error(f\"Could not encrypt API key: {e}\")\n\n return {\n \"embedding_provider\": embedding_provider,\n \"embedding_model\": embedding_model,\n \"api_key\": encrypted_api_key,\n \"api_key_used\": bool(api_key),\n \"chunk_size\": self.chunk_size,\n \"created_at\": datetime.now(timezone.utc).isoformat(),\n }\n\n def _save_embedding_metadata(self, kb_path: Path, embedding_model: str, api_key: str) -> None:\n \"\"\"Save embedding model metadata.\"\"\"\n embedding_metadata = self._build_embedding_metadata(embedding_model, api_key)\n metadata_path = kb_path / \"embedding_metadata.json\"\n metadata_path.write_text(json.dumps(embedding_metadata, indent=2))\n\n def _save_kb_files(\n self,\n kb_path: Path,\n config_list: list[dict[str, Any]],\n ) -> None:\n \"\"\"Save KB files using File Component storage patterns.\"\"\"\n try:\n # Create directory (following File Component patterns)\n kb_path.mkdir(parents=True, exist_ok=True)\n\n # Save column configuration\n # Only do this if the file doesn't exist already\n cfg_path = kb_path / \"schema.json\"\n if not cfg_path.exists():\n cfg_path.write_text(json.dumps(config_list, indent=2))\n\n except (OSError, TypeError, ValueError) as e:\n self.log(f\"Error saving KB files: {e}\")\n\n def _build_column_metadata(self, config_list: list[dict[str, Any]], df_source: pd.DataFrame) -> dict[str, Any]:\n \"\"\"Build detailed column metadata.\"\"\"\n metadata: dict[str, Any] = {\n \"total_columns\": len(df_source.columns),\n \"mapped_columns\": len(config_list),\n \"unmapped_columns\": len(df_source.columns) - len(config_list),\n \"columns\": [],\n \"summary\": {\"vectorized_columns\": [], \"identifier_columns\": []},\n }\n\n for config in config_list:\n col_name = config.get(\"column_name\")\n vectorize = config.get(\"vectorize\") == \"True\" or config.get(\"vectorize\") is True\n identifier = config.get(\"identifier\") == \"True\" or config.get(\"identifier\") is True\n\n # Add to columns list\n metadata[\"columns\"].append(\n {\n \"name\": col_name,\n \"vectorize\": vectorize,\n \"identifier\": identifier,\n }\n )\n\n # Update summary\n if vectorize:\n metadata[\"summary\"][\"vectorized_columns\"].append(col_name)\n if identifier:\n metadata[\"summary\"][\"identifier_columns\"].append(col_name)\n\n return metadata\n\n def _create_vector_store(\n self, df_source: pd.DataFrame, config_list: list[dict[str, Any]], embedding_model: str, api_key: str\n ) -> None:\n \"\"\"Create vector store following Local DB component pattern.\"\"\"\n try:\n # Set up vector store directory\n base_dir = self._get_kb_root()\n\n vector_store_dir = base_dir / self.knowledge_base\n vector_store_dir.mkdir(parents=True, exist_ok=True)\n\n # Create embeddings model\n embedding_function = self._build_embeddings(embedding_model, api_key)\n\n # Convert DataFrame to Data objects (following Local DB pattern)\n data_objects = self._convert_df_to_data_objects(df_source, config_list)\n\n # Create vector store\n chroma = Chroma(\n persist_directory=str(vector_store_dir),\n embedding_function=embedding_function,\n collection_name=self.knowledge_base,\n )\n\n # Convert Data objects to LangChain Documents\n documents = []\n for data_obj in data_objects:\n doc = data_obj.to_lc_document()\n documents.append(doc)\n\n # Add documents to vector store\n if documents:\n chroma.add_documents(documents)\n self.log(f\"Added {len(documents)} documents to vector store '{self.knowledge_base}'\")\n\n except (OSError, ValueError, RuntimeError) as e:\n self.log(f\"Error creating vector store: {e}\")\n\n def _convert_df_to_data_objects(self, df_source: pd.DataFrame, config_list: list[dict[str, Any]]) -> list[Data]:\n \"\"\"Convert DataFrame to Data objects for vector store.