langchainrb 0.6.13 → 0.6.14

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 9a8dc8c16a235328e6122725804fc8dface37910d2014ecf44410631d3ec63cb
-  data.tar.gz: 5d69e6b1dda419d2834f9041a18eae48b2c63303cc901515cd883153635f0742
+  metadata.gz: 7beb48b4b2bc88c4a25bef1cbc9eb0e95c0bd3eaeb02af6a12e0026c9081dd6d
+  data.tar.gz: ebdb7816bf0e37e221a13ecf6cb620a335d3d49b564be76f4f714cb9849ebeb6
 SHA512:
-  metadata.gz: 0c176e717986c0bb0b74761858bf0a6aac8e84fff179791fd67b68441760d5a419ae767bb678bab8a783bfaf6a17f53b7177b24d96fd8f1b076fd4f091c5443e
-  data.tar.gz: a0457aaf411f8932ada4bb3683fef81396334186047a5bb26040c0c03634443662806ad3d2d37ec81f4121f519e387f77688e395cc38882b7f8ab85c0795fba6
+  metadata.gz: 1cb2c147ffbe999eb1f027161e6cda3beea76e31b821bdec564eb36cc6a2d96e31c5d450be8d744738fabef07a9f519c8b96ab2e6dc9585fb05ceea7ebc494a2
+  data.tar.gz: 9f3a1d015de4f568bea1e08637a07ed6bf2ef93bb68068ebe51a50c16ca5a1d5d3f850cf19439ad785b6078305a7dfbd740f7bf7916c1e3466efdb04060f360e

data/CHANGELOG.md

@@ -1,5 +1,12 @@
 ## [Unreleased]
 
+## [0.6.14] - 2023-09-11
+- Add `find` method to `Langchain::Vectorsearch::Qdrant`
+- Enhance Google search output
+- Raise ApiError when OpenAI returns an error
+- Update OpenAI `complete` method to use chat completion api
+  - Deprecate legacy completion models. See https://platform.openai.com/docs/deprecations/2023-07-06-gpt-and-embeddings
+
 ## [0.6.13] - 2023-08-23
 - Add `k:` parameter to all `ask()` vector search methods
 - Bump Faraday to 2.x 

data/lib/langchain/agent/react_agent.rb

@@ -83,9 +83,9 @@ module Langchain::Agent
           else
             "\nObservation: #{result}\nThought:"
           end
-        else
+        elsif response.include?("Final Answer:")
           # Return the final answer
-          final_response = response.match(/Final Answer: (.*)/)&.send(:[], -1)
+          final_response = response.split("Final Answer:")[-1]
           break
         end
       end

data/lib/langchain/dependency_helper.rb

@@ -2,6 +2,8 @@
 
 module Langchain
   module DependencyHelper
+    class LoadError < ::LoadError; end
+
     class VersionError < ScriptError; end
 
     # This method requires and loads the given gem, and then checks to see if the version of the gem meets the requirements listed in `langchain.gemspec`
@@ -12,7 +14,7 @@ module Langchain
     # @raise [LoadError] If the gem is not installed
     # @raise [VersionError] If the gem is installed, but the version does not meet the requirements
     #
-    def depends_on(gem_name)
+    def depends_on(gem_name, req: true)
       gem(gem_name) # require the gem
 
       return(true) unless defined?(Bundler) # If we're in a non-bundler environment, we're no longer able to determine if we'll meet requirements
@@ -26,8 +28,13 @@ module Langchain
         raise VersionError, "The #{gem_name} gem is installed, but version #{gem_requirement} is required. You have #{gem_version}."
       end
 
+      lib_name = gem_name if req == true
+      lib_name = req if req.is_a?(String)
+
+      require(lib_name) if lib_name
+
       true
-    rescue LoadError
+    rescue ::LoadError
       raise LoadError, "Could not load #{gem_name}. Please ensure that the #{gem_name} gem is installed."
     end
   end

data/lib/langchain/llm/ai21.rb

@@ -20,7 +20,6 @@ module Langchain::LLM
 
     def initialize(api_key:, default_options: {})
       depends_on "ai21"
-      require "ai21"
 
