omniai 1.1.2 → 1.1.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: facf65abfc03bc2fedc0572bbdebe5baca776da08b3eba9f4bde4b082bf037a7
-  data.tar.gz: 265613b4ab126f34d68d079241d8f4d2eae96c3e5c4f083aa9ff2761003a5678
+  metadata.gz: a82c1b8f8724f5dd7f35239ca2454e4ec7631fa1b8b0544249a49f8b6c52962f
+  data.tar.gz: 85e9a9ee74f30ad4c4ba6ef8fbbaa88dc0d31fea6927774d5c7a421e2d38e341
 SHA512:
-  metadata.gz: a689d6591eabb97d47fea58a73c2b7fcc133c62aea51180bffe846976086b3819099708757cae85e20c5db3e38964b9dc7c0dc1cbc9fa8b076bc322a56dec4cb
-  data.tar.gz: bc80184100a4ca2888fb8a0ee3b93ac8d9d301e875e541e692b9a08564eb85fa59d58fcd88e35c4d72142624621ef73174b276c8da95ae53bfcabd1fc1ae776d
+  metadata.gz: 42c6324488864562480cf898c8e087ac841d035468a0f4b4dbcd7c86f4239d61ca8361c80e8dcd06ddb9dc565c2e04faa047c2e1342bacf1ac8d2d3825d85b25
+  data.tar.gz: cbeef3a03122130ce8c2bf634700df52a2a44aa1c013e8d0f2862379940965326f693f4e87f2704d0dd0a197cddf136cc5cd440d4c36a5a573ba00910845d044

data/README.md

@@ -1,6 +1,8 @@
 # OmniAI
 
-OmniAI is a flexible AI library that standardizes the APIs for multipe AI providers:
+[![CircleCI](https://circleci.com/gh/ksylvest/omniai.svg?style=svg)](https://circleci.com/gh/ksylvest/omniai)
+
+OmniAI is a flexible AI library that standardizes the APIs for multiple AI providers:
 
 - [OmniAI::Anthropic](https://github.com/ksylvest/omniai-anthropic)
 - [OmniAI::Google](https://github.com/ksylvest/omniai-google)
@@ -29,7 +31,7 @@ OmniAI implements APIs for a number of popular clients by default. A client can
 #### [OmniAI::Anthropic](https://github.com/ksylvest/omniai-anthropic)
 
 ```ruby
-require 'omniai-anthropic'
+require 'omniai/anthropic'
 
 client = OmniAI::Anthropic::Client.new
 ```
@@ -37,7 +39,7 @@ client = OmniAI::Anthropic::Client.new
 #### [OmniAI::Google](https://github.com/ksylvest/omniai-google)
 
 ```ruby
-require 'omniai-google'
+require 'omniai/google'
 
 client = OmniAI::Google::Client.new
 ```
@@ -45,7 +47,7 @@ client = OmniAI::Google::Client.new
 #### [OmniAI::Mistral](https://github.com/ksylvest/omniai-mistral)
 
 ```ruby
-require 'omniai-mistral'
+require 'omniai/mistral'
 
 client = OmniAI::Mistral::Client.new
 ```
@@ -53,7 +55,7 @@ client = OmniAI::Mistral::Client.new
 #### [OmniAI::OpenAI](https://github.com/ksylvest/omniai-openai)
 
 ```ruby
-require 'omniai-openai'
+require 'omniai/openai'
 
 client = OmniAI::OpenAI::Client.new
 ```
@@ -62,19 +64,19 @@ client = OmniAI::OpenAI::Client.new
 
 Clients that support chat (e.g. Anthropic w/ "Claude", Google w/ "Gemini", Mistral w/ "LeChat", OpenAI w/ "ChatGPT", etc) generate completions using the following calls:
 
-#### w/ a Simple Prompt
+#### Completions using Single Message
 
 ```ruby
 completion = client.chat('Tell me a joke.')
 completion.choice.message.content # '...'
 ```
 
