Simplified input for modelfit_chat

main.py

@@ -160,6 +160,34 @@ async def _fetch_url_to_records(data_url: str) -> list:
             df[col] = df[col].astype(str)
     return df.to_dict(orient="records")
 
+
+async def _fetch_url_to_dataframe(data_url: str) -> pd.DataFrame:
+    """Fetch data_url (https only), parse CSV/Excel/JSON to DataFrame. Raises HTTPException on error."""
+    if not data_url.strip().lower().startswith("https://"):
+        raise HTTPException(status_code=400, detail="data_url must be an HTTPS URL.")
+    async with httpx.AsyncClient(timeout=DATA_URL_FETCH_TIMEOUT, follow_redirects=True) as client:
+        r = await client.get(data_url)
+        r.raise_for_status()
+        raw = r.content
+        content_type = (r.headers.get("content-type") or "").lower()
+    if len(raw) > DATA_URL_MAX_BYTES:
+        raise HTTPException(status_code=413, detail=f"Data at URL exceeds {DATA_URL_MAX_BYTES // (1024*1024)}MB limit.")
+    path_lower = data_url.split("?")[0].lower()
+    try:
+        if "json" in content_type or path_lower.endswith(".json"):
+            df = pd.read_json(io.BytesIO(raw))
+        elif "spreadsheet" in content_type or "excel" in content_type or path_lower.endswith((".xlsx", ".xls")):
+            df = pd.read_excel(io.BytesIO(raw))
+        else:
+            df = pd.read_csv(io.BytesIO(raw))
+    except Exception as e:
+        raise HTTPException(status_code=400, detail=f"Could not parse data from URL: {str(e)[:200]}")
+    for col in df.columns:
+        if pd.api.types.is_datetime64_any_dtype(df[col]):
+            df[col] = df[col].astype(str)
+    return df
+
+
 async def _forward(path: str, method: str = "GET", json_body=None, user_token: str | None = None):
     """
     Forward request to the PRIVATE Space:
@@ -554,32 +582,37 @@ async def modelforecast(req: Request):
 async def modelfit_chat(
     req: Request,
     data_url: str = Body(..., embed=True),
-    id_col: str = Body(..., embed=True),
-    time_col: str = Body(..., embed=True),
-    y: str = Body(..., embed=True),
-    lag_y: object = Body(None, embed=True),
-    lagged_features: object = Body({}, embed=True),
-    current_features: object = Body([], embed=True),
-    filter_by_significance: bool = Body(False, embed=True),
-    meanvar_test: bool = Body(False, embed=True),
-    signif: object = Body(0.05, embed=True),
 ):
-    """Fetch training data from data_url (HTTPS), then call /modelfit/ on the API. Returns JSON fit result."""
+    """
+    Fetch training data from data_url (HTTPS), infer schema from column order, then call /modelfit/ on the API.
+    Column order: 1st = id_col, 2nd = time_col, 3rd..second-to-last = current_features, last = y.
+    All other parameters use defaults. Returns JSON fit result.
+    """
     user_token = _extract_user_token(req)
     if not user_token:
         raise HTTPException(status_code=401, detail="Missing Authorization Bearer token (dt+...)")
-    df_records = await _fetch_url_to_records(data_url)
+    df = await _fetch_url_to_dataframe(data_url)
+    cols = list(df.columns)
+    if len(cols) < 3:
+        raise HTTPException(
+            status_code=400,
+            detail="Data must have at least 3 columns (order: id_col, time_col, ...current_features..., y).",
+        )
+    id_col = cols[0]
+    time_col = cols[1]
+    y = cols[-1]
+    current_features = cols[2:-1]  # empty if exactly 3 columns
     payload = {
-        "df": df_records,
+        "df": df.to_dict(orient="records"),
         "id_col": id_col,
         "time_col": time_col,
         "y": y,
-        "lag_y": lag_y,
-        "lagged_features": lagged_features,
+        "lag_y": None,
+        "lagged_features": {},
         "current_features": current_features,
-        "filter_by_significance": filter_by_significance,
-        "meanvar_test": meanvar_test,
-        "signif": signif,
+        "filter_by_significance": False,
+        "meanvar_test": False,
+        "signif": 0.05,
     }
     return await _forward("/modelfit/", "POST", json_body=payload, user_token=user_token)