       @client = ::AI21::Client.new(api_key)
       @defaults = DEFAULTS.merge(default_options)

data/lib/langchain/llm/anthropic.rb

@@ -22,7 +22,6 @@ module Langchain::LLM
 
     def initialize(api_key:, llm_options: {}, default_options: {})
       depends_on "anthropic"
-      require "anthropic"
 
       @client = ::Anthropic::Client.new(access_token: api_key, **llm_options)
       @defaults = DEFAULTS.merge(default_options)

data/lib/langchain/llm/cohere.rb

@@ -20,8 +20,7 @@ module Langchain::LLM
     }.freeze
 
     def initialize(api_key:, default_options: {})
-      depends_on "cohere-ruby"
-      require "cohere"
+      depends_on "cohere-ruby", req: "cohere"
 
       @client = ::Cohere::Client.new(api_key: api_key)
       @defaults = DEFAULTS.merge(default_options)

data/lib/langchain/llm/google_palm.rb

@@ -25,7 +25,6 @@ module Langchain::LLM
 
     def initialize(api_key:, default_options: {})
       depends_on "google_palm_api"
-      require "google_palm_api"
 
       @client = ::GooglePalmApi::Client.new(api_key: api_key)
       @defaults = DEFAULTS.merge(default_options)

data/lib/langchain/llm/hugging_face.rb

@@ -25,8 +25,7 @@ module Langchain::LLM
     # @param api_key [String] The API key to use
     #
     def initialize(api_key:)
-      depends_on "hugging-face"
-      require "hugging_face"
+      depends_on "hugging-face", req: "hugging_face"
 
       @client = ::HuggingFace::InferenceApi.new(api_token: api_key)
     end

data/lib/langchain/llm/llama_cpp.rb

@@ -24,7 +24,6 @@ module Langchain::LLM
     # @param seed [Integer] The seed to use
     def initialize(model_path:, n_gpu_layers: 1, n_ctx: 2048, n_threads: 1, seed: -1)
       depends_on "llama_cpp"
-      require "llama_cpp"
 
       @model_path = model_path
       @n_gpu_layers = n_gpu_layers

data/lib/langchain/llm/openai.rb

@@ -12,11 +12,19 @@ module Langchain::LLM
   class OpenAI < Base
     DEFAULTS = {
       temperature: 0.0,
-      completion_model_name: "text-davinci-003",
+      completion_model_name: "gpt-3.5-turbo",
       chat_completion_model_name: "gpt-3.5-turbo",
       embeddings_model_name: "text-embedding-ada-002",
       dimension: 1536
     }.freeze
+
+    LEGACY_COMPLETION_MODELS = %w[
+      ada
+      babbage
+      curie
+      davinci
+    ].freeze
+
     LENGTH_VALIDATOR = Langchain::Utils::TokenLength::OpenAIValidator
     ROLE_MAPPING = {
       "ai" => "assistant",
@@ -26,8 +34,7 @@ module Langchain::LLM
     attr_accessor :functions
 
     def initialize(api_key:, llm_options: {}, default_options: {})
-      depends_on "ruby-openai"
-      require "openai"
+      depends_on "ruby-openai", req: "openai"
 
       @client = ::OpenAI::Client.new(access_token: api_key, **llm_options)
       @defaults = DEFAULTS.merge(default_options)
@@ -45,7 +52,10 @@ module Langchain::LLM
 
       validate_max_tokens(text, parameters[:model])
 
-      response = client.embeddings(parameters: parameters.merge(params))
+      response = with_api_error_handling do
+        client.embeddings(parameters: parameters.merge(params))
+      end
+
       response.dig("data").first.dig("embedding")
     end
 
