langchainrb 0.19.3 → 0.19.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 5567aa14daf63f120a6c2b2b2991fbd85b650b4cb4e2ac49703736dacd46197e
-  data.tar.gz: ee71c45590998793c343c9620565cfcfdff5cd2d715091f3ad6804d54b814444
+  metadata.gz: 5851ca8d3c014468081f957ecb1f76ab7d50e6d1519c265ab4d52ec39c7bc8b9
+  data.tar.gz: 4850f1636545fedca6a1c7f4d17bc4a07309708219d2a06024b14adb17a275a5
 SHA512:
-  metadata.gz: a5d212aadfef3ca07a6c9dc517141f0e6dedb1e414b0e8bbcbc0b7abddd5e2405a8a07ab6cc10efd66ee8997c41b5c8bd82ea8755d0f8e5fea7728db94f82379
-  data.tar.gz: a3a5a5c62cc075e8d09c0e6b629e1ec8b118d35d98e4de107c48d779d8b35261de3ad0061e22a4be4b2aefc15bbb25c20f658fe1f0a694ecc87279cd21309ff8
+  metadata.gz: b46baf32e44f9134aceea57957045b37466a23a28c709c03df867071de9615bcf8803a97ec64fa3194ee4d3e13a686389a7c0b9d870c5f9716b5ff361fb8e873
+  data.tar.gz: 4e7112d5cb46857e89bbd25f10dae57ec38085027ee6c5ea523e6637218fa97899a0e9deeb2fbf7a116cd6a70a6f17a4c4b294d31c4edf703cce401477ae3a0a

data/CHANGELOG.md

@@ -9,7 +9,19 @@
 - [DOCS]: Documentation changes. No changes to the library's behavior.
 - [SECURITY]: A change which fixes a security vulnerability.
 
-## [Unreleased]
+## [0.19.5]
+- [BREAKING] [https://github.com/patterns-ai-core/langchainrb/pull/859] Add metadata support to PgVector storage
+- [BUGFIX] [https://github.com/patterns-ai-core/langchainrb/pull/939] Fix Langchain::Vectorsearch::Milvus initializer by passing :api_key
+- [BUGFIX] [https://github.com/patterns-ai-core/langchainrb/pull/953] Handle nil response in OpenAI LLM streaming
+- [BREAKING] [https://github.com/patterns-ai-core/langchainrb/pull/956] Deprecate `Langchain::Vectorsearch::Epsilla` class
+- [BREAKING] [https://github.com/patterns-ai-core/langchainrb/pull/961] Deprecate `Langchain::LLM::LlamaCpp` class
+- [BREAKING] [https://github.com/patterns-ai-core/langchainrb/pull/962] Deprecate `Langchain::LLM::AI21` class
+- [BREAKING] [https://github.com/patterns-ai-core/langchainrb/pull/971] Exclude `temperature` from being automatically added to OpenAI LLM parameters
+- [OPTIM] [https://github.com/patterns-ai-core/langchainrb/pull/977] Enable `Langchain::Tool::RubyCodeInterpreter` on Ruby 3.3+
+
+## [0.19.4] - 2025-02-17
+- [BREAKING] [https://github.com/patterns-ai-core/langchainrb/pull/894] Tools can now output image_urls, and all tool output must be wrapped by a tool_response() method
+- [BUGFIX] [https://github.com/patterns-ai-core/langchainrb/pull/921] Fix for Assistant when OpenAI o1/o3 models are used
 
 ## [0.19.3] - 2025-01-13
 - [BUGFIX] [https://github.com/patterns-ai-core/langchainrb/pull/900] Empty text content should not be set when content is nil when using AnthropicMessage

data/README.md

@@ -2,14 +2,14 @@
 ---
 ⚡ Building LLM-powered applications in Ruby ⚡
 
-For deep Rails integration see: [langchainrb_rails](https://github.com/andreibondarev/langchainrb_rails) gem.
+For deep Rails integration see: [langchainrb_rails](https://github.com/patterns-ai-core/langchainrb_rails) gem.
 
 Available for paid consulting engagements! [Email me](mailto:andrei@sourcelabs.io).
 
-![Tests status](https://github.com/andreibondarev/langchainrb/actions/workflows/ci.yml/badge.svg?branch=main)
+![Tests status](https://github.com/patterns-ai-core/langchainrb/actions/workflows/ci.yml/badge.svg?branch=main)
 [![Gem Version](https://badge.fury.io/rb/langchainrb.svg)](https://badge.fury.io/rb/langchainrb)
 [![Docs](http://img.shields.io/badge/yard-docs-blue.svg)](http://rubydoc.info/gems/langchainrb)
-[![License](https://img.shields.io/badge/license-MIT-green.svg)](https://github.com/andreibondarev/langchainrb/blob/main/LICENSE.txt)
+[![License](https://img.shields.io/badge/license-MIT-green.svg)](https://github.com/patterns-ai-core/langchainrb/blob/main/LICENSE.txt)
 [![](https://dcbadge.vercel.app/api/server/WDARp7J2n8?compact=true&style=flat)](https://discord.gg/WDARp7J2n8)
 [![X](https://img.shields.io/twitter/url/https/twitter.com/cloudposse.svg?style=social&label=Follow%20%40rushing_andrei)](https://twitter.com/rushing_andrei)
 
@@ -57,7 +57,6 @@ The `Langchain::LLM` module provides a unified interface for interacting with va
 
