feat: integración de Asistente IA contextual con Llama.cpp Qwen3 Coder 30B (SDD-009, v3.8.24)

This commit is contained in:
2026-10-27 07:37:00 -03:00
parent 502865e45b
commit 4125f7f7bd
20 changed files with 1692 additions and 16 deletions
+123 -1
View File
@@ -852,6 +852,18 @@ def get_all_settings(current_user: User = Depends(require_roles(["ADMIN"]))):
'show_role_banner': SystemSetting.get_value('dashboard_role_banner_enabled', 'true') in ['true', 'True', '1', True]
}
ai_config = {
'enabled': SystemSetting.get_value('ai_chat_enabled', 'true') in ['true', 'True', '1', True],
'base_url': SystemSetting.get_value('ai_chat_base_url', 'http://10.0.0.214:8080/v1'),
'api_key': SystemSetting.get_masked_value('ai_chat_api_key', '••••••••••••') if SystemSetting.get_value('ai_chat_api_key') else 'llama',
'model': SystemSetting.get_value('ai_chat_model', 'qwen3-coder:30b'),
'model_name': SystemSetting.get_value('ai_chat_model_name', 'Qwen3 Coder 30B-A3B (Llama.cpp)'),
'temperature': float(SystemSetting.get_value('ai_chat_temperature', 0.3) or 0.3),
'timeout': int(SystemSetting.get_value('ai_chat_timeout', 60) or 60),
'context_limit': int(SystemSetting.get_value('ai_chat_context_limit', 65536) or 65536),
'output_limit': int(SystemSetting.get_value('ai_chat_output_limit', 65536) or 65536)
}
return {
'status': 'success',
'settings': {
@@ -859,7 +871,9 @@ def get_all_settings(current_user: User = Depends(require_roles(["ADMIN"]))):
'auth_providers': auth_providers,
'google_oauth': google_oauth,
'moodle': moodle_config,
'ui': ui_config
'ui': ui_config,
'ai': ai_config,
'ai_assistant': ai_config
}
}
@@ -1052,6 +1066,114 @@ async def update_ui_settings(
}
@admin_router.post("/settings/ai")
@admin_router.post("/admin/settings/ai")
async def update_ai_settings(
request: Request,
current_user: User = Depends(require_roles(["ADMIN"]))
):
"""Actualiza la configuración del proveedor de IA local (Llama.cpp / Qwen3)."""
data = await get_request_payload(request)
enabled_input = data.get('enabled')
enabled = 'true' if enabled_input in [True, 'true', '1', 'on', 1] else 'false'
base_url = str(data.get('base_url', 'http://10.0.0.214:8080/v1')).strip().rstrip('/')
api_key = str(data.get('api_key', '')).strip()
model = str(data.get('model', 'qwen3-coder:30b')).strip()
model_name = str(data.get('model_name', 'Qwen3 Coder 30B-A3B (Llama.cpp)')).strip()
temperature = str(data.get('temperature', '0.3')).strip()
timeout = str(data.get('timeout', '60')).strip()
context_limit = str(data.get('context_limit', '65536')).strip()
output_limit = str(data.get('output_limit', '65536')).strip()
SystemSetting.set_value('ai_chat_enabled', enabled, 'Habilitación de Asistente IA', category='ai')
SystemSetting.set_value('ai_chat_base_url', base_url, 'URL base del servidor Llama.cpp', category='ai')
if api_key and api_key != '••••••••••••':
SystemSetting.set_encrypted_value('ai_chat_api_key', api_key, 'Clave API del servidor de IA', category='ai')
SystemSetting.set_value('ai_chat_model', model, 'ID del modelo Llama.cpp', category='ai')
SystemSetting.set_value('ai_chat_model_name', model_name, 'Nombre descriptivo del modelo', category='ai')
SystemSetting.set_value('ai_chat_temperature', temperature, 'Temperatura de generación', category='ai')
SystemSetting.set_value('ai_chat_timeout', timeout, 'Timeout de inferencia en segundos', category='ai')
SystemSetting.set_value('ai_chat_context_limit', context_limit, 'Límite de tokens de contexto', category='ai')
SystemSetting.set_value('ai_chat_output_limit', output_limit, 'Límite de tokens de respuesta', category='ai')
AuditService.log(
action='UPDATE_AI_SETTINGS',
module='settings',
user=current_user,
details=f"Configuración de Asistente IA actualizada (Modelo: {model}, URL: {base_url})"
)
return {
'status': 'success',
'message': 'Configuración de Asistente IA guardada exitosamente.',
'settings': {
'enabled': enabled == 'true',
'base_url': base_url,
'model': model,
'model_name': model_name,
'temperature': float(temperature) if temperature else 0.3,
'timeout': int(timeout) if timeout else 60
}
}
@admin_router.post("/settings/ai/test")
@admin_router.post("/admin/settings/ai/test")
async def test_ai_settings_connection(
request: Request,
current_user: User = Depends(require_roles(["ADMIN"]))
):
"""Prueba en tiempo real la conectividad con el servidor de IA local."""
import time
import httpx
data = await get_request_payload(request)
base_url = str(data.get('base_url') or SystemSetting.get_value('ai_chat_base_url', 'http://10.0.0.214:8080/v1')).strip().rstrip('/')
api_key = data.get('api_key')
if not api_key or api_key == '••••••••••••':
api_key = SystemSetting.get_decrypted_value('ai_chat_api_key', 'llama') or 'llama'
target_url = f"{base_url}/models"
headers = {"Authorization": f"Bearer {api_key}"}
start_time = time.time()
async with httpx.AsyncClient(timeout=10.0) as client:
try:
resp = await client.get(target_url, headers=headers)
latency_ms = round((time.time() - start_time) * 1000, 2)
if resp.status_code == 200:
data_json = resp.json()
models = [m.get("id") for m in data_json.get("data", []) if isinstance(m, dict) and "id" in m]
return {
'status': 'success',
'message': f"Conexión exitosa con Llama.cpp ({base_url}). Latencia: {latency_ms} ms.",
'latency_ms': latency_ms,
'models': models
}
else:
return {
'status': 'warning',
'message': f"Respuesta inesperada del servidor ({resp.status_code}): {resp.text[:200]}",
'latency_ms': latency_ms,
'models': []
}
except httpx.TimeoutException:
return {
'status': 'error',
'message': f"Tiempo de espera agotado al conectar a {base_url} (Timeout: 10s).",
'latency_ms': None,
'models': []
}
except httpx.RequestError as e:
return {
'status': 'error',
'message': f"No se pudo conectar a {base_url}: {str(e)}",
'latency_ms': None,
'models': []
}
@admin_router.get("/settings/demo-mode")
@admin_router.get("/admin/settings/demo-mode")
def get_demo_mode_status(current_user: User = Depends(require_roles(["ADMIN"]))):