langchainrb 0.6.5 → 0.6.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 3404535e036c3efe68fd12706d2ebb269caed87b562fc38434122b1be01a356d
-  data.tar.gz: e3be77b32cf754235e8895fb1af60edca54cb5acb84278bfa2e39b6ed7c2abbe
+  metadata.gz: 5e8dc16375bc94045a728b37d6d388a8670699c35af4213bc4d16163fbb0047c
+  data.tar.gz: c9695ac8ce5d6bf351afd3148d281a26733f37a43674fae71c3c81c7530cdc2e
 SHA512:
-  metadata.gz: b3fae04c73176c758c2d2d32c3ac538f3e094eb10f378b9a8befbbdcc62b60e55941a1bfefcb61eac7daca43ef91d0e57306dbc26bd59afbdad6ab4efff2ba89
-  data.tar.gz: 626bb4a226112ee6fe709077a6d49ba91c0483fee657848153e9cff61693183709aede5844237c24cc02c561f59be82ea1fd296fe1c3f4ee4d971494ee4dcd75
+  metadata.gz: 67a532ea9c46961a8636f1332590876f8d5d26e67384789c9a4ab7907c4a2150240a7c9d63ae188d19a269e345ff0dfafcf24ab1cb4a987e4eab5cc2439bbf06
+  data.tar.gz: 333888fdd41e2ce21b369fc0844ca3b181f2db603833bf8cbf8a46e59b8049dcb31a216fb4adcb12fe5a69dc33ec21725131e6e8ce8e7e7633814f5dc7c51c7b

data/CHANGELOG.md

@@ -1,5 +1,18 @@
 ## [Unreleased]
 
+## [0.6.7] - 2023-07-19
+- Support for OpenAI functions
+- Streaming vectorsearch ask() responses
+
+## [0.6.6] - 2023-07-13
+- Langchain::Chunker::RecursiveText
+- Fixes
+
+## [0.6.5] - 2023-07-06
+- 🗣️ LLMs
+  - Introducing Llama.cpp support
+- Langchain::OutputParsers::OutputFixingParser to wrap a Langchain::OutputParser and handle invalid response
+
 ## [0.6.4] - 2023-07-01
 - Fix `Langchain::Vectorsearch::Qdrant#add_texts()`
 - Introduce `ConversationMemory`

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    langchainrb (0.6.5)
+    langchainrb (0.6.7)
       baran (~> 0.1.6)
       colorize (~> 0.8.1)
       json-schema (~> 4.0.0)
@@ -133,7 +133,7 @@ GEM
       faraday (>= 1.0)
       faraday_middleware
       graphql-client
-    graphql (2.0.23)
+    graphql (2.0.24)
     graphql-client (0.18.0)
       activesupport (>= 3.0)
       graphql
@@ -264,7 +264,7 @@ GEM
       rubocop (>= 1.7.0, < 2.0)
       rubocop-ast (>= 0.4.0)
     ruby-next-core (0.15.3)
-    ruby-openai (4.0.0)
+    ruby-openai (4.1.0)
       faraday (>= 1)
       faraday-multipart (>= 1)
     ruby-progressbar (1.13.0)
@@ -299,7 +299,7 @@ GEM
     tzinfo (2.0.6)
       concurrent-ruby (~> 1.0)
     unicode-display_width (2.4.2)
-    weaviate-ruby (0.8.3)
+    weaviate-ruby (0.8.4)
       faraday (~> 1)
       faraday_middleware (~> 1)
       graphlient (~> 0.6.0)
@@ -344,7 +344,7 @@ DEPENDENCIES
   roo (~> 2.10.0)
   rspec (~> 3.0)
   rubocop
-  ruby-openai (~> 4.0.0)
+  ruby-openai (~> 4.1.0)
   safe_ruby (~> 1.0.4)
   sequel (~> 5.68.0)
   standardrb

data/README.md

@@ -39,7 +39,7 @@ require "langchain"
 | [Hnswlib](https://github.com/nmslib/hnswlib/) | :white_check_mark: | :white_check_mark: | :white_check_mark: | WIP     | WIP               |
 | [Milvus](https://milvus.io/) | :white_check_mark: | :white_check_mark: | :white_check_mark: | WIP     | WIP               |
 | [Pinecone](https://www.pinecone.io/) | :white_check_mark: | :white_check_mark: | :white_check_mark: | WIP     | :white_check_mark: |
-| [Pgvector](https://github.com/pgvector/pgvector) | :white_check_mark: | :white_check_mark: | :white_check_mark: | WIP     | WIP               |
+| [Pgvector](https://github.com/pgvector/pgvector) | :white_check_mark: | :white_check_mark: | :white_check_mark: | WIP     | :white_check_mark: |
 | [Qdrant](https://qdrant.tech/) | :white_check_mark: | :white_check_mark: | :white_check_mark: | WIP     | :white_check_mark: |
 | [Weaviate](https://weaviate.io/) | :white_check_mark: | :white_check_mark: | :white_check_mark: | WIP     | :white_check_mark: |
 
@@ -54,7 +54,7 @@ Pick the vector search database you'll be using and instantiate the client:
 client = Langchain::Vectorsearch::Weaviate.new(
     url: ENV["WEAVIATE_URL"],
     api_key: ENV["WEAVIATE_API_KEY"],
-    index: "",
+    index_name: "",
     llm: Langchain::LLM::OpenAI.new(api_key: ENV["OPENAI_API_KEY"])
 )
 
@@ -145,6 +145,27 @@ openai.embed(text: "foo bar")
 openai.complete(prompt: "What is the meaning of life?")