 ## Supported LLM Providers
 
-- AI21
 - Anthropic
 - AWS Bedrock
 - Azure OpenAI
@@ -65,7 +64,6 @@ The `Langchain::LLM` module provides a unified interface for interacting with va
 - Google Gemini
 - Google Vertex AI
 - HuggingFace
-- LlamaCpp
 - Mistral AI
 - Ollama
 - OpenAI
@@ -369,7 +367,7 @@ fix_parser = Langchain::OutputParsers::OutputFixingParser.from_llm(
 fix_parser.parse(llm_response)
 ```
 
-See [here](https://github.com/andreibondarev/langchainrb/tree/main/examples/create_and_manage_prompt_templates_using_structured_output_parser.rb) for a concrete example
+See [here](https://github.com/patterns-ai-core/langchainrb/tree/main/examples/create_and_manage_prompt_templates_using_structured_output_parser.rb) for a concrete example
 
 ## Building Retrieval Augment Generation (RAG) system
 RAG is a methodology that assists LLMs generate accurate and up-to-date information.
@@ -387,7 +385,6 @@ Langchain.rb provides a convenient unified interface on top of supported vectors
 | Database                                                                                   | Open-source        | Cloud offering     |
 | --------                                                                                   |:------------------:| :------------:     |
 | [Chroma](https://trychroma.com/?utm_source=langchainrb&utm_medium=github)                  | ✅                 | ✅                 |
-| [Epsilla](https://epsilla.com/?utm_source=langchainrb&utm_medium=github)                   | ✅                 | ✅                 |
 | [Hnswlib](https://github.com/nmslib/hnswlib/?utm_source=langchainrb&utm_medium=github)     | ✅                 | ❌                 |
 | [Milvus](https://milvus.io/?utm_source=langchainrb&utm_medium=github)                      | ✅                 | ✅ Zilliz Cloud    |
 | [Pinecone](https://www.pinecone.io/?utm_source=langchainrb&utm_medium=github)              | ❌                 | ✅                 |
@@ -420,7 +417,6 @@ client = Langchain::Vectorsearch::Weaviate.new(
 You can instantiate any other supported vector search database:
 ```ruby
 client = Langchain::Vectorsearch::Chroma.new(...)   # `gem "chroma-db", "~> 0.6.0"`
-client = Langchain::Vectorsearch::Epsilla.new(...)  # `gem "epsilla-ruby", "~> 0.0.3"`
 client = Langchain::Vectorsearch::Hnswlib.new(...)  # `gem "hnswlib", "~> 0.8.1"`
 client = Langchain::Vectorsearch::Milvus.new(...)   # `gem "milvus", "~> 0.9.3"`
 client = Langchain::Vectorsearch::Pinecone.new(...) # `gem "pinecone", "~> 0.1.6"`
@@ -580,11 +576,11 @@ class MovieInfoTool
   end
 
   def search_movie(query:)
-    ...
+    tool_response(...)
   end
 
   def get_movie_details(movie_id:)
-    ...
+    tool_response(...)
   end
 end
 ```
@@ -639,7 +635,7 @@ ragas.score(answer: "", question: "", context: "")
 ```
 
 ## Examples
-Additional examples available: [/examples](https://github.com/andreibondarev/langchainrb/tree/main/examples)
+Additional examples available: [/examples](https://github.com/patterns-ai-core/langchainrb/tree/main/examples)
 
 ## Logging
 
@@ -665,7 +661,7 @@ gem install unicode -- --with-cflags="-Wno-incompatible-function-pointer-types"
 
 ## Development
 
-1. `git clone https://github.com/andreibondarev/langchainrb.git`
+1. `git clone https://github.com/patterns-ai-core/langchainrb.git`
 2. `cp .env.example .env`, then fill out the environment variables in `.env`
 3. `bundle exec rake` to ensure that the tests pass and to run standardrb
 4. `bin/console` to load the gem in a REPL session. Feel free to add your own instances of LLMs, Tools, Agents, etc. and experiment with them.
@@ -680,7 +676,7 @@ Join us in the [Langchain.rb](https://discord.gg/WDARp7J2n8) Discord server.
 
 ## Contributing
 
-Bug reports and pull requests are welcome on GitHub at https://github.com/andreibondarev/langchainrb.
+Bug reports and pull requests are welcome on GitHub at https://github.com/patterns-ai-core/langchainrb.
 
 ## License
 

data/lib/langchain/assistant/llm/adapters/openai.rb

@@ -24,7 +24,9 @@ module Langchain
             if tools.any?
               params[:tools] = build_tools(tools)
               params[:tool_choice] = build_tool_choice(tool_choice)
-              params[:parallel_tool_calls] = parallel_tool_calls
+              # Temporary fix because OpenAI o1/o3/reasoning models don't support `parallel_tool_calls` parameter.
+              # Set `Assistant.new(parallel_tool_calls: nil, ...)` to avoid the error.
+              params[:parallel_tool_calls] = parallel_tool_calls unless parallel_tool_calls.nil?
             end
             params
           end

data/lib/langchain/assistant.rb

@@ -371,9 +371,15 @@ module Langchain
 
       # Call the callback if set
       tool_execution_callback.call(tool_call_id, tool_name, method_name, tool_arguments) if tool_execution_callback # rubocop:disable Style/SafeNavigation
+
       output = tool_instance.send(method_name, **tool_arguments)
 
