Improve validation/sanitization precision for WIndows platform

Author: thombashi

pathvalidate/_common.py

@@ -9,7 +9,6 @@
 import re
 import string
 import sys
-from os.path import normpath
 
 from ._six import text_type
 from .error import NullNameError
@@ -53,7 +52,7 @@ def preprocess(name):
     if is_pathlike_obj(name):
         name = text_type(name)
 
-    return normpath(name)
+    return name
 
 
 def is_null_string(value):

pathvalidate/_file.py

@@ -8,6 +8,7 @@
 
 import enum
 import itertools
+import ntpath
 import os.path
 import platform
 import re
@@ -295,20 +296,23 @@ def sanitize(self, value, replacement_text=""):
         except AttributeError as e:
             raise ValueError(e)
 
+        drive, unicode_file_path = ntpath.splitdrive(unicode_file_path)
         sanitized_path = self._sanitize_regexp.sub(replacement_text, unicode_file_path)
-
-        if len(sanitized_path.split("/")) > len(sanitized_path.split("\\")):
-            path_separator = "/"
-        else:
+        if self._is_windows():
             path_separator = "\\"
+        else:
+            path_separator = "/"
 
         sanitized_entries = []
+        if drive:
+            sanitized_entries.append(drive)
         for entry in sanitized_path.replace("\\", "/").split("/"):
             try:
                 sanitized_entries.append(self.__fname_sanitizer.sanitize(entry))
             except ValidationError as e:
                 if e.reason == ErrorReason.NULL_NAME:
-                    sanitized_entries.append("")
+                    if not sanitized_entries:
+                        sanitized_entries.append("")
                 else:
                     raise
         sanitized_path = path_separator.join(sanitized_entries)
@@ -326,7 +330,7 @@ def sanitize(self, value, replacement_text=""):
     def validate(self, value):
         self._validate_null_string(value)
 
-        value = os.path.splitdrive(value)[1]
+        value = ntpath.splitdrive(value)[1]
         if not value:
             return
 

test/test_filepath.py

@@ -278,6 +278,7 @@ def test_exception_invalid_win_char(self, value, platform):
             for reserved_keyword, platform in product(
                 WIN_RESERVED_FILE_NAMES, ["windows", "universal"]
             )
+            if reserved_keyword not in [".", ".."]
         ]
         + [
             ["/foo/abc/{}.txt".format(reserved_keyword), platform, ReservedNameError]
@@ -336,23 +337,23 @@ class Test_sanitize_filepath(object):
     @pytest.mark.parametrize(
         ["platform", "value", "replace_text", "expected"],
         [
-            ["universal", "A" + c + "B", rep, "A" + rep + "B"]
+            ["universal", "AA" + c + "B", rep, "AA" + rep + "B"]
             for c, rep in product(SANITIZE_CHARS, REPLACE_TEXT_LIST)
         ]
         + [
-            ["universal", "A" + c + "B", rep, "A" + c + "B"]
+            ["universal", "AA" + c + "B", rep, "AA" + c + "B"]
             for c, rep in product(NOT_SANITIZE_CHARS, REPLACE_TEXT_LIST)
         ]
         + [
             ["universal", "あ" + c + "い", rep, "あ" + c + "い"]
             for c, rep in product(NOT_SANITIZE_CHARS, REPLACE_TEXT_LIST)
         ]
         + [
-            ["linux", "A" + c + "B", rep, "A" + rep + "B"]
+            ["linux", "AA" + c + "B", rep, "AA" + rep + "B"]
             for c, rep in product(INVALID_PATH_CHARS + unprintable_ascii_chars, REPLACE_TEXT_LIST)
         ]
         + [
-            ["linux", "A" + c + "B", rep, "A" + c + "B"]
+            ["linux", "AA" + c + "B", rep, "AA" + c + "B"]
             for c, rep in product([":", "*", "?", '"', "<", ">", "|"], REPLACE_TEXT_LIST)
         ],
     )
@@ -371,9 +372,16 @@ def test_normal_str(self, platform, value, replace_text, expected):
                 platform,
                 "/abc/{}_/xyz".format(reserved_keyword),
             ]
-            for reserved_keyword, platform in product(
-                WIN_RESERVED_FILE_NAMES, ["windows", "universal"]
-            )
+            for reserved_keyword, platform in product(WIN_RESERVED_FILE_NAMES, ["universal"])
+            if reserved_keyword not in [".", ".."]
+        ]
+        + [
+            [
+                "/abc/{}/xyz".format(reserved_keyword),
+                platform,
+                "/abc/{}/xyz".format(reserved_keyword),
+            ]
+            for reserved_keyword, platform in product(WIN_RESERVED_FILE_NAMES, ["linux"])
             if reserved_keyword not in [".", ".."]
         ]
         + [
@@ -382,9 +390,34 @@ def test_normal_str(self, platform, value, replace_text, expected):
                 platform,
                 "/abc/{}_.txt".format(reserved_keyword),
             ]
-            for reserved_keyword, platform in product(
-                WIN_RESERVED_FILE_NAMES, ["windows", "universal"]
-            )
+            for reserved_keyword, platform in product(WIN_RESERVED_FILE_NAMES, ["universal"])
+            if reserved_keyword not in [".", ".."]
+        ]
+        + [
+            [
+                "/abc/{}.txt".format(reserved_keyword),
+                platform,
+                "/abc/{}.txt".format(reserved_keyword),
+            ]
+            for reserved_keyword, platform in product(WIN_RESERVED_FILE_NAMES, ["linux"])
+            if reserved_keyword not in [".", ".."]
+        ]
+        + [
+            [
+                "C:\\abc\\{}.txt".format(reserved_keyword),
+                platform,
+                "C:/abc/{}_.txt".format(reserved_keyword),
+            ]
+            for reserved_keyword, platform in product(WIN_RESERVED_FILE_NAMES, ["universal"])
+            if reserved_keyword not in [".", ".."]
+        ]
+        + [
+            [
+                "C:\\abc\\{}.txt".format(reserved_keyword),
+                platform,
+                "C:\\abc\\{}_.txt".format(reserved_keyword),
+            ]
+            for reserved_keyword, platform in product(WIN_RESERVED_FILE_NAMES, ["windows"])
             if reserved_keyword not in [".", ".."]
         ],
     )
@@ -397,11 +430,11 @@ def test_normal_reserved_name(self, value, test_platform, expected):
     @pytest.mark.parametrize(
         ["value", "replace_text", "expected"],
         [
-            [Path("A" + c + "B"), rep, Path("A" + rep + "B")]
+            [Path("AA" + c + "B"), rep, Path("AA" + rep + "B")]
             for c, rep in product(SANITIZE_CHARS, REPLACE_TEXT_LIST)
         ]
         + [
-            [Path("A" + c + "B"), rep, Path("A" + c + "B")]
+            [Path("AA" + c + "B"), rep, Path("AA" + c + "B")]
             for c, rep in product(NOT_SANITIZE_CHARS, REPLACE_TEXT_LIST)
         ]
         + [