Class: RailsAi::Providers::GeminiAdapter

Inherits:
Base
  • Object
show all
Defined in:
lib/rails_ai/providers/gemini_adapter.rb

Constant Summary collapse

GEMINI_API_BASE = 
"https://generativelanguage.googleapis.com/v1beta"

Instance Method Summary collapse

Constructor Details

#initialize ⇒ GeminiAdapter

Returns a new instance of GeminiAdapter.



12
13
14
15
 
# File 'lib/rails_ai/providers/gemini_adapter.rb', line 12

def initialize
  @api_key = ENV.fetch("GEMINI_API_KEY")
  super
end

Instance Method Details

#analyze_image!(image:, prompt:, model: "gemini-2.0-flash-exp", **opts) ⇒ Object

Multimodal analysis - Gemini supports image and video analysis



351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
 
# File 'lib/rails_ai/providers/gemini_adapter.rb', line 351

def analyze_image!(image:, prompt:, model: "gemini-2.0-flash-exp", **opts)
  return "[stubbed] Image analysis: #{prompt}" if RailsAi.config.stub_responses
  
  contents = [
    {
      parts: [
        { text: prompt },
        {
          inlineData: {
            mimeType: detect_image_type(image),
            data: extract_base64_data(image)
          }
        }
      ]
    }
  ]
  
  response = make_request(
    "models/#{model}:generateContent",
    {
      contents: contents,
      generationConfig: {
        maxOutputTokens: opts[:max_tokens] || RailsAi.config.token_limit,
        temperature: opts[:temperature] || 0.7,
        topP: opts[:top_p] || 0.8,
        topK: opts[:top_k] || 40,
        **opts.except(:max_tokens, :temperature, :top_p, :top_k)
      }
    }
  )
  
  response.dig("candidates", 0, "content", "parts", 0, "text")
end

#analyze_video!(video:, prompt:, model: "gemini-2.0-flash-exp", **opts) ⇒ Object 



385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
 
# File 'lib/rails_ai/providers/gemini_adapter.rb', line 385

def analyze_video!(video:, prompt:, model: "gemini-2.0-flash-exp", **opts)
  return "[stubbed] Video analysis: #{prompt}" if RailsAi.config.stub_responses
  
  contents = [
    {
      parts: [
        { text: prompt },
        {
          inlineData: {
            mimeType: "video/mp4",
            data: extract_base64_data(video)
          }
        }
      ]
    }
  ]
  
  response = make_request(
    "models/#{model}:generateContent",
    {
      contents: contents,
      generationConfig: {
        maxOutputTokens: opts[:max_tokens] || RailsAi.config.token_limit,
        temperature: opts[:temperature] || 0.7,
        topP: opts[:top_p] || 0.8,
        topK: opts[:top_k] || 40,
        **opts.except(:max_tokens, :temperature, :top_p, :top_k)
      }
    }
  )
  
  response.dig("candidates", 0, "content", "parts", 0, "text")
end

#chat!(messages:, model:, **opts) ⇒ Object

Text-based operations



18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
 
# File 'lib/rails_ai/providers/gemini_adapter.rb', line 18

def chat!(messages:, model:, **opts)
  return "(stubbed) #{messages.last[:content]}" if RailsAi.config.stub_responses
  
  # Convert OpenAI format to Gemini format
  gemini_messages = convert_messages_to_gemini(messages)
  
  response = make_request(
    "models/#{model}:generateContent",
    {
      contents: gemini_messages,
      generationConfig: {
        maxOutputTokens: opts[:max_tokens] || RailsAi.config.token_limit,
        temperature: opts[:temperature] || 0.7,
        topP: opts[:top_p] || 0.8,
        topK: opts[:top_k] || 40,
        **opts.except(:max_tokens, :temperature, :top_p, :top_k)
      }
    }
  )
  
  response.dig("candidates", 0, "content", "parts", 0, "text")
end

#create_variation!(image:, **opts) ⇒ Object 



173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
 
# File 'lib/rails_ai/providers/gemini_adapter.rb', line 173

def create_variation!(image:, **opts)
  return "data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAYAAAAfFcSJAAAADUlEQVR42mNkYPhfDwAChwGA60e6kgAAAABJRU5ErkJggg==" if RailsAi.config.stub_responses
  
  # Create variations using Gemini 2.0 Flash
  variation_prompt = "Create a variation of this image with similar style but different composition."
  
  contents = [
    {
      parts: [
        { text: variation_prompt },
        {
          inlineData: {
            mimeType: detect_image_type(image),
            data: extract_base64_data(image)
          }
        }
      ]
    }
  ]
  
  response = make_request(
    "models/gemini-2.0-flash-exp:generateContent",
    {
      contents: contents,
      generationConfig: {
        maxOutputTokens: 1000,
        temperature: opts[:temperature] || 0.8,
        **opts
      }
    }
  )
  
  image_data = response.dig("candidates", 0, "content", "parts", 0, "inlineData", "data")
  if image_data
    "data:image/png;base64,#{image_data}"
  else
    raise "Image variation failed: No image data in response"
  end
end