-      submit_tool_output(tool_call_id: tool_call_id, output: output)
+      # Handle both ToolResponse and legacy return values
+      if output.is_a?(ToolResponse)
+        add_message(role: @llm_adapter.tool_role, content: output.content, image_url: output.image_url, tool_call_id: tool_call_id)
+      else
+        submit_tool_output(tool_call_id: tool_call_id, output: output)
+      end
     end
 
     # Build a message

data/lib/langchain/llm/ai21.rb

@@ -17,6 +17,8 @@ module Langchain::LLM
     }.freeze
 
     def initialize(api_key:, default_options: {})
+      Langchain.logger.warn "DEPRECATED: `Langchain::LLM::AI21` is deprecated, and will be removed in the next major version. Please use another LLM provider."
+
       depends_on "ai21"
 
       @client = ::AI21::Client.new(api_key)

data/lib/langchain/llm/anthropic.rb

@@ -22,10 +22,15 @@ module Langchain::LLM
     #
     # @param api_key [String] The API key to use
     # @param llm_options [Hash] Options to pass to the Anthropic client
-    # @param default_options [Hash] Default options to use on every call to LLM, e.g.: { temperature:, completion_model:, chat_model:, max_tokens: }
+    # @param default_options [Hash] Default options to use on every call to LLM, e.g.: { temperature:, completion_model:, chat_model:, max_tokens:, thinking: }
     # @return [Langchain::LLM::Anthropic] Langchain::LLM::Anthropic instance
     def initialize(api_key:, llm_options: {}, default_options: {})
-      depends_on "anthropic"
+      begin
+        depends_on "ruby-anthropic", req: "anthropic"
+      rescue Langchain::DependencyHelper::LoadError
+        # Falls back to the older `anthropic` gem if `ruby-anthropic` gem cannot be loaded.
+        depends_on "anthropic"
+      end
 
       @client = ::Anthropic::Client.new(access_token: api_key, **llm_options)
       @defaults = DEFAULTS.merge(default_options)
@@ -34,7 +39,8 @@ module Langchain::LLM
         temperature: {default: @defaults[:temperature]},
         max_tokens: {default: @defaults[:max_tokens]},
         metadata: {},
-        system: {}
+        system: {},
+        thinking: {default: @defaults[:thinking]}
       )
       chat_parameters.ignore(:n, :user)
       chat_parameters.remap(stop: :stop_sequences)
@@ -97,6 +103,7 @@ module Langchain::LLM
     # @option params [String] :system System prompt
     # @option params [Float] :temperature Amount of randomness injected into the response
     # @option params [Array<String>] :tools Definitions of tools that the model may use
+    # @option params [Hash] :thinking Enable extended thinking mode, e.g. { type: "enabled", budget_tokens: 4000 }
     # @option params [Integer] :top_k Only sample from the top K options for each subsequent token
     # @option params [Float] :top_p Use nucleus sampling.
     # @return [Langchain::LLM::AnthropicResponse] The chat completion
@@ -170,7 +177,7 @@ module Langchain::LLM
           "id" => first_block.dig("content_block", "id"),
           "type" => "tool_use",
           "name" => first_block.dig("content_block", "name"),
-          "input" => JSON.parse(input).transform_keys(&:to_sym)
+          "input" => input.empty? ? nil : JSON.parse(input).transform_keys(&:to_sym)
         }
       end.compact
     end

data/lib/langchain/llm/aws_bedrock.rb

@@ -167,7 +167,7 @@ module Langchain::LLM
 
     def parse_model_id(model_id)
       model_id
-        .gsub("us.", "") # Meta append "us." to their model ids
+        .gsub(/^(us|eu|apac|us-gov)\./, "") # Meta append "us." to their model ids, and AWS region prefixes are used by Bedrock cross-region model IDs.
         .split(".")
     end
 

data/lib/langchain/llm/base.rb

@@ -7,14 +7,12 @@ module Langchain::LLM
   #
   # Langchain.rb provides a common interface to interact with all supported LLMs:
   #
-  # - {Langchain::LLM::AI21}
   # - {Langchain::LLM::Anthropic}
   # - {Langchain::LLM::Azure}
   # - {Langchain::LLM::Cohere}
   # - {Langchain::LLM::GoogleGemini}
   # - {Langchain::LLM::GoogleVertexAI}
   # - {Langchain::LLM::HuggingFace}
-  # - {Langchain::LLM::LlamaCpp}
   # - {Langchain::LLM::OpenAI}
   # - {Langchain::LLM::Replicate}
   #

data/lib/langchain/llm/llama_cpp.rb

@@ -23,6 +23,8 @@ module Langchain::LLM
     # @param n_threads [Integer] The CPU number of threads to use
     # @param seed [Integer] The seed to use
     def initialize(model_path:, n_gpu_layers: 1, n_ctx: 2048, n_threads: 1, seed: 0)
+      Langchain.logger.warn "DEPRECATED: `Langchain::LLM::LlamaCpp` is deprecated, and will be removed in the next major version. Please use `Langchain::LLM::Ollama` for self-hosted LLM inference."
+
       depends_on "llama_cpp"
 
       @model_path = model_path

data/lib/langchain/llm/openai.rb

@@ -15,7 +15,6 @@ module Langchain::LLM
   class OpenAI < Base
     DEFAULTS = {
       n: 1,
-      temperature: 0.0,
       chat_model: "gpt-4o-mini",
       embedding_model: "text-embedding-3-small"
     }.freeze
@@ -173,7 +172,7 @@ module Langchain::LLM
 
     def with_api_error_handling
       response = yield
-      return if response.empty?
+      return if response.nil? || response.empty?
 
