smart_prompt 0.4.4 → 0.5.0

data/config/image_generation_config.yml

@@ -0,0 +1,22 @@
+# Configuration for SiliconFlow image generation.
+#
+# Get an API key from https://siliconflow.cn and export it as SILICONFLOW_API_KEY.
+# Available image models: Kwai-Kolors/Kolors, Qwen/Qwen-Image,
+# Qwen/Qwen-Image-Edit (image editing). See:
+# https://api-docs.siliconflow.cn/docs/api/images-generations-post
+
+adapters:
+  image_generation: "ImageGenerationAdapter"
+
+llms:
+  image_gen:
+    adapter: "image_generation"
+    url: "https://api.siliconflow.cn/v1/"
+    api_key: ENV["SILICONFLOW_API_KEY"]
+    # Kolors supports batch_size, guidance_scale and a range of image_size values.
+    model: "Kwai-Kolors/Kolors"
+
+default_llm: "image_gen"
+template_path: "./templates"
+worker_path: "./workers"
+logger_file: "./logs/smart_prompt.log"

data/config/multimodal_config.yml

@@ -0,0 +1,85 @@
+# Multimodal Configuration for SmartPrompt
+# This configuration enables multimodal capabilities with SiliconFlow
+
+# Adapter definitions
+adapters:
+  openai: "OpenAIAdapter"
+  multimodal: "MultimodalAdapter"
+
+# LLM configurations
+llms:
+  # Multimodal models for vision and video understanding
+  qwen_vl:
+    adapter: "multimodal"
+    url: "https://api.siliconflow.cn/v1/"
+    api_key: ENV["SILICONFLOW_API_KEY"]
+    default_model: "Qwen/Qwen2.5-VL-7B-Instruct"
+    temperature: 0.7
+
+  qwen_omni:
+    adapter: "multimodal"
+    url: "https://api.siliconflow.cn/v1/"
+    api_key: ENV["SILICONFLOW_API_KEY"]
+    default_model: "Qwen/Qwen3-Omni-7B-Instruct"
+    temperature: 0.7
+
+  deepseek_vl:
+    adapter: "multimodal"
+    url: "https://api.siliconflow.cn/v1/"
+    api_key: ENV["SILICONFLOW_API_KEY"]
+    default_model: "deepseek-ai/DeepSeek-VL2"
+    temperature: 0.7
+
+  # Text-only models for comparison
+  siliconflow_text:
+    adapter: "openai"
+    url: "https://api.siliconflow.cn/v1/"
+    api_key: ENV["SILICONFLOW_API_KEY"]
+    default_model: "Qwen/Qwen2.5-7B-Instruct"
+    temperature: 0.7
+
+# Default settings
+default_llm: "qwen_vl"
+
+# Path configurations
+template_path: "./templates"
+worker_path: "./workers"
+logger_file: "./logs/smart_prompt.log"
+
+# Multimodal specific settings
+multimodal:
+  # Default image detail level ("low", "high", "auto")
+  default_image_detail: "auto"
+
+  # Default video extraction settings
+  default_max_frames: 10
+  default_fps: 1
+
+  # Supported file formats
+  supported_image_formats:
+    - "jpg"
+    - "jpeg"
+    - "png"
+    - "gif"
+    - "bmp"
+    - "webp"
+
+  supported_video_formats:
+    - "mp4"
+    - "mov"
+    - "avi"
+    - "mkv"
+    - "webm"
+
+# Advanced settings
+advanced:
+  # Timeout settings (in seconds)
+  request_timeout: 240
+  connection_timeout: 30
+
+  # Retry settings
+  max_retries: 3
+  retry_delay: 2
+
+  # Rate limiting
+  requests_per_minute: 60