@@ -59,11 +69,16 @@ module Langchain::LLM
     def complete(prompt:, **params)
       parameters = compose_parameters @defaults[:completion_model_name], params
 
-      parameters[:prompt] = prompt
-      parameters[:max_tokens] = validate_max_tokens(prompt, parameters[:model])
+      return legacy_complete(prompt, parameters) if is_legacy_model?(parameters[:model])
 
-      response = client.completions(parameters: parameters)
-      response.dig("choices", 0, "text")
+      parameters[:messages] = compose_chat_messages(prompt: prompt)
+      parameters[:max_tokens] = validate_max_tokens(parameters[:messages], parameters[:model])
+
+      response = with_api_error_handling do
+        client.chat(parameters: parameters)
+      end
+
+      response.dig("choices", 0, "message", "content")
     end
 
     #
@@ -131,8 +146,10 @@ module Langchain::LLM
         end
       end
 
-      response = client.chat(parameters: parameters)
-      raise Langchain::LLM::ApiError.new "Chat completion failed: #{response.dig("error", "message")}" if !response.empty? && response.dig("error")
+      response = with_api_error_handling do
+        client.chat(parameters: parameters)
+      end
+
       unless streaming
         message = response.dig("choices", 0, "message")
         content = message["content"]
@@ -158,6 +175,22 @@ module Langchain::LLM
 
     private
 
+    def is_legacy_model?(model)
+      LEGACY_COMPLETION_MODELS.any? { |legacy_model| model.include?(legacy_model) }
+    end
+
+    def legacy_complete(prompt, parameters)
+      Langchain.logger.warn "DEPRECATION WARNING: The model #{parameters[:model]} is deprecated. Please use gpt-3.5-turbo instead. Details: https://platform.openai.com/docs/deprecations/2023-07-06-gpt-and-embeddings"
+
+      parameters[:prompt] = prompt
+      parameters[:max_tokens] = validate_max_tokens(prompt, parameters[:model])
+
+      response = with_api_error_handling do
+        client.completions(parameters: parameters)
+      end
+      response.dig("choices", 0, "text")
+    end
+
     def compose_parameters(model, params)
       default_params = {model: model, temperature: @defaults[:temperature]}
 
@@ -166,7 +199,7 @@ module Langchain::LLM
       default_params.merge(params)
     end
 
-    def compose_chat_messages(prompt:, messages:, context:, examples:)
+    def compose_chat_messages(prompt:, messages: [], context: "", examples: [])
       history = []
 
       history.concat transform_messages(examples) unless examples.empty?
@@ -198,6 +231,13 @@ module Langchain::LLM
       end
     end
 
+    def with_api_error_handling
+      response = yield
+      raise Langchain::LLM::ApiError.new "OpenAI API error: #{response.dig("error", "message")}" if response&.dig("error")
+
+      response
+    end
+
     def validate_max_tokens(messages, model)
       LENGTH_VALIDATOR.validate_max_tokens!(messages, model)
     end

data/lib/langchain/llm/replicate.rb

@@ -33,8 +33,7 @@ module Langchain::LLM
     # @param api_key [String] The API key to use
     #
     def initialize(api_key:, default_options: {})
-      depends_on "replicate-ruby"
-      require "replicate"
+      depends_on "replicate-ruby", req: "replicate"
 
       ::Replicate.configure do |config|
         config.api_token = api_key

data/lib/langchain/loader.rb

@@ -98,8 +98,8 @@ module Langchain
       Dir.glob(File.join(@path, "**/*")).map do |file|
         # Only load and add to result files with supported extensions
         Langchain::Loader.new(file, @options).load(&block)
-      rescue => e
-        UnknownFormatError.new(e)
+      rescue
+        UnknownFormatError nil
       end.flatten.compact
     end
 
