llm.rb 4.15.0 → 4.16.0

data/lib/llm/active_record.rb

@@ -0,0 +1,3 @@
+# frozen_string_literal: true
+
+require "llm/active_record/acts_as_llm"

data/lib/llm/provider.rb

@@ -22,15 +22,18 @@ class LLM::Provider
   #  The number of seconds to wait for a response
   # @param [Boolean] ssl
   #  Whether to use SSL for the connection
+  # @param [String] base_path
+  #  Optional base path prefix for HTTP API routes.
   # @param [Boolean] persistent
   #  Whether to use a persistent connection.
   #  Requires the net-http-persistent gem.
-  def initialize(key:, host:, port: 443, timeout: 60, ssl: true, persistent: false)
+  def initialize(key:, host:, port: 443, timeout: 60, ssl: true, base_path: "", persistent: false)
     @key = key
     @host = host
     @port = port
     @timeout = timeout
     @ssl = ssl
+    @base_path = normalize_base_path(base_path)
     @base_uri = URI("#{ssl ? "https" : "http"}://#{host}:#{port}/")
     @headers = {"User-Agent" => "llm.rb v#{LLM::VERSION}"}
     @transport = Transport::HTTP.new(host:, port:, timeout:, ssl:, persistent:)
@@ -330,6 +333,18 @@ class LLM::Provider
 
   private
 
+  def path(suffix)
+    return suffix if @base_path.empty?
+    "#{@base_path}#{suffix}"
+  end
+
+  def normalize_base_path(path)
+    path = path.to_s.strip
+    return "" if path.empty? || path == "/"
+    path = "/#{path}" unless path.start_with?("/")
+    path.sub(%r{/+\z}, "")
+  end
+
   attr_reader :base_uri, :host, :port, :timeout, :ssl, :transport
 
   ##

data/lib/llm/providers/openai/audio.rb

@@ -32,7 +32,7 @@ class LLM::OpenAI
     # @raise (see LLM::Provider#request)
     # @return [LLM::Response]
     def create_speech(input:, voice: "alloy", model: "gpt-4o-mini-tts", response_format: "mp3", **params)
-      req = Net::HTTP::Post.new("/v1/audio/speech", headers)
+      req = Net::HTTP::Post.new(path("/audio/speech"), headers)
       req.body = LLM.json.dump({input:, voice:, model:, response_format:}.merge!(params))
       io = StringIO.new("".b)
       res, span, tracer = execute(request: req, operation: "request") { _1.read_body { |chunk| io << chunk } }
@@ -55,7 +55,7 @@ class LLM::OpenAI
     # @return [LLM::Response]
     def create_transcription(file:, model: "whisper-1", **params)
       multi = LLM::Multipart.new(params.merge!(file: LLM.File(file), model:))
-      req = Net::HTTP::Post.new("/v1/audio/transcriptions", headers)
+      req = Net::HTTP::Post.new(path("/audio/transcriptions"), headers)
       req["content-type"] = multi.content_type
       set_body_stream(req, multi.body)
       res, span, tracer = execute(request: req, operation: "request")
@@ -79,7 +79,7 @@ class LLM::OpenAI
     # @return [LLM::Response]
     def create_translation(file:, model: "whisper-1", **params)
       multi = LLM::Multipart.new(params.merge!(file: LLM.File(file), model:))
-      req = Net::HTTP::Post.new("/v1/audio/translations", headers)
+      req = Net::HTTP::Post.new(path("/audio/translations"), headers)
       req["content-type"] = multi.content_type
       set_body_stream(req, multi.body)
       res, span, tracer = execute(request: req, operation: "request")
@@ -90,7 +90,7 @@ class LLM::OpenAI
 
     private
 
-    [:headers, :execute, :set_body_stream].each do |m|
+    [:path, :headers, :execute, :set_body_stream].each do |m|
       define_method(m) { |*args, **kwargs, &b| @provider.send(m, *args, **kwargs, &b) }
     end
   end

