Class: Intelligence::ChatRequest

Inherits:
Object
  • Object
show all
Defined in:
lib/intelligence/chat_request.rb

Overview

The ChatRequest class encapsulates a request to an LLM. After creating a new ChatRequest instance you can make the actual request by calling the chat or stream methods. In order to construct a ChatRequest you must first construct and configure an adapter.

example

adapter = Intelligence::Adapter.build( :open_ai ) do

key ENV[ 'OPENAI_API_KEY' ]
chat_options do 
  model 'gpt-4o'
  max_tokens 512
end

end

request = Intelligence::ChatRequest.new( adapter: adapter ) response = request.chat( ‘Hello!’ )

if response.success?

puts response.result.text

else

puts response.result.error_description

end

Constant Summary collapse

DEFAULT_CONNECTION = 
Faraday.new { | builder | builder.adapter Faraday.default_adapter }

Instance Method Summary collapse

Constructor Details

#initialize(connection: nil, adapter:, **options) ⇒ ChatRequest

The initialize method initializes the ChatRequest instance. You MUST pass a previously constructed and configured adapter and optionally a (Faraday) connection.

Raises:

  • (ArgumentError) 


57
58
59
60
61
62
63
64
 
# File 'lib/intelligence/chat_request.rb', line 57

def initialize( connection: nil, adapter: , **options )
  @connection = connection || DEFAULT_CONNECTION
  @adapter = adapter
  @options = options || {}

  raise ArgumentError.new( 'An adapter must be configured before a request is constructed.' ) \
    if @adapter.nil?
end

Instance Method Details

#chat(conversation, *options) ⇒ Object

The chat method leverages the adapter associated with this ChatRequest instance to construct and make an HTTP request - through Faraday - to an LLM service. The chat method always returns a Faraday::Respose which is augmented with a result method.

If the response is successful ( if response.success? returns true ) the result method returns a ChatResponse instance. If the response is not successful a ChatErrorResult instance is returned.

arguments

  • conversation - an instance of Intelligence::Conversation or String; this encapsulates

    the content to be sent to the LLM

  • options - one or more Hashes with options; these options overide any of the

    configuration options used to configure the adapter; you can, for 
example, pass +{ chat_options: { max_tokens: 1024 }+ to limit the 
response to 1024 tokens.



82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
 
# File 'lib/intelligence/chat_request.rb', line 82

def chat( conversation, *options )

  conversation = build_quick_conversation( conversation ) if conversation.is_a?( String )
  options = options.compact.reduce( {} ) { | accumulator, o | accumulator.merge( o ) }
  options = @options.merge( options )

  # conversation and tools are presented as simple Hashes to the adapter
  conversation = conversation.to_h
  options[ :tools ] = options[ :tools ].to_a.map!( &:to_h ) if options[ :tools ]

  uri = @adapter.chat_request_uri( options )
  headers = @adapter.chat_request_headers( options )
  payload = @adapter.chat_request_body( conversation, options )

  result_callback = nil 
  response = @connection.post( uri ) do | request |
    headers.each { | key, value | request.headers[ key ] = value }
    request.body = payload          
    yield request.extend( ChatRequestMethods ) if block_given?
    result_callback = request.instance_variable_get( "@_intelligence_result_callback" )
  end

  result = nil 

  if response.success?  
    chat_result_attributes = @adapter.chat_result_attributes( response )
    result = ChatResult.new( chat_result_attributes )
  else 
    error_result_attributes = @adapter.chat_result_error_attributes( response )
    result = ChatErrorResult.new( error_result_attributes )
  end
  
  response.instance_variable_set( "@_intelligence_result", result )
  response.extend( ChatResponseMethods ) 

end

#stream(conversation, *options) ⇒ Object 



119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
 
# File 'lib/intelligence/chat_request.rb', line 119

def stream( conversation, *options )

  conversation = build_quick_conversation( conversation ) if conversation.is_a?( String )
  options = options.compact.reduce( {} ) { | accumulator, o | accumulator.merge( o ) }
  options = @options.merge( options )
  
  # conversation and tools are presented as simple Hashes to the adapter
  conversation = conversation.to_h
  options[ :tools ] = options[ :tools ].to_a.map!( &:to_h ) if options[ :tools ]

  uri = @adapter.chat_request_uri( options )
  headers = @adapter.chat_request_headers( @options.merge( options ) )
  payload = @adapter.chat_request_body( conversation, options )

  context = nil
  response = @connection.post( uri ) do | request |

    headers.each { | key, value | request.headers[ key ] = value }
    request.body = payload
    yield request.extend( ChatRequestMethods ) 

    result_callback = request.instance_variable_get( "@_intelligence_result_callback" )
    request.options.on_data = Proc.new do | chunk, received_bytes |
      context, attributes = @adapter.stream_result_chunk_attributes( context, chunk )
      result_callback.call( ChatResult.new( attributes ) ) unless attributes.nil?
    end 

  end

  result = nil 
  if response.success?  
    stream_result_attributes = @adapter.stream_result_attributes( context )
    result = ChatResult.new( stream_result_attributes )
  else 
    error_result_attributes = @adapter.stream_result_error_attributes( response )
    result = ChatErrorResult.new( error_result_attributes )
  end
  
  response.instance_variable_set( "@_intelligence_result", result )
  response.extend( ChatResponseMethods )

end