oafuncs 0.0.98.2__py3-none-any.whl → 0.0.98.3__py3-none-any.whl

oafuncs/_script/netcdf_write.py

@@ -1,11 +1,15 @@
 import os
+
 import netCDF4 as nc
 import numpy as np
 import xarray as xr
+import warnings
+
+warnings.filterwarnings("ignore", category=RuntimeWarning)
 
 
 def _numpy_to_nc_type(numpy_type):
-    """将NumPy数据类型映射到NetCDF数据类型"""
+    """将 NumPy 数据类型映射到 NetCDF 数据类型"""
     numpy_to_nc = {
         "float32": "f4",
         "float64": "f8",
@@ -23,6 +27,12 @@ def _numpy_to_nc_type(numpy_type):
 
 
 def _calculate_scale_and_offset(data, n=16):
+    """
+    计算数值型数据的 scale_factor 与 add_offset，
+    将数据映射到 [0, 2**n - 1] 的范围。
+
+    要求 data 为数值型的 NumPy 数组，不允许全 NaN 值。
+    """
     if not isinstance(data, np.ndarray):
         raise ValueError("Input data must be a NumPy array.")
 
@@ -32,85 +42,162 @@ def _calculate_scale_and_offset(data, n=16):
     if np.isnan(data_min) or np.isnan(data_max):
         raise ValueError("Input data contains NaN values.")
 
-    scale_factor = (data_max - data_min) / (2**n - 1)
-    add_offset = data_min + 2 ** (n - 1) * scale_factor
+    if data_max == data_min:
+        scale_factor = 1.0
+        add_offset = data_min
+    else:
+        scale_factor = (data_max - data_min) / (2**n - 1)
+        add_offset = data_min + 2 ** (n - 1) * scale_factor
     return scale_factor, add_offset
 
 
+def _data_to_scale_offset(data, scale, offset):
+    """
+    将数据转换为 scale_factor 和 add_offset 的形式。
+    此处同时替换 NaN、正无穷和负无穷为填充值 -32767，
+    以确保转换后的数据可安全转为 int16。
+    """
+    if not isinstance(data, np.ndarray):
+        raise ValueError("Input data must be a NumPy array.")
+
+    # 先计算转换后的数据
+    result = np.around((data - offset) / scale)
+    # 替换 NaN, 正负无穷（posinf, neginf）为 -32767
+    result = np.nan_to_num(result, nan=-32767, posinf=-32767, neginf=-32767)
+    result = np.clip(result, -32767, 32767)  # 限制范围在 int16 的有效范围内
+    result = np.where(np.isfinite(result), result, -32767)  # 替换无效值为 -32767
+    new_data = result.astype(np.int16)
+    return new_data
+
+
 def save_to_nc(file, data, varname=None, coords=None, mode="w", scale_offset_switch=True, compile_switch=True):
-    # 处理xarray对象的情况（当varname和coords都为None时）
-    if varname is None and coords is None:
-        if not isinstance(data, (xr.DataArray, xr.Dataset)):
-            raise ValueError("When varname and coords are not provided, data must be an xarray object")
+    """
+    保存数据到 NetCDF 文件，支持 xarray 对象（DataArray 或 Dataset）和 numpy 数组。
+
+    仅对数据变量中数值型数据进行压缩转换（利用 scale_factor/add_offset 转换后转为 int16），
+    非数值型数据以及所有坐标变量将禁用任何压缩，直接保存原始数据。
+
+    参数：
+      - file: 保存文件的路径
+      - data: xarray.DataArray、xarray.Dataset 或 numpy 数组
+      - varname: 变量名（仅适用于传入 numpy 数组或 DataArray 时）
+      - coords: 坐标字典（numpy 数组分支时使用），所有坐标变量均不压缩
+      - mode: "w"（覆盖）或 "a"（追加）
+      - scale_offset_switch: 是否对数值型数据变量进行压缩转换
+      - compile_switch: 是否启用 NetCDF4 的 zlib 压缩（仅针对数值型数据有效）
+    """
+    # 处理 xarray 对象（DataArray 或 Dataset）的情况
+    if isinstance(data, (xr.DataArray, xr.Dataset)):
+        encoding = {}  # 用于保存数据变量的编码信息
 