       raise Langchain::LLM::ApiError.new "OpenAI API error: #{response.dig("error", "message")}" if response&.dig("error")
 

data/lib/langchain/llm/parameters/chat.rb

@@ -30,6 +30,7 @@ module Langchain::LLM::Parameters
       presence_penalty: {}, # Range: [-2, 2]
       repetition_penalty: {}, # Range: (0, 2]
       seed: {}, # OpenAI only
+      store: {}, # store messages with OpenAI
 
       # Function-calling
       tools: {default: []},

data/lib/langchain/output_parsers/base.rb

@@ -28,15 +28,4 @@ module Langchain::OutputParsers
       raise NotImplementedError
     end
   end
-
-  class OutputParserException < StandardError
-    def initialize(message, text)
-      @message = message
-      @text = text
-    end
-
-    def to_s
-      "#{@message}\nText: #{@text}"
-    end
-  end
 end

data/lib/langchain/output_parsers/output_parser_exception.rb

@@ -0,0 +1,10 @@
+class Langchain::OutputParsers::OutputParserException < StandardError
+  def initialize(message, text)
+    @message = message
+    @text = text
+  end
+
+  def to_s
+    "#{@message}\nText: #{@text}"
+  end
+end

data/lib/langchain/tool/calculator.rb

@@ -26,13 +26,14 @@ module Langchain::Tool
     # Evaluates a pure math expression or if equation contains non-math characters (e.g.: "12F in Celsius") then it uses the google search calculator to evaluate the expression
     #
     # @param input [String] math expression
-    # @return [String] Answer
+    # @return [Langchain::Tool::Response] Answer
     def execute(input:)
       Langchain.logger.debug("#{self.class} - Executing \"#{input}\"")
 
-      Eqn::Calculator.calc(input)
+      result = Eqn::Calculator.calc(input)
+      tool_response(content: result)
     rescue Eqn::ParseError, Eqn::NoVariableValueError
-      "\"#{input}\" is an invalid mathematical expression"
+      tool_response(content: "\"#{input}\" is an invalid mathematical expression")
     end
   end
 end

data/lib/langchain/tool/database.rb

@@ -49,50 +49,53 @@ module Langchain::Tool
 
     # Database Tool: Returns a list of tables in the database
     #
-    # @return [Array<Symbol>] List of tables in the database
+    # @return [Langchain::Tool::Response] List of tables in the database
     def list_tables
-      db.tables
+      tool_response(content: db.tables)
     end
 
     # Database Tool: Returns the schema for a list of tables
     #
     # @param tables [Array<String>] The tables to describe.
-    # @return [String] The schema for the tables
+    # @return [Langchain::Tool::Response] The schema for the tables
     def describe_tables(tables: [])
       return "No tables specified" if tables.empty?
 
       Langchain.logger.debug("#{self.class} - Describing tables: #{tables}")
 
-      tables
+      result = tables
         .map do |table|
           describe_table(table)
         end
         .join("\n")
+
+      tool_response(content: result)
     end
 
     # Database Tool: Returns the database schema
     #
-    # @return [String] Database schema
+    # @return [Langchain::Tool::Response] Database schema
     def dump_schema
       Langchain.logger.debug("#{self.class} - Dumping schema tables and keys")
 
       schemas = db.tables.map do |table|
         describe_table(table)
       end
-      schemas.join("\n")
+
+      tool_response(content: schemas.join("\n"))
     end
 
     # Database Tool: Executes a SQL query and returns the results
     #
     # @param input [String] SQL query to be executed
-    # @return [Array] Results from the SQL query
+    # @return [Langchain::Tool::Response] Results from the SQL query
     def execute(input:)
       Langchain.logger.debug("#{self.class} - Executing \"#{input}\"")
 
-      db[input].to_a
+      tool_response(content: db[input].to_a)
     rescue Sequel::DatabaseError => e
       Langchain.logger.error("#{self.class} - #{e.message}")
-      e.message # Return error to LLM
+      tool_response(content: e.message)
     end
 
     private
@@ -100,7 +103,7 @@ module Langchain::Tool
     # Describes a table and its schema
     #
     # @param table [String] The table to describe
-    # @return [String] The schema for the table
+    # @return [Langchain::Tool::Response] The schema for the table
     def describe_table(table)
       # TODO: There's probably a clear way to do all of this below
 
@@ -127,6 +130,8 @@ module Langchain::Tool
         schema << ",\n" unless fk == db.foreign_key_list(table).last
       end
       schema << ");\n"
+
+      tool_response(content: schema)
     end
   end
 end

data/lib/langchain/tool/file_system.rb

@@ -24,21 +24,22 @@ module Langchain::Tool
     end
 
     def list_directory(directory_path:)
-      Dir.entries(directory_path)
+      tool_response(content: Dir.entries(directory_path))
     rescue Errno::ENOENT
-      "No such directory: #{directory_path}"
+      tool_response(content: "No such directory: #{directory_path}")
     end
 
     def read_file(file_path:)
-      File.read(file_path)
+      tool_response(content: File.read(file_path))
     rescue Errno::ENOENT
-      "No such file: #{file_path}"
+      tool_response(content: "No such file: #{file_path}")
     end
 
     def write_to_file(file_path:, content:)
       File.write(file_path, content)
+      tool_response(content: "File written successfully")
     rescue Errno::EACCES
-      "Permission denied: #{file_path}"
+      tool_response(content: "Permission denied: #{file_path}")
     end
   end
 end

data/lib/langchain/tool/google_search.rb

@@ -36,7 +36,7 @@ module Langchain::Tool
     # Executes Google Search and returns the result
     #
     # @param input [String] search query
-    # @return [String] Answer
+    # @return [Langchain::Tool::Response] Answer
     def execute(input:)
       Langchain.logger.debug("#{self.class} - Executing \"#{input}\"")
 