\"\"\"\n data_objects: list[Data] = []\n\n # Set up vector store directory\n base_dir = self._get_kb_root()\n\n # If we don't allow duplicates, we need to get the existing hashes\n chroma = Chroma(\n persist_directory=str(base_dir / self.knowledge_base),\n collection_name=self.knowledge_base,\n )\n\n # Get all documents and their metadata\n all_docs = chroma.get()\n\n # Extract all _id values from metadata\n id_list = [metadata.get(\"_id\") for metadata in all_docs[\"metadatas\"] if metadata.get(\"_id\")]\n\n # Get column roles\n content_cols = []\n identifier_cols = []\n\n for config in config_list:\n col_name = config.get(\"column_name\")\n vectorize = config.get(\"vectorize\") == \"True\" or config.get(\"vectorize\") is True\n identifier = config.get(\"identifier\") == \"True\" or config.get(\"identifier\") is True\n\n if vectorize:\n content_cols.append(col_name)\n elif identifier:\n identifier_cols.append(col_name)\n\n # Convert each row to a Data object\n for _, row in df_source.iterrows():\n # Build content text from vectorized columns using list comprehension\n content_parts = [str(row[col]) for col in content_cols if col in row and pd.notna(row[col])]\n\n page_content = \" \".join(content_parts)\n\n # Build metadata from NON-vectorized columns only (simple key-value pairs)\n data_dict = {\n \"text\": page_content, # Main content for vectorization\n }\n\n # Add metadata columns as simple key-value pairs\n for col in df_source.columns:\n if col not in content_cols and col in row and pd.notna(row[col]):\n # Convert to simple types for Chroma metadata\n value = row[col]\n data_dict[col] = str(value) # Convert complex types to string\n\n # Hash the page_content for unique ID\n page_content_hash = hashlib.sha256(page_content.encode()).hexdigest()\n data_dict[\"_id\"] = page_content_hash\n\n # If duplicates are disallowed, and hash exists, prevent adding this row\n if not self.allow_duplicates and page_content_hash in id_list:\n self.log(f\"Skipping duplicate row with hash {page_content_hash}\")\n continue\n\n # Create Data object - everything except \"text\" becomes metadata\n data_obj = Data(data=data_dict)\n data_objects.append(data_obj)\n\n return data_objects\n\n def is_valid_collection_name(self, name, min_length: int = 3, max_length: int = 63) -> bool:\n \"\"\"Validates collection name against conditions 1-3.\n\n 1. Contains 3-63 characters\n 2. Starts and ends with alphanumeric character\n 3. Contains only alphanumeric characters, underscores, or hyphens.\n\n Args:\n name (str): Collection name to validate\n min_length (int): Minimum length of the name\n max_length (int): Maximum length of the name\n\n Returns:\n bool: True if valid, False otherwise\n \"\"\"\n # Check length (condition 1)\n if not (min_length <= len(name) <= max_length):\n return False\n\n # Check start/end with alphanumeric (condition 2)\n if not (name[0].isalnum() and name[-1].isalnum()):\n return False\n\n # Check allowed characters (condition 3)\n return re.match(r\"^[a-zA-Z0-9_-]+$\", name) is not None\n\n # ---------------------------------------------------------------------\n # OUTPUT METHODS\n # ---------------------------------------------------------------------\n def build_kb_info(self) -> Data:\n \"\"\"Main ingestion routine → returns a dict with KB metadata.\"\"\"\n try:\n # Get source DataFrame\n df_source: pd.DataFrame = self.input_df\n\n # Validate column configuration (using Structured Output patterns)\n config_list = self._validate_column_config(df_source)\n column_metadata = self._build_column_metadata(config_list, df_source)\n\n # Prepare KB folder (using File Component patterns)\n kb_root = self._get_kb_root()\n kb_path = kb_root / self.knowledge_base\n\n # Read the embedding info from the knowledge base folder\n metadata_path = kb_path / \"embedding_metadata.json\"\n\n # If the API key is not provided, try to read it from the metadata file\n if metadata_path.exists():\n settings_service = get_settings_service()\n metadata = json.loads(metadata_path.read_text())\n embedding_model = metadata.get(\"embedding_model\")\n try:\n api_key = decrypt_api_key(metadata[\"api_key\"], settings_service)\n except (InvalidToken, TypeError, ValueError) as e:\n logger.error(f\"Could not decrypt API key. Please provide it manually. Error: {e}\")\n\n # Check if a custom API key was provided, update metadata if so\n if self.api_key:\n api_key = self.api_key\n self._save_embedding_metadata(\n kb_path=kb_path,\n embedding_model=embedding_model,\n api_key=api_key,\n )\n\n # Create vector store following Local DB component pattern\n self._create_vector_store(df_source, config_list, embedding_model=embedding_model, api_key=api_key)\n\n # Save KB files (using File Component storage patterns)\n self._save_kb_files(kb_path, config_list)\n\n # Build metadata response\n meta: dict[str, Any] = {\n \"kb_id\": str(uuid.uuid4()),\n \"kb_name\": self.knowledge_base,\n \"rows\": len(df_source),\n \"column_metadata\": column_metadata,\n \"path\": str(kb_path),\n \"config_columns\": len(config_list),\n \"timestamp\": datetime.now(tz=timezone.utc).isoformat(),\n }\n\n # Set status message\n self.status = f\"✅ KB **{self.knowledge_base}** saved · {len(df_source)} chunks.\"\n\n return Data(data=meta)\n\n except (OSError, ValueError, RuntimeError, KeyError) as e:\n self.log(f\"Error in KB ingestion: {e}\")\n self.status = f\"❌ KB ingestion failed: {e}\"\n return Data(data={\"error\": str(e), \"kb_name\": self.knowledge_base})\n\n def _get_knowledge_bases(self) -> list[str]:\n \"\"\"Retrieve a list of available knowledge bases.\n\n Returns:\n A list of knowledge base names.\n \"\"\"\n # Return the list of directories in the knowledge base root path\n kb_root_path = self._get_kb_root()\n\n if not kb_root_path.exists():\n return []\n\n return [str(d.name) for d in kb_root_path.iterdir() if not d.name.startswith(\".\") and d.is_dir()]\n\n def update_build_config(self, build_config: dotdict, field_value: Any, field_name: str | None = None) -> dotdict:\n \"\"\"Update build configuration based on provider selection.\"\"\"\n # Create a new knowledge base\n if field_name == \"knowledge_base\":\n if isinstance(field_value, dict) and \"01_new_kb_name\" in field_value:\n # Validate the knowledge base name - Make sure it follows these rules:\n if not self.is_valid_collection_name(field_value[\"01_new_kb_name\"]):\n msg = f\"Invalid knowledge base name: {field_value['01_new_kb_name']}\"\n raise ValueError(msg)\n\n # We need to test the API Key one time against the embedding model\n embed_model = self._build_embeddings(\n embedding_model=field_value[\"02_embedding_model\"], api_key=field_value[\"03_api_key\"]\n )\n\n # Try to generate a dummy embedding to validate the API key\n embed_model.embed_query(\"test\")\n\n # Create the new knowledge base directory\n kb_path = KNOWLEDGE_BASES_ROOT_PATH / field_value[\"01_new_kb_name\"]\n kb_path.mkdir(parents=True, exist_ok=True)\n\n # Save the embedding metadata\n build_config[\"knowledge_base\"][\"value\"] = field_value[\"01_new_kb_name\"]\n self._save_embedding_metadata(\n kb_path=kb_path,\n embedding_model=field_value[\"02_embedding_model\"],\n api_key=field_value[\"03_api_key\"],\n )\n\n # Update the knowledge base options dynamically\n build_config[\"knowledge_base\"][\"options\"] = self._get_knowledge_bases()\n if build_config[\"knowledge_base\"][\"value\"] not in build_config[\"knowledge_base\"][\"options\"]:\n build_config[\"knowledge_base\"][\"value\"] = None\n\n return build_config\n" + "value": "from __future__ import annotations\n\nimport hashlib\nimport json\nimport re\nimport uuid\nfrom dataclasses import asdict, dataclass, field\nfrom datetime import datetime, timezone\nfrom pathlib import Path\nfrom typing import Any\n\nimport pandas as pd\nfrom cryptography.fernet import InvalidToken\nfrom langchain_chroma import Chroma\nfrom loguru import logger\n\nfrom langflow.base.models.openai_constants import OPENAI_EMBEDDING_MODEL_NAMES\nfrom langflow.custom import Component\nfrom