#edit_image!(image:, prompt:, **opts) ⇒ Object 



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
161
162
163
164
165
166
167
168
169
170
171
 
# File 'lib/rails_ai/providers/gemini_adapter.rb', line 132

def edit_image!(image:, prompt:, **opts)
  return "data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAYAAAAfFcSJAAAADUlEQVR42mNkYPhfDwAChwGA60e6kgAAAABJRU5ErkJggg==" if RailsAi.config.stub_responses
  
  # Gemini doesn't have direct image editing, but we can use it to generate variations
  image_prompt = "Edit this image: #{prompt}. Show the edited version."
  
  contents = [
    {
      parts: [
        { text: image_prompt },
        {
          inlineData: {
            mimeType: detect_image_type(image),
            data: extract_base64_data(image)
          }
        }
      ]
    }
  ]
  
  response = make_request(
    "models/gemini-2.0-flash-exp:generateContent",
    {
      contents: contents,
      generationConfig: {
        maxOutputTokens: 1000,
        temperature: opts[:temperature] || 0.7,
        **opts
      }
    }
  )
  
  # Extract generated image data
  image_data = response.dig("candidates", 0, "content", "parts", 0, "inlineData", "data")
  if image_data
    "data:image/png;base64,#{image_data}"
  else
    raise "Image editing failed: No image data in response"
  end
end

#edit_video!(video:, prompt:, **opts) ⇒ Object 



245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
 
# File 'lib/rails_ai/providers/gemini_adapter.rb', line 245

def edit_video!(video:, prompt:, **opts)
  return "data:video/mp4;base64,AAAAIGZ0eXBpc29tAAACAGlzb21pc28yYXZjMW1wNDEAAAAIZnJlZQAAAB1tZGF0AQAAARxtYXNrAAAAAG1wNDEAAAAAIG1kYXQ=" if RailsAi.config.stub_responses
  
  video_prompt = "Edit this video: #{prompt}. Show the edited version."
  
  contents = [
    {
      parts: [
        { text: video_prompt },
        {
          inlineData: {
            mimeType: "video/mp4",
            data: extract_base64_data(video)
          }
        }
      ]
    }
  ]
  
  response = make_request(
    "models/gemini-2.0-flash-exp:generateContent",
    {
      contents: contents,
      generationConfig: {
        maxOutputTokens: 1000,
        temperature: opts[:temperature] || 0.7,
        **opts
      }
    }
  )
  
  video_data = response.dig("candidates", 0, "content", "parts", 0, "inlineData", "data")
  if video_data
    "data:video/mp4;base64,#{video_data}"
  else
    raise "Video editing failed: No video data in response"
  end
end

#embed!(texts:, model:, **opts) ⇒ Object 



65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
 
# File 'lib/rails_ai/providers/gemini_adapter.rb', line 65

def embed!(texts:, model:, **opts)
  return Array.new(texts.length) { [0.0] * 768 } if RailsAi.config.stub_responses
  
  # Gemini has embedding models
  response = make_request(
    "models/#{model}:embedContent",
    {
      content: {
        parts: texts.map { |text| { text: text } }
      }
    }
  )
  
  # Handle both single and batch embedding responses
  if texts.length == 1
    [response.dig("embedding", "values")]
  else
    # For multiple texts, we need to make separate requests or use batch embedding
    texts.map do |text|
      single_response = make_request(
        "models/#{model}:embedContent",
        {
          content: {
            parts: [{ text: text }]
          }
        }
      )
      single_response.dig("embedding", "values")
    end
  end
end

#generate_image!(prompt:, model: "gemini-2.0-flash-exp", **opts) ⇒ Object

Image generation - Gemini 2.0 Flash supports image generation



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
129
130
 
# File 'lib/rails_ai/providers/gemini_adapter.rb', line 98

def generate_image!(prompt:, model: "gemini-2.0-flash-exp", **opts)
  return "data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAYAAAAfFcSJAAAADUlEQVR42mNkYPhfDwAChwGA60e6kgAAAABJRU5ErkJggg==" if RailsAi.config.stub_responses
  
  # Use Gemini 2.0 Flash for image generation
  response = make_request(
    "models/#{model}:generateContent",
    {
      contents: [
        {
          parts: [
            {
              text: "Generate an image: #{prompt}"
            }
          ]
        }
      ],
      generationConfig: {
        maxOutputTokens: 1000,
        temperature: opts[:temperature] || 0.7,
        **opts
      }
    }
  )
  