@@ -134,7 +134,6 @@ module Langchain
     end
 
     def source_type
-      binding.pry
       url? ? @raw_data.content_type : File.extname(@path)
     end
 

data/lib/langchain/processors/docx.rb

@@ -8,7 +8,6 @@ module Langchain
 
       def initialize(*)
         depends_on "docx"
-        require "docx"
       end
 
       # Parse the document and return the text

data/lib/langchain/processors/html.rb

@@ -11,7 +11,6 @@ module Langchain
 
       def initialize(*)
         depends_on "nokogiri"
-        require "nokogiri"
       end
 
       # Parse the document and return the text

data/lib/langchain/processors/pdf.rb

@@ -8,7 +8,6 @@ module Langchain
 
       def initialize(*)
         depends_on "pdf-reader"
-        require "pdf-reader"
       end
 
       # Parse the document and return the text

data/lib/langchain/processors/xlsx.rb

@@ -8,7 +8,6 @@ module Langchain
 
       def initialize(*)
         depends_on "roo"
-        require "roo"
       end
 
       # Parse the document and return the text

data/lib/langchain/tool/calculator.rb

@@ -25,7 +25,6 @@ module Langchain::Tool
 
     def initialize
       depends_on "eqn"
-      require "eqn"
     end
 
     # Evaluates a pure math expression or if equation contains non-math characters (e.g.: "12F in Celsius") then

data/lib/langchain/tool/database.rb

@@ -27,7 +27,6 @@ module Langchain::Tool
     #
     def initialize(connection_string:, tables: [], exclude_tables: [])
       depends_on "sequel"
-      require "sequel"
 
       raise StandardError, "connection_string parameter cannot be blank" if connection_string.empty?
 

data/lib/langchain/tool/google_search.rb

@@ -3,7 +3,7 @@
 module Langchain::Tool
   class GoogleSearch < Base
     #
-    # Wrapper around Google Serp SPI
+    # Wrapper around SerpApi's Google Search API
     #
     # Gem requirements: gem "google_search_results", "~> 2.0.0"
     #
@@ -15,7 +15,7 @@ module Langchain::Tool
     NAME = "google_search"
 
     description <<~DESC
-      A wrapper around Google Search.
+      A wrapper around SerpApi's Google Search API.
 
       Useful for when you need to answer questions about current events.
       Always one of the first options when you need to find information on internet.
@@ -33,7 +33,7 @@ module Langchain::Tool
     #
     def initialize(api_key:)
       depends_on "google_search_results"
-      require "google_search_results"
+
       @api_key = api_key
     end
 
@@ -56,13 +56,78 @@ module Langchain::Tool
     def execute(input:)
       Langchain.logger.info("Executing \"#{input}\"", for: self.class)
 
