RISHIT JARIWALA

1# vsftpd configuration file for SevenSins Lab
2# WARNING: INTENTIONALLY VULNERABLE CONFIGURATION
3listen=YES
4anonymous_enable=YES
5anon_root=/var/ftp/pub
6local_enable=YES
7write_enable=YES
8anon_upload_enable=YES
9anon_mkdir_write_enable=YES
10dirmessage_enable=YES
11use_localtime=YES
12xferlog_enable=YES
13connect_from_port_20=YES
14chroot_local_user=NO
15secure_chroot_dir=/var/run/vsftpd/empty
16pam_service_name=vsftpd
17rsa_cert_file=/etc/ssl/certs/ssl-cert-snakeoil.pem
18rsa_private_key_file=/etc/ssl/private/ssl-cert-snakeoil.key
19ssl_enable=NO

1import os
2from langchain_community.document_loaders import PyPDFLoader
3from langchain_text_splitter import RecursiveCharacterTextSplitter
4from langchain_community.vectorstores import FAISS
5from langchain_huggingface import HuggingFaceEmbeddings
6from langchain_community.llms import Ollama
7from langchain.chains import RetrievalQA
8
9def initialize_rag(pdf_path):
10    # Load unstructured multilingual police reports
11    loader = PyPDFLoader(pdf_path)
12    documents = loader.load()
13    
14    # Split text with overlap to preserve semantic context
15    splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
16    texts = splitter.split_documents(documents)
17    
18    # Generate local embeddings and load vector store
19    embeddings = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")
20    db = FAISS.from_documents(texts, embeddings)
21    
22    # Setup Ollama with local Llama3 model
23    llm = Ollama(model="llama3")
24    qa = RetrievalQA.from_chain_type(llm=llm, chain_type="stuff", retriever=db.as_retriever())
25    return qa