-#### w/ a Collection of Messages
+#### Completions using Multiple Messages
 
 ```ruby
 messages = [
   {
-    role: 'system',
+    role: OmniAI::Chat::Role::SYSTEM,
     content: 'You are a helpful assistant with an expertise in geography.',
   },
   'What is the capital of Canada?'
@@ -83,40 +85,53 @@ completion = client.chat(messages, model: '...', temperature: 0.7, format: :json
 completion.choice.message.content  # '...'
 ```
 
-#### w/ a Collection of Files
+#### Completions using Real-Time Streaming
 
 ```ruby
+stream = proc do |chunk|
+  print(chunk.choice.delta.content) # '...'
+end
+client.chat('Tell me a joke.', stream:)
+```
 
-image_a_url = "https://images.unsplash.com/photo-1517849845537-4d257902454a?w=800&h=800&format=jpeg&fit=crop"
-image_b_url = "https://images.unsplash.com/photo-1537151625747-768eb6cf92b2?q=80&w=1024&h=1024&format=jpeg"
+### Transcribe
 
-message = {
-  role: 'user',
-  content: [
-    OmniAI::Chat::Content.new('What are in these images and are they different?'),
-    OmniAI::Chat::Content.new(image_a_url, type: :image),
-    OmniAI::Chat::Content.new(image_b_url, type: :image),
-  ]
-}
+Clients that support transcribe (e.g. OpenAI w/ "Whisper") convert recordings to text via the following calls:
 
-completion = client.chat(message)
-completion.choice.message.content  # '...'
+#### Transcriptions with Path
+
+```ruby
+transcription = client.transcribe("example.ogg")
+transcription.text # '...'
 ```
 
-#### Streaming
+#### Transcriptions with Files
 
 ```ruby
-stream = proc do |chunk|
-  print(chunk.choice.delta.content) # '...'
+File.open("example.ogg", "rb") do |file|
+  transcription = client.transcribe(file)
+  transcription.text # '...'
 end
-client.chat('Tell me a joke.', stream:)
 ```
 
-### Transcribe
+### Speak
+
+Clients that support speak (e.g. OpenAI w/ "Whisper") convert text to recordings via the following calls:
 
-Clients that support chat (e.g. OpenAI w/ "Whisper", etc) convert recordings to text via the following calls:
+#### Speech with Stream
 
 ```ruby
-transcription = client.transcribe(file.path)
-transcription.text # '...'
+File.open('example.ogg', 'wb') do |file|
+  client.speak('The quick brown fox jumps over a lazy dog.', voice: 'HAL') do |chunk|
+    file << chunk
+  end
+end
+```
+
+#### Speech with File
+
+```ruby
+tempfile = client.speak('The quick brown fox jumps over a lazy dog.', voice: 'HAL')
+tempfile.close
+tempfile.unlink
 ```

data/lib/omniai/chat/choice.rb

@@ -16,7 +16,9 @@ module OmniAI
         new(index:, delta:, message:)
       end
 
-      # @param data [Hash]
+      # @param index [Integer]
+      # @param delta [OmniAI::Chat::Delta] optional
+      # @param message [OmniAI::Chat::Message] optional
       def initialize(index:, delta:, message:)
         @index = index
         @delta = delta

data/lib/omniai/chat/chunk.rb

@@ -36,7 +36,7 @@ module OmniAI
         @choices ||= @data['choices'].map { |data| Choice.for(data:) }
       end
 
-      # @param [index] [Integer]
+      # @param index [Integer]
       # @return [OmniAI::Chat::Delta]
       def choice(index: 0)
         choices[index]

data/lib/omniai/chat/completion.rb

@@ -43,7 +43,7 @@ module OmniAI
         @choices ||= @data['choices'].map { |data| Choice.for(data:) }
       end
 
-      # @param [index] [Integer] optional - default is 0
+      # @param index [Integer] optional - default is 0
       # @return [OmniAI::Chat::Choice]
       def choice(index: 0)
         choices[index]

data/lib/omniai/chat/content.rb

@@ -6,8 +6,7 @@ module OmniAI
     class Content
       attr_accessor :type, :value
 