langflow.io import BoolInput, DataFrameInput, DropdownInput, IntInput, Output, SecretStrInput, StrInput, TableInput\nfrom langflow.schema.data import Data\nfrom langflow.schema.dotdict import dotdict # noqa: TC001\nfrom langflow.schema.table import EditMode\nfrom langflow.services.auth.utils import decrypt_api_key, encrypt_api_key\nfrom langflow.services.deps import get_settings_service\n\nHUGGINGFACE_MODEL_NAMES = [\"sentence-transformers/all-MiniLM-L6-v2\", \"sentence-transformers/all-mpnet-base-v2\"]\nCOHERE_MODEL_NAMES = [\"embed-english-v3.0\", \"embed-multilingual-v3.0\"]\n\nsettings = get_settings_service().settings\nknowledge_directory = settings.knowledge_bases_dir\nif not knowledge_directory:\n msg = \"Knowledge bases directory is not set in the settings.\"\n raise ValueError(msg)\nKNOWLEDGE_BASES_ROOT_PATH = Path(knowledge_directory).expanduser()\n\n\nclass KBIngestionComponent(Component):\n \"\"\"Create or append to Langflow Knowledge from a DataFrame.\"\"\"\n\n # ------ UI metadata ---------------------------------------------------\n display_name = \"Knowledge Ingestion\"\n description = \"Create or update knowledge in Langflow.\"\n icon = \"database\"\n name = \"KBIngestion\"\n\n @dataclass\n class NewKnowledgeBaseInput:\n functionality: str = \"create\"\n fields: dict[str, dict] = field(\n default_factory=lambda: {\n \"data\": {\n \"node\": {\n \"name\": \"create_knowledge_base\",\n \"description\": \"Create new knowledge in Langflow.\",\n \"display_name\": \"Create new knowledge\",\n \"field_order\": [\"01_new_kb_name\", \"02_embedding_model\", \"03_api_key\"],\n \"template\": {\n \"01_new_kb_name\": StrInput(\n name=\"new_kb_name\",\n display_name=\"Knowledge Name\",\n info=\"Name of the new knowledge to create.\",\n required=True,\n ),\n \"02_embedding_model\": DropdownInput(\n name=\"embedding_model\",\n display_name=\"Model Name\",\n info=\"Select the embedding model to use for this knowledge base.\",\n required=True,\n options=OPENAI_EMBEDDING_MODEL_NAMES + HUGGINGFACE_MODEL_NAMES + COHERE_MODEL_NAMES,\n options_metadata=[{\"icon\": \"OpenAI\"} for _ in OPENAI_EMBEDDING_MODEL_NAMES]\n + [{\"icon\": \"HuggingFace\"} for _ in HUGGINGFACE_MODEL_NAMES]\n + [{\"icon\": \"Cohere\"} for _ in COHERE_MODEL_NAMES],\n ),\n \"03_api_key\": SecretStrInput(\n name=\"api_key\",\n display_name=\"API Key\",\n info=\"Provider API key for embedding model\",\n required=True,\n load_from_db=True,\n ),\n },\n },\n }\n }\n )\n\n # ------ Inputs --------------------------------------------------------\n inputs = [\n DropdownInput(\n name=\"knowledge_base\",\n display_name=\"Knowledge\",\n info=\"Select the knowledge to load data from.\",\n required=True,\n options=[\n str(d.name) for d in KNOWLEDGE_BASES_ROOT_PATH.iterdir() if not d.name.startswith(\".\") and d.is_dir()\n ]\n if KNOWLEDGE_BASES_ROOT_PATH.exists()\n else [],\n refresh_button=True,\n dialog_inputs=asdict(NewKnowledgeBaseInput()),\n ),\n DataFrameInput(\n name=\"input_df\",\n display_name=\"Data\",\n info=\"Table with all original columns (already chunked / processed).\",\n required=True,\n ),\n TableInput(\n name=\"column_config\",\n display_name=\"Column Configuration\",\n info=\"Configure column behavior for the knowledge base.\",\n required=True,\n table_schema=[\n {\n \"name\": \"column_name\",\n \"display_name\": \"Column Name\",\n \"type\": \"str\",\n \"description\": \"Name of the column in the source DataFrame\",\n \"edit_mode\": EditMode.INLINE,\n },\n {\n \"name\": \"vectorize\",\n \"display_name\": \"Vectorize\",\n \"type\": \"boolean\",\n \"description\": \"Create embeddings for this column\",\n \"default\": False,\n \"edit_mode\": EditMode.INLINE,\n },\n {\n \"name\": \"identifier\",\n \"display_name\": \"Identifier\",\n \"type\": \"boolean\",\n \"description\": \"Use this column as unique identifier\",\n \"default\": False,\n \"edit_mode\": EditMode.INLINE,\n },\n ],\n value=[\n {\n \"column_name\": \"text\",\n \"vectorize\": True,\n \"identifier\": True,\n },\n ],\n ),\n IntInput(\n name=\"chunk_size\",\n display_name=\"Chunk Size\",\n info=\"Batch size for processing embeddings\",\n advanced=True,\n value=1000,\n ),\n SecretStrInput(\n name=\"api_key\",\n display_name=\"Embedding Provider API Key\",\n info=\"API key for the embedding provider to generate embeddings.