langchainrb 0.13.5 → 0.15.0

data/lib/langchain/tool/{wikipedia/wikipedia.rb → wikipedia.rb}

@@ -1,19 +1,23 @@
 # frozen_string_literal: true
 
 module Langchain::Tool
-  class Wikipedia < Base
-    #
-    # Tool that adds the capability to search using the Wikipedia API
-    #
-    # Gem requirements:
-    #     gem "wikipedia-client", "~> 1.17.0"
-    #
-    # Usage:
-    #     wikipedia = Langchain::Tool::Wikipedia.new
-    #     wikipedia.execute(input: "The Roman Empire")
-    #
-    NAME = "wikipedia"
-    ANNOTATIONS_PATH = Langchain.root.join("./langchain/tool/#{NAME}/#{NAME}.json").to_path
+  #
+  # Tool that adds the capability to search using the Wikipedia API
+  #
+  # Gem requirements:
+  #     gem "wikipedia-client", "~> 1.17.0"
+  #
+  # Usage:
+  #     wikipedia = Langchain::Tool::Wikipedia.new
+  #     wikipedia.execute(input: "The Roman Empire")
+  #
+  class Wikipedia
+    extend Langchain::ToolDefinition
+    include Langchain::DependencyHelper
+
+    define_function :execute, description: "Executes Wikipedia API search and returns the answer" do
+      property :input, type: "string", description: "Search query", required: true
+    end
 