-      # @param url [String]
-      # @param text [String]
+      # @param value [String]
       # @param type [Symbol] :image / :video / :audio / :text
       def initialize(value, type: :text)
         @value = value

data/lib/omniai/chat/stream.rb

@@ -5,7 +5,6 @@ module OmniAI
     # A stream given when streaming.
     class Stream
       # @param response [HTTP::Response]
-      # @param block [Proc]
       def initialize(response:)
         @response = response
         @parser = EventStreamParser::Parser.new

data/lib/omniai/chat.rb

@@ -56,7 +56,7 @@ module OmniAI
       @format = format
     end
 
-    # @raise [ExecutionError]
+    # @raise [HTTPError]
     def process!
       response = request!
       raise HTTPError, response.flush unless response.status.ok?

data/lib/omniai/client.rb

@@ -48,7 +48,7 @@ module OmniAI
 
     # @raise [OmniAI::Error]
     #
-    # @param file [IO]
+    # @param io [String, Pathname, IO] required
     # @param model [String]
     # @param language [String, nil] optional
     # @param prompt [String, nil] optional
@@ -56,7 +56,28 @@ module OmniAI
     # @param format [Symbol] :text, :srt, :vtt, or :json (default)
     #
     # @return text [OmniAI::Transcribe::Transcription]
-    def transcribe(file, model:, language: nil, prompt: nil, temperature: nil, format: nil)
+    def transcribe(io, model:, language: nil, prompt: nil, temperature: nil, format: nil)
+      raise NotImplementedError, "#{self.class.name}#speak undefined"
+    end
+
+    # @raise [OmniAI::Error]
+    #
+    # @param input [String] required
+    # @param model [String] required
+    # @param voice [String] required
+    # @param speed [Float] optional
+    # @param format [String] optional (default "aac"):
+    #   - "aac"
+    #   - "mp3"
+    #   - "flac"
+    #   - "opus"
+    #   - "pcm"
+    #   - "wav"
+    #
+    # @yield [output] optional
+    #
+    # @return [Tempfile``]
+    def speak(input, model:, voice:, speed: nil, format: nil, &stream)
       raise NotImplementedError, "#{self.class.name}#speak undefined"
     end
   end

data/lib/omniai/speak.rb

@@ -0,0 +1,138 @@
+# frozen_string_literal: true
+
+module OmniAI
+  # An abstract class that provides a consistent interface for processing speak requests.
+  #
+  # Usage:
+  #
+  #   class OmniAI::OpenAI::Speak < OmniAI::Speakw
+  #     module Model
+  #       WHISPER_1 = "whisper-1"
+  #     end
+  #
+  #     protected
+  #
+  #     # @return [Hash]
+  #     def payload
+  #       raise NotImplementedError, "#{self.class.name}#payload undefined"
+  #     end
+  #
+  #     # @return [String]
+  #     def path
+  #       raise NotImplementedError, "#{self.class.name}#path undefined"
+  #     end
+  #   end
+  #
+  #   client.transcribe(File.open("..."), model: "...", format: :json)
+  class Speak
+    module Format
+      AAC = 'aac'
+      FLAC = 'flac'
+      MP3 = 'mp3'
+      OPUS = 'opus'
+      PCM = 'pcm'
+      WAV = 'wav'
+    end
+
+    # @raise [HTTPError]
+    #
+    # @param client [OmniAI::Client] required
+    # @param input [String] required
+    # @param model [String] required
+    # @param voice [String] required
+    # @param speed [Float] optional
+    # @param format [String] optional (default "aac"):
+    #   - "aac"
+    #   - "mp3"
+    #   - "flac"
+    #   - "opus"
+    #   - "pcm"
+    #   - "wav"
+    #
+    # @yield [chunk]
+    #
+    # @return [Tempfile]
+    def self.process!(input, client:, model:, voice:, speed: nil, format: nil, &)
+      new(input, client:, model:, voice:, speed:, format:).process!(&)
+    end
+
+    # @param client [OmniAI::Client] required
+    # @param input [String] required
+    # @param model [String] required
+    # @param voice [String] required
+    # @param speed [Float] optional
+    # @param format [String] optional (default "aac"):
+    #   - "aac"
+    #   - "mp3"
+    #   - "flac"
+    #   - "opus"
+    #   - "pcm"
+    #   - "wav"
+    def initialize(input, client:, model:, voice:, speed: nil, format: nil)
+      @input = input
+      @client = client
+      @model = model
+      @voice = voice
+      @speed = speed
+      @format = format
+    end
+
+    # @raise [HTTPError]
+    #
+    # @yield [chunk]
+    #
+    # @return [Tempfile]
+    def process!(&block)
+      response = request!
+      raise HTTPError, response.flush unless response.status.ok?
+
+      if block
+        stream!(response:, &block)
+      else
+        fetch!(response:)
+      end
+    end
+
+    protected
+
+    # @param response [HTTP::Response]
+    #
+    # @yield [chunk]
+    def stream!(response:, &block)
+      response.body.each { |chunk| block.call(chunk) }
+    end
+
+    # @param response [HTTP::Response]
+    #
+    # @return [Tempfile]
+    def fetch!(response:)
+      tempfile = Tempfile.new(['', ".#{@format}"])
+      tempfile.binmode
+      response.body.each { |chunk| tempfile << chunk }
+      tempfile.rewind
+      tempfile
+    end
+
+    # @return [Hash]
+    def payload
+      {
+        model: @model,
+        voice: @voice,
+        input: @input,
+        speed: @speed,
+      }.compact
+    end
+
+    # @return [String]
+    def path
+      raise NotImplementedError, "#{self.class.name}#path undefined"
+    end
+
+    # @return [HTTP::Response]
+    def request!
+      @client
+        .connection
+        .post(path, json: payload)
+    end
+  end
+end