data/config/sensenova_config.yml

@@ -0,0 +1,63 @@
+# SenseNova (商汤 日日新) Configuration for SmartPrompt
+#
+# Get an API key from https://platform.sensenova.cn/console and export it as
+# SENSENOVA_API_KEY. A single SenseNovaAdapter covers all four model categories —
+# just point a different llm entry at each model.
+#
+# Two base hosts are in play (verified 2026-06-19):
+#   * token.sensenova.cn/v1          — the free-tier models (sensenova-6.7-flash-lite
+#                                       chat/vision, sensenova-u1-fast image). Works
+#                                       out-of-the-box with a free key.
+#   * api.sensenova.cn/compatible-mode/v2 — paid models (SenseChat-5, SenseNova-V6-Pro).
+#                                       Returns 403 if your key lacks that model.
+#   * api.sensenova.cn/v1/llm/embeddings — Cupido embeddings (native; paid).
+#
+# The image endpoint is NOT under compatible-mode (it 404s there) — it lives on
+# token.sensenova.cn/v1/images/generations and only accepts specific sizes
+# (see VALID_IMAGE_SIZES in sensenova_adapter.rb; default 2048x2048).
+
+adapters:
+  sensenova: "SenseNovaAdapter"
+
+llms:
+  # 1. 商量 文本对话 (free-tier; works with a free key)
+  sensechat:
+    adapter: "sensenova"
+    url: "https://token.sensenova.cn/v1"
+    api_key: ENV["SENSENOVA_API_KEY"]
+    model: "sensenova-6.7-flash-lite"
+    temperature: 0.7
+    # Optional SenseNova sampling extras (forwarded to /chat/completions):
+    # reasoning_effort: "medium"
+    # max_completion_tokens: 4096
+    # Paid alternative: url https://api.sensenova.cn/compatible-mode/v2, model SenseChat-5
+
+  # 2. 商量 图文多模态 (sensenova-6.7-flash-lite is natively multimodal)
+  sensevision:
+    adapter: "sensenova"
+    url: "https://token.sensenova.cn/v1"
+    api_key: ENV["SENSENOVA_API_KEY"]
+    model: "sensenova-6.7-flash-lite"
+    # Paid alternative: url https://api.sensenova.cn/compatible-mode/v2, model SenseNova-V6-Pro
+
+  # 3. Cupido 向量模型 (paid; native endpoint, non-OpenAI response shape)
+  senseembedding:
+    adapter: "sensenova"
+    url: "https://api.sensenova.cn/compatible-mode/v2"
+    embeddings_url: "https://api.sensenova.cn/v1/llm/embeddings"
+    api_key: ENV["SENSENOVA_API_KEY"]
+    model: "Cupido"
+
+  # 4. 秒画 文生图 (free-tier sensenova-u1-fast; native token.sensenova.cn base)
+  senseimage:
+    adapter: "sensenova"
+    url: "https://token.sensenova.cn/v1"
+    image_url: "https://token.sensenova.cn/v1/images/generations"
+    api_key: ENV["SENSENOVA_API_KEY"]
+    model: "sensenova-u1-fast"
+
+default_llm: "sensechat"
+
+template_path: "./templates"
+worker_path: "./workers"
+logger_file: "./logs/smart_prompt.log"

data/config/zhipu_config.yml

@@ -0,0 +1,73 @@
+# 智谱 AI (BigModel / GLM) Configuration for SmartPrompt
+#
+# Get an API key from https://open.bigmodel.cn/ and export it as ZHIPUAI_API_KEY.
+# A single ZhipuAIAdapter covers all REST model categories — just point a different
+# llm entry at each model. Defaults use the free-tier models so it runs out-of-box.
+#
+# Base URL: https://open.bigmodel.cn/api/paas/v4  (coding models: .../api/coding/paas/v4)
+# Auth:     Authorization: Bearer <API_KEY>
+
+adapters:
+  zhipu: "ZhipuAIAdapter"
+
+llms:
+  # 1. 文本对话 (free-tier glm-4-flash; paid: glm-4-plus / glm-4-long / glm-5.2)
+  glm:
+    adapter: "zhipu"
+    url: "https://open.bigmodel.cn/api/paas/v4"
+    api_key: ENV["ZHIPUAI_API_KEY"]
+    model: "glm-4-flash"
+    temperature: 0.7
+    # Optional sampling extras (forwarded to /chat/completions):
+    # top_p: 0.7
+    # max_tokens: 2048
+    # CodeGeeX-4: set `coding: true` (uses the coding base) and model: codegeex-4
+
+  # 2. 图文多模态 (free-tier glm-4v-flash; paid: glm-4v-plus)
+  glm_vision:
+    adapter: "zhipu"
+    url: "https://open.bigmodel.cn/api/paas/v4"
+    api_key: ENV["ZHIPUAI_API_KEY"]
+    model: "glm-4v-flash"
+
+  # 3. 向量模型 (embedding-3 supports custom dimensions: 256/512/1024/2048)
+  embedding:
+    adapter: "zhipu"
+    url: "https://open.bigmodel.cn/api/paas/v4"
+    api_key: ENV["ZHIPUAI_API_KEY"]
+    model: "embedding-3"
+    dimensions: 1024
+
+  # 4. 文生图 (free-tier cogview-3-flash; paid: cogview-4 / glm-image)
+  cogview:
+    adapter: "zhipu"
+    url: "https://open.bigmodel.cn/api/paas/v4"
+    api_key: ENV["ZHIPUAI_API_KEY"]
+    model: "cogview-3-flash"
+
+  # 5. 文生视频 (async: submit -> poll -> download; free-tier cogvideox-flash)
+  cogvideo:
+    adapter: "zhipu"
+    url: "https://open.bigmodel.cn/api/paas/v4"
+    api_key: ENV["ZHIPUAI_API_KEY"]
+    model: "cogvideox-flash"
+
+  # 6. 语音合成 (GLM-TTS)
+  glm_tts:
+    adapter: "zhipu"
+    url: "https://open.bigmodel.cn/api/paas/v4"
+    api_key: ENV["ZHIPUAI_API_KEY"]
+    model: "glm-tts"
+
+  # 7. 语音识别 (GLM-ASR-2512)
+  glm_asr:
+    adapter: "zhipu"
+    url: "https://open.bigmodel.cn/api/paas/v4"
+    api_key: ENV["ZHIPUAI_API_KEY"]
+    model: "glm-asr-2512"
+
+default_llm: "glm"
+
+template_path: "./templates"
+worker_path: "./workers"
+logger_file: "./logs/smart_prompt.log"