data/lib/llm/providers/openai/files.rb

@@ -40,7 +40,7 @@ class LLM::OpenAI
     # @return [LLM::Response]
     def all(**params)
       query = URI.encode_www_form(params)
-      req = Net::HTTP::Get.new("/v1/files?#{query}", headers)
+      req = Net::HTTP::Get.new(path("/files?#{query}"), headers)
       res, span, tracer = execute(request: req, operation: "request")
       res = ResponseAdapter.adapt(res, type: :enumerable)
       tracer.on_request_finish(operation: "request", res:, span:)
@@ -60,7 +60,7 @@ class LLM::OpenAI
     # @return [LLM::Response]
     def create(file:, purpose: "assistants", **params)
       multi = LLM::Multipart.new(params.merge!(file: LLM.File(file), purpose:))
-      req = Net::HTTP::Post.new("/v1/files", headers)
+      req = Net::HTTP::Post.new(path("/files"), headers)
       req["content-type"] = multi.content_type
       set_body_stream(req, multi.body)
       res, span, tracer = execute(request: req, operation: "request")
@@ -83,7 +83,7 @@ class LLM::OpenAI
     def get(file:, **params)
       file_id = file.respond_to?(:id) ? file.id : file
       query = URI.encode_www_form(params)
-      req = Net::HTTP::Get.new("/v1/files/#{file_id}?#{query}", headers)
+      req = Net::HTTP::Get.new(path("/files/#{file_id}?#{query}"), headers)
       res, span, tracer = execute(request: req, operation: "request")
       res = ResponseAdapter.adapt(res, type: :file)
       tracer.on_request_finish(operation: "request", res:, span:)
@@ -105,7 +105,7 @@ class LLM::OpenAI
     def download(file:, **params)
       query = URI.encode_www_form(params)
       file_id = file.respond_to?(:id) ? file.id : file
-      req = Net::HTTP::Get.new("/v1/files/#{file_id}/content?#{query}", headers)
+      req = Net::HTTP::Get.new(path("/files/#{file_id}/content?#{query}"), headers)
       io = StringIO.new("".b)
       res, span, tracer = execute(request: req, operation: "request") { |res| res.read_body { |chunk| io << chunk } }
       res = LLM::Response.new(res).tap { _1.define_singleton_method(:file) { io } }
@@ -125,7 +125,7 @@ class LLM::OpenAI
     # @return [LLM::Response]
     def delete(file:)
       file_id = file.respond_to?(:id) ? file.id : file
-      req = Net::HTTP::Delete.new("/v1/files/#{file_id}", headers)
+      req = Net::HTTP::Delete.new(path("/files/#{file_id}"), headers)
       res, span, tracer = execute(request: req, operation: "request")
       res = LLM::Response.new(res)
       tracer.on_request_finish(operation: "request", res:, span:)
@@ -134,7 +134,7 @@ class LLM::OpenAI
 
     private
 
-    [:headers, :execute, :set_body_stream].each do |m|
+    [:path, :headers, :execute, :set_body_stream].each do |m|
       define_method(m) { |*args, **kwargs, &b| @provider.send(m, *args, **kwargs, &b) }
     end
   end

data/lib/llm/providers/openai/images.rb

@@ -50,7 +50,7 @@ class LLM::OpenAI
     # @raise (see LLM::Provider#request)
     # @return [LLM::Response]
     def create(prompt:, model: "dall-e-3", response_format: "b64_json", **params)
-      req = Net::HTTP::Post.new("/v1/images/generations", headers)
+      req = Net::HTTP::Post.new(path("/images/generations"), headers)
       req.body = LLM.json.dump({prompt:, n: 1, model:, response_format:}.merge!(params))
       res, span, tracer = execute(request: req, operation: "request")
       res = ResponseAdapter.adapt(res, type: :image)
@@ -76,7 +76,7 @@ class LLM::OpenAI
     def create_variation(image:, model: "dall-e-2", response_format: "b64_json", **params)
       image = LLM.File(image)
       multi = LLM::Multipart.new(params.merge!(image:, model:, response_format:))
-      req = Net::HTTP::Post.new("/v1/images/variations", headers)
+      req = Net::HTTP::Post.new(path("/images/variations"), headers)
       req["content-type"] = multi.content_type
       set_body_stream(req, multi.body)
       res, span, tracer = execute(request: req, operation: "request")
@@ -102,7 +102,7 @@ class LLM::OpenAI
     def edit(image:, prompt:, model: "dall-e-2", response_format: "b64_json", **params)
       image = LLM.File(image)
       multi = LLM::Multipart.new(params.merge!(image:, prompt:, model:, response_format:))
-      req = Net::HTTP::Post.new("/v1/images/edits", headers)
+      req = Net::HTTP::Post.new(path("/images/edits"), headers)
       req["content-type"] = multi.content_type
       set_body_stream(req, multi.body)
       res, span, tracer = execute(request: req, operation: "request")
@@ -113,7 +113,7 @@ class LLM::OpenAI
 