  # Extract image data from response
  image_data = response.dig("candidates", 0, "content", "parts", 0, "inlineData", "data")
  if image_data
    "data:image/png;base64,#{image_data}"
  else
    # Fallback: return a placeholder or raise error
    raise "Image generation failed: No image data in response"
  end
end

#generate_speech!(text:, model: "gemini-2.0-flash-exp", **opts) ⇒ Object

Audio generation - Gemini 2.0 Flash supports audio generation



285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
 
# File 'lib/rails_ai/providers/gemini_adapter.rb', line 285

def generate_speech!(text:, model: "gemini-2.0-flash-exp", **opts)
  return "data:audio/mp3;base64,SUQzBAAAAAAAI1RTU0UAAAAPAAADTGF2ZjU4Ljc2LjEwMAAAAAAAAAAAAAAA//tQxAADB8AhSmAhIIEVWWWU" if RailsAi.config.stub_responses
  
  response = make_request(
    "models/#{model}:generateContent",
    {
      contents: [
        {
          parts: [
            {
              text: "Generate speech for: #{text}"
            }
          ]
        }
      ],
      generationConfig: {
        maxOutputTokens: 1000,
        temperature: opts[:temperature] || 0.7,
        **opts
      }
    }
  )
  
  audio_data = response.dig("candidates", 0, "content", "parts", 0, "inlineData", "data")
  if audio_data
    "data:audio/mp3;base64,#{audio_data}"
  else
    raise "Speech generation failed: No audio data in response"
  end
end

#generate_video!(prompt:, model: "gemini-2.0-flash-exp", **opts) ⇒ Object

Video generation - Gemini 2.0 Flash supports video generation



214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
 
# File 'lib/rails_ai/providers/gemini_adapter.rb', line 214

def generate_video!(prompt:, model: "gemini-2.0-flash-exp", **opts)
  return "data:video/mp4;base64,AAAAIGZ0eXBpc29tAAACAGlzb21pc28yYXZjMW1wNDEAAAAIZnJlZQAAAB1tZGF0AQAAARxtYXNrAAAAAG1wNDEAAAAAIG1kYXQ=" if RailsAi.config.stub_responses
  
  response = make_request(
    "models/#{model}:generateContent",
    {
      contents: [
        {
          parts: [
            {
              text: "Generate a video: #{prompt}"
            }
          ]
        }
      ],
      generationConfig: {
        maxOutputTokens: 1000,
        temperature: opts[:temperature] || 0.7,
        **opts
      }
    }
  )
  
  video_data = response.dig("candidates", 0, "content", "parts", 0, "inlineData", "data")
  if video_data
    "data:video/mp4;base64,#{video_data}"
  else
    raise "Video generation failed: No video data in response"
  end
end

#stream_chat!(messages:, model:, **opts, &on_token) ⇒ Object 



41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
 
# File 'lib/rails_ai/providers/gemini_adapter.rb', line 41

def stream_chat!(messages:, model:, **opts, &on_token)
  return on_token.call("(stubbed stream)") if RailsAi.config.stub_responses
  
  # Convert OpenAI format to Gemini format
  gemini_messages = convert_messages_to_gemini(messages)
  
  make_streaming_request(
    "models/#{model}:streamGenerateContent",
    {
      contents: gemini_messages,
      generationConfig: {
        maxOutputTokens: opts[:max_tokens] || RailsAi.config.token_limit,
        temperature: opts[:temperature] || 0.7,
        topP: opts[:top_p] || 0.8,
        topK: opts[:top_k] || 40,
        **opts.except(:max_tokens, :temperature, :top_p, :top_k)
      }
    }
  ) do |chunk|
    text = chunk.dig("candidates", 0, "content", "parts", 0, "text")
    on_token.call(text) if text
  end
end

#transcribe_audio!(audio:, model: "gemini-2.0-flash-exp", **opts) ⇒ Object 



316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
 
# File 'lib/rails_ai/providers/gemini_adapter.rb', line 316

def transcribe_audio!(audio:, model: "gemini-2.0-flash-exp", **opts)
  return "[stubbed transcription]" if RailsAi.config.stub_responses
  
  contents = [
    {
      parts: [
        {
          text: "Transcribe this audio:"
        },
        {
          inlineData: {
            mimeType: "audio/mp3",
            data: extract_base64_data(audio)
          }
        }
      ]
    }
  ]
  
  response = make_request(
    "models/#{model}:generateContent",
    {
      contents: contents,
      generationConfig: {
        maxOutputTokens: 1000,
        temperature: opts[:temperature] || 0.1,
        **opts
      }
    }
  )
  
  response.dig("candidates", 0, "content", "parts", 0, "text")
end