data/examples/anthropic_basic_chat.rb

@@ -0,0 +1,143 @@
+#!/usr/bin/env ruby
+# frozen_string_literal: true
+
+require './lib/smart_prompt'
+
+# Example: Basic Chat with Anthropic Claude
+# This example demonstrates simple text-based conversations with Claude
+
+puts "=" * 60
+puts "Anthropic Claude - Basic Chat Example"
+puts "=" * 60
+
+# Initialize the engine with Anthropic configuration
+engine = SmartPrompt::Engine.new('config/anthropic_config.yml')
+
+# Example 1: Simple Question-Answer
+puts "\n1. Simple Question-Answer"
+puts "-" * 60
+
+SmartPrompt.define_worker :simple_chat do
+  use "claude"
+  sys_msg("You are a helpful AI assistant.")
+  prompt(params[:message])
+  send_msg
+end
+
+response = engine.call_worker(:simple_chat, {
+  message: "What is the capital of France?"
+})
+puts "User: What is the capital of France?"
+puts "Claude: #{response}\n"
+
+# Example 2: Multi-turn Conversation with History
+puts "\n2. Multi-turn Conversation"
+puts "-" * 60
+
+SmartPrompt.define_worker :conversation do
+  use "claude"
+  sys_msg("You are a knowledgeable history teacher.")
+  prompt(params[:message], with_history: true)
+  send_msg
+end
+
+response1 = engine.call_worker(:conversation, {
+  message: "Who was the first president of the United States?"
+})
+puts "User: Who was the first president of the United States?"
+puts "Claude: #{response1}\n"
+
+response2 = engine.call_worker(:conversation, {
+  message: "What were his major accomplishments?"
+})
+puts "User: What were his major accomplishments?"
+puts "Claude: #{response2}\n"
+
+# Example 3: Using Different Models
+puts "\n3. Comparing Different Claude Models"
+puts "-" * 60
+
+question = "Explain quantum entanglement in one sentence."
+
+# Claude 3.5 Sonnet (balanced)
+SmartPrompt.define_worker :sonnet_chat do
+  use "claude"
+  model "claude-3-5-sonnet-20241022"
+  prompt(params[:message])
+  send_msg
+end
+
+sonnet_response = engine.call_worker(:sonnet_chat, { message: question })
+puts "Question: #{question}"
+puts "\nClaude 3.5 Sonnet: #{sonnet_response}\n"
+
+# Claude 3.5 Haiku (faster)
+SmartPrompt.define_worker :haiku_chat do
+  use "claude_haiku"
+  model "claude-3-5-haiku-20241022"
+  prompt(params[:message])
+  send_msg
+end
+
+haiku_response = engine.call_worker(:haiku_chat, { message: question })
+puts "Claude 3.5 Haiku: #{haiku_response}\n"
+
+# Example 4: Temperature Control
+puts "\n4. Temperature Control for Creativity"
+puts "-" * 60
+
+# Creative response (high temperature)
+SmartPrompt.define_worker :creative_chat do
+  use "claude_creative"  # temperature: 0.9
+  sys_msg("You are a creative writer.")
+  prompt("Write a creative tagline for a coffee shop.")
+  send_msg
+end
+
+creative = engine.call_worker(:creative_chat, {})
+puts "Creative (temp=0.9): #{creative}\n"
+
+# Precise response (low temperature)
+SmartPrompt.define_worker :precise_chat do
+  use "claude_precise"  # temperature: 0.3
+  sys_msg("You are a precise analyst.")
+  prompt("Write a tagline for a coffee shop.")
+  send_msg
+end
+
+precise = engine.call_worker(:precise_chat, {})
+puts "Precise (temp=0.3): #{precise}\n"
+
+# Example 5: System Message Variations
+puts "\n5. System Message Variations"
+puts "-" * 60
+
+question = "Should I invest in cryptocurrency?"
+
+# As a financial advisor
+SmartPrompt.define_worker :advisor_chat do
+  use "claude"
+  sys_msg("You are a conservative financial advisor.")
+  prompt(params[:message])
+  send_msg
+end
+
+advisor_response = engine.call_worker(:advisor_chat, { message: question })
+puts "As Financial Advisor:"
+puts advisor_response[0..150] + "...\n"
+
+# As a tech enthusiast
+SmartPrompt.define_worker :enthusiast_chat do
+  use "claude"
+  sys_msg("You are an enthusiastic technology advocate.")
+  prompt(params[:message])
+  send_msg
+end
+
+enthusiast_response = engine.call_worker(:enthusiast_chat, { message: question })
+puts "\nAs Tech Enthusiast:"
+puts enthusiast_response[0..150] + "...\n"
+
+puts "\n" + "=" * 60
+puts "Basic chat examples completed!"
+puts "=" * 60