     # Initializes the Wikipedia tool
     def initialize

data/lib/langchain/tool_definition.rb

@@ -0,0 +1,212 @@
+# frozen_string_literal: true
+
+require "json"
+
+#
+# Extends a class to be used as a tool in the assistant.
+# A tool is a collection of functions (methods) used to perform specific tasks.
+#
+# == Usage
+#
+# 1. Extend your class with {Langchain::ToolDefinition}
+# 2. Use {#define_function} to define each function of the tool
+#
+# == Key Concepts
+#
+# - {#define_function}: Defines a new function (method) for the tool
+# - {ParameterBuilder#property}: Defines properties for the function parameters
+# - {ParameterBuilder#item}: Alias for {ParameterBuilder#property}, used for array items
+#
+# These methods support various data types and nested structures, allowing for flexible and expressive tool definitions.
+#
+# @example Defining a tool with various property types and configurations
+#   define_function :sample_function, description: "Demonstrates various property types and configurations" do
+#     property :string_prop, type: "string", description: "A simple string property"
+#     property :number_prop, type: "number", description: "A number property"
+#     property :integer_prop, type: "integer", description: "An integer property"
+#     property :boolean_prop, type: "boolean", description: "A boolean property"
+#     property :enum_prop, type: "string", description: "An enum property", enum: ["option1", "option2", "option3"]
+#     property :required_prop, type: "string", description: "A required property", required: true
+#     property :array_prop, type: "array", description: "An array property" do
+#       item type: "string", description: "Array item"
+#     end
+#     property :object_prop, type: "object", description: "An object property" do
+#       property :nested_string, type: "string", description: "Nested string property"
+#       property :nested_number, type: "number", description: "Nested number property"
+#     end
+#   end
+#
+module Langchain::ToolDefinition
+  # Defines a function for the tool
+  #
+  # @param method_name [Symbol] Name of the method to define
+  # @param description [String] Description of the function
+  # @yield Block that defines the parameters for the function
+  def define_function(method_name, description:, &)
+    function_schemas.add_function(method_name:, description:, &)
+  end
+
+  # Returns the FunctionSchemas instance for this tool
+  #
+  # @return [FunctionSchemas] The FunctionSchemas instance
+  def function_schemas
+    @function_schemas ||= FunctionSchemas.new(tool_name)
+  end
+
+  # Returns the snake_case version of the class name as the tool's name
+  #
+  # @return [String] The snake_case version of the class name
+  def tool_name
+    @tool_name ||= name
+      .gsub("::", "_")
+      .gsub(/(?<=[A-Z])(?=[A-Z][a-z])|(?<=[a-z\d])(?=[A-Z])/, "_")
+      .downcase
+  end
+
+  # Manages schemas for functions
+  class FunctionSchemas
+    def initialize(tool_name)
+      @schemas = {}
+      @tool_name = tool_name
+    end
+
+    # Adds a function to the schemas
+    #
+    # @param method_name [Symbol] Name of the method to add
+    # @param description [String] Description of the function
+    # @yield Block that defines the parameters for the function
+    # @raise [ArgumentError] If a block is defined and no parameters are specified for the function
+    def add_function(method_name:, description:, &)
+      name = "#{@tool_name}__#{method_name}"
+
+      if block_given?
+        parameters = ParameterBuilder.new(parent_type: "object").build(&)
+
+        if parameters[:properties].empty?
+          raise ArgumentError, "Function parameters must have at least one property defined within it, if a block is provided"
+        end
+      end
+
+      @schemas[method_name] = {
+        type: "function",
+        function: {name:, description:, parameters:}.compact
+      }
+    end
+
+    # Converts schemas to OpenAI-compatible format
+    #
+    # @return [String] JSON string of schemas in OpenAI format
+    def to_openai_format
+      @schemas.values
+    end
+
+    # Converts schemas to Anthropic-compatible format
+    #
+    # @return [String] JSON string of schemas in Anthropic format
+    def to_anthropic_format
+      @schemas.values.map do |schema|
+        schema[:function].transform_keys(parameters: :input_schema)
+      end
+    end
+
+    # Converts schemas to Google Gemini-compatible format
+    #
+    # @return [String] JSON string of schemas in Google Gemini format
+    def to_google_gemini_format
+      @schemas.values.map { |schema| schema[:function] }
+    end
+  end
+
+  # Builds parameter schemas for functions
+  class ParameterBuilder
+    VALID_TYPES = %w[object array string number integer boolean].freeze
+
+    def initialize(parent_type:)
+      @schema = (parent_type == "object") ? {type: "object", properties: {}, required: []} : {}
+      @parent_type = parent_type
+    end
+
+    # Builds the parameter schema
+    #
+    # @yield Block that defines the properties of the schema
+    # @return [Hash] The built schema
+    def build(&)
+      instance_eval(&)
+      @schema
+    end
+
+    # Defines a property in the schema
+    #
+    # @param name [Symbol] Name of the property (required only for a parent of type object)
+    # @param type [String] Type of the property
+    # @param description [String] Description of the property
+    # @param enum [Array] Array of allowed values
+    # @param required [Boolean] Whether the property is required
+    # @yield [Block] Block for nested properties (only for object and array types)
+    # @raise [ArgumentError] If any parameter is invalid
+    def property(name = nil, type:, description: nil, enum: nil, required: false, &)
+      validate_parameters(name:, type:, enum:, required:)
+
+      prop = {type:, description:, enum:}.compact
+
+      if block_given?
+        nested_schema = ParameterBuilder.new(parent_type: type).build(&)
+
+        case type
+        when "object"
+          if nested_schema[:properties].empty?
+            raise ArgumentError, "Object properties must have at least one property defined within it"
+          end
+          prop = nested_schema
+        when "array"
+          if nested_schema.empty?
+            raise ArgumentError, "Array properties must have at least one item defined within it"
+          end
+          prop[:items] = nested_schema
+        end
+      end
+
+      if @parent_type == "object"
+        @schema[:properties][name] = prop
+        @schema[:required] << name.to_s if required
+      else
+        @schema = prop
+      end
+    end
+
+    # Alias for property method, used for defining array items
+    alias_method :item, :property
+
+    private
+
+    # Validates the parameters for a property
+    #
+    # @param name [Symbol] Name of the property
+    # @param type [String] Type of the property
+    # @param enum [Array] Array of allowed values
+    # @param required [Boolean] Whether the property is required
+    # @raise [ArgumentError] If any parameter is invalid
+    def validate_parameters(name:, type:, enum:, required:)
+      if @parent_type == "object"
+        if name.nil?
+          raise ArgumentError, "Name must be provided for properties of an object"
+        end
+        unless name.is_a?(Symbol)
+          raise ArgumentError, "Invalid name '#{name}'. Name must be a symbol"
+        end
+      end
+
+      unless VALID_TYPES.include?(type)
+        raise ArgumentError, "Invalid type '#{type}'. Valid types are: #{VALID_TYPES.join(", ")}"
+      end
+
+      unless enum.nil? || enum.is_a?(Array)
+        raise ArgumentError, "Invalid enum '#{enum}'. Enum must be nil or an array"
+      end
+
+      unless [true, false].include?(required)
+        raise ArgumentError, "Invalid required '#{required}'. Required must be a boolean"
+      end
+    end
+  end
+end

