Class: SimpleRag::Engine

Inherits:
Object
  • Object
show all
Defined in:
lib/simple_rag.rb

Constant Summary collapse

DEFAULT_URL =
"https://raw.githubusercontent.com/run-llama/llama_index/main/docs/docs/examples/data/paul_graham/paul_graham_essay.txt"

Instance Method Summary collapse

Instance Method Details

#get_urlObject



43
44
45
46
47
48
49
50
# File 'lib/simple_rag.rb', line 43

def get_url
  url = prompt_user_for_url
  until valid_url?(url)
    puts "The URL provided is invalid. Please try again."
    url = prompt_user_for_url
  end
  url
end

#prompt_user_for_urlObject



30
31
32
33
34
# File 'lib/simple_rag.rb', line 30

def prompt_user_for_url
  print "Specify a URL to an HTML document you would like to ask questions of (Default: What I Worked On by Paul Graham): "
  input_url = gets.chomp
  input_url.empty? ? DEFAULT_URL : input_url
end

#runObject



52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
# File 'lib/simple_rag.rb', line 52

def run
  url = get_url
  puts "Document Downloaded"

  # Setup LLM of choice
  api_key = ENV["MISTRAL_AI_KEY"] || STDIN.getpass("Type your API Key: ")
  raise "Missing API Key" unless api_key

  client = Mistral.new(
    credentials: {api_key: api_key},
    options: {server_sent_events: true}
  )

  # Indexing
  puts "Initialize indexing"
  index_instance = SimpleRag::Index.new(client)
  puts "Loading url"
  text = index_instance.load(url)
  puts "Chunk text"
  chunks = index_instance.chunk(text)
  puts "Embed chunks"
  text_embeddings = index_instance.embed_chunks(chunks)
  index = index_instance.save(text_embeddings)

  retrieve_instance = SimpleRag::Retrieve.new(client)
  retrieve_instance.save_index(index)
  retrieve_instance.save_chunks(chunks)

  loop do
    print "Enter your query (or type 'exit' to quit): "
    query = gets.chomp
    break if query.downcase == "exit"
    puts

    # Retrieval/Search
    question_embedding = retrieve_instance.embed_query(query)
    retrieved_chunks = retrieve_instance.similarity_search(question_embedding, 2)

    # Generation
    prompt = SimpleRag::Generate.new.prompt(query, retrieved_chunks)

    puts run_mistral(client, prompt)
    puts
  end
end

#run_mistral(client, user_message, model: "mistral-medium-latest") ⇒ Object



24
25
26
27
28
# File 'lib/simple_rag.rb', line 24

def run_mistral(client, user_message, model: "mistral-medium-latest")
  messages = [{role: "user", content: user_message}]
  chat_response = client.chat_completions({model: model, messages: messages})
  chat_response.dig("choices", 0, "message", "content")
end

#valid_url?(url) ⇒ Boolean

Returns:

  • (Boolean)


36
37
38
39
40
41
# File 'lib/simple_rag.rb', line 36

def valid_url?(url)
  uri = URI.parse(url)
  uri.is_a?(URI::HTTP) || uri.is_a?(URI::HTTPS)
rescue URI::InvalidURIError
  false
end