-        encoding = {}
         if isinstance(data, xr.DataArray):
             if data.name is None:
                 data = data.rename("data")
-            varname = data.name
-            encoding[varname] = {"zlib": compile_switch, "complevel": 4}
-            if scale_offset_switch:
+            varname = data.name if varname is None else varname
+            # 判断数据是否为数值型
+            if np.issubdtype(data.values.dtype, np.number) and scale_offset_switch:
                 scale, offset = _calculate_scale_and_offset(data.values)
-                encoding[varname].update({"dtype": "int16", "scale_factor": scale, "add_offset": offset, "_FillValue": -32767})
+                new_values = _data_to_scale_offset(data.values, scale, offset)
+                # 生成新 DataArray，保留原坐标和属性，同时写入转换参数到属性中
+                new_da = data.copy(data=new_values)
+                new_da.attrs["scale_factor"] = float(scale)
+                new_da.attrs["add_offset"] = float(offset)
+                encoding[varname] = {
+                    "zlib": compile_switch,
+                    "complevel": 4,
+                    "dtype": "int16",
+                    "_FillValue": -32767,
+                }
+                new_da.to_dataset(name=varname).to_netcdf(file, mode=mode, encoding=encoding)
             else:
-                encoding[varname].update({"dtype": "float32", "_FillValue": np.nan})
-        else:  # Dataset情况
+                data.to_dataset(name=varname).to_netcdf(file, mode=mode)
+            return
+
+        else:
+            # 处理 Dataset 的情况，仅处理 data_vars 数据变量，坐标变量保持原样
+            new_vars = {}
+            encoding = {}
             for var in data.data_vars:
-                encoding[var] = {"zlib": compile_switch, "complevel": 4}
-                if scale_offset_switch:
-                    scale, offset = _calculate_scale_and_offset(data[var].values)
-                    encoding[var].update({"dtype": "int16", "scale_factor": scale, "add_offset": offset, "_FillValue": -32767})
+                da = data[var]
+                if np.issubdtype(np.asarray(da.values).dtype, np.number) and scale_offset_switch:
+                    scale, offset = _calculate_scale_and_offset(da.values)
+                    new_values = _data_to_scale_offset(da.values, scale, offset)
+                    new_da = xr.DataArray(new_values, dims=da.dims, coords=da.coords, attrs=da.attrs)
+                    new_da.attrs["scale_factor"] = float(scale)
+                    new_da.attrs["add_offset"] = float(offset)
+                    new_vars[var] = new_da
+                    encoding[var] = {
+                        "zlib": compile_switch,
+                        "complevel": 4,
+                        "dtype": "int16",
+                        "_FillValue": -32767,
+                    }
                 else:
-                    encoding[var].update({"dtype": "float32", "_FillValue": np.nan})
-
-        try:
-            data.to_netcdf(file, mode=mode, encoding=encoding)
-            return
-        except Exception as e:
-            raise RuntimeError(f"Failed to save xarray object: {str(e)}") from e
+                    new_vars[var] = da
+            new_ds = xr.Dataset(new_vars, coords=data.coords)
+            if encoding:
+                new_ds.to_netcdf(file, mode=mode, encoding=encoding)
+            else:
+                new_ds.to_netcdf(file, mode=mode)
+        return
 