     private
 
-    [:headers, :execute, :set_body_stream].each do |m|
+    [:path, :headers, :execute, :set_body_stream].each do |m|
       define_method(m) { |*args, **kwargs, &b| @provider.send(m, *args, **kwargs, &b) }
     end
   end

data/lib/llm/providers/openai/models.rb

@@ -39,7 +39,7 @@ class LLM::OpenAI
     # @return [LLM::Response]
     def all(**params)
       query = URI.encode_www_form(params)
-      req = Net::HTTP::Get.new("/v1/models?#{query}", headers)
+      req = Net::HTTP::Get.new(path("/models?#{query}"), headers)
       res, span, tracer = execute(request: req, operation: "request")
       res = ResponseAdapter.adapt(res, type: :models)
       tracer.on_request_finish(operation: "request", res:, span:)
@@ -48,7 +48,7 @@ class LLM::OpenAI
 
     private
 
-    [:headers, :execute, :set_body_stream].each do |m|
+    [:path, :headers, :execute, :set_body_stream].each do |m|
       define_method(m) { |*args, **kwargs, &b| @provider.send(m, *args, **kwargs, &b) }
     end
   end

data/lib/llm/providers/openai/moderations.rb

@@ -47,7 +47,7 @@ class LLM::OpenAI
     # @param [String, LLM::Model] model The model to use
     # @return [LLM::Response]
     def create(input:, model: "omni-moderation-latest", **params)
-      req = Net::HTTP::Post.new("/v1/moderations", headers)
+      req = Net::HTTP::Post.new(path("/moderations"), headers)
       input = RequestAdapter::Moderation.new(input).adapt
       req.body = LLM.json.dump({input:, model:}.merge!(params))
       res, span, tracer = execute(request: req, operation: "request")
@@ -58,7 +58,7 @@ class LLM::OpenAI
 
     private
 
-    [:headers, :execute].each do |m|
+    [:path, :headers, :execute].each do |m|
       define_method(m) { |*args, **kwargs, &b| @provider.send(m, *args, **kwargs, &b) }
     end
   end

data/lib/llm/providers/openai/responses.rb

@@ -40,7 +40,7 @@ class LLM::OpenAI
       params = [params, adapt_schema(params), adapt_tools(tools)].inject({}, &:merge!).compact
       role, stream = params.delete(:role), params.delete(:stream)
       params[:stream] = true if @provider.streamable?(stream) || stream == true
-      req = Net::HTTP::Post.new("/v1/responses", headers)
+      req = Net::HTTP::Post.new(path("/responses"), headers)
       messages = build_complete_messages(prompt, params, role)
       @provider.tracer.set_request_metadata(user_input: extract_user_input(messages, fallback: prompt))
       body = LLM.json.dump({input: [adapt(messages, mode: :response)].flatten}.merge!(params))
@@ -61,7 +61,7 @@ class LLM::OpenAI
     def get(response, **params)
       response_id = response.respond_to?(:id) ? response.id : response
       query = URI.encode_www_form(params)
-      req = Net::HTTP::Get.new("/v1/responses/#{response_id}?#{query}", headers)
+      req = Net::HTTP::Get.new(path("/responses/#{response_id}?#{query}"), headers)
       res, span, tracer = execute(request: req, operation: "request")
       res = ResponseAdapter.adapt(res, type: :responds)
       tracer.on_request_finish(operation: "request", res:, span:)
@@ -76,7 +76,7 @@ class LLM::OpenAI
     # @return [LLM::Object] Response body
     def delete(response)
       response_id = response.respond_to?(:id) ? response.id : response
-      req = Net::HTTP::Delete.new("/v1/responses/#{response_id}", headers)
+      req = Net::HTTP::Delete.new(path("/responses/#{response_id}"), headers)
       res, span, tracer = execute(request: req, operation: "request")
       res = LLM::Response.new(res)
       tracer.on_request_finish(operation: "request", res:, span:)
@@ -85,7 +85,7 @@ class LLM::OpenAI
 
     private
 
-    [:headers, :execute, :set_body_stream, :resolve_tools].each do |m|
+    [:path, :headers, :execute, :set_body_stream, :resolve_tools].each do |m|
       define_method(m) { |*args, **kwargs, &b| @provider.send(m, *args, **kwargs, &b) }
     end
 