data/lib/omniai/transcribe.rb

@@ -96,15 +96,15 @@ module OmniAI
       new(...).process!
     end
 
-    # @param path [String] required
+    # @param io [String, Pathname, IO] required
     # @param client [OmniAI::Client] the client
     # @param model [String] required
     # @param language [String, nil] optional
     # @param prompt [String, nil] optional
     # @param temperature [Float, nil] optional
     # @param format [String, nil] optional
-    def initialize(path, client:, model:, language: nil, prompt: nil, temperature: nil, format: Format::JSON)
-      @path = path
+    def initialize(io, client:, model:, language: nil, prompt: nil, temperature: nil, format: Format::JSON)
+      @io = io
       @model = model
       @language = language
       @prompt = prompt
@@ -113,8 +113,8 @@ module OmniAI
       @client = client
     end
 
+    # @raise [HTTPError]
     # @return [OmniAI::Transcribe::Transcription]
-    # @raise [ExecutionError]
     def process!
       response = request!
       raise HTTPError, response.flush unless response.status.ok?
@@ -128,7 +128,7 @@ module OmniAI
     # @return [Hash]
     def payload
       {
-        file: HTTP::FormData::File.new(@path),
+        file: HTTP::FormData::File.new(@io),
         model: @model,
         language: @language,
         prompt: @prompt,

data/lib/omniai/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module OmniAI
-  VERSION = '1.1.2'
+  VERSION = '1.1.5'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: omniai
 version: !ruby/object:Gem::Version
-  version: 1.1.2
+  version: 1.1.5
 platform: ruby
 authors:
 - Kevin Sylvestre
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2024-06-17 00:00:00.000000000 Z
+date: 2024-06-20 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: event_stream_parser
@@ -76,11 +76,13 @@ files:
 - lib/omniai/chat/usage.rb
 - lib/omniai/client.rb
 - lib/omniai/config.rb
+- lib/omniai/speak.rb
 - lib/omniai/transcribe.rb
 - lib/omniai/transcribe/transcription.rb
 - lib/omniai/version.rb
 homepage: https://github.com/ksylvest/omniai
-licenses: []
+licenses:
+- MIT
 metadata:
   homepage_uri: https://github.com/ksylvest/omniai
   changelog_uri: https://github.com/ksylvest/omniai/releases