feat: add multi-server control panel and host-aware sync

app/api/models.py

@@ -2,12 +2,13 @@
 Models endpoints - Gestione dei modelli Ollama
 """
 
-from fastapi import APIRouter, HTTPException
+from fastapi import APIRouter, HTTPException, Query
 from pydantic import BaseModel
 from typing import Any, Dict, List, Optional
 from datetime import datetime
 import requests
 import logging
+from urllib.parse import urlparse
 from app.config import settings
 
 logger = logging.getLogger(__name__)
@@ -22,6 +23,18 @@ def ensure_rw_api_enabled() -> None:
             detail="Endpoint non disponibile"
         )
 
+
+def resolve_ollama_host(host: Optional[str]) -> str:
+    """Resolve target Ollama host, optionally overridden by query parameter."""
+    if not host:
+        return settings.OLLAMA_HOST
+
+    parsed = urlparse(host.strip())
+    if parsed.scheme not in {"http", "https"} or not parsed.netloc:
+        raise HTTPException(status_code=422, detail="Invalid Ollama host URL")
+
+    return host.rstrip("/")
+
 class ModelInfo(BaseModel):
     """Informazioni su un modello"""
     name: str
@@ -60,7 +73,7 @@ class ModelsResponse(BaseModel):
         }
 
 @router.get("/models", response_model=ModelsResponse)
-async def get_models():
+async def get_models(host: Optional[str] = Query(default=None)):
     """
     Recupera l'elenco di tutti i modelli caricati in Ollama
     
@@ -70,9 +83,10 @@ async def get_models():
     Raises:
         HTTPException: Se Ollama non è disponibile
     """
+    target_host = resolve_ollama_host(host)
     try:
         response = requests.get(
-            f"{settings.OLLAMA_HOST}/api/tags",
+            f"{target_host}/api/tags",
             timeout=settings.OLLAMA_TIMEOUT
         )
         
@@ -121,16 +135,17 @@ async def get_models():
 
 
 @router.get("/models/running")
-async def get_running_models() -> Dict[str, Any]:
+async def get_running_models(host: Optional[str] = Query(default=None)) -> Dict[str, Any]:
     """
     Recupera i modelli attualmente residenti in memoria, equivalenti a `ollama ps`.
 
     Returns:
         Dict[str, Any]: Payload con modelli running e conteggio
     """
+    target_host = resolve_ollama_host(host)
     try:
         response = requests.get(
-            f"{settings.OLLAMA_HOST}/api/ps",
+            f"{target_host}/api/ps",
             timeout=settings.OLLAMA_TIMEOUT
         )
 
@@ -168,7 +183,7 @@ async def get_running_models() -> Dict[str, Any]:
         )
 
 @router.get("/models/{model_name}", response_model=ModelInfo)
-async def get_model(model_name: str):
+async def get_model(model_name: str, host: Optional[str] = Query(default=None)):
     """
     Recupera le informazioni di un modello specifico
     
@@ -181,9 +196,10 @@ async def get_model(model_name: str):
     Raises:
         HTTPException: Se il modello non esiste o Ollama non è disponibile
     """
+    target_host = resolve_ollama_host(host)
     try:
         response = requests.get(
-            f"{settings.OLLAMA_HOST}/api/tags",
+            f"{target_host}/api/tags",
             timeout=settings.OLLAMA_TIMEOUT
         )
         
@@ -224,7 +240,7 @@ async def get_model(model_name: str):
 
 
 @router.get("/models/{model_name}/show")
-async def get_model_show(model_name: str) -> Dict[str, Any]:
+async def get_model_show(model_name: str, host: Optional[str] = Query(default=None)) -> Dict[str, Any]:
     """
     Recupera le informazioni estese di un modello tramite endpoint Ollama /api/show.
 
@@ -234,9 +250,10 @@ async def get_model_show(model_name: str) -> Dict[str, Any]:
     Returns:
         Dict[str, Any]: Dati estesi del modello
     """
+    target_host = resolve_ollama_host(host)
     try:
         response = requests.post(
-            f"{settings.OLLAMA_HOST}/api/show",
+            f"{target_host}/api/show",
             json={"model": model_name},
             timeout=settings.OLLAMA_TIMEOUT
         )