fix(nodeenv): preserve surrogateescaped activation content

CHANGES

@@ -4,6 +4,9 @@ Nodeenv changelog
 Version [unreleased]
 --------------------
 
+- Preserve non-UTF-8 filesystem bytes when writing activation scripts, so
+  environments in such paths can be activated on Linux
+  `#442 <https://github.com/ekalinin/nodeenv/issues/442>`_
 - `nodeenv -p` no longer puts the python venv's `bin` directory on `PATH`
   a second time: the venv's own `activate` and `activate.fish` put it there
   already. Envs without those scripts, such as conda envs, and the git-bash

nodeenv.py

@@ -653,7 +653,7 @@ def writefile(dest, content, overwrite=True, append=False):
     Create file and write content in it
     """
     if not isinstance(content, bytes):
-        content = bytes(content, 'utf-8')
+        content = content.encode('utf-8', 'surrogateescape')
     if not os.path.exists(dest):
         logger.debug(' * Writing %s ... ', dest, extra=dict(continued=True))
         with open(dest, 'wb') as f:

tests/nodeenv_test.py

@@ -3007,6 +3007,37 @@ def test_with_certifi_is_configurable(self):
 class TestWritefile:
     """Tests for writefile()"""
 
+    @pytest.mark.parametrize('append', [False, True])
+    def test_preserves_surrogateescaped_bytes(self, tmpdir, append):
+        """Filesystem bytes in activation content must round-trip."""
+        dest = tmpdir.join('activate')
+        dest.write_binary(b'# existing\n')
+        raw = b"NODE_VIRTUAL_ENV='/tmp/caf\xe9/env'\n"
+
+        nodeenv.writefile(
+            str(dest), raw.decode('utf-8', 'surrogateescape'), append=append)
+
+        expected = b'# existing\n' + raw if append else raw
+        assert dest.read_binary() == expected
+        nodeenv.writefile(
+            str(dest), raw.decode('utf-8', 'surrogateescape'), append=append)
+        assert dest.read_binary() == expected
+
+    @pytest.mark.parametrize('content, expected', [
+        ('caf\u00e9', b'caf\xc3\xa9'),
+        (b'caf\xe9', b'caf\xe9'),
+    ])
+    def test_preserves_utf8_text_and_bytes(self, tmpdir, content, expected):
+        """Regular Unicode stays UTF-8 and bytes stay unchanged."""
+        dest = tmpdir.join('activate')
+        nodeenv.writefile(str(dest), content)
+        assert dest.read_binary() == expected
+
+    def test_rejects_non_filesystem_surrogate(self, tmpdir):
+        """surrogateescape must not conceal arbitrary invalid Unicode."""
+        with pytest.raises(UnicodeEncodeError):
+            nodeenv.writefile(str(tmpdir.join('activate')), '\ud800')
+
     def test_append_starts_on_its_own_line(self, tmpdir):
         """A file not ending with a newline must not swallow the first
         appended line: `:END` plus `@echo off` is a label, not a command

tests/test_install_activate.py

@@ -62,6 +62,29 @@ def test_write(tmpdir, name, content_var):
     assert bin_dir.join(name).read() == fix_content(content, tmpdir)
 
 
+@pytest.mark.skipif(not sys.platform.startswith('linux'),
+                    reason='non-UTF-8 filesystem paths need Linux')
+def test_activate_with_surrogateescaped_path(tmp_path):
+    """Keep a raw filesystem path intact in generated scripts and sh."""
+    env_dir = tmp_path / os.fsdecode(b'caf\xe9')
+    bin_dir = env_dir / 'bin'
+    bin_dir.mkdir(parents=True)
+    node = bin_dir / 'node'
+    node.write_bytes(b'#!/bin/sh\necho stub\n')
+    node.chmod(0o755)
+    with mock.patch.object(sys, 'argv', ['nodeenv', str(env_dir)]):
+        opts = nodeenv.parse_args()
+        nodeenv.install_activate(str(env_dir), opts)
+
+    activate = bin_dir / 'activate'
+    assert os.fsencode(str(env_dir)) in activate.read_bytes()
+    output = subprocess.check_output([
+        'sh', '-c', '. "$1"; printf "%s" "$NODE_VIRTUAL_ENV"',
+        'sh', str(activate),
+    ])
+    assert output == os.fsencode(str(env_dir))
+
+
 @pytest.mark.skipif(nodeenv.is_WIN, reason='activate.fish is POSIX only')
 def test_activate_fish_has_no_caret_redirect(tmpdir):
     """Generated activate.fish must not contain caret stderr redirects.