ollama_ai_service.h

Go to the documentation of this file.

#ifndef YAZE_SRC_CLI_OLLAMA_AI_SERVICE_H_
#define YAZE_SRC_CLI_OLLAMA_AI_SERVICE_H_
 
#include <cstdint>
#include <string>
#include <vector>
 
#include "absl/status/status.h"
#include "absl/status/statusor.h"
#include "absl/time/time.h"
#include "cli/service/ai/ai_service.h"
#include "cli/service/ai/provider_ids.h"
 
#ifdef YAZE_AI_RUNTIME_AVAILABLE
#include "cli/service/ai/prompt_builder.h"
#endif
 
namespace yaze {
namespace cli {
 
// Ollama configuration for local LLM inference
struct OllamaConfig {
  std::string base_url = "http://localhost:11434";  // Default Ollama endpoint
  std::string model =
      "qwen2.5-coder:0.5b";            // Lightweight default with tool-calling
  float temperature = 0.1;             // Low temp for deterministic commands
  int max_tokens = 2048;               // Sufficient for command lists
  std::string system_prompt;           // Injected from resource catalogue
  bool use_enhanced_prompting = true;  // Enable few-shot examples
  float top_p = 0.92f;
  int top_k = 40;
  int num_ctx = 4096;
  bool stream = false;
  bool use_chat_completions = true;
  std::vector<std::string> favorite_models;
};
 
#ifdef YAZE_AI_RUNTIME_AVAILABLE
 
class OllamaAIService : public AIService {
 public:
  explicit OllamaAIService(const OllamaConfig& config);
 
  struct OllamaModelDetails {
    std::string name;
    std::string digest;
    std::string family;
    std::string parameter_size;
    std::string quantization_level;
    uint64_t size_bytes = 0;
    absl::Time modified_at = absl::InfinitePast();
  };
 
  void SetRomContext(Rom* rom) override;
 
  // Generate z3ed commands from natural language prompt
  absl::StatusOr<AgentResponse> GenerateResponse(
      const std::string& prompt) override;
  absl::StatusOr<AgentResponse> GenerateResponse(
      const std::vector<agent::ChatMessage>& history) override;
 
  // Health check: verify Ollama server is running and model is available
  absl::Status CheckAvailability();
 
  // List available models on Ollama server
  absl::StatusOr<std::vector<ModelInfo>> ListAvailableModels() override;
 
  std::string GetProviderName() const override { return kProviderOllama; }
 
 private:
  OllamaConfig config_;
  PromptBuilder prompt_builder_;
 
  // Build system prompt from resource catalogue
  std::string BuildSystemPrompt();
 
  // Parse JSON response from Ollama API
  absl::StatusOr<std::string> ParseOllamaResponse(
      const std::string& json_response);
};
 
#else  // !YAZE_AI_RUNTIME_AVAILABLE
 
class OllamaAIService : public AIService {
 public:
  struct OllamaModelDetails {
    std::string name;
    std::string digest;
    std::string family;
    std::string parameter_size;
    std::string quantization_level;
    uint64_t size_bytes = 0;
    absl::Time modified_at = absl::InfinitePast();
  };
 
  explicit OllamaAIService(const OllamaConfig&) {}
  void SetRomContext(Rom*) override {}
  absl::StatusOr<AgentResponse> GenerateResponse(const std::string&) override {
    return absl::FailedPreconditionError("Ollama AI runtime is disabled");
  }
  absl::StatusOr<AgentResponse> GenerateResponse(
      const std::vector<agent::ChatMessage>&) override {
    return absl::FailedPreconditionError("Ollama AI runtime is disabled");
  }
  absl::Status CheckAvailability() {
    return absl::FailedPreconditionError("Ollama AI runtime is disabled");
  }
  absl::StatusOr<std::vector<ModelInfo>> ListAvailableModels() override {
    return absl::FailedPreconditionError("Ollama AI runtime is disabled");
  }
  std::string GetProviderName() const override { return kProviderOllama; }
};
 
#endif  // YAZE_AI_RUNTIME_AVAILABLE
 
}  // namespace cli
}  // namespace yaze
 
#endif  // YAZE_SRC_CLI_OLLAMA_AI_SERVICE_H_