CoolFace
Apppublic

Sameer360/Document_based_chatbot

sourceHugging Faceupdated 2y agoView on Hugging Face
0likes
document_processor.cpython-39.pyc17 linesDownload Raw Back to __pycache__
1a

2X��g��@sbddlmZmZddlmZddlmZddlmZddl	Z	ddl3mZmZedd�Z
d	d4�ZdS)�)�PyPDFLoader�5TextLoader)�RecursiveCharacterTextSplitter)�Chroma)�HuggingFaceEmbeddingsN)�PERSIST_DIRECTORY�
DOCUMENTS_DIRz&sentence-transformers/all-MiniLM-L6-v2)�6model_namecCs.tj�t�r:tttd�}dd�|jdgd��dg�D�}n7d}t�}g}t||�D]�\}}||vrrt	d|���qRt	d|���tj�8|�d}|d	kr�t|���}n*|d9kr�t
|dd���}nt	d
|���qRtddd��|�}	|	D]}10||11jd<q�|�|	�qR|�r*|�r|�|�ntj|ttd�}|S)N)�persist_directory�embedding_functioncSsh|]}d|vr|d�qS)�source�)�.0�metar
r
�)D:\LangChain\backup\document_processor.py�	<setcomp>s�z$process_documents.<locals>.<setcomp>�	metadatas)�includez!Skipping already processed file: zProcessing �z.pdfz.txtzutf-8)�encodingz Skipping unsupported file type: i���)�12chunk_size�
chunk_overlapr)r13)�os�path�existsrr�14embeddings�get�set�zip�print�splitextr�loadrr�split_documents�metadata�extend�
add_documents�from_documents)�files�15file_names�db�existing_source�16all_chunks�	file_path�	file_name�ext�doc�chunks�chunkr
r
r�process_documents
sJ�����r3)�$langchain_community.document_loadersrr�langchain.text_splitterr�langchain_chromar�langchain_huggingfacerr�configrrrr3r
r
r
r�<module>s17