Class: Prescient::Provider::OpenAI

Inherits:
Base
  • Object
show all
Includes:
HTTParty
Defined in:
lib/prescient/provider/openai.rb

Constant Summary collapse

EMBEDDING_DIMENSIONS = 
{
  'text-embedding-3-small' => 1536,
  'text-embedding-3-large' => 3072,
  'text-embedding-ada-002' => 1536,
}.freeze

Instance Attribute Summary

Attributes inherited from Base

#options

Instance Method Summary collapse

Methods inherited from Base

#available?

Constructor Details

#initialize(**options) ⇒ OpenAI

Returns a new instance of OpenAI.



16
17
18
19
 
# File 'lib/prescient/provider/openai.rb', line 16

def initialize(**options)
  super
  self.class.default_timeout(@options[:timeout] || 60)
end

Instance Method Details

#generate_embedding(text, **_options) ⇒ Object 



21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
 
# File 'lib/prescient/provider/openai.rb', line 21

def generate_embedding(text, **_options)
  handle_errors do
    clean_text_input = clean_text(text)

    response = self.class.post('/v1/embeddings',
                               headers: {
                                 'Content-Type'  => 'application/json',
                                 'Authorization' => "Bearer #{@options[:api_key]}",
                               },
                               body:    {
                                 model:           @options[:embedding_model],
                                 input:           clean_text_input,
                                 encoding_format: 'float',
                               }.to_json)

    validate_response!(response, 'embedding generation')

    embedding_data = response.parsed_response.dig('data', 0, 'embedding')
    raise Prescient::InvalidResponseError, 'No embedding returned' unless embedding_data

    expected_dimensions = EMBEDDING_DIMENSIONS[@options[:embedding_model]] || 1536
    normalize_embedding(embedding_data, expected_dimensions)
  end
end

#generate_response(prompt, context_items = [], **options) ⇒ Object 



46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
 
# File 'lib/prescient/provider/openai.rb', line 46

def generate_response(prompt, context_items = [], **options)
  handle_errors do
    formatted_prompt = build_prompt(prompt, context_items)

    response = self.class.post('/v1/chat/completions',
                               headers: {
                                 'Content-Type'  => 'application/json',
                                 'Authorization' => "Bearer #{@options[:api_key]}",
                               },
                               body:    {
                                 model:       @options[:chat_model],
                                 messages:    [
                                   {
                                     role:    'user',
                                     content: formatted_prompt,
                                   },
                                 ],
                                 max_tokens:  options[:max_tokens] || 2000,
                                 temperature: options[:temperature] || 0.7,
                                 top_p:       options[:top_p] || 0.9,
                               }.to_json)

    validate_response!(response, 'text generation')

    content = response.parsed_response.dig('choices', 0, 'message', 'content')
    raise Prescient::InvalidResponseError, 'No response generated' unless content

    {
      response:        content.strip,
      model:           @options[:chat_model],
      provider:        'openai',
      processing_time: nil,
      metadata:        {
        usage:         response.parsed_response['usage'],
        finish_reason: response.parsed_response.dig('choices', 0, 'finish_reason'),
      },
    }
  end
end

#health_check ⇒ Object 



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
118
119
120
121
122
123
124
125
126
127
128
 
# File 'lib/prescient/provider/openai.rb', line 86

def health_check
  handle_errors do
    response = self.class.get('/v1/models',
                              headers: {
                                'Authorization' => "Bearer #{@options[:api_key]}",
                              })

    if response.success?
      models = response.parsed_response['data'] || []
      embedding_available = models.any? { |m| m['id'] == @options[:embedding_model] }
      chat_available = models.any? { |m| m['id'] == @options[:chat_model] }

      {
        status:           'healthy',
        provider:         'openai',
        models_available: models.map { |m| m['id'] },
        embedding_model:  {
          name:      @options[:embedding_model],
          available: embedding_available,
        },
        chat_model:       {
          name:      @options[:chat_model],
          available: chat_available,
        },
        ready:            embedding_available && chat_available,
      }
    else
      {
        status:   'unhealthy',
        provider: 'openai',
        error:    "HTTP #{response.code}",
        message:  response.message,
      }
    end
  end
rescue Prescient::Error => e
  {
    status:   'unavailable',
    provider: 'openai',
    error:    e.class.name,
    message:  e.message,
  }
end

#list_models ⇒ Object 



130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
 
# File 'lib/prescient/provider/openai.rb', line 130

def list_models
  handle_errors do
    response = self.class.get('/v1/models',
                              headers: {
                                'Authorization' => "Bearer #{@options[:api_key]}",
                              })
    validate_response!(response, 'model listing')

    models = response.parsed_response['data'] || []
    models.map do |model|
      {
        name:     model['id'],
        created:  model['created'],
        owned_by: model['owned_by'],
      }
    end
  end
end