data/lib/llm/providers/openai/vector_stores.rb

@@ -31,7 +31,7 @@ class LLM::OpenAI
     # @return [LLM::Response]
     def all(**params)
       query = URI.encode_www_form(params)
-      req = Net::HTTP::Get.new("/v1/vector_stores?#{query}", headers)
+      req = Net::HTTP::Get.new(path("/vector_stores?#{query}"), headers)
       res, span, tracer = execute(request: req, operation: "request")
       res = ResponseAdapter.adapt(res, type: :enumerable)
       tracer.on_request_finish(operation: "request", res:, span:)
@@ -47,7 +47,7 @@ class LLM::OpenAI
     # @return [LLM::Response]
     # @see https://platform.openai.com/docs/api-reference/vector_stores/create OpenAI docs
     def create(name:, file_ids: nil, **params)
-      req = Net::HTTP::Post.new("/v1/vector_stores", headers)
+      req = Net::HTTP::Post.new(path("/vector_stores"), headers)
       req.body = LLM.json.dump(params.merge({name:, file_ids:}).compact)
       res, span, tracer = execute(request: req, operation: "request")
       res = LLM::Response.new(res)
@@ -72,7 +72,7 @@ class LLM::OpenAI
     # @see https://platform.openai.com/docs/api-reference/vector_stores/retrieve OpenAI docs
     def get(vector:)
       vector_id = vector.respond_to?(:id) ? vector.id : vector
-      req = Net::HTTP::Get.new("/v1/vector_stores/#{vector_id}", headers)
+      req = Net::HTTP::Get.new(path("/vector_stores/#{vector_id}"), headers)
       res, span, tracer = execute(request: req, operation: "request")
       res = LLM::Response.new(res)
       tracer.on_request_finish(operation: "request", res:, span:)
@@ -89,7 +89,7 @@ class LLM::OpenAI
     # @see https://platform.openai.com/docs/api-reference/vector_stores/modify OpenAI docs
     def modify(vector:, name: nil, **params)
       vector_id = vector.respond_to?(:id) ? vector.id : vector
-      req = Net::HTTP::Post.new("/v1/vector_stores/#{vector_id}", headers)
+      req = Net::HTTP::Post.new(path("/vector_stores/#{vector_id}"), headers)
       req.body = LLM.json.dump(params.merge({name:}).compact)
       res, span, tracer = execute(request: req, operation: "request")
       res = LLM::Response.new(res)
@@ -105,7 +105,7 @@ class LLM::OpenAI
     # @see https://platform.openai.com/docs/api-reference/vector_stores/delete OpenAI docs
     def delete(vector:)
       vector_id = vector.respond_to?(:id) ? vector.id : vector
-      req = Net::HTTP::Delete.new("/v1/vector_stores/#{vector_id}", headers)
+      req = Net::HTTP::Delete.new(path("/vector_stores/#{vector_id}"), headers)
       res, span, tracer = execute(request: req, operation: "request")
       res = LLM::Response.new(res)
       tracer.on_request_finish(operation: "request", res:, span:)
@@ -122,7 +122,7 @@ class LLM::OpenAI
     # @see https://platform.openai.com/docs/api-reference/vector_stores/search OpenAI docs
     def search(vector:, query:, **params)
       vector_id = vector.respond_to?(:id) ? vector.id : vector
-      req = Net::HTTP::Post.new("/v1/vector_stores/#{vector_id}/search", headers)
+      req = Net::HTTP::Post.new(path("/vector_stores/#{vector_id}/search"), headers)
       req.body = LLM.json.dump(params.merge({query:}).compact)
       res, span, tracer = execute(request: req, operation: "retrieval")
       res = ResponseAdapter.adapt(res, type: :enumerable)
@@ -140,7 +140,7 @@ class LLM::OpenAI
     def all_files(vector:, **params)
       vector_id = vector.respond_to?(:id) ? vector.id : vector
       query = URI.encode_www_form(params)
-      req = Net::HTTP::Get.new("/v1/vector_stores/#{vector_id}/files?#{query}", headers)
+      req = Net::HTTP::Get.new(path("/vector_stores/#{vector_id}/files?#{query}"), headers)
       res, span, tracer = execute(request: req, operation: "request")
       res = ResponseAdapter.adapt(res, type: :enumerable)
       tracer.on_request_finish(operation: "request", res:, span:)
@@ -159,7 +159,7 @@ class LLM::OpenAI
     def add_file(vector:, file:, attributes: nil, **params)
       vector_id = vector.respond_to?(:id) ? vector.id : vector
       file_id = file.respond_to?(:id) ? file.id : file
-      req = Net::HTTP::Post.new("/v1/vector_stores/#{vector_id}/files", headers)
+      req = Net::HTTP::Post.new(path("/vector_stores/#{vector_id}/files"), headers)
       req.body = LLM.json.dump(params.merge({file_id:, attributes:}).compact)
       res, span, tracer = execute(request: req, operation: "request")
       res = LLM::Response.new(res)
@@ -190,7 +190,7 @@ class LLM::OpenAI
     def update_file(vector:, file:, attributes:, **params)
       vector_id = vector.respond_to?(:id) ? vector.id : vector
       file_id = file.respond_to?(:id) ? file.id : file
-      req = Net::HTTP::Post.new("/v1/vector_stores/#{vector_id}/files/#{file_id}", headers)
+      req = Net::HTTP::Post.new(path("/vector_stores/#{vector_id}/files/#{file_id}"), headers)
       req.body = LLM.json.dump(params.merge({attributes:}).compact)
       res, span, tracer = execute(request: req, operation: "request")
       res = LLM::Response.new(res)
@@ -209,7 +209,7 @@ class LLM::OpenAI
       vector_id = vector.respond_to?(:id) ? vector.id : vector
       file_id = file.respond_to?(:id) ? file.id : file
       query = URI.encode_www_form(params)
-      req = Net::HTTP::Get.new("/v1/vector_stores/#{vector_id}/files/#{file_id}?#{query}", headers)
+      req = Net::HTTP::Get.new(path("/vector_stores/#{vector_id}/files/#{file_id}?#{query}"), headers)
       res, span, tracer = execute(request: req, operation: "request")
       res = LLM::Response.new(res)
       tracer.on_request_finish(operation: "request", res:, span:)
@@ -226,7 +226,7 @@ class LLM::OpenAI
     def delete_file(vector:, file:)
       vector_id = vector.respond_to?(:id) ? vector.id : vector
       file_id = file.respond_to?(:id) ? file.id : file
-      req = Net::HTTP::Delete.new("/v1/vector_stores/#{vector_id}/files/#{file_id}", headers)
+      req = Net::HTTP::Delete.new(path("/vector_stores/#{vector_id}/files/#{file_id}"), headers)
       res, span, tracer = execute(request: req, operation: "request")
       res = LLM::Response.new(res)
       tracer.on_request_finish(operation: "request", res:, span:)
@@ -259,7 +259,7 @@ class LLM::OpenAI
 