\",\n advanced=True,\n required=False,\n ),\n BoolInput(\n name=\"allow_duplicates\",\n display_name=\"Allow Duplicates\",\n info=\"Allow duplicate rows in the knowledge base\",\n advanced=True,\n value=False,\n ),\n ]\n\n # ------ Outputs -------------------------------------------------------\n outputs = [Output(display_name=\"DataFrame\", name=\"dataframe\", method=\"build_kb_info\")]\n\n # ------ Internal helpers ---------------------------------------------\n def _get_kb_root(self) -> Path:\n \"\"\"Return the root directory for knowledge bases.\"\"\"\n return KNOWLEDGE_BASES_ROOT_PATH\n\n def _validate_column_config(self, df_source: pd.DataFrame) -> list[dict[str, Any]]:\n \"\"\"Validate column configuration using Structured Output patterns.\"\"\"\n if not self.column_config:\n msg = \"Column configuration cannot be empty\"\n raise ValueError(msg)\n\n # Convert table input to list of dicts (similar to Structured Output)\n config_list = self.column_config if isinstance(self.column_config, list) else []\n\n # Validate column names exist in DataFrame\n df_columns = set(df_source.columns)\n for config in config_list:\n col_name = config.get(\"column_name\")\n if col_name not in df_columns:\n msg = f\"Column '{col_name}' not found in DataFrame. Available columns: {sorted(df_columns)}\"\n raise ValueError(msg)\n\n return config_list\n\n def _get_embedding_provider(self, embedding_model: str) -> str:\n \"\"\"Get embedding provider by matching model name to lists.\"\"\"\n if embedding_model in OPENAI_EMBEDDING_MODEL_NAMES:\n return \"OpenAI\"\n if embedding_model in HUGGINGFACE_MODEL_NAMES:\n return \"HuggingFace\"\n if embedding_model in COHERE_MODEL_NAMES:\n return \"Cohere\"\n return \"Custom\"\n\n def _build_embeddings(self, embedding_model: str, api_key: str):\n \"\"\"Build embedding model using provider patterns.\"\"\"\n # Get provider by matching model name to lists\n provider = self._get_embedding_provider(embedding_model)\n\n # Validate provider and model\n if provider == \"OpenAI\":\n from langchain_openai import OpenAIEmbeddings\n\n if not api_key:\n msg = \"OpenAI API key is required when using OpenAI provider\"\n raise ValueError(msg)\n return OpenAIEmbeddings(\n model=embedding_model,\n api_key=api_key,\n chunk_size=self.chunk_size,\n )\n if provider == \"HuggingFace\":\n from langchain_huggingface import HuggingFaceEmbeddings\n\n return HuggingFaceEmbeddings(\n model=embedding_model,\n )\n if provider == \"Cohere\":\n from langchain_cohere import CohereEmbeddings\n\n if not api_key:\n msg = \"Cohere API key is required when using Cohere provider\"\n raise ValueError(msg)\n return CohereEmbeddings(\n model=embedding_model,\n cohere_api_key=api_key,\n )\n if provider == \"Custom\":\n # For custom embedding models, we would need additional configuration\n msg = \"Custom embedding models not yet supported\"\n raise NotImplementedError(msg)\n msg = f\"Unknown provider: {provider}\"\n raise ValueError(msg)\n\n def _build_embedding_metadata(self, embedding_model, api_key) -> dict[str, Any]:\n \"\"\"Build embedding model metadata.