@@ -44,31 +44,31 @@ module Langchain::Tool
 
       answer_box = results[:answer_box_list] ? results[:answer_box_list].first : results[:answer_box]
       if answer_box
-        return answer_box[:result] ||
+        return tool_response(content: 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") }
+            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)
+        return tool_response(content: events_results.take(10))
       elsif (sports_results = results[:sports_results])
-        return sports_results
+        return tool_response(content: sports_results)
       elsif (top_stories = results[:top_stories])
-        return top_stories
+        return tool_response(content: top_stories)
       elsif (news_results = results[:news_results])
-        return news_results
+        return tool_response(content: news_results)
       elsif (jobs_results = results.dig(:jobs_results, :jobs))
-        return jobs_results
+        return tool_response(content: jobs_results)
       elsif (shopping_results = results[:shopping_results]) && shopping_results.first.key?(:title)
-        return shopping_results.take(3)
+        return tool_response(content: shopping_results.take(3))
       elsif (questions_and_answers = results[:questions_and_answers])
-        return questions_and_answers
+        return tool_response(content: questions_and_answers)
       elsif (popular_destinations = results.dig(:popular_destinations, :destinations))
-        return popular_destinations
+        return tool_response(content: popular_destinations)
       elsif (top_sights = results.dig(:top_sights, :sights))
-        return top_sights
+        return tool_response(content: top_sights)
       elsif (images_results = results[:images_results]) && images_results.first.key?(:thumbnail)
-        return images_results.map { |h| h[:thumbnail] }.take(10)
+        return tool_response(content: images_results.map { |h| h[:thumbnail] }.take(10))
       end
 
       snippets = []
@@ -110,8 +110,8 @@ module Langchain::Tool
         snippets << local_results
       end
 
-      return "No good search result found" if snippets.empty?
-      snippets
+      return tool_response(content: "No good search result found") if snippets.empty?
+      tool_response(content: snippets)
     end
 
     #

data/lib/langchain/tool/news_retriever.rb

@@ -57,7 +57,7 @@ module Langchain::Tool
     # @param page_size [Integer] The number of results to return per page. 20 is the API's default, 100 is the maximum. Our default is 5.
     # @param page [Integer] Use this to page through the results.
     #
-    # @return [String] JSON response
+    # @return [Langchain::Tool::Response] JSON response
     def get_everything(
       q: nil,
       search_in: nil,
@@ -86,7 +86,8 @@ module Langchain::Tool
       params[:pageSize] = page_size if page_size
       params[:page] = page if page
 
-      send_request(path: "everything", params: params)
+      response = send_request(path: "everything", params: params)
+      tool_response(content: response)
     end
 
     # Retrieve top headlines
@@ -98,7 +99,7 @@ module Langchain::Tool
     # @param page_size [Integer] The number of results to return per page. 20 is the API's default, 100 is the maximum. Our default is 5.
     # @param page [Integer] Use this to page through the results.
     #
-    # @return [String] JSON response
+    # @return [Langchain::Tool::Response] JSON response
     def get_top_headlines(
       country: nil,
       category: nil,
@@ -117,7 +118,8 @@ module Langchain::Tool
       params[:pageSize] = page_size if page_size
       params[:page] = page if page
 
-      send_request(path: "top-headlines", params: params)
+      response = send_request(path: "top-headlines", params: params)
+      tool_response(content: response)
     end
 
     # Retrieve news sources
@@ -126,7 +128,7 @@ module Langchain::Tool
     # @param language [String] The 2-letter ISO-639-1 code of the language you want to get headlines for. Possible options: ar, de, en, es, fr, he, it, nl, no, pt, ru, se, ud, zh.
     # @param country [String] The 2-letter ISO 3166-1 code of the country you want to get headlines for. Possible options: ae, ar, at, au, be, bg, br, ca, ch, cn, co, cu, cz, de, eg, fr, gb, gr, hk, hu, id, ie, il, in, it, jp, kr, lt, lv, ma, mx, my, ng, nl, no, nz, ph, pl, pt, ro, rs, ru, sa, se, sg, si, sk, th, tr, tw, ua, us, ve, za.
     #
-    # @return [String] JSON response
+    # @return [Langchain::Tool::Response] JSON response
     def get_sources(
       category: nil,
       language: nil,
@@ -139,7 +141,8 @@ module Langchain::Tool
       params[:category] = category if category
       params[:language] = language if language
 