data/examples/anthropic_example.rb

@@ -0,0 +1,232 @@
+#!/usr/bin/env ruby
+# frozen_string_literal: true
+
+require "./lib/smart_prompt"
+
+# Example: Using Anthropic Claude with SmartPrompt
+# This example demonstrates various features of the AnthropicAdapter
+
+# Initialize the engine with Anthropic configuration
+engine = SmartPrompt::Engine.new("config/anthropic_config.yml")
+
+puts "=" * 60
+puts "Anthropic Claude Examples with SmartPrompt"
+puts "=" * 60
+
+=begin
+# Example 1: Basic Chat
+puts "\n1. Basic Chat Example"
+puts "-" * 60
+
+SmartPrompt.define_worker :basic_chat do
+  use "deepseek_anthropic"
+  model "deepseek-chat"
+  sys_msg("You are a helpful AI assistant.")
+  prompt(params[:message])
+  send_msg
+end
+
+response = engine.call_worker(:basic_chat, {
+  message: "What is the capital of France?",
+})
+puts "User: What is the capital of France?"
+puts "DeepSeek: #{response}"
+
+# Example 2: Multi-turn Conversation
+puts "\n2. Multi-turn Conversation Example"
+puts "-" * 60
+
+SmartPrompt.define_worker :conversation do
+  use "deepseek_anthropic"
+  model "deepseek-chat"
+  sys_msg("You are a knowledgeable history teacher.")
+  prompt(params[:message], with_history: true)
+  send_msg
+end
+
+response1 = engine.call_worker(:conversation, {  
+  message: "Who was the first president of the United States?",
+  with_history: true,
+})
+puts "User: Who was the first president of the United States?"
+puts "Claude: #{response1}"
+
+response2 = engine.call_worker(:conversation, {
+  message: "What were his major accomplishments?",
+  with_history: true,
+})
+puts "\nUser: What were his major accomplishments?"
+puts "Claude: #{response2}"
+
+# Example 3: Code Generation
+puts "\n3. Code Generation Example"
+puts "-" * 60
+
+SmartPrompt.define_worker :code_generator do
+  use "deepseek_anthropic"
+  model "deepseek-chat"
+  sys_msg("You are an expert programmer. Generate clean, well-documented code.")
+  prompt("Write a Ruby function that #{params[:task]}")
+  send_msg
+end
+
+code = engine.call_worker(:code_generator, {
+  task: "calculates the factorial of a number using recursion",
+})
+puts "Task: Write a Ruby function that calculates the factorial of a number using recursion"
+puts "Generated Code:\n#{code}"
+
+# Example 4: Streaming Response
+puts "\n4. Streaming Response Example"
+puts "-" * 60
+
+SmartPrompt.define_worker :streaming_chat do
+  use "deepseek_anthropic"
+  model "deepseek-chat"
+  sys_msg("You are a storyteller.")
+  prompt(params[:message])
+  send_msg
+end
+
+print "User: Tell me a short story about a brave knight.\n"
+print "Claude (streaming): "
+
+engine.call_worker_by_stream(:streaming_chat, {
+  message: "Tell me a short story about a brave knight.",
+}) do |chunk, bytesize|
+  # Handle Anthropic streaming format
+  if chunk.is_a?(Hash)
+    if chunk["type"] == "content_block_delta"
+      text = chunk.dig("delta", "text")
+      print text if text
+    end
+  end
+end
+
+puts "\n"
+=end
+
+# Example 7: Tool Calling (Function Calling)
+puts "\n7. Tool Calling Example"
+puts "-" * 60
+
+# Define tools
+weather_tools = [
+  {
