|
| 1 | +# frozen_string_literal: true |
| 2 | + |
| 3 | +require "langchain" |
| 4 | +require "dotenv/load" |
| 5 | +require "ruby/openai" |
| 6 | + |
| 7 | +# Initialize the Pgvector client |
| 8 | +pgvector = Langchain::Vectorsearch::Pgvector.new( |
| 9 | + url: ENV["POSTGRES_URL"], |
| 10 | + index_name: "documents", |
| 11 | + llm: Langchain::LLM::OpenAI.new(api_key: ENV["OPENAI_API_KEY"]) |
| 12 | +) |
| 13 | + |
| 14 | +# Create the default schema if it doesn't exist |
| 15 | +pgvector.create_default_schema |
| 16 | + |
| 17 | +# Add documents with metadata |
| 18 | +documents = [ |
| 19 | + { |
| 20 | + text: "The quick brown fox jumps over the lazy dog", |
| 21 | + metadata: { |
| 22 | + source: "fables", |
| 23 | + author: "Aesop", |
| 24 | + category: "animals" |
| 25 | + } |
| 26 | + }, |
| 27 | + { |
| 28 | + text: "To be or not to be, that is the question", |
| 29 | + metadata: { |
| 30 | + source: "hamlet", |
| 31 | + author: "Shakespeare", |
| 32 | + category: "drama" |
| 33 | + } |
| 34 | + }, |
| 35 | + { |
| 36 | + text: "It was the best of times, it was the worst of times", |
| 37 | + metadata: { |
| 38 | + source: "tale_of_two_cities", |
| 39 | + author: "Dickens", |
| 40 | + category: "novel" |
| 41 | + } |
| 42 | + } |
| 43 | +] |
| 44 | + |
| 45 | +# Add texts with metadata |
| 46 | +ids = pgvector.add_texts( |
| 47 | + texts: documents.map { |doc| doc[:text] }, |
| 48 | + metadata: documents.map { |doc| doc[:metadata] } |
| 49 | +) |
| 50 | + |
| 51 | +puts "Added documents with IDs: #{ids.inspect}" |
| 52 | + |
| 53 | +# Perform a similarity search |
| 54 | +puts "\nSearching for documents similar to 'fox':" |
| 55 | +results = pgvector.similarity_search(query: "fox", k: 2) |
| 56 | +results.each do |result| |
| 57 | + puts "Content: #{result.content}" |
| 58 | + puts "Metadata: #{JSON.parse(result.metadata)}" |
| 59 | + puts "---" |
| 60 | +end |
| 61 | + |
| 62 | +# Search with specific metadata filter |
| 63 | +puts "\nSearching for Shakespeare's works:" |
| 64 | +results = pgvector.similarity_search(query: "drama", k: 1) |
| 65 | +results.each do |result| |
| 66 | + metadata = JSON.parse(result.metadata) |
| 67 | + if metadata["author"] == "Shakespeare" |
| 68 | + puts "Found Shakespeare's work:" |
| 69 | + puts "Content: #{result.content}" |
| 70 | + puts "Metadata: #{metadata}" |
| 71 | + end |
| 72 | +end |
| 73 | + |
| 74 | +# Clean up (optional) |
| 75 | +# pgvector.destroy_default_schema |
0 commit comments