-      send_request(path: "top-headlines/sources", params: params)
+      response = send_request(path: "top-headlines/sources", params: params)
+      tool_response(content: response)
     end
 
     private

data/lib/langchain/tool/ruby_code_interpreter.rb

@@ -5,7 +5,7 @@ module Langchain::Tool
   # A tool that execute Ruby code in a sandboxed environment.
   #
   # Gem requirements:
-  #     gem "safe_ruby", "~> 1.0.4"
+  #     gem "safe_ruby", "~> 1.0.5"
   #
   # Usage:
   #    interpreter = Langchain::Tool::RubyCodeInterpreter.new
@@ -27,11 +27,11 @@ module Langchain::Tool
     # Executes Ruby code in a sandboxes environment.
     #
     # @param input [String] ruby code expression
-    # @return [String] Answer
+    # @return [Langchain::Tool::Response] Answer
     def execute(input:)
       Langchain.logger.debug("#{self.class} - Executing \"#{input}\"")
 
-      safe_eval(input)
+      tool_response(content: safe_eval(input))
     end
 
     def safe_eval(code)

data/lib/langchain/tool/tavily.rb

@@ -41,7 +41,7 @@ module Langchain::Tool
     # @param include_domains [Array<String>] A list of domains to specifically include in the search results. Default is None, which includes all domains.
     # @param exclude_domains [Array<String>] A list of domains to specifically exclude from the search results. Default is None, which doesn't exclude any domains.
     #
-    # @return [String] The search results in JSON format.
+    # @return [Langchain::Tool::Response] The search results in JSON format.
     def search(
       query:,
       search_depth: "basic",
@@ -70,7 +70,7 @@ module Langchain::Tool
       response = Net::HTTP.start(uri.hostname, uri.port, use_ssl: uri.scheme == "https") do |http|
         http.request(request)
       end
-      response.body
+      tool_response(content: response.body)
     end
   end
 end

data/lib/langchain/tool/vectorsearch.rb

@@ -33,8 +33,10 @@ module Langchain::Tool
     #
     # @param query [String] The query to search for
     # @param k [Integer] The number of results to return
+    # @return [Langchain::Tool::Response] The response from the server
     def similarity_search(query:, k: 4)
-      vectorsearch.similarity_search(query:, k: 4)
+      result = vectorsearch.similarity_search(query:, k: 4)
+      tool_response(content: result)
     end
   end
 end

data/lib/langchain/tool/weather.rb

@@ -55,15 +55,15 @@ module Langchain::Tool
       params = {appid: @api_key, q: [city, state_code, country_code].compact.join(","), units: units}
 
       location_response = send_request(path: "geo/1.0/direct", params: params.except(:units))
-      return location_response if location_response.is_a?(String) # Error occurred
+      return tool_response(content: location_response) if location_response.is_a?(String) # Error occurred
 
       location = location_response.first
-      return "Location not found" unless location
+      return tool_response(content: "Location not found") unless location
 
       params = params.merge(lat: location["lat"], lon: location["lon"]).except(:q)
       weather_data = send_request(path: "data/2.5/weather", params: params)
 
-      parse_weather_response(weather_data, units)
+      tool_response(content: parse_weather_response(weather_data, units))
     end
 
     def send_request(path:, params:)

data/lib/langchain/tool/wikipedia.rb

@@ -27,13 +27,13 @@ module Langchain::Tool
     # Executes Wikipedia API search and returns the answer
     #
     # @param input [String] search query
-    # @return [String] Answer
+    # @return [Langchain::Tool::Response] Answer
     def execute(input:)
       Langchain.logger.debug("#{self.class} - Executing \"#{input}\"")
 
       page = ::Wikipedia.find(input)
       # It would be nice to figure out a way to provide page.content but the LLM token limit is an issue
-      page.summary
+      tool_response(content: page.summary)
     end
   end
 end

data/lib/langchain/tool_definition.rb

@@ -61,6 +61,20 @@ module Langchain::ToolDefinition
       .downcase
   end
 
+  def self.extended(base)
+    base.include(InstanceMethods)
+  end
+
+  module InstanceMethods
+    # Create a tool response
+    # @param content [String, nil] The content of the tool response
+    # @param image_url [String, nil] The URL of an image
+    # @return [Langchain::ToolResponse] The tool response
+    def tool_response(content: nil, image_url: nil)
+      Langchain::ToolResponse.new(content: content, image_url: image_url)
+    end
+  end
+
   # Manages schemas for functions
   class FunctionSchemas
     def initialize(tool_name)

data/lib/langchain/tool_response.rb

@@ -0,0 +1,24 @@
+# frozen_string_literal: true
+
+module Langchain
+  # ToolResponse represents the standardized output of a tool.
+  # It can contain either text content or an image URL.
+  class ToolResponse
+    attr_reader :content, :image_url
+
+    # Initializes a new ToolResponse.
+    #
+    # @param content [String] The text content of the response.
+    # @param image_url [String, nil] Optional URL to an image.
+    def initialize(content: nil, image_url: nil)
+      raise ArgumentError, "Either content or image_url must be provided" if content.nil? && image_url.nil?
+
+      @content = content
+      @image_url = image_url
+    end
+
+    def to_s
+      content.to_s
+    end
+  end
+end

data/lib/langchain/vectorsearch/base.rb

@@ -7,7 +7,6 @@ module Langchain::Vectorsearch
   # == Available vector databases
   #
   # - {Langchain::Vectorsearch::Chroma}
-  # - {Langchain::Vectorsearch::Epsilla}
   # - {Langchain::Vectorsearch::Elasticsearch}
   # - {Langchain::Vectorsearch::Hnswlib}
   # - {Langchain::Vectorsearch::Milvus}
@@ -30,7 +29,6 @@ module Langchain::Vectorsearch
   #     )
   #
   #     # You can instantiate other supported vector databases the same way:
-  #     epsilla  = Langchain::Vectorsearch::Epsilla.new(...)
   #     milvus   = Langchain::Vectorsearch::Milvus.new(...)
   #     qdrant   = Langchain::Vectorsearch::Qdrant.new(...)
   #     pinecone = Langchain::Vectorsearch::Pinecone.new(...)