data/lib/langchain/utils/hash_transformer.rb

@@ -1,24 +1,16 @@
 module Langchain
   module Utils
     class HashTransformer
-      # Converts a string to camelCase
-      def self.camelize_lower(str)
-        str.split("_").inject([]) { |buffer, e| buffer.push(buffer.empty? ? e : e.capitalize) }.join
-      end
-
-      # Recursively transforms the keys of a hash to camel case
-      def self.deep_transform_keys(hash, &block)
-        case hash
-        when Hash
-          hash.each_with_object({}) do |(key, value), result|
-            new_key = block.call(key)
-            result[new_key] = deep_transform_keys(value, &block)
+      def self.symbolize_keys(hash)
+        hash.map do |k, v|
+          new_key = begin
+            k.to_sym
+          rescue
+            k
           end
-        when Array
-          hash.map { |item| deep_transform_keys(item, &block) }
-        else
-          hash
-        end
+          new_value = v.is_a?(Hash) ? symbolize_keys(v) : v
+          [new_key, new_value]
+        end.to_h
       end
     end
   end

data/lib/langchain/vectorsearch/chroma.rb

@@ -124,7 +124,7 @@ module Langchain::Vectorsearch
     # @param k [Integer] The number of results to have in context
     # @yield [String] Stream responses back one String at a time
     # @return [String] The answer to the question
-    def ask(question:, k: 4, &)
+    def ask(question:, k: 4, &block)
       search_results = similarity_search(query: question, k: k)
 
       context = search_results.map do |result|
@@ -136,7 +136,7 @@ module Langchain::Vectorsearch
       prompt = generate_rag_prompt(question: question, context: context)
 
       messages = [{role: "user", content: prompt}]
-      response = llm.chat(messages: messages, &)
+      response = llm.chat(messages: messages, &block)
 
       response.context = context
       response

data/lib/langchain/vectorsearch/elasticsearch.rb

@@ -143,7 +143,7 @@ module Langchain::Vectorsearch
     # @param k [Integer] The number of results to have in context
     # @yield [String] Stream responses back one String at a time
     # @return [String] The answer to the question
-    def ask(question:, k: 4, &)
+    def ask(question:, k: 4, &block)
       search_results = similarity_search(query: question, k: k)
 
       context = search_results.map do |result|
@@ -153,7 +153,7 @@ module Langchain::Vectorsearch
       prompt = generate_rag_prompt(question: question, context: context)
 
       messages = [{role: "user", content: prompt}]
-      response = llm.chat(messages: messages, &)
+      response = llm.chat(messages: messages, &block)
 
       response.context = context
       response

data/lib/langchain/vectorsearch/epsilla.rb

@@ -96,7 +96,7 @@ module Langchain::Vectorsearch
 
       status_code, response = @client.database.insert(@table_name, data)
       raise "Failed to insert texts: #{response}" if status_code != 200
-      response
+      JSON.parse(response)
     end
 
     # Search for similar texts
@@ -129,7 +129,7 @@ module Langchain::Vectorsearch
     # @param k [Integer] The number of results to have in context
     # @yield [String] Stream responses back one String at a time
     # @return [String] The answer to the question
-    def ask(question:, k: 4, &)
+    def ask(question:, k: 4, &block)
       search_results = similarity_search(query: question, k: k)
 
       context = search_results.map do |result|
@@ -140,7 +140,7 @@ module Langchain::Vectorsearch
       prompt = generate_rag_prompt(question: question, context: context)
 
       messages = [{role: "user", content: prompt}]
-      response = llm.chat(messages: messages, &)
+      response = llm.chat(messages: messages, &block)
 
       response.context = context
       response

data/lib/langchain/vectorsearch/milvus.rb

@@ -140,7 +140,7 @@ module Langchain::Vectorsearch
 
       client.search(
         collection_name: index_name,
-        output_fields: ["id", "content", "vectors"],
+        output_fields: ["id", "content"], # Add "vectors" if need to have full vectors returned.
         top_k: k.to_s,
         vectors: [embedding],
         dsl_type: 1,
@@ -156,7 +156,7 @@ module Langchain::Vectorsearch
     # @param k [Integer] The number of results to have in context
     # @yield [String] Stream responses back one String at a time
     # @return [String] The answer to the question
-    def ask(question:, k: 4, &)
+    def ask(question:, k: 4, &block)
       search_results = similarity_search(query: question, k: k)
 
