feat(excel2xml): make_bitstream_prop(): make file existence check opt…

…-in (DEV-3113) (#709)

src/dsp_tools/commands/excel2xml/excel2xml_cli.py

@@ -200,14 +200,13 @@ def _append_bitstream_to_resource(
                 f"'{row['file']}' (Resource ID '{row['id']}', Excel row {row_number}). "
                 f"An attempt to deduce them from the resource permissions failed."
             )
-    with warnings.catch_warnings():  # ignore warnings about non-existing files
-        resource.append(
-            make_bitstream_prop(
-                path=str(row["file"]),
-                permissions=str(file_permissions),
-                calling_resource=row["id"],
-            )
+    resource.append(
+        make_bitstream_prop(
+            path=str(row["file"]),
+            permissions=str(file_permissions),
+            calling_resource=row["id"],
         )
+    )
     return resource
 
 

src/dsp_tools/commands/excel2xml/excel2xml_lib.py

@@ -447,6 +447,7 @@ def make_resource(  # noqa: D417 (undocumented-param)
 def make_bitstream_prop(
     path: Union[str, os.PathLike[Any]],
     permissions: str = "prop-default",
+    check: bool = False,
     calling_resource: str = "",
 ) -> etree._Element:
     """
@@ -455,10 +456,11 @@ def make_bitstream_prop(
     Args:
         path: path to a valid file that will be uploaded
         permissions: permissions string
+        check: if True, issue a warning if the path doesn't point to an existing file
         calling_resource: the name of the parent resource (for better error messages)
 
     Raises:
-        Warning: if the path doesn't point to an existing file
+        Warning: if the path doesn't point to an existing file (only if check=True)
 
     Returns:
         an etree._Element that can be appended to the parent resource with resource.append(make_*_prop(...))
@@ -471,7 +473,7 @@ def make_bitstream_prop(
     See https://docs.dasch.swiss/latest/DSP-TOOLS/file-formats/xml-data-file/#bitstream
     """
 
-    if not Path(path).is_file():
+    if check and not Path(path).is_file():
         warnings.warn(
             f"Failed validation in bitstream tag of resource '{calling_resource}': "
             f"The following path doesn't point to a file: {path}",

test/unittests/commands/excel2xml/test_excel2xml_lib.py

@@ -1,4 +1,5 @@
 import unittest
+import warnings
 from pathlib import Path
 from typing import Any, Callable, Optional, Sequence, Union
 
@@ -341,32 +342,41 @@ def test_prepare_value(self) -> None:
         values_output = excel2xml.prepare_value([excel2xml.PropertyElement(x) for x in values_with_nas])
         self.assertEqual([x.value for x in values_output], values_with_nas)
 
-    def test_make_bitstream_prop(self) -> None:
-        path_string = "foo/bar/baz.txt"
-        path_path = Path(path_string)
-        test_cases: list[tuple[str, Callable[..., etree._Element]]] = [
-            (
-                '<bitstream permissions="prop-default">foo/bar/baz.txt</bitstream>',
-                lambda: excel2xml.make_bitstream_prop(path_string),
-            ),
-            (
-                '<bitstream permissions="prop-default">foo/bar/baz.txt</bitstream>',
-                lambda: excel2xml.make_bitstream_prop(path_path),
-            ),
-            (
-                '<bitstream permissions="prop-restricted">foo/bar/baz.txt</bitstream>',
-                lambda: excel2xml.make_bitstream_prop(path_string, "prop-restricted"),
-            ),
-            (
-                '<bitstream permissions="prop-restricted">foo/bar/baz.txt</bitstream>',
-                lambda: excel2xml.make_bitstream_prop(path_path, "prop-restricted"),
-            ),
-        ]
-        for expected, method_call in test_cases:
-            with self.assertWarnsRegex(UserWarning, ".*Failed validation in bitstream tag.*"):
-                result = etree.tostring(method_call(), encoding="unicode")
-                result = regex.sub(r" xmlns(:.+?)?=\".+?\"", "", result)
-                self.assertEqual(result, expected)
+    def test_make_bitstream_prop_from_string(self) -> None:
+        res = excel2xml.make_bitstream_prop("foo/bar/baz.txt")
+        assert res.tag.endswith("bitstream")
+        assert res.attrib["permissions"] == "prop-default"
+        assert res.text == "foo/bar/baz.txt"
+
+    def test_make_bitstream_prop_from_path(self) -> None:
+        res = excel2xml.make_bitstream_prop(Path("foo/bar/baz.txt"))
+        assert res.tag.endswith("bitstream")
+        assert res.attrib["permissions"] == "prop-default"
+        assert res.text == "foo/bar/baz.txt"
+
+    def test_make_bitstream_prop_custom_permissions(self) -> None:
+        res = excel2xml.make_bitstream_prop("foo/bar/baz.txt", "prop-restricted")
+        assert res.tag.endswith("bitstream")
+        assert res.attrib["permissions"] == "prop-restricted"
+        assert res.text == "foo/bar/baz.txt"
+
+    def test_make_bitstream_prop_valid_file(self) -> None:
+        with warnings.catch_warnings():
+            warnings.filterwarnings("error", category=UserWarning)
+            try:
+                res = excel2xml.make_bitstream_prop("testdata/bitstreams/test.jpg", check=True)
+            except UserWarning as e:
+                raise AssertionError from e
+        assert res.tag.endswith("bitstream")
+        assert res.attrib["permissions"] == "prop-default"
+        assert res.text == "testdata/bitstreams/test.jpg"
+
+    def test_make_bitstream_prop_invalid_file(self) -> None:
+        with pytest.warns(UserWarning, match=".*Failed validation in bitstream tag.*"):
+            res = excel2xml.make_bitstream_prop("foo/bar/baz.txt", check=True)
+        assert res.tag.endswith("bitstream")
+        assert res.attrib["permissions"] == "prop-default"
+        assert res.text == "foo/bar/baz.txt"
 
     def test_make_boolean_prop(self) -> None:
         # prepare true_values