 ```
 
+##### Open AI Function calls support
+
+Conversation support
+
+```ruby
+chat = Langchain::Conversation.new(llm: openai)
+```
+```ruby
+chat.set_context("You are the climate bot")
+chat.set_functions(functions)
+```
+
+qdrant:
+
+```ruby
+client.llm.functions = functions
+client.llm.complete_response = true
+```
+
+`complete_response` will return the entire choices data from the gpt response
+
 #### Cohere
 Add `gem "cohere-ruby", "~> 0.9.3"` to your Gemfile.
 
@@ -427,7 +448,7 @@ agent.run(question: "How many users have a name with length greater than 5 in th
 | "database"   | Useful for querying a SQL database |                                                               | `gem "sequel", "~> 5.68.0"`                   |
 | "ruby_code_interpreter" | Interprets Ruby expressions             |                                                               | `gem "safe_ruby", "~> 1.0.4"`             |
 | "google_search"     | A wrapper around Google Search                     | `ENV["SERPAPI_API_KEY"]` (https://serpapi.com/manage-api-key) | `gem "google_search_results", "~> 2.0.0"` |
-| "weather"  | Calls Open Weather API to retrieve the current weather        |      `ENV["OPEN_WEATHER_API_KEY]` (https://home.openweathermap.org/api_keys)               | `gem "open-weather-ruby-client", "~> 0.3.0"`    |
+| "weather"  | Calls Open Weather API to retrieve the current weather        |      `ENV["OPEN_WEATHER_API_KEY"]` (https://home.openweathermap.org/api_keys)               | `gem "open-weather-ruby-client", "~> 0.3.0"`    |
 | "wikipedia"  | Calls Wikipedia API to retrieve the summary        |                                                               | `gem "wikipedia-client", "~> 1.17.0"`     |
 
 #### Loaders 🚚

data/examples/open_ai_function_calls.rb

@@ -0,0 +1,41 @@
+require "langchain"
+require "dotenv/load"
+
+functions = [
+  {
+    name: "get_current_weather",
+    description: "Get the current weather in a given location",
+    parameters: {
+      type: :object,
+      properties: {
+        location: {
+          type: :string,
+          description: "The city and state, e.g. San Francisco, CA"
+        },
+        unit: {
+          type: "string",
+          enum: %w[celsius fahrenheit]
+        }
+      },
+      required: ["location"]
+    }
+  }
+]
+
+openai = Langchain::LLM::OpenAI.new(
+  api_key: ENV["OPENAI_API_KEY"],
+  default_options: {
+    chat_completion_model_name: "gpt-3.5-turbo-16k"
+  }
+)
+
+chat = Langchain::Conversation.new(llm: openai)
+
+chat.set_context("You are the climate bot")
+chat.set_functions(functions)
+
+DONE = %w[done end eof exit].freeze
+
+user_message = "what's the weather in NYC?"
