Quickstart

Here are 5 easy steps to build your RAG system with PurrfectKit.

from purrfectmeow.meow.felis import DocTemplate, MetaFile
from purrfectmeow import Suphalak, Malet, WichienMaat, KhaoManee, Kornja

# Step 1: Read content from a PDF
file_path = 'test/test.pdf'
with open(file_path, 'rb') as file:
    content = Suphalak.reading(file.read(), 'test.pdf', loader='PYMUPDF')

# Step 2: Split content into chunks
chunks = Malet.chunking(content, chunk_method='token', chunk_size=500, chunk_overlap=25)

# Step 3: Create document template with metadata
metadata = MetaFile.get_metadata(file_path)
docs = DocTemplate.create_template(chunks, metadata)

# Step 4: Embed chunks and query
chunk_embeddings = WichienMaat.embedding(chunks)
query = "What is the main topic of the document?"
query_embedding = WichienMaat.embedding(query)

# Step 5: Search for relevant chunks
results = KhaoManee.searching(query_embedding, chunk_embeddings, docs, top_k=2)

# (Optional) Step 6: Generate answer (if implemented)
# answer = Kornja.generating(results, query)
# print(answer)

print("Top search results:", results)

This example reads a PDF, chunks it, embeds the chunks, searches for relevant content, and (if implemented) generates an answer.