       content_field = search_results.dig("results", "fields_data").select { |field| field.dig("field_name") == "content" }
@@ -167,7 +167,7 @@ module Langchain::Vectorsearch
       prompt = generate_rag_prompt(question: question, context: context)
 
       messages = [{role: "user", content: prompt}]
-      response = llm.chat(messages: messages, &)
+      response = llm.chat(messages: messages, &block)
 
       response.context = context
       response

data/lib/langchain/vectorsearch/pgvector.rb

@@ -146,7 +146,7 @@ module Langchain::Vectorsearch
     # @param k [Integer] The number of results to have in context
     # @yield [String] Stream responses back one String at a time
     # @return [String] The answer to the question
-    def ask(question:, k: 4, &)
+    def ask(question:, k: 4, &block)
       search_results = similarity_search(query: question, k: k)
 
       context = search_results.map do |result|
@@ -157,7 +157,7 @@ module Langchain::Vectorsearch
       prompt = generate_rag_prompt(question: question, context: context)
 
       messages = [{role: "user", content: prompt}]
-      response = llm.chat(messages: messages, &)
+      response = llm.chat(messages: messages, &block)
 
       response.context = context
       response

data/lib/langchain/vectorsearch/pinecone.rb

@@ -171,7 +171,7 @@ module Langchain::Vectorsearch
     # @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, k: 4, &)
+    def ask(question:, namespace: "", filter: nil, k: 4, &block)
       search_results = similarity_search(query: question, namespace: namespace, filter: filter, k: k)
 
       context = search_results.map do |result|
@@ -182,7 +182,7 @@ module Langchain::Vectorsearch
       prompt = generate_rag_prompt(question: question, context: context)
 
       messages = [{role: "user", content: prompt}]
-      response = llm.chat(messages: messages, &)
+      response = llm.chat(messages: messages, &block)
 
       response.context = context
       response

data/lib/langchain/vectorsearch/qdrant.rb

@@ -137,7 +137,7 @@ module Langchain::Vectorsearch
     # @param k [Integer] The number of results to have in context
     # @yield [String] Stream responses back one String at a time
     # @return [String] The answer to the question
-    def ask(question:, k: 4, &)
+    def ask(question:, k: 4, &block)
       search_results = similarity_search(query: question, k: k)
 
       context = search_results.map do |result|
@@ -148,7 +148,7 @@ module Langchain::Vectorsearch
       prompt = generate_rag_prompt(question: question, context: context)
 
       messages = [{role: "user", content: prompt}]
-      response = llm.chat(messages: messages, &)
+      response = llm.chat(messages: messages, &block)
 
       response.context = context
       response

data/lib/langchain/vectorsearch/weaviate.rb

@@ -6,7 +6,7 @@ module Langchain::Vectorsearch
     # Wrapper around Weaviate
     #
     # Gem requirements:
-    #     gem "weaviate-ruby", "~> 0.8.9"
+    #     gem "weaviate-ruby", "~> 0.9.0"
     #
     # Usage:
     #     weaviate = Langchain::Vectorsearch::Weaviate.new(url: ENV["WEAVIATE_URL"], api_key: ENV["WEAVIATE_API_KEY"], index_name: "Docs", llm: llm)
@@ -17,7 +17,7 @@ module Langchain::Vectorsearch
     # @param api_key [String] The API key to use
     # @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:)
+    def initialize(url:, index_name:, llm:, api_key: nil)
       depends_on "weaviate-ruby", req: "weaviate"
 
       @client = ::Weaviate::Client.new(
@@ -143,7 +143,7 @@ module Langchain::Vectorsearch
     # @param k [Integer] The number of results to have in context
     # @yield [String] Stream responses back one String at a time
     # @return [Hash] The answer
-    def ask(question:, k: 4, &)
+    def ask(question:, k: 4, &block)
       search_results = similarity_search(query: question, k: k)
 
       context = search_results.map do |result|
@@ -154,7 +154,7 @@ module Langchain::Vectorsearch
       prompt = generate_rag_prompt(question: question, context: context)
 
       messages = [{role: "user", content: prompt}]
-      response = llm.chat(messages: messages, &)
+      response = llm.chat(messages: messages, &block)
 