+
+puts chat.message(user_message)

data/examples/open_ai_qdrant_function_calls.rb

@@ -0,0 +1,43 @@
+require "langchain"
+require "dotenv/load"
+
+functions = [
+  {
+    name: "create_rails_controller",
+    description: "gives a command to create a rails controller",
+    parameters: {
+      type: :object,
+      properties: {
+        controller_name: {
+          type: :string,
+          description: "the controller name, e.g. users_controller"
+        },
+        unit: {
+          type: "string",
+          enum: %w[celsius fahrenheit]
+        }
+      },
+      required: ["controller_name"]
+    }
+  }
+]
+
+openai = Langchain::LLM::OpenAI.new(
+  api_key: ENV["OPENAI_API_KEY"],
+  default_options: {
+    chat_completion_model_name: "gpt-3.5-turbo-16k"
+  }
+)
+
+client = Langchain::Vectorsearch::Qdrant.new(
+  url: ENV["QDRANT_URL"],
+  api_key: ENV["QDRANT_API_KEY"],
+  index_name: ENV["QDRANT_INDEX"],
+  llm: openai
+)
+
+client.llm.functions = functions
+client.llm.complete_response = true
+chat = client.ask(question: "create a users_controller")
+
+puts chat

data/lib/langchain/chunker/recursive_text.rb

@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+require "baran"
+
+module Langchain
+  module Chunker
+    #
+    # Recursive text chunker. Preferentially splits on separators.
+    #
+    # Usage:
+    #     Langchain::Chunker::RecursiveText.new(text).chunks
+    #
+    class RecursiveText < Base
+      attr_reader :text, :chunk_size, :chunk_overlap, :separators
+
+      # @param [String] text
+      # @param [Integer] chunk_size
+      # @param [Integer] chunk_overlap
+      # @param [Array<String>] separators
+      def initialize(text, chunk_size: 1000, chunk_overlap: 200, separators: ["\n\n"])
+        @text = text
+        @chunk_size = chunk_size
+        @chunk_overlap = chunk_overlap
+        @separators = separators
+      end
+
+      # @return [Array<String>]
+      def chunks
+        splitter = Baran::RecursiveCharacterTextSplitter.new(
+          chunk_size: chunk_size,
+          chunk_overlap: chunk_overlap,
+          separators: separators
+        )
+        splitter.chunks(text)
+      end
+    end
+  end
+end

data/lib/langchain/conversation.rb

@@ -37,6 +37,11 @@ module Langchain
       @block = block
     end
 
+    def set_functions(functions)
+      @llm.functions = functions
+      @llm.complete_response = true
+    end
+
     # Set the context of the conversation. Usually used to set the model's persona.
     # @param message [String] The context of the conversation
     def set_context(message)

data/lib/langchain/llm/openai.rb

@@ -19,6 +19,8 @@ module Langchain::LLM
     }.freeze
     LENGTH_VALIDATOR = Langchain::Utils::TokenLength::OpenAIValidator
 
+    attr_accessor :functions, :complete_response
+
     def initialize(api_key:, llm_options: {}, default_options: {})
       depends_on "ruby-openai"
       require "openai"
@@ -115,20 +117,25 @@ module Langchain::LLM
 
       parameters = compose_parameters @defaults[:chat_completion_model_name], options
       parameters[:messages] = compose_chat_messages(prompt: prompt, messages: messages, context: context, examples: examples)
-      parameters[:max_tokens] = validate_max_tokens(parameters[:messages], parameters[:model])
+
+      if functions
+        parameters[:functions] = functions
+      else
+        parameters[:max_tokens] = validate_max_tokens(parameters[:messages], parameters[:model])
+      end
 
       if (streaming = block_given?)
         parameters[:stream] = proc do |chunk, _bytesize|
-          yield chunk.dig("choices", 0, "delta", "content")
+          yield chunk if complete_response
+          yield chunk.dig("choices", 0, "delta", "content") if !complete_response
         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")
-
       unless streaming
-        response.dig("choices", 0, "message", "content")
+        return response.dig("choices", 0, "message", "content") if !complete_response
+        return response if complete_response
       end
     end
 

data/lib/langchain/vectorsearch/base.rb

@@ -161,12 +161,16 @@ module Langchain::Vectorsearch
     end
 
     def add_data(paths:)
-      raise ArgumentError, "Paths must be provided" if paths.to_a.empty?
+      raise ArgumentError, "Paths must be provided" if Array(paths).empty?
