/**
 * Model Recommendation Command
 *
 * Uses the ModelSupervisor to recommend optimal models for different tasks
 */

import chalk from 'chalk';
import inquirer from 'inquirer';
import { OllamaClient, ModelSupervisor, TaskType, TaskContext } from '@fsfalmansour/neohub-core';
import { ConfigManager } from '../config';

export async function recommendCommand(): Promise<void> {
  console.log(chalk.bold.cyan('\n🎯 Model Recommendation\n'));

  const config = new ConfigManager();
  const client = new OllamaClient(config.getOllamaConfig());
  const supervisor = new ModelSupervisor(client);

  try {
    // Initialize supervisor
    console.log(chalk.gray('Analyzing available models...'));
    await supervisor.initialize();

    const flagshipModels = supervisor.getAvailableFlagshipModels();

    if (flagshipModels.length === 0) {
      console.log(
        chalk.yellow('\n⚠️  No flagship models detected. Please install at least one:\n')
      );
      console.log(chalk.gray('  ollama pull deepseek-coder:33b'));
      console.log(chalk.gray('  ollama pull codellama:34b\n'));
      return;
    }

    console.log(
      chalk.green(
        `\n✓ Found ${flagshipModels.length} flagship model(s): ${flagshipModels.join(', ')}\n`
      )
    );

    // Prompt for task type
    const { taskType } = await inquirer.prompt<{ taskType: TaskType }>([
      {
        type: 'list',
        name: 'taskType',
        message: 'What type of task do you need help with?',
        choices: [
          { name: '📝 Code Generation', value: 'code-generation' },
          { name: '🔍 Code Review', value: 'code-review' },
          { name: '♻️  Refactoring', value: 'refactoring' },
          { name: '🐛 Debugging', value: 'debugging' },
          { name: '📖 Code Explanation', value: 'explanation' },
          { name: '🏗️  Architecture Design', value: 'architecture' },
          { name: '✏️  Quick Edit', value: 'quick-edit' },
          { name: '🔬 Code Analysis', value: 'analysis' },
        ],
      },
    ]);

    // Prompt for complexity
    const { complexity } = await inquirer.prompt<{
      complexity: 'low' | 'medium' | 'high';
    }>([
      {
        type: 'list',
        name: 'complexity',
        message: 'How complex is the task?',
        choices: [
          { name: 'Low - Simple, straightforward task', value: 'low' },
          { name: 'Medium - Moderate complexity', value: 'medium' },
          { name: 'High - Complex, requires deep understanding', value: 'high' },
        ],
      },
    ]);

    // Optional: file size context
    const { hasFileContext } = await inquirer.prompt<{ hasFileContext: boolean }>([
      {
        type: 'confirm',
        name: 'hasFileContext',
        message: 'Are you working with a large file (>500 lines)?',
        default: false,
      },
    ]);

    const lineCount = hasFileContext ? 1000 : 100;

    // Build context and get recommendation
    const context: TaskContext = {
      type: taskType,
      complexity,
      lineCount,
    };

    const recommendation = await supervisor.recommendModel(context);

    // Display recommendation
    console.log(chalk.bold.green('\n🎯 Recommended Model\n'));
    console.log(chalk.cyan('Model:      ') + chalk.bold(recommendation.model));
    console.log(
      chalk.cyan('Confidence: ') + chalk.bold((recommendation.confidence * 100).toFixed(0) + '%')
    );
    console.log(chalk.cyan('Reasoning:  ') + recommendation.reasoning);

    if (recommendation.alternatives && recommendation.alternatives.length > 0) {
      console.log(
        chalk.cyan('\nAlternatives: ') + chalk.gray(recommendation.alternatives.join(', '))
      );
    }

    // Display model stats
    const stats = supervisor.getModelStats();
    if (stats.performanceMetrics.size > 0) {
      console.log(chalk.bold('\n📊 Performance History\n'));

      for (const [model, metrics] of stats.performanceMetrics.entries()) {
        console.log(
          chalk.gray(`  ${model}: `) +
            chalk.white(`${metrics.avgResponseTime.toFixed(0)}ms avg `) +
            chalk.gray(`(${metrics.sampleSize} samples)`)
        );
      }
    }

    // Suggest command to use
    console.log(chalk.bold.cyan('\n💡 Suggested Usage\n'));
    console.log(
      chalk.gray('  To use this model, run:\n  ') +
        chalk.white(`neohub chat --model ${recommendation.model}`)
    );
    console.log(
      chalk.gray('  Or set it as default:\n  ') +
        chalk.white(`neohub config set ollama.model ${recommendation.model}\n`)
    );
  } catch (error) {
    console.log(chalk.red('\n✗ Error: ' + (error as Error).message + '\n'));
  }
}