       response.context = context
       response

data/lib/langchain/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Langchain
-  VERSION = "0.13.5"
+  VERSION = "0.15.0"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: langchainrb
 version: !ruby/object:Gem::Version
-  version: 0.13.5
+  version: 0.15.0
 platform: ruby
 authors:
 - Andrei Bondarev
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2024-07-01 00:00:00.000000000 Z
+date: 2024-08-14 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: baran
@@ -212,14 +212,14 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '0.2'
+        version: '0.3'
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '0.2'
+        version: '0.3'
 - !ruby/object:Gem::Dependency
   name: aws-sdk-bedrockruntime
   requirement: !ruby/object:Gem::Requirement
@@ -682,20 +682,6 @@ dependencies:
     - - "~>"
       - !ruby/object:Gem::Version
         version: 0.1.0
-- !ruby/object:Gem::Dependency
-  name: tiktoken_ruby
-  requirement: !ruby/object:Gem::Requirement
-    requirements:
-    - - "~>"
-      - !ruby/object:Gem::Version
-        version: 0.0.9
-  type: :development
-  prerelease: false
-  version_requirements: !ruby/object:Gem::Requirement
-    requirements:
-    - - "~>"
-      - !ruby/object:Gem::Version
-        version: 0.0.9
 description: Build LLM-backed Ruby applications with Ruby's Langchain.rb
 email:
 - andrei.bondarev13@gmail.com
@@ -711,6 +697,7 @@ files:
 - lib/langchain/assistants/messages/anthropic_message.rb
 - lib/langchain/assistants/messages/base.rb
 - lib/langchain/assistants/messages/google_gemini_message.rb
+- lib/langchain/assistants/messages/ollama_message.rb
 - lib/langchain/assistants/messages/openai_message.rb
 - lib/langchain/assistants/thread.rb
 - lib/langchain/chunk.rb
@@ -787,35 +774,19 @@ files:
 - lib/langchain/prompt/few_shot_prompt_template.rb
 - lib/langchain/prompt/loading.rb
 - lib/langchain/prompt/prompt_template.rb
-- lib/langchain/tool/base.rb
-- lib/langchain/tool/calculator/calculator.json
-- lib/langchain/tool/calculator/calculator.rb
-- lib/langchain/tool/database/database.json
-- lib/langchain/tool/database/database.rb
-- lib/langchain/tool/file_system/file_system.json
-- lib/langchain/tool/file_system/file_system.rb
-- lib/langchain/tool/google_search/google_search.json
-- lib/langchain/tool/google_search/google_search.rb
-- lib/langchain/tool/news_retriever/news_retriever.json
-- lib/langchain/tool/news_retriever/news_retriever.rb
-- lib/langchain/tool/ruby_code_interpreter/ruby_code_interpreter.json
-- lib/langchain/tool/ruby_code_interpreter/ruby_code_interpreter.rb
-- lib/langchain/tool/tavily/tavily.json
-- lib/langchain/tool/tavily/tavily.rb
-- lib/langchain/tool/vectorsearch/vectorsearch.json
-- lib/langchain/tool/vectorsearch/vectorsearch.rb
-- lib/langchain/tool/weather/weather.json
-- lib/langchain/tool/weather/weather.rb
-- lib/langchain/tool/wikipedia/wikipedia.json
-- lib/langchain/tool/wikipedia/wikipedia.rb
+- lib/langchain/tool/calculator.rb
+- lib/langchain/tool/database.rb
+- lib/langchain/tool/file_system.rb
+- lib/langchain/tool/google_search.rb
+- lib/langchain/tool/news_retriever.rb
+- lib/langchain/tool/ruby_code_interpreter.rb
+- lib/langchain/tool/tavily.rb
+- lib/langchain/tool/vectorsearch.rb
+- lib/langchain/tool/weather.rb
+- lib/langchain/tool/wikipedia.rb
+- lib/langchain/tool_definition.rb
 - lib/langchain/utils/cosine_similarity.rb
 - lib/langchain/utils/hash_transformer.rb
-- lib/langchain/utils/token_length/ai21_validator.rb
-- lib/langchain/utils/token_length/base_validator.rb
-- lib/langchain/utils/token_length/cohere_validator.rb
-- lib/langchain/utils/token_length/google_palm_validator.rb
-- lib/langchain/utils/token_length/openai_validator.rb
-- lib/langchain/utils/token_length/token_limit_exceeded.rb
 - lib/langchain/vectorsearch/base.rb
 - lib/langchain/vectorsearch/chroma.rb
 - lib/langchain/vectorsearch/elasticsearch.rb