 
       texts = Array(paths)
         .flatten
-        .map { |path| Langchain::Loader.new(path)&.load&.value }
-        .compact
+        .map do |path|
+          data = Langchain::Loader.new(path)&.load&.chunks
+          data.map { |chunk| chunk[:text] }
+        end
+
+      texts.flatten!
 
       add_texts(texts: texts)
     end

data/lib/langchain/vectorsearch/chroma.rb

@@ -113,8 +113,9 @@ module Langchain::Vectorsearch
 
     # Ask a question and return the answer
     # @param question [String] The question to ask
+    # @yield [String] Stream responses back one String at a time
     # @return [String] The answer to the question
-    def ask(question:)
+    def ask(question:, &block)
       search_results = similarity_search(query: question)
 
       context = search_results.map do |result|
@@ -125,7 +126,7 @@ module Langchain::Vectorsearch
 
       prompt = generate_prompt(question: question, context: context)
 
-      llm.chat(prompt: prompt)
+      llm.chat(prompt: prompt, &block)
     end
 
     private

data/lib/langchain/vectorsearch/pgvector.rb

@@ -40,20 +40,53 @@ module Langchain::Vectorsearch
       super(llm: llm)
     end
 
-    # Add a list of texts to the index
+    # Upsert a list of texts to the index
     # @param texts [Array<String>] The texts to add to the index
-    # @return [PG::Result] The response from the database
-    def add_texts(texts:)
-      data = texts.flat_map do |text|
-        [text, llm.embed(text: text)]
+    # @param ids [Array<Integer>] The ids of the objects to add to the index, in the same order as the texts
+    # @return [PG::Result] The response from the database including the ids of
+    # the added or updated texts.
+    def upsert_texts(texts:, ids:)
+      data = texts.zip(ids).flat_map do |(text, id)|
+        [id, text, llm.embed(text: text)]
       end
-      values = texts.length.times.map { |i| "($#{2 * i + 1}, $#{2 * i + 2})" }.join(",")
+      values = texts.length.times.map { |i| "($#{3 * i + 1}, $#{3 * i + 2}, $#{3 * i + 3})" }.join(",")
+      # see https://github.com/pgvector/pgvector#storing
       client.exec_params(
-        "INSERT INTO #{quoted_table_name} (content, vectors) VALUES #{values};",
+        "INSERT INTO #{quoted_table_name} (id, content, vectors) VALUES
+#{values} ON CONFLICT (id) DO UPDATE SET content = EXCLUDED.content, vectors = EXCLUDED.vectors RETURNING id;",
         data
       )
     end
 
+    # Add a list of texts to the index
+    # @param texts [Array<String>] The texts to add to the index
+    # @param ids [Array<String>] The ids to add to the index, in the same order as the texts
+    # @return [PG::Result] The response from the database including the ids of
+    # the added texts.
+    def add_texts(texts:, ids: nil)
+      if ids.nil? || ids.empty?
+        data = texts.flat_map do |text|
+          [text, llm.embed(text: text)]
+        end
+        values = texts.length.times.map { |i| "($#{2 * i + 1}, $#{2 * i + 2})" }.join(",")
+        client.exec_params(
+          "INSERT INTO #{quoted_table_name} (content, vectors) VALUES #{values} RETURNING id;",
+          data
+        )
+      else
+        upsert_texts(texts: texts, ids: ids)
+      end
+    end
+
+    # Update a list of ids and corresponding texts to the index
+    # @param texts [Array<String>] The texts to add to the index
+    # @param ids [Array<String>] The ids to add to the index, in the same order as the texts
+    # @return [PG::Result] The response from the database including the ids of
+    # the updated texts.