-      hash_results = execute_search(input: input)
+      results = execute_search(input: input)
+
+      answer_box = results[:answer_box_list] ? results[:answer_box_list].first : results[:answer_box]
+      if answer_box
+        return answer_box[:result] ||
+            answer_box[:answer] ||
+            answer_box[:snippet] ||
+            answer_box[:snippet_highlighted_words] ||
+            answer_box.reject { |_k, v| v.is_a?(Hash) || v.is_a?(Array) || v.start_with?("http") }
+      elsif (events_results = results[:events_results])
+        return events_results.take(10)
+      elsif (sports_results = results[:sports_results])
+        return sports_results
+      elsif (top_stories = results[:top_stories])
+        return top_stories
+      elsif (news_results = results[:news_results])
+        return news_results
+      elsif (jobs_results = results.dig(:jobs_results, :jobs))
+        return jobs_results
+      elsif (shopping_results = results[:shopping_results]) && shopping_results.first.key?(:title)
+        return shopping_results.take(3)
+      elsif (questions_and_answers = results[:questions_and_answers])
+        return questions_and_answers
+      elsif (popular_destinations = results.dig(:popular_destinations, :destinations))
+        return popular_destinations
+      elsif (top_sights = results.dig(:top_sights, :sights))
+        return top_sights
+      elsif (images_results = results[:images_results]) && images_results.first.key?(:thumbnail)
+        return images_results.map { |h| h[:thumbnail] }.take(10)
+      end
+
+      snippets = []
+      if (knowledge_graph = results[:knowledge_graph])
+        snippets << knowledge_graph[:description] if knowledge_graph[:description]
+
+        title = knowledge_graph[:title] || ""
+        knowledge_graph.each do |k, v|
+          if v.is_a?(String) &&
+              k != :title &&
+              k != :description &&
+              !k.to_s.end_with?("_stick") &&
+              !k.to_s.end_with?("_link") &&
+              !k.to_s.start_with?("http")
+            snippets << "#{title} #{k}: #{v}"
+          end
+        end
+      end
+
+      if (first_organic_result = results.dig(:organic_results, 0))
+        if (snippet = first_organic_result[:snippet])
+          snippets << snippet
+        elsif (snippet_highlighted_words = first_organic_result[:snippet_highlighted_words])
+          snippets << snippet_highlighted_words
+        elsif (rich_snippet = first_organic_result[:rich_snippet])
+          snippets << rich_snippet
+        elsif (rich_snippet_table = first_organic_result[:rich_snippet_table])
+          snippets << rich_snippet_table
+        elsif (link = first_organic_result[:link])
+          snippets << link
+        end
+      end
+
+      if (buying_guide = results[:buying_guide])
+        snippets << buying_guide
+      end
+
+      if (local_results = results.dig(:local_results, :places))
+        snippets << local_results
+      end
 
-      # TODO: Glance at all of the fields that langchain Python looks through: https://github.com/hwchase17/langchain/blob/v0.0.166/langchain/utilities/serpapi.py#L128-L156
-      # We may need to do the same thing here.
-      hash_results.dig(:answer_box, :answer) ||
-        hash_results.dig(:answer_box, :snippet) ||
-        hash_results.dig(:organic_results, 0, :snippet)
+      return "No good search result found" if snippets.empty?
+      snippets
     end
 
     #

data/lib/langchain/tool/ruby_code_interpreter.rb

@@ -13,9 +13,9 @@ module Langchain::Tool
     DESC
 
     def initialize(timeout: 30)
-      @timeout = timeout
       depends_on "safe_ruby"
-      require "safe_ruby"
+
+      @timeout = timeout
     end
 
     # @param input [String] ruby code expression

data/lib/langchain/tool/wikipedia.rb

@@ -18,8 +18,7 @@ module Langchain::Tool
     DESC
 
     def initialize
-      depends_on "wikipedia-client"
-      require "wikipedia"
+      depends_on "wikipedia-client", req: "wikipedia"
     end
 
     # Executes Wikipedia API search and returns the answer

data/lib/langchain/vectorsearch/chroma.rb

@@ -18,7 +18,6 @@ module Langchain::Vectorsearch
     # @param llm [Object] The LLM client to use
     def initialize(url:, index_name:, llm:, api_key: nil)
       depends_on "chroma-db"
-      require "chroma-db"
 
       ::Chroma.connect_host = url
       ::Chroma.logger = Langchain.logger

data/lib/langchain/vectorsearch/hnswlib.rb

@@ -23,7 +23,6 @@ module Langchain::Vectorsearch
     #
     def initialize(llm:, path_to_index:)
       depends_on "hnswlib"
-      require "hnswlib"
 
       super(llm: llm)
 

data/lib/langchain/vectorsearch/milvus.rb

@@ -13,7 +13,6 @@ module Langchain::Vectorsearch
 
     def initialize(url:, index_name:, llm:, api_key: nil)
       depends_on "milvus"
-      require "milvus"
 