data/lib/langchain/vectorsearch/elasticsearch.rb

@@ -144,7 +144,7 @@ module Langchain::Vectorsearch
     # @yield [String] Stream responses back one String at a time
     # @return [String] The answer to the question
     def ask(question:, k: 4, &block)
-      search_results = similarity_search(query: question, k: k)
+      search_results = similarity_search(text: question, k: k)
 
       context = search_results.map do |result|
         result[:input]

data/lib/langchain/vectorsearch/epsilla.rb

@@ -21,6 +21,8 @@ module Langchain::Vectorsearch
     # @param index_name [String] The name of the Epsilla table to use
     # @param llm [Object] The LLM client to use
     def initialize(url:, db_name:, db_path:, index_name:, llm:)
+      Langchain.logger.warn "DEPRECATED: `Langchain::Vectorsearch::Epsilla` is deprecated, and will be removed in the next major version. Please use other vector storage engines."
+
       depends_on "epsilla-ruby", req: "epsilla"
 
       uri = URI.parse(url)

data/lib/langchain/vectorsearch/milvus.rb

@@ -16,6 +16,7 @@ module Langchain::Vectorsearch
 
       @client = ::Milvus::Client.new(
         url: url,
+        api_key: api_key,
         logger: Langchain.logger
       )
       @index_name = index_name

data/lib/langchain/vectorsearch/pgvector.rb

@@ -51,17 +51,30 @@ module Langchain::Vectorsearch
     # Upsert a list of texts to the index
     # @param texts [Array<String>] The texts to add to the index
     # @param ids [Array<Integer>] The ids of the objects to add to the index, in the same order as the texts
+    # @param metadata [Array<Hash>] The metadata to associate with each text, 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: id, content: text, vectors: llm.embed(text: text).embedding.to_s, namespace: namespace}
+    def upsert_texts(texts:, ids:, metadata: nil)
+      metadata = Array.new(texts.size, {}) if metadata.nil?
+
+      data = texts.zip(ids, metadata).flat_map do |text, id, meta|
+        {
+          id: id,
+          content: text,
+          vectors: llm.embed(text: text).embedding.to_s,
+          namespace: namespace,
+          metadata: meta.to_json
+        }
       end
       # @db[table_name.to_sym].multi_insert(data, return: :primary_key)
       @db[table_name.to_sym]
         .insert_conflict(
           target: :id,
-          update: {content: Sequel[:excluded][:content], vectors: Sequel[:excluded][:vectors]}
+          update: {
+            content: Sequel[:excluded][:content],
+            vectors: Sequel[:excluded][:vectors],
+            metadata: Sequel[:excluded][:metadata]
+          }
         )
         .multi_insert(data, return: :primary_key)
     end
@@ -69,25 +82,34 @@ module Langchain::Vectorsearch
     # 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
+    # @param metadata [Array<Hash>] The metadata to associate with each text, in the same order as the texts
     # @return [Array<Integer>] The the ids of the added texts.
-    def add_texts(texts:, ids: nil)
+    def add_texts(texts:, ids: nil, metadata: nil)
+      metadata = Array.new(texts.size, {}) if metadata.nil?
+
       if ids.nil? || ids.empty?
-        data = texts.map do |text|
-          {content: text, vectors: llm.embed(text: text).embedding.to_s, namespace: namespace}
+        data = texts.zip(metadata).map do |text, meta|
+          {
+            content: text,
+            vectors: llm.embed(text: text).embedding.to_s,
+            namespace: namespace,
+            metadata: meta.to_json
+          }
         end
 
         @db[table_name.to_sym].multi_insert(data, return: :primary_key)
       else
-        upsert_texts(texts: texts, ids: ids)
+        upsert_texts(texts: texts, ids: ids, metadata: metadata)
       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
+    # @param metadata [Array<Hash>] The metadata to associate with each text, in the same order as the texts
     # @return [Array<Integer>] The ids of the updated texts.
-    def update_texts(texts:, ids:)
-      upsert_texts(texts: texts, ids: ids)
+    def update_texts(texts:, ids:, metadata: nil)
+      upsert_texts(texts: texts, ids: ids, metadata: metadata)
     end
 
     # Remove a list of texts from the index
@@ -107,6 +129,7 @@ module Langchain::Vectorsearch
         text :content
         column :vectors, "vector(#{vector_dimensions})"
         text namespace_column.to_sym, default: nil
+        jsonb :metadata, default: "{}"
       end
     end
 
@@ -136,6 +159,7 @@ module Langchain::Vectorsearch
     def similarity_search_by_vector(embedding:, k: 4)
       db.transaction do # BEGIN
         documents_model
+          .select(:content, :metadata)
           .nearest_neighbors(:vectors, embedding, distance: operator).limit(k)
           .where(namespace_column.to_sym => namespace)
       end

data/lib/langchain/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Langchain
-  VERSION = "0.19.3"
+  VERSION = "0.19.5"
 end

data/lib/langchain.rb

@@ -29,10 +29,6 @@ loader.inflector.inflect(
 
 loader.collapse("#{__dir__}/langchain/llm/response")
 
