Momento/memento-note/app/api/ai/models/route.ts

import { NextRequest, NextResponse } from 'next/server'
import { getSystemConfig } from '@/lib/config'
import { auth } from '@/auth'

// Modèles populaires pour chaque provider (2025)
const PROVIDER_MODELS = {
  ollama: {
    tags: [
      'llama3:latest',
      'llama3.2:latest',
      'granite4:latest',
      'mistral:latest',
      'mixtral:latest',
      'phi3:latest',
      'gemma2:latest',
      'qwen2:latest'
    ],
    embeddings: [
      'embeddinggemma:latest',
      'mxbai-embed-large:latest',
      'nomic-embed-text:latest'
    ]
  },
  openai: {
    tags: [
      'gpt-4o',
      'gpt-4o-mini',
      'gpt-4-turbo',
      'gpt-4',
      'gpt-3.5-turbo'
    ],
    embeddings: [
      'text-embedding-3-small',
      'text-embedding-3-large',
      'text-embedding-ada-002'
    ]
  },
  custom: {
    tags: [], // Will be loaded dynamically
    embeddings: [] // Will be loaded dynamically
  }
}

export async function GET(request: NextRequest) {
  const session = await auth()
  if (!session?.user?.id) {
    return NextResponse.json({ error: 'Unauthorized' }, { status: 401 })
  }
  if ((session.user as any).role !== 'ADMIN') {
    return NextResponse.json({ error: 'Forbidden' }, { status: 403 })
  }

  try {
    const config = await getSystemConfig()
    const provider = (config.AI_PROVIDER || 'ollama').toLowerCase()

    let models = PROVIDER_MODELS[provider as keyof typeof PROVIDER_MODELS] || { tags: [], embeddings: [] }

    // Pour Ollama, essayer de récupérer la liste réelle depuis l'API locale
    if (provider === 'ollama') {
      try {
        const ollamaBaseUrl = config.OLLAMA_BASE_URL || process.env.OLLAMA_BASE_URL || 'http://localhost:11434'
        const response = await fetch(`${ollamaBaseUrl}/api/tags`, {
          method: 'GET',
          headers: { 'Content-Type': 'application/json' }
        })

        if (response.ok) {
          const data = await response.json()
          const allModels = data.models || []

          // Séparer les modèles de tags et d'embeddings
          const tagModels = allModels
            .filter((m: any) => !m.name.includes('embed') && !m.name.includes('Embedding'))
            .map((m: any) => m.name)
            .slice(0, 20) // Limiter à 20 modèles

          const embeddingModels = allModels
            .filter((m: any) => m.name.includes('embed') || m.name.includes('Embedding'))
            .map((m: any) => m.name)

          models = {
            tags: tagModels.length > 0 ? tagModels : models.tags,
            embeddings: embeddingModels.length > 0 ? embeddingModels : models.embeddings
          }
        }
      } catch (error) {
        // Garder les modèles par défaut
      }
    }

    return NextResponse.json({
      provider,
      models: models || { tags: [], embeddings: [] }
    })
  } catch (error: any) {
    return NextResponse.json(
      {
        error: error.message || 'Failed to fetch models',
        models: { tags: [], embeddings: [] }
      },
      { status: 500 }
    )
  }
}