       @client = ::Milvus::Client.new(url: url)
       @index_name = index_name

data/lib/langchain/vectorsearch/pgvector.rb

@@ -26,9 +26,7 @@ module Langchain::Vectorsearch
     # @param namespace [String] The namespace to use for the index when inserting/querying
     def initialize(url:, index_name:, llm:, namespace: nil)
       depends_on "sequel"
-      require "sequel"
       depends_on "pgvector"
-      require "pgvector"
 
       @db = Sequel.connect(url)
 

data/lib/langchain/vectorsearch/pinecone.rb

@@ -18,7 +18,6 @@ module Langchain::Vectorsearch
     # @param llm [Object] The LLM client to use
     def initialize(environment:, api_key:, index_name:, llm:)
       depends_on "pinecone"
-      require "pinecone"
 
       ::Pinecone.configure do |config|
         config.api_key = api_key
@@ -31,6 +30,19 @@ module Langchain::Vectorsearch
       super(llm: llm)
     end
 
+    # Find records by ids
+    # @param ids [Array] The ids to find
+    # @param namespace String The namespace to search through
+    # @return [Hash] The response from the server
+    def find(ids: [], namespace: "")
+      raise ArgumentError, "Ids must be provided" if Array(ids).empty?
+
+      client.index(index_name).fetch(
+        ids: ids,
+        namespace: namespace
+      )
+    end
+
     # Add a list of texts to the index
     # @param texts [Array] The list of texts to add
     # @param ids [Array] The list of IDs to add

data/lib/langchain/vectorsearch/qdrant.rb

@@ -17,8 +17,7 @@ module Langchain::Vectorsearch
     # @param index_name [String] The name of the index to use
     # @param llm [Object] The LLM client to use
     def initialize(url:, api_key:, index_name:, llm:)
-      depends_on "qdrant-ruby"
-      require "qdrant"
+      depends_on "qdrant-ruby", req: "qdrant"
 
       @client = ::Qdrant::Client.new(
         url: url,
@@ -29,6 +28,18 @@ module Langchain::Vectorsearch
       super(llm: llm)
     end
 
+    # Find records by ids
+    # @param ids [Array] The ids to find
+    # @return [Hash] The response from the server
+    def find(ids: [])
+      client.points.get_all(
+        collection_name: index_name,
+        ids: ids,
+        with_payload: true,
+        with_vector: true
+      )
+    end
+
     # Add a list of texts to the index
     # @param texts [Array] The list of texts to add
     # @return [Hash] The response from the server

data/lib/langchain/vectorsearch/weaviate.rb

@@ -17,8 +17,7 @@ module Langchain::Vectorsearch
     # @param index_name [String] The capitalized name of the index to use
     # @param llm [Object] The LLM client to use
     def initialize(url:, api_key:, index_name:, llm:)
-      depends_on "weaviate-ruby"
-      require "weaviate"
+      depends_on "weaviate-ruby", req: "weaviate"
 
       @client = ::Weaviate::Client.new(
         url: url,

data/lib/langchain/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Langchain
-  VERSION = "0.6.13"
+  VERSION = "0.6.14"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: langchainrb
 version: !ruby/object:Gem::Version
-  version: 0.6.13
+  version: 0.6.14
 platform: ruby
 authors:
 - Andrei Bondarev
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2023-08-23 00:00:00.000000000 Z
+date: 2023-09-11 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: baran
@@ -16,14 +16,14 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 0.1.6
+        version: 0.1.8
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 0.1.6
+        version: 0.1.8
 - !ruby/object:Gem::Dependency
   name: colorize
   requirement: !ruby/object:Gem::Requirement
@@ -408,14 +408,14 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 0.9.3
+        version: 0.9.4
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 0.9.3
+        version: 0.9.4
 - !ruby/object:Gem::Dependency
   name: roo
   requirement: !ruby/object:Gem::Requirement
@@ -606,7 +606,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.3.7
+rubygems_version: 3.2.33
 signing_key:
 specification_version: 4
 summary: Build LLM-backed Ruby applications with Ruby's LangChain