-    # 处理普通numpy数组的情况
+    # 处理纯 numpy 数组情况
     if mode == "w" and os.path.exists(file):
         os.remove(file)
     elif mode == "a" and not os.path.exists(file):
         mode = "w"
-
+    data = np.asarray(data)
+    is_numeric = np.issubdtype(data.dtype, np.number)
     try:
         with nc.Dataset(file, mode, format="NETCDF4") as ncfile:
-            # 创建维度并写入坐标
+            # 坐标变量直接写入，不做压缩
             if coords is not None:
                 for dim, values in coords.items():
                     if dim not in ncfile.dimensions:
                         ncfile.createDimension(dim, len(values))
-                        var = ncfile.createVariable(dim, _numpy_to_nc_type(values.dtype), (dim,))
-                        var[:] = values
+                        var_obj = ncfile.createVariable(dim, _numpy_to_nc_type(np.asarray(values).dtype), (dim,))
+                        var_obj[:] = values
 
-            # 创建变量
             dims = list(coords.keys()) if coords else []
-            if scale_offset_switch:
+            if is_numeric and scale_offset_switch:
                 scale, offset = _calculate_scale_and_offset(data)
+                new_data = _data_to_scale_offset(data, scale, offset)
                 var = ncfile.createVariable(varname, "i2", dims, fill_value=-32767, zlib=compile_switch)
                 var.scale_factor = scale
                 var.add_offset = offset
+                # Ensure no invalid values in new_data before assignment
+                var[:] = new_data
             else:
+                # 非数值型数据，禁止压缩
                 dtype = _numpy_to_nc_type(data.dtype)
-                var = ncfile.createVariable(varname, dtype, dims, zlib=compile_switch)
-
-            var[:] = data
+                var = ncfile.createVariable(varname, dtype, dims, zlib=False)
+                var[:] = data
     except Exception as e:
-        raise RuntimeError(f"Failed to save netCDF4 file: {str(e)}") from e
+        raise RuntimeError(f"netCDF4 保存失败: {str(e)}") from e
 
 
+# 测试用例
 if __name__ == "__main__":
-    # Example usage
-    data = xr.open_dataset(r"F:\roms_rst.nc")["u"]
-    save_to_nc(r"F:\test.nc", data)
-
-    # xarray测试
-    data = xr.DataArray(np.random.rand(10, 20), dims=("x", "y"), name="temperature")
-    save_to_nc(r"F:\test_xarray.nc", data)
-
-    # numpy测试
-    arr = np.random.rand(5, 3)
-    coords = {"x": np.arange(5), "y": np.arange(3)}
-    save_to_nc(r"F:\test_numpy.nc", arr, varname="data", coords=coords)
+    # --------------------------------
+    # dataset
+    file = r"F:\roms_rst.nc"
+    ds = xr.open_dataset(file)
+    outfile = r"F:\roms_rst_test.nc"
+    save_to_nc(outfile, ds)
+    ds.close()
+    # --------------------------------
+    # dataarray
+    data = np.random.rand(4, 3, 2)
+    coords = {"x": np.arange(4), "y": np.arange(3), "z": np.arange(2)}
+    varname = "test_var"
+    data = xr.DataArray(data, dims=("x", "y", "z"), coords=coords, name=varname)
+    outfile = r"F:\test_dataarray.nc"
+    save_to_nc(outfile, data)
+    # --------------------------------
+    # numpy array
+    data = np.random.rand(4, 3, 2)
+    coords = {"x": np.arange(4), "y": np.arange(3), "z": np.arange(2)}
+    varname = "test_var"
+    outfile = r"F:\test_numpy.nc"
+    save_to_nc(outfile, data, varname=varname, coords=coords)
+    # --------------------------------

{oafuncs-0.0.98.2.dist-info → oafuncs-0.0.98.3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: oafuncs
-Version: 0.0.98.2
+Version: 0.0.98.3
 Summary: Oceanic and Atmospheric Functions
 Home-page: https://github.com/Industry-Pays/OAFuncs
 Author: Kun Liu

