fix: expand single filename legal characters

src/syrupy/extensions/single_file.py

@@ -1,4 +1,3 @@
-import re
 from gettext import gettext
 from pathlib import Path
 from typing import (
@@ -7,6 +6,7 @@
     Optional,
     Set,
 )
+from unicodedata import category
 
 from syrupy.data import (
     Snapshot,
@@ -69,5 +69,12 @@ def _write_snapshot_fossil(self, *, snapshot_fossil: "SnapshotFossil") -> None:
 
     def __clean_filename(self, filename: str) -> str:
         filename = str(filename).strip().replace(" ", "_")
-        max_filename_length = 255 - len(self._file_extension or "")
-        return re.sub(r"(?u)[^-\w.]", "", filename)[:max_filename_length]
+        exclude_chars = '\\/:*"<>|?'
+        exclude_categ = ("C",)
+        cleaned_filename = "".join(
+            c
+            for c in filename
+            if c not in exclude_chars
+            and not any(categ in category(c) for categ in exclude_categ)
+        )
+        return cleaned_filename[:255]  # maximum filename length

tests/syrupy/extensions/__snapshots__/test_single_file/test_____underscore.raw

@@ -0,0 +1 @@
+orange

tests/syrupy/extensions/__snapshots__/test_single_file/test_special_characters[.123~!@#$%^&()[]{}].raw

@@ -0,0 +1 @@
+.123~!@#$%^&*()/[]{}|

tests/syrupy/extensions/__snapshots__/test_single_file/test_special_characters[_].raw

@@ -0,0 +1 @@
+_

tests/syrupy/extensions/__snapshots__/test_single_file/test_special_characters[a].raw

@@ -0,0 +1 @@
+a?

tests/syrupy/extensions/__snapshots__/test_single_file/test_special_characters[space_space].raw

@@ -0,0 +1 @@
+space space

tests/syrupy/extensions/__snapshots__/test_single_file/test_underscore.raw

@@ -0,0 +1 @@
+apple

tests/syrupy/extensions/__snapshots__/test_single_file/test_unicode[greek_u1ff4].raw

@@ -0,0 +1 @@
+apple

tests/syrupy/extensions/test_single_file.py

@@ -18,6 +18,16 @@ def snapshot_single(snapshot):
     return snapshot.use_extension(SingleFileSnapshotExtension)
 
 
+class SingleFileUTF8SnapshotExtension(SingleFileSnapshotExtension):
+    def serialize(self, data, **kwargs) -> bytes:
+        return bytes(data, "utf8")
+
+
+@pytest.fixture
+def snapshot_utf8(snapshot):
+    return snapshot.use_extension(SingleFileUTF8SnapshotExtension)
+
+
 def test_does_not_write_non_binary(testdir, snapshot_single: "SnapshotAssertion"):
     snapshot_fossil = SnapshotFossil(
         location=str(Path(testdir.tmpdir).joinpath("snapshot_fossil.raw")),
@@ -37,3 +47,23 @@ def test_class_method_name(self, snapshot_single):
     @pytest.mark.parametrize("content", [b"x", b"y", b"z"])
     def test_class_method_parametrized(self, snapshot_single, content):
         assert snapshot_single == content
+
+
+def test_underscore(snapshot_single):
+    assert snapshot_single == b"apple"
+
+
+def test_____underscore(snapshot_single):
+    assert snapshot_single == b"orange"
+
+
+@pytest.mark.parametrize(
+    "content", [b"", b"_", b"a?", b"space space", b".123~!@#$%^&*()/[]{}|"]
+)
+def test_special_characters(snapshot_single, content):
+    assert snapshot_single == content
+
+
+@pytest.mark.parametrize("content", ["greek ῴ"])
+def test_unicode(snapshot_utf8, content):
+    assert snapshot_utf8 == "apple"