+    type: "function",
+    function: {
+      name: "get_weather",
+      description: "Get the current weather for a location",
+      parameters: {
+        type: "object",
+        properties: {
+          location: {
+            type: "string",
+            description: "The city and state, e.g. San Francisco, CA",
+          },
+          unit: {
+            type: "string",
+            enum: ["celsius", "fahrenheit"],
+            description: "The temperature unit",
+          },
+        },
+        required: ["location"],
+      },
+    },
+  },
+]
+
+SmartPrompt.define_worker :weather_assistant do
+  use "deepseek_anthropic"
+  model "deepseek-chat"
+  sys_msg("You are a helpful weather assistant. Use the get_weather tool when users ask about weather.")
+  prompt(params[:message])
+  send_msg
+end
+
+response = engine.call_worker(:weather_assistant, {
+  message: "What's the weather like in Tokyo?",
+  tools: weather_tools,
+})
+puts "User: What's the weather like in Tokyo?"
+puts "Claude: #{response}"
+
+=begin
+
+# Example 8: Image Analysis (Multimodal)
+puts "\n8. Image Analysis Example (Multimodal)"
+puts "-" * 60
+
+SmartPrompt.define_worker :image_analyzer do
+  use "claude"
+  sys_msg("You are an expert at analyzing images and describing what you see.")
+  prompt(params[:message])
+  send_msg
+end
+
+# Using a public image URL
+image_url = "https://upload.wikimedia.org/wikipedia/commons/thumb/3/3a/Cat03.jpg/1200px-Cat03.jpg"
+
+response = engine.call_worker(:image_analyzer, {
+  message: [
+    { type: "text", text: "What do you see in this image? Describe it in detail." },
+    { type: "image_url", image_url: image_url }
+  ]
+})
+puts "User: [Sends image of a cat]"
+puts "Claude: #{response}"
+
+# Example 9: Different Models Comparison
+puts "\n9. Different Models Comparison"
+puts "-" * 60
+
+question = "Explain quantum entanglement in simple terms."
+
+# Using Claude 3.5 Sonnet
+SmartPrompt.define_worker :sonnet_chat do
+  use "claude"
+  model "claude-3-5-sonnet-20241022"
+  prompt(params[:message])
+  send_msg
+end
+
+sonnet_response = engine.call_worker(:sonnet_chat, { message: question })
+puts "Question: #{question}"
+puts "\nClaude 3.5 Sonnet Response:"
+puts sonnet_response[0..200] + "..."
+
+# Using Claude 3.5 Haiku (faster)
+SmartPrompt.define_worker :haiku_chat do
+  use "claude_haiku"
+  model "claude-3-5-haiku-20241022"
+  prompt(params[:message])
+  send_msg
+end
+
+haiku_response = engine.call_worker(:haiku_chat, { message: question })
+puts "\nClaude 3.5 Haiku Response:"
+puts haiku_response[0..200] + "..."
+
+# Example 10: Error Handling
+puts "\n10. Error Handling Example"
+puts "-" * 60
+
+begin
+  # Try to use a non-existent model
+  SmartPrompt.define_worker :error_test do
+    use "claude"
+    model "non-existent-model"
+    prompt(params[:message])
+    send_msg
+  end
+
+  engine.call_worker(:error_test, { message: "Hello" })
+rescue SmartPrompt::LLMAPIError => e
+  puts "Caught API Error: #{e.message}"
+  puts "Error handling works correctly!"
+end
+
+puts "\n" + "=" * 60
+puts "All examples completed!"
+puts "=" * 60
+=end