+    def update_texts(texts:, ids:)
+      upsert_texts(texts: texts, ids: ids)
+    end
+
     # Create default schema
     # @return [PG::Result] The response from the database
     def create_default_schema
@@ -103,8 +136,9 @@ module Langchain::Vectorsearch
 
     # Ask a question and return the answer
     # @param question [String] The question to ask
+    # @yield [String] Stream responses back one String at a time
     # @return [String] The answer to the question
-    def ask(question:)
+    def ask(question:, &block)
       search_results = similarity_search(query: question)
 
       context = search_results.map do |result|
@@ -114,7 +148,7 @@ module Langchain::Vectorsearch
 
       prompt = generate_prompt(question: question, context: context)
 
-      llm.chat(prompt: prompt)
+      llm.chat(prompt: prompt, &block)
     end
   end
 end

data/lib/langchain/vectorsearch/pinecone.rb

@@ -139,8 +139,9 @@ module Langchain::Vectorsearch
     # @param question [String] The question to ask
     # @param namespace [String] The namespace to search in
     # @param filter [String] The filter to use
+    # @yield [String] Stream responses back one String at a time
     # @return [String] The answer to the question
-    def ask(question:, namespace: "", filter: nil)
+    def ask(question:, namespace: "", filter: nil, &block)
       search_results = similarity_search(query: question, namespace: namespace, filter: filter)
 
       context = search_results.map do |result|
@@ -150,7 +151,7 @@ module Langchain::Vectorsearch
 
       prompt = generate_prompt(question: question, context: context)
 
-      llm.chat(prompt: prompt)
+      llm.chat(prompt: prompt, &block)
     end
 
     # Pinecone index

data/lib/langchain/vectorsearch/qdrant.rb

@@ -112,8 +112,9 @@ module Langchain::Vectorsearch
 
     # Ask a question and return the answer
     # @param question [String] The question to ask
+    # @yield [String] Stream responses back one String at a time
     # @return [String] The answer to the question
-    def ask(question:)
+    def ask(question:, &block)
       search_results = similarity_search(query: question)
 
       context = search_results.map do |result|
@@ -123,7 +124,7 @@ module Langchain::Vectorsearch
 
       prompt = generate_prompt(question: question, context: context)
 
-      llm.chat(prompt: prompt)
+      llm.chat(prompt: prompt, &block)
     end
   end
 end

data/lib/langchain/vectorsearch/weaviate.rb

@@ -124,8 +124,9 @@ module Langchain::Vectorsearch
 
     # Ask a question and return the answer
     # @param question [String] The question to ask
+    # @yield [String] Stream responses back one String at a time
     # @return [Hash] The answer
-    def ask(question:)
+    def ask(question:, &block)
       search_results = similarity_search(query: question)
 
       context = search_results.map do |result|
@@ -135,7 +136,7 @@ module Langchain::Vectorsearch
 
       prompt = generate_prompt(question: question, context: context)
 
-      llm.chat(prompt: prompt)
+      llm.chat(prompt: prompt, &block)
     end
 
     private

data/lib/langchain/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Langchain
-  VERSION = "0.6.5"
+  VERSION = "0.6.7"
 end

data/lib/langchain.rb

@@ -82,6 +82,7 @@ module Langchain
   module Chunker
     autoload :Base, "langchain/chunker/base"
     autoload :Text, "langchain/chunker/text"
+    autoload :RecursiveText, "langchain/chunker/recursive_text"
   end
 
   module Tool

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: langchainrb
 version: !ruby/object:Gem::Version
-  version: 0.6.5
+  version: 0.6.7
 platform: ruby
 authors:
 - Andrei Bondarev
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2023-07-06 00:00:00.000000000 Z
+date: 2023-07-20 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: baran
@@ -408,14 +408,14 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 4.0.0
+        version: 4.1.0
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 4.0.0
+        version: 4.1.0
 - !ruby/object:Gem::Dependency
   name: safe_ruby
   requirement: !ruby/object:Gem::Requirement
@@ -493,6 +493,8 @@ files:
 - examples/create_and_manage_prompt_templates.rb
 - examples/create_and_manage_prompt_templates_using_structured_output_parser.rb
 - examples/llama_cpp.rb
+- examples/open_ai_function_calls.rb
+- examples/open_ai_qdrant_function_calls.rb
 - examples/pdf_store_and_query_with_chroma.rb
 - examples/store_and_query_with_pinecone.rb
 - examples/store_and_query_with_qdrant.rb
@@ -507,6 +509,7 @@ files:
 - lib/langchain/agent/sql_query_agent/sql_query_agent_answer_prompt.yaml
 - lib/langchain/agent/sql_query_agent/sql_query_agent_sql_prompt.yaml
 - lib/langchain/chunker/base.rb
+- lib/langchain/chunker/recursive_text.rb
 - lib/langchain/chunker/text.rb
 - lib/langchain/contextual_logger.rb
 - lib/langchain/conversation.rb
@@ -588,7 +591,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.2.3
+rubygems_version: 3.3.7
 signing_key:
 specification_version: 4
 summary: Build LLM-backed Ruby applications with Ruby's LangChain