\"\"\"\n # Get provider by matching model name to lists\n embedding_provider = self._get_embedding_provider(embedding_model)\n\n api_key_to_save = None\n if api_key and hasattr(api_key, \"get_secret_value\"):\n api_key_to_save = api_key.get_secret_value()\n elif isinstance(api_key, str):\n api_key_to_save = api_key\n\n encrypted_api_key = None\n if api_key_to_save:\n settings_service = get_settings_service()\n try:\n encrypted_api_key = encrypt_api_key(api_key_to_save, settings_service=settings_service)\n except (TypeError, ValueError) as e:\n self.log(f\"Could not encrypt API key: {e}\")\n logger.error(f\"Could not encrypt API key: {e}\")\n\n return {\n \"embedding_provider\": embedding_provider,\n \"embedding_model\": embedding_model,\n \"api_key\": encrypted_api_key,\n \"api_key_used\": bool(api_key),\n \"chunk_size\": self.chunk_size,\n \"created_at\": datetime.now(timezone.utc).isoformat(),\n }\n\n def _save_embedding_metadata(self, kb_path: Path, embedding_model: str, api_key: str) -> None:\n \"\"\"Save embedding model metadata.\"\"\"\n embedding_metadata = self._build_embedding_metadata(embedding_model, api_key)\n metadata_path = kb_path / \"embedding_metadata.json\"\n metadata_path.write_text(json.dumps(embedding_metadata, indent=2))\n\n def _save_kb_files(\n self,\n kb_path: Path,\n config_list: list[dict[str, Any]],\n ) -> None:\n \"\"\"Save KB files using File Component storage patterns.\"\"\"\n try:\n # Create directory (following File Component patterns)\n kb_path.mkdir(parents=True, exist_ok=True)\n\n # Save column configuration\n # Only do this if the file doesn't exist already\n cfg_path = kb_path / \"schema.json\"\n if not cfg_path.exists():\n cfg_path.write_text(json.dumps(config_list, indent=2))\n\n except (OSError, TypeError, ValueError) as e:\n self.log(f\"Error saving KB files: {e}\")\n\n def _build_column_metadata(self, config_list: list[dict[str, Any]], df_source: pd.DataFrame) -> dict[str, Any]:\n \"\"\"Build detailed column metadata.\"\"\"\n metadata: dict[str, Any] = {\n \"total_columns\": len(df_source.columns),\n \"mapped_columns\": len(config_list),\n \"unmapped_columns\": len(df_source.columns) - len(config_list),\n \"columns\": [],\n \"summary\": {\"vectorized_columns\": [], \"identifier_columns\": []},\n }\n\n for config in config_list:\n col_name = config.get(\"column_name\")\n vectorize = config.get(\"vectorize\") == \"True\" or config.get(\"vectorize\") is True\n identifier = config.get(\"identifier\") == \"True\" or config.get(\"identifier\") is True\n\n # Add to columns list\n metadata[\"columns\"].append(\n {\n \"name\": col_name,\n \"vectorize\": vectorize,\n \"identifier\": identifier,\n }\n )\n\n # Update summary\n if vectorize:\n metadata[\"summary\"][\"vectorized_columns\"].append(col_name)\n if identifier:\n metadata[\"summary\"][\"identifier_columns\"].append(col_name)\n\n return metadata\n\n def _create_vector_store(\n self, df_source: pd.DataFrame, config_list: list[dict[str, Any]], embedding_model: str, api_key: str\n ) -> None:\n \"\"\"Create vector store following Local DB component pattern.\"\"\"\n try:\n # Set up vector store directory\n base_dir = self._get_kb_root()\n\n vector_store_dir = base_dir / self.knowledge_base\n vector_store_dir.mkdir(parents=True, exist_ok=True)\n\n # Create embeddings model\n embedding_function = self._build_embeddings(embedding_model, api_key)\n\n # Convert DataFrame to Data objects (following Local DB pattern)\n data_objects = self._convert_df_to_data_objects(df_source, config_list)\n\n # Create vector store\n chroma = Chroma(\n persist_directory=str(vector_store_dir),\n embedding_function=embedding_function,\n collection_name=self.knowledge_base,\n )\n\n # Convert Data objects to LangChain Documents\n documents = []\n for data_obj in data_objects:\n doc = data_obj.to_lc_document()\n documents.append(doc)\n\n # Add documents to vector store\n if documents:\n chroma.add_documents(documents)\n self.log(f\"Added {len(documents)} documents to vector store '{self.knowledge_base}'\")\n\n except (OSError, ValueError, RuntimeError) as e:\n self.log(f\"Error creating vector store: {e}\")\n\n def _convert_df_to_data_objects(self, df_source: pd.DataFrame, config_list: list[dict[str, Any]]) -> list[Data]:\n \"\"\"Convert DataFrame to Data objects for vector store.