norasalem88/rag_assistant_devops
0
1�
2 (�i � �� � S SK r S SKJr S SKJr S SKJr \" \5 r\ R R \ R R \ R R \ 5 5 5 r
\ R R \
5 S:w a+ \ R R \ R " 5 S5 r
\ R R \
S5 rS rSS\S \S3\S \4S jjr\S
:X a� \" 5 r\( az Sr\" \\/ SS9u rr\ " S\ " \5 35 \ HT r \ " S\ R> RA S5 S\ R> RA S5 S\ RB SS S35 MV ggg)� N)�Chroma)�4get_logger)�get_embeddingszdevops-rag-assistant�vectorstore_v3c �* � [ R R [ 5 ( a [ R " [ 5 ( d [5 R
S[ S35 g[6 R S5 [ 5 n [ [ U S9nU$ )zQReturns the Chroma vectorstore directly, so we can do custom similarity searches.zVector store directory z1 is empty or missing. Please run ingestion first.Nz Loading Chroma vector store...)�persist_directory�embedding_function)7�os�path�exists�VECTORSTORE_DIR�listdir�logger�warning�infor r )�8embeddings� vectorstores �FE:\Devops chatbot\Devops-chatbot\devops-rag-assistant\src\retrieval.py�
get_retrieverr sk � �
�7�7�>�>�/� *� *�"�*�*�_�2M�2M����0� �0A�Ar�s�t��9�K�K�0�1� �!�J��)�%��K� �� �query�active_doc_titles�k�distance_thresholdc
�r � [ R SU SU 35 SnU( a SSU00nU R UUUS9n[ R S[ U5 S35 / n/ n[ U5 H� u n u p�UR U105 [ R S U S11- S U S S
U12R R S5 SU13R R S5 35 X�:: a UR U145 M� [ R SU S SU 35 M� [ R S[ U5 35 Xx4$ )z�15Retrieves the top_k chunks from the vector store, explicitly filtering by active_doc_titles.16Returns both the threshold-filtered docs and the raw docs (for fallback/short-circuit logic).17zQuerying for: 'z' within active docs: N� doc_titlez$in)r �filterzActive docs have z" chunks returned before threshold.zRetrieved chunk � z18 | Score: z.3fz19 | Title: z | Page: �pagezChunk REJECTED: Score z is above threshold z Top chunks AFTER threshold: ) r r � similarity_search_with_score�len� enumerate�append�metadata�getr ) r r r r r � filter_dict�results�20valid_docs�raw_docs�i�doc�scores r �retrieve_relevant_contextr- sc � �21 �K�K�/�%��(>�?P�>Q�R�S��K��"�U�,=�$>�?� ��6�6�
�22 �� 7� �G� �K�K�#�C��L�>�1S�T�U��J��H�$�W�-���<�C����� �� � �&�q��s�e�:�e�C�[�23�3�<�<�K[�K[�\g�Kh�Ji�ir�sv�s�s� tD� tD� EK� tL� sM� N� O� � &� � � �c� "� �N�N� 3�E�#�;�>R�Se�Rf�g� h� .� �K�K�.�s�:��.?�@�A� � � r �__main__zWhat is a Kubernetes pod?� )r r z Total valid docs retrieved: z- r z (Page r z): �2 z...)� g�������?)"r24 � langchain_community.vectorstoresr � src.utilsr �
src.ingestionr �__name__r r �dirname�abspath�__file__�BASE_DIR�basename�join�getcwdr
r �str�list�int�floatr- r r �valid�raw�printr! r+ r$ r% � page_content� r r �<module>rF s_ �� � 3� � (� �H� ��
�7�7�?�?�2�7�7�?�?�2�7�7�?�?�8�+D�E� F���7�7���H� �!7�7��w�w�|�|�B�I�I�K�)?�@�H��'�'�,�,�x�)9�:�� � �#� �$� �SV� �pu� �D �z�� �/�K��+��.�{�E�UW�[\�]�25��s�
�,�S��Z�L�9�:��C� �B�s�|�|�'�'� �4�5�W�S�\�\�=M�=M�f�=U�<V�VY�Z]�Zj�Zj�kn�ln�Zo�Yp�ps�t� u� � � r