     private
 
-    [:headers, :execute, :set_body_stream].each do |m|
+    [:path, :headers, :execute, :set_body_stream].each do |m|
       define_method(m) { |*args, **kwargs, &b| @provider.send(m, *args, **kwargs, &b) }
     end
   end

data/lib/llm/providers/openai.rb

@@ -32,8 +32,8 @@ module LLM
 
     ##
     # @param key (see LLM::Provider#initialize)
-    def initialize(**)
-      super(host: HOST, **)
+    def initialize(base_path: "/v1", **)
+      super(host: HOST, base_path:, **)
     end
 
     ##
@@ -52,7 +52,7 @@ module LLM
     # @raise (see LLM::Provider#request)
     # @return (see LLM::Provider#embed)
     def embed(input, model: "text-embedding-3-small", **params)
-      req = Net::HTTP::Post.new("/v1/embeddings", headers)
+      req = Net::HTTP::Post.new(path("/embeddings"), headers)
       req.body = LLM.json.dump({input:, model:}.merge!(params))
       res, span, tracer = execute(request: req, operation: "embeddings", model:)
       res = ResponseAdapter.adapt(res, type: :embedding)
@@ -187,7 +187,7 @@ module LLM
     private
 
     def completions_path
-      "/v1/chat/completions"
+      path("/chat/completions")
     end
 
     def headers

data/lib/llm/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module LLM
-  VERSION = "4.15.0"
+  VERSION = "4.16.0"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: llm.rb
 version: !ruby/object:Gem::Version
-  version: 4.15.0
+  version: 4.16.0
 platform: ruby
 authors:
 - Antar Azri
@@ -228,6 +228,8 @@ files:
 - data/xai.json
 - data/zai.json
 - lib/llm.rb
+- lib/llm/active_record.rb
+- lib/llm/active_record/acts_as_llm.rb
 - lib/llm/agent.rb
 - lib/llm/bot.rb
 - lib/llm/buffer.rb