\"\"\"\n data_objects: list[Data] = []\n\n # Set up vector store directory\n base_dir = self._get_kb_root()\n\n # If we don't allow duplicates, we need to get the existing hashes\n chroma = Chroma(\n persist_directory=str(base_dir / self.knowledge_base),\n collection_name=self.knowledge_base,\n )\n\n # Get all documents and their metadata\n all_docs = chroma.get()\n\n # Extract all _id values from metadata\n id_list = [metadata.get(\"_id\") for metadata in all_docs[\"metadatas\"] if metadata.get(\"_id\")]\n\n # Get column roles\n content_cols = []\n identifier_cols = []\n\n for config in config_list:\n col_name = config.get(\"column_name\")\n vectorize = config.get(\"vectorize\") == \"True\" or config.get(\"vectorize\") is True\n identifier = config.get(\"identifier\") == \"True\" or config.get(\"identifier\") is True\n\n if vectorize:\n content_cols.append(col_name)\n elif identifier:\n identifier_cols.append(col_name)\n\n # Convert each row to a Data object\n for _, row in df_source.iterrows():\n # Build content text from identifier columns using list comprehension\n identifier_parts = [str(row[col]) for col in content_cols if col in row and pd.notna(row[col])]\n\n # Join all parts into a single string\n page_content = \" \".join(identifier_parts)\n\n # Build metadata from NON-vectorized columns only (simple key-value pairs)\n data_dict = {\n \"text\": page_content, # Main content for vectorization\n }\n\n # Add identifier columns if they exist\n if identifier_cols:\n identifier_parts = [str(row[col]) for col in identifier_cols if col in row and pd.notna(row[col])]\n page_content = \" \".join(identifier_parts)\n\n # Add metadata columns as simple key-value pairs\n for col in df_source.columns:\n if col not in content_cols and col in row and pd.notna(row[col]):\n # Convert to simple types for Chroma metadata\n value = row[col]\n data_dict[col] = str(value) # Convert complex types to string\n\n # Hash the page_content for unique ID\n page_content_hash = hashlib.sha256(page_content.encode()).hexdigest()\n data_dict[\"_id\"] = page_content_hash\n\n # If duplicates are disallowed, and hash exists, prevent adding this row\n if not self.allow_duplicates and page_content_hash in id_list:\n self.log(f\"Skipping duplicate row with hash {page_content_hash}\")\n continue\n\n # Create Data object - everything except \"text\" becomes metadata\n data_obj = Data(data=data_dict)\n data_objects.append(data_obj)\n\n return data_objects\n\n def is_valid_collection_name(self, name, min_length: int = 3, max_length: int = 63) -> bool:\n \"\"\"Validates collection name against conditions 1-3.\n\n 1. Contains 3-63 characters\n 2. Starts and ends with alphanumeric character\n 3. Contains only alphanumeric characters, underscores, or hyphens.\n\n Args:\n name (str): Collection name to validate\n min_length (int): Minimum length of the name\n max_length (int): Maximum length of the name\n\n Returns:\n bool: True if valid, False otherwise\n \"\"\"\n # Check length (condition 1)\n if not (min_length <= len(name) <= max_length):\n return False\n\n # Check start/end with alphanumeric (condition 2)\n if not (name[0].isalnum() and name[-1].isalnum()):\n return False\n\n # Check allowed characters (condition 3)\n return re.match(r\"^[a-zA-Z0-9_-]+$\", name) is not None\n\n # ---------------------------------------------------------------------\n # OUTPUT METHODS\n # ---------------------------------------------------------------------\n def build_kb_info(self) -> Data:\n \"\"\"Main ingestion routine → returns a dict with KB metadata.