{oafuncs-0.0.98.2.dist-info → oafuncs-0.0.98.3.dist-info}/RECORD

@@ -14,7 +14,7 @@ oafuncs/_script/cprogressbar.py,sha256=wRU3SFPFtMI7ER26tTzg223kVKNo5RDWE9CzdIgUs
 oafuncs/_script/email.py,sha256=lL4HGKrr524-g0xLlgs-4u7x4-u7DtgNoD9AL8XJKj4,3058
 oafuncs/_script/netcdf_merge.py,sha256=_EPF9Xj4HOVC9sZpi1lt62-Aq6pMlgsgwaajEBLhW6g,5092
 oafuncs/_script/netcdf_modify.py,sha256=sGRUYNhfGgf9JV70rnBzw3bzuTRSXzBTL_RMDnDPeLQ,4552
-oafuncs/_script/netcdf_write.py,sha256=8bVsJKRPLWJawUTAsMUiG5D9lOg_-sRMgshgtRjo_us,4598
+oafuncs/_script/netcdf_write.py,sha256=iO1Qv9bp6RLiw1D8Nrv7tX_8X-diUZaX3Nxhk6pJ5Nw,8556
 oafuncs/_script/parallel.py,sha256=FS9FgaByq2yb9j6nL-Y0xP1VLvp4USMLBFMRsJDoqeQ,21848
 oafuncs/_script/parallel_example_usage.py,sha256=uLvE7iwkMn9Cyq6-wk5_RpbQk7PXM9d16-26lTknW9s,2646
 oafuncs/_script/plot_dataset.py,sha256=zkSEnO_-biyagorwWXPoihts_cwuvripzEt-l9bHJ2E,13989
@@ -36,8 +36,8 @@ oafuncs/oa_sign/__init__.py,sha256=QKqTFrJDFK40C5uvk48GlRRbGFzO40rgkYwu6dYxatM,5
 oafuncs/oa_sign/meteorological.py,sha256=8091SHo2L8kl4dCFmmSH5NGVHDku5i5lSiLEG5DLnOQ,6489
 oafuncs/oa_sign/ocean.py,sha256=xrW-rWD7xBWsB5PuCyEwQ1Q_RDKq2KCLz-LOONHgldU,5932
 oafuncs/oa_sign/scientific.py,sha256=a4JxOBgm9vzNZKpJ_GQIQf7cokkraV5nh23HGbmTYKw,5064
-oafuncs-0.0.98.2.dist-info/licenses/LICENSE.txt,sha256=rMtLpVg8sKiSlwClfR9w_Dd_5WubTQgoOzE2PDFxzs4,1074
-oafuncs-0.0.98.2.dist-info/METADATA,sha256=2-IktPjhTVHt-GLLgJk7J2tYMBYaNA0eY7IZyB0kxlk,4220
-oafuncs-0.0.98.2.dist-info/WHEEL,sha256=CmyFI0kx5cdEMTLiONQRbGQwjIoR1aIYB7eCAQ4KPJ0,91
-oafuncs-0.0.98.2.dist-info/top_level.txt,sha256=bgC35QkXbN4EmPHEveg_xGIZ5i9NNPYWqtJqaKqTPsQ,8
-oafuncs-0.0.98.2.dist-info/RECORD,,
+oafuncs-0.0.98.3.dist-info/licenses/LICENSE.txt,sha256=rMtLpVg8sKiSlwClfR9w_Dd_5WubTQgoOzE2PDFxzs4,1074
+oafuncs-0.0.98.3.dist-info/METADATA,sha256=1-5l6aAaOFaZ9U0pSE6f1944hrE01bLGCBLPonbEXeU,4220
+oafuncs-0.0.98.3.dist-info/WHEEL,sha256=CmyFI0kx5cdEMTLiONQRbGQwjIoR1aIYB7eCAQ4KPJ0,91
+oafuncs-0.0.98.3.dist-info/top_level.txt,sha256=bgC35QkXbN4EmPHEveg_xGIZ5i9NNPYWqtJqaKqTPsQ,8
+oafuncs-0.0.98.3.dist-info/RECORD,,