geminai 0.1.1 → 1.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: fd8310e9cec4ec0a99b4fd40b509475f4f4f444d60bc6ee25864ebbca6361192
-  data.tar.gz: 9854c80f0e3e319ba6cbfd8c5cbdc14e4ae4b0a80609ada5e99423fdc2c1c75d
+  metadata.gz: 182008113410d8953e5d9dbb32d0b3d31e400f0b64dc6de753bd2b0a0ecd0e3b
+  data.tar.gz: 5e4f6e4c304a109ec19e1bb00e98d7c60cd27b3f9c36d88fd51721aef9f424a2
 SHA512:
-  metadata.gz: 2c9cbf33c85d33bff8b5104a53817e77429c62ab9d0e873c1c8cf2a5a45d212441f6d6626efbb3814641a16bacd67e8a3e2f3cae2713f666c20f178ed8d746be
-  data.tar.gz: 514038195a974de0cc62d953540f3ea370c5d03ab9c270a921bbed232b9e299ce63a43cdd1481b76f1504e286108c2aa5939ba077620a8e1fd4ab8f5e8b7ef1c
+  metadata.gz: f1c57d71925d784ff071a79f207b2f5fff3615774929b39a1dced79408c05dbb2bebbc561edc1a6b7b725d394bd9082c0590184ffe131b118d8d6c8edd02143d
+  data.tar.gz: bb3cf7288bfba9404ea343f75f931ba31b75154a17db76d435703d09acc989eaf7b586d39a8e22f7c5458ac2ebbcde3ad9144701e5a2074100366849b6001e29

data/README.md

@@ -2,12 +2,6 @@
 
 A gem for gemini's interactions api
 
-* **Built-in Tools**: Native search grounding (`google_search`), mapping citations, and suggestions.
-* **Image Generation**: High-fidelity native image generation (`gemini-3.1-flash-image`) with direct config like `aspect_ratio` and `image_size`.
-* **Stateful Conversations**: Seamless multi-turn conversations managed server-side via `previous_interaction_id`.
-* **Stateless Conversations**: Client-managed conversation paths using `store: false`.
-* **Zero Dependencies**: Pure Ruby standard library (`net/http` and `json`)
-
 ## Install
 
 ```ruby
@@ -65,7 +59,7 @@ Geminai includes a helper method `generate_image` which calls the Interactions A
 ```ruby
 require 'base64'
 
-interaction = client.generate_image(
+interaction = client.interact(
   "An image of a pelican riding a bicycle",
   model: 'gemini-3.1-flash-image',
   aspect_ratio: '1:1',
@@ -74,7 +68,7 @@ interaction = client.generate_image(
 )
 