\"\"\"\n try:\n # Get source DataFrame\n df_source: pd.DataFrame = self.input_df\n\n # Validate column configuration (using Structured Output patterns)\n config_list = self._validate_column_config(df_source)\n column_metadata = self._build_column_metadata(config_list, df_source)\n\n # Prepare KB folder (using File Component patterns)\n kb_root = self._get_kb_root()\n kb_path = kb_root / self.knowledge_base\n\n # Read the embedding info from the knowledge base folder\n metadata_path = kb_path / \"embedding_metadata.json\"\n\n # If the API key is not provided, try to read it from the metadata file\n if metadata_path.exists():\n settings_service = get_settings_service()\n metadata = json.loads(metadata_path.read_text())\n embedding_model = metadata.get(\"embedding_model\")\n try:\n api_key = decrypt_api_key(metadata[\"api_key\"], settings_service)\n except (InvalidToken, TypeError, ValueError) as e:\n logger.error(f\"Could not decrypt API key. Please provide it manually. Error: {e}\")\n\n # Check if a custom API key was provided, update metadata if so\n if self.api_key:\n api_key = self.api_key\n self._save_embedding_metadata(\n kb_path=kb_path,\n embedding_model=embedding_model,\n api_key=api_key,\n )\n\n # Create vector store following Local DB component pattern\n self._create_vector_store(df_source, config_list, embedding_model=embedding_model, api_key=api_key)\n\n # Save KB files (using File Component storage patterns)\n self._save_kb_files(kb_path, config_list)\n\n # Build metadata response\n meta: dict[str, Any] = {\n \"kb_id\": str(uuid.uuid4()),\n \"kb_name\": self.knowledge_base,\n \"rows\": len(df_source),\n \"column_metadata\": column_metadata,\n \"path\": str(kb_path),\n \"config_columns\": len(config_list),\n \"timestamp\": datetime.now(tz=timezone.utc).isoformat(),\n }\n\n # Set status message\n self.status = f\"✅ KB **{self.knowledge_base}** saved · {len(df_source)} chunks.\"\n\n return Data(data=meta)\n\n except (OSError, ValueError, RuntimeError, KeyError) as e:\n self.log(f\"Error in KB ingestion: {e}\")\n self.status = f\"❌ KB ingestion failed: {e}\"\n return Data(data={\"error\": str(e), \"kb_name\": self.knowledge_base})\n\n def _get_knowledge_bases(self) -> list[str]:\n \"\"\"Retrieve a list of available knowledge bases.\n\n Returns:\n A list of knowledge base names.\n \"\"\"\n # Return the list of directories in the knowledge base root path\n kb_root_path = self._get_kb_root()\n\n if not kb_root_path.exists():\n return []\n\n return [str(d.name) for d in kb_root_path.iterdir() if not d.name.startswith(\".\") and d.is_dir()]\n\n def update_build_config(self, build_config: dotdict, field_value: Any, field_name: str | None = None) -> dotdict:\n \"\"\"Update build configuration based on provider selection.\"\"\"\n # Create a new knowledge base\n if field_name == \"knowledge_base\":\n if isinstance(field_value, dict) and \"01_new_kb_name\" in field_value:\n # Validate the knowledge base name - Make sure it follows these rules:\n if not self.is_valid_collection_name(field_value[\"01_new_kb_name\"]):\n msg = f\"Invalid knowledge base name: {field_value['01_new_kb_name']}\"\n raise ValueError(msg)\n\n # We need to test the API Key one time against the embedding model\n embed_model = self._build_embeddings(\n embedding_model=field_value[\"02_embedding_model\"], api_key=field_value[\"03_api_key\"]\n )\n\n # Try to generate a dummy embedding to validate the API key\n embed_model.embed_query(\"test\")\n\n # Create the new knowledge base directory\n kb_path = KNOWLEDGE_BASES_ROOT_PATH / field_value[\"01_new_kb_name\"]\n kb_path.mkdir(parents=True, exist_ok=True)\n\n # Save the embedding metadata\n build_config[\"knowledge_base\"][\"value\"] = field_value[\"01_new_kb_name\"]\n self._save_embedding_metadata(\n kb_path=kb_path,\n embedding_model=field_value[\"02_embedding_model\"],\n api_key=field_value[\"03_api_key\"],\n )\n\n # Update the knowledge base options dynamically\n build_config[\"knowledge_base\"][\"options\"] = self._get_knowledge_bases()\n if build_config[\"knowledge_base\"][\"value\"] not in build_config[\"knowledge_base\"][\"options\"]:\n build_config[\"knowledge_base\"][\"value\"] = None\n\n return build_config\n" }, "column_config": { "_input_type": "TableInput",