-# RubyCodeInterpreter does not work with Ruby 3.3;
-# https://github.com/ukutaht/safe_ruby/issues/4
-loader.ignore("#{__dir__}/langchain/tool/ruby_code_interpreter") if RUBY_VERSION >= "3.3.0"
-
 loader.setup
 
 # Langchain.rb a is library for building LLM-backed Ruby applications. It is an abstraction layer that sits on top of the emerging AI-related tools that makes it easy for developers to consume and string those services together.

metadata

@@ -1,14 +1,13 @@
 --- !ruby/object:Gem::Specification
 name: langchainrb
 version: !ruby/object:Gem::Version
-  version: 0.19.3
+  version: 0.19.5
 platform: ruby
 authors:
 - Andrei Bondarev
-autorequire:
 bindir: exe
 cert_chain: []
-date: 2025-01-13 00:00:00.000000000 Z
+date: 1980-01-02 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: baran
@@ -42,16 +41,22 @@ dependencies:
   name: json-schema
   requirement: !ruby/object:Gem::Requirement
     requirements:
-    - - "~>"
+    - - ">="
       - !ruby/object:Gem::Version
         version: '4'
+    - - "<"
+      - !ruby/object:Gem::Version
+        version: '6'
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
-    - - "~>"
+    - - ">="
       - !ruby/object:Gem::Version
         version: '4'
+    - - "<"
+      - !ruby/object:Gem::Version
+        version: '6'
 - !ruby/object:Gem::Dependency
   name: zeitwerk
   requirement: !ruby/object:Gem::Requirement
@@ -114,14 +119,14 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 3.10.0
+        version: 3.11.0
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 3.10.0
+        version: 3.11.0
 - !ruby/object:Gem::Dependency
   name: yard
   requirement: !ruby/object:Gem::Requirement
@@ -193,19 +198,19 @@ dependencies:
       - !ruby/object:Gem::Version
         version: 0.2.1
 - !ruby/object:Gem::Dependency
-  name: anthropic
+  name: ruby-anthropic
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '0.3'
+        version: '0.4'
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '0.3'
+        version: '0.4'
 - !ruby/object:Gem::Dependency
   name: aws-sdk-bedrockruntime
   requirement: !ruby/object:Gem::Requirement
@@ -276,20 +281,6 @@ dependencies:
     - - "~>"
       - !ruby/object:Gem::Version
         version: 8.2.0
-- !ruby/object:Gem::Dependency
-  name: epsilla-ruby
-  requirement: !ruby/object:Gem::Requirement
-    requirements:
-    - - "~>"
-      - !ruby/object:Gem::Version
-        version: 0.0.4
-  type: :development
-  prerelease: false
-  version_requirements: !ruby/object:Gem::Requirement
-    requirements:
-    - - "~>"
-      - !ruby/object:Gem::Version
-        version: 0.0.4
 - !ruby/object:Gem::Dependency
   name: eqn
   requirement: !ruby/object:Gem::Requirement
@@ -388,20 +379,6 @@ dependencies:
     - - "~>"
       - !ruby/object:Gem::Version
         version: 0.10.3
-- !ruby/object:Gem::Dependency
-  name: llama_cpp
-  requirement: !ruby/object:Gem::Requirement
-    requirements:
-    - - "~>"
-      - !ruby/object:Gem::Version
-        version: 0.9.4
-  type: :development
-  prerelease: false
-  version_requirements: !ruby/object:Gem::Requirement
-    requirements:
-    - - "~>"
-      - !ruby/object:Gem::Version
-        version: 0.9.4
 - !ruby/object:Gem::Dependency
   name: nokogiri
   requirement: !ruby/object:Gem::Requirement
@@ -576,14 +553,14 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 1.0.4
+        version: 1.0.5
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 1.0.4
+        version: 1.0.5
 - !ruby/object:Gem::Dependency
   name: sequel
   requirement: !ruby/object:Gem::Requirement
@@ -718,6 +695,7 @@ files:
 - lib/langchain/loader.rb
 - lib/langchain/output_parsers/base.rb
 - lib/langchain/output_parsers/output_fixing_parser.rb
+- lib/langchain/output_parsers/output_parser_exception.rb
 - lib/langchain/output_parsers/prompts/naive_fix_prompt.yaml
 - lib/langchain/output_parsers/structured_output_parser.rb
 - lib/langchain/processors/base.rb
@@ -749,6 +727,7 @@ files:
 - lib/langchain/tool/weather.rb
 - lib/langchain/tool/wikipedia.rb
 - lib/langchain/tool_definition.rb
+- lib/langchain/tool_response.rb
 - lib/langchain/utils/cosine_similarity.rb
 - lib/langchain/utils/hash_transformer.rb
 - lib/langchain/utils/image_wrapper.rb
@@ -775,7 +754,6 @@ metadata:
   source_code_uri: https://github.com/patterns-ai-core/langchainrb
   changelog_uri: https://github.com/patterns-ai-core/langchainrb/blob/main/CHANGELOG.md
   documentation_uri: https://rubydoc.info/gems/langchainrb
-post_install_message:
 rdoc_options: []
 require_paths:
 - lib
@@ -790,8 +768,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.5.3
-signing_key:
+rubygems_version: 3.6.7
 specification_version: 4
 summary: Build LLM-backed Ruby applications with Ruby's Langchain.rb
 test_files: []