 # Extract and save generated images
-interaction.output_images.each_with_index do |img, index|
+interaction.output_files.each_with_index do |img, index|
   File.binwrite("pelican#{index}.png", Base64.decode64(img[:data]))
   puts "Saved pelican_#{index}.png"
 end
@@ -104,7 +98,29 @@ puts interaction_2.output_text
 # => "Your favorite jihad is a butlerian jihad."
 ```
 
----
+### 4. Video generation
+
+```ruby
+# Generate a video
+interaction = client.generate_video(
+  "A beautiful sunset over a calm ocean.",
+  model: "gemini-omni-flash-preview"
+)
+
+if interaction.output_video
+  # Access base64 data
+  video_data = interaction.output_video[:data]
+  # Or access the URI if delivery: "uri" was used
+  video_uri = interaction.output_video[:uri]
+end
+
+# Stateful video editing
+edit_interaction = client.interact(
+  model: "gemini-omni-flash-preview",
+  input: "Make the sun more vibrant and red.",
+  previous_interaction_id: interaction.id
+)
+```
 
 ## Running Tests
 

data/lib/geminai/client.rb

@@ -23,23 +23,36 @@ module Geminai
         input: input
       }
 
-      # Forward all recognized interactions API request parameters
-      [
-        :system_instruction,
-        :generation_config,
-        :response_format,
-        :tools,
-        :previous_interaction_id,
-        :stream,
-        :store
-      ].each do |opt|
-        body[opt] = options[opt] if options.key?(opt)
+      # Infer output shortcut from model name if not explicitly provided
+      output = options.delete(:output)
+      unless output
+        if model.include?("-tts-") || model.include?("-audio")
+          output = :audio
+        elsif model.include?("-image")
+          output = :image
+        elsif model.include?("-video") || model.include?("-generate")
+          output = :video
+        end
       end
 
-      # Handle any extra keys passed in options as well to be fully future-proof
+      # Handle output shortcut
+      if output
+        body[:response_format] = {type: output.to_s}
+        body[:response_format][:aspect_ratio] = options.delete(:aspect_ratio) if options.key?(:aspect_ratio)
+        body[:response_format][:image_size] = options.delete(:image_size) if options.key?(:image_size)
+      end
+
+      if (voice = options.delete(:voice))
+        body[:generation_config] = {speech_config: [{voice: voice}]}
+      end
+
+      # Merge other options, allowing them to override shortcuts if explicitly provided
       options.each do |k, v|
-        next if body.key?(k)
-        body[k] = v
+        if body[k].is_a?(Hash) && v.is_a?(Hash)
+          body[k] = body[k].merge(v)
+        else
+          body[k] = v
+        end
       end
 
       http = Net::HTTP.new(uri.host, uri.port)
@@ -65,20 +78,6 @@ module Geminai
       data = JSON.parse(response.body, symbolize_names: true)
       Interaction.new(data)
     end
-
-    # Helper method for image generation using response_format
-    def generate_image(prompt, model:, aspect_ratio: "1:1", image_size: "1K", **options)
-      interact(
-        model: model,
-        input: prompt,
-        response_format: {
-          type: "image",
-          aspect_ratio: aspect_ratio,
-          image_size: image_size
-        },
-        **options
-      )
-    end
   end
 
   class ApiError < StandardError

data/lib/geminai/interaction.rb

@@ -26,29 +26,67 @@ module Geminai
       text_parts.join("")
     end
 
-    def output_images
-      images = []
+    def files
+      files = []
       @steps.each do |step|
         next unless step.model_output?
         (step.content || []).each do |part|
-          if part[:type] == "image"
-            images <<
-              {
-                data: part[:data],
-                mime_type: part[:mime_type]
-              }
-          elsif part[:image]
-            # Handle other possible image shapes in responses if applicable
-            images <<
-              {
-                data: part[:image][:data] || part[:image][:image_bytes],
-                mime_type: part[:image][:mime_type]
-              }
-          end
+          file_data = extract_file_data(part)
+          files << file_data if file_data
+        end
+      end
+
+      # Also include top-level outputs (like output_audio) if present in raw data
+      @raw_data.each do |key, value|
+        if key.to_s.start_with?("output_") && value.is_a?(Hash) && (value[:data] || value[:uri])
+          files << value.merge(type: key.to_s.sub("output_", ""))
         end
       end
 
-      images
+      files
+    end
+
+    alias_method :output_files, :files
+
+    def base64(type = nil)
+      file = if type
+        files.find { |f| f[:type] == type.to_s && f[:data] }
+      else
+        files.find { |f| f[:data] }
+      end
+
+      file ? file[:data] : nil
+    end
+
+    alias_method :base64_file, :base64
+
+    def uri(type = nil)
+      file = if type
+        files.find { |f| f[:type] == type.to_s && f[:uri] }
+      else
+        files.find { |f| f[:uri] }
+      end
+
+      file ? file[:uri] : nil
+    end
+
+    alias_method :uri_file, :uri
+
+    # Deprecated specific helpers kept for compatibility but powered by generic methods
+    def output_images
+      files.select { |f| f[:type] == "image" }
+    end
+
+    def output_videos
+      files.select { |f| f[:type] == "video" }
+    end
+
+    def output_video
+      output_videos.first
+    end
+
+    def output_audio
+      files.select { |f| f[:type] == "audio" }.first
     end
 
     def grounding_metadata
@@ -93,5 +131,34 @@ module Geminai
         search_suggestions: suggestions
       )
     end
+
+    private
+
+    def extract_file_data(part)
+      # Check part[:type]
+      if %w[image video audio file].include?(part[:type])
+        return {
+          type: part[:type],
+          data: part[:data],
+          uri: part[:uri],
+          mime_type: part[:mime_type]
+        }
+      end
+
+      # Check for keys like :image, :video, :audio
+      [:image, :video, :audio].each do |key|
+        if part[key]
+          data = part[key]
+          return {
+            type: key.to_s,
+            data: data[:data] || data[:image_bytes] || data[:video_bytes] || data[:audio_bytes],
+            uri: data[:uri],
+            mime_type: data[:mime_type]
+          }
+        end
+      end
+
+      nil
+    end
   end
 end

data/lib/geminai/version.rb

@@ -1,3 +1,3 @@
 module Geminai
-  VERSION = "0.1.1"
+  VERSION = "1.0.0"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: geminai
 version: !ruby/object:Gem::Version
-  version: 0.1.1
+  version: 1.0.0
 platform: ruby
 authors:
 - swlkr