FIX: Open metadata with utf-8 encoding

When calling 'open' without specifying the encoding Python will use
`locale.getencoding()` for Python 3.11+ and
`locale.getpreferredencoding(False)` (with or without the `False`)
for Python 3.8 to determine which encoding to open the file with.

RMS in interactive mode sets this to ANSI_X3.4-1968 (a weird way of
saying ascii). This can cause opening the file to fail when non-ascii
characters are present within it.

src/fmu/dataio/_utils.py

@@ -130,7 +130,7 @@ def export_file(
             serialized_json = json.dumps(obj)
 
         if isinstance(file, Path):
-            with open(file, "w") as stream:
+            with open(file, "w", encoding="utf-8") as stream:
                 stream.write(serialized_json)
         else:
             file.write(serialized_json.encode("utf-8"))
@@ -220,7 +220,7 @@ def read_parameters_txt(pfile: Path | str) -> types.Parameters:
 
     res: types.Parameters = {}
 
-    with open(pfile) as f:
+    with open(pfile, encoding="utf-8") as f:
         for line in f:
             line_parts = shlex.split(line)
             if len(line_parts) == 2:
@@ -346,7 +346,7 @@ def read_metadata_from_file(filename: str | Path) -> dict:
     metafilepath = Path(metafile)
     if not metafilepath.exists():
         raise OSError(f"Cannot find requested metafile: {metafile}")
-    with open(metafilepath) as stream:
+    with open(metafilepath, encoding="utf-8") as stream:
         return yaml.safe_load(stream)
 
 

tests/test_units/test_rms_context.py

@@ -4,9 +4,11 @@
 interactive or from ERT. Hence the rootpath will be ../../
 """
 
+import builtins
 import logging
 import os
 import shutil
+from copy import deepcopy
 from pathlib import Path
 
 import pandas as pd
@@ -616,6 +618,41 @@ def test_gridproperty_export_with_geometry(inside_rms_setup, grid, gridproperty)
     assert "this_is_parent" in output
 
 
+@inside_rms
+def test_gridproperty_export_with_geometry_and_bad_character(
+    inside_rms_setup, grid, gridproperty, monkeypatch
+):
+    """Ensures a non-ascii character in masterdata does not cause encoding parsing
+    failures"""
+    original_open = builtins.open
+
+    def open_with_ansi(file, mode="r", *args, **kwargs):
+        if "r" in mode and "b" not in mode and "encoding" not in kwargs:
+            kwargs["encoding"] = "ANSI_X3.4-1968"
+        return original_open(file, mode, *args, **kwargs)
+
+    monkeypatch.setattr(builtins, "open", open_with_ansi)
+
+    cfg = deepcopy(inside_rms_setup["config"])
+
+    cfg["masterdata"]["smda"]["field"][0]["identifier"] = "Drogøn"
+
+    grd_edata = dataio.ExportData(
+        config=cfg,
+        name="geogrid",
+        content={"property": {"is_discrete": False}},
+    )
+    outgrid = grd_edata.export(grid)
+
+    dataio.ExportData(
+        config=cfg,
+        name="geogrid",
+        content={"property": {"is_discrete": False}},
+        geometry=outgrid,
+    ).export(gridproperty)
+    # Will raise an exception if decoding fails
+
+
 # ======================================================================================
 # Dataframe and PyArrow
 # ======================================================================================