Add URL encoding helpers

  - Added is_encoded_url(), encode_url() and decode_url()

tests/test_locations.py

@@ -20,7 +20,7 @@
 import xmlschema.locations
 from xmlschema.locations import LocationPath, LocationPosixPath, LocationWindowsPath, \
     is_url, is_local_url, is_remote_url, url_path_is_file, normalize_url, \
-    normalize_locations, match_location
+    normalize_locations, match_location, is_encoded_url, encode_url, decode_url
 
 TEST_CASES_DIR = str(pathlib.Path(__file__).absolute().parent.joinpath('test_cases'))
 
@@ -85,6 +85,33 @@ def check_url(self, url, expected):
             expected_path = PurePath(expected_parts.path)
         self.assertEqual(path, expected_path, "%r: Paths differ." % url)
 
+    def test_urlsplit(self):
+        url = "https://xmlschema.test/schema/test.xsd"
+        self.assertEqual(
+            urlsplit(url), ("https", "xmlschema.test", "/schema/test.xsd", '', '')
+        )
+
+        url = "https://xmlschema.test/xs:schema/test.xsd"
+        self.assertEqual(
+            urlsplit(url), ("https", "xmlschema.test", "/xs:schema/test.xsd", '', '')
+        )
+
+        url = "https://xmlschema.test/schema/test.xsd#xs:element"
+        self.assertEqual(
+            urlsplit(url), ("https", "xmlschema.test", "/schema/test.xsd", '', 'xs:element')
+        )
+
+        url = "https://xmlschema.test@username:password/schema/test.xsd"
+        self.assertEqual(
+            urlsplit(url),
+            ("https", "xmlschema.test@username:password", "/schema/test.xsd", '', '')
+        )
+
+        url = "https://xmlschema.test/schema/test.xsd?id=10"
+        self.assertEqual(
+            urlsplit(url), ("https", "xmlschema.test", "/schema/test.xsd", 'id=10', '')
+        )
+
     def test_path_from_uri(self):
         with self.assertRaises(ValueError) as ec:
             LocationPath.from_uri('')
@@ -346,6 +373,37 @@ def test_url_path_is_file_function(self):
         with patch('platform.system', MagicMock(return_value="Windows")):
             self.assertFalse(url_path_is_file('file:///c:/Windows/unknown'))
 
+    def test_is_encoded_url(self):
+        self.assertFalse(is_encoded_url("https://xmlschema.test/schema/test.xsd"))
+        self.assertTrue(is_encoded_url("https://xmlschema.test/schema/issue%20%231999.xsd"))
+        self.assertFalse(is_encoded_url("a b c"))
+        self.assertFalse(is_encoded_url("a+b+c"))
+        self.assertFalse(is_encoded_url("a b+c"))
+
+    def test_encode_and_decode_url(self):
+        url = "https://xmlschema.test/schema/test.xsd"
+        self.assertEqual(encode_url(url), url)
+        self.assertEqual(decode_url(encode_url(url)), url)
+
+        url = "https://xmlschema.test/schema 2/test.xsd"
+        self.assertEqual(encode_url(url), "https://xmlschema.test/schema%202/test.xsd")
+        self.assertEqual(decode_url(encode_url(url)), url)
+
+        url = "https://xmlschema.test@u:p/xs:schema@2/test.xsd"
+        self.assertEqual(encode_url(url), "https://xmlschema.test@u:p/xs%3Aschema%402/test.xsd")
+        self.assertEqual(decode_url(encode_url(url)), url)
+
+        url = "https://xmlschema.test/schema 2/test.xsd?name=2 id=3"
+        self.assertEqual(
+            encode_url(url), "https://xmlschema.test/schema%202/test.xsd?name=2%20id=3")
+        self.assertEqual(decode_url(encode_url(url)), url)
+
+        self.assertEqual(encode_url(url, method='html'),
+                         "https://xmlschema.test/schema%202/test.xsd?name=2+id=3")
+        self.assertEqual(decode_url(encode_url(url, method='html'), method='html'), url)
+        self.assertEqual(decode_url(encode_url(url), method='html'), url)
+        self.assertNotEqual(decode_url(encode_url(url, method='html')), url)
+
     def test_normalize_locations_function(self):
         locations = normalize_locations(
             [('tns0', 'alpha'), ('tns1', 'http://example.com/beta')], base_url='/home/user'

xmlschema/locations.py

@@ -15,7 +15,7 @@
 from collections.abc import MutableMapping
 from pathlib import Path, PurePath, PurePosixPath, PureWindowsPath
 from typing import Optional, Iterable
-from urllib.parse import urlsplit, urlunsplit, unquote, quote_from_bytes
+from urllib.parse import urlsplit, urlunsplit, quote, quote_plus, unquote, unquote_plus, quote_from_bytes
 
 from .exceptions import XMLSchemaValueError
 from .aliases import NormalizedLocationsType, LocationsType
@@ -239,6 +239,50 @@ def url_path_is_file(url: str) -> bool:
     return os.path.isfile(path)
 
 
+def is_encoded_url(url: str) -> bool:
+    """
+    Determines whether the given URL is encoded. The case with '+' and without
+    spaces is not univocal and the plus signs are ignored for the result.
+    """
+    return unquote(url) != url or \
+        '+' in url and ' ' not in url and \
+        unquote(url.replace('+', '$')) != url.replace('+', '$')
+
+
+def encode_url(url: str, method: str = 'xml') -> str:
+    """Encode the given url, if necessary."""
+    if is_encoded_url(url):
+        return url
+
+    query_quote = quote_plus if method == 'html' else quote
+
+    parts = urlsplit(url)
+    return urlunsplit((
+        parts.scheme,
+        quote(parts.netloc, safe='@:'),
+        quote(parts.path, safe='/'),
+        query_quote(parts.query, safe=';/?:@=&'),
+        query_quote(parts.fragment, safe=';/?:@=&'),
+    ))
+
+
+def decode_url(url: str, method: str = 'xml') -> str:
+    """Decode the given url, if necessary."""
+    if not is_encoded_url(url):
+        return url
+
+    query_unquote = unquote_plus if method == 'html' else unquote
+
+    parts = urlsplit(url)
+    return urlunsplit((
+        parts.scheme,
+        unquote(parts.netloc),
+        unquote(parts.path),
+        query_unquote(parts.query),
+        query_unquote(parts.fragment),
+    ))
+
+
 def normalize_locations(locations: LocationsType,
                         base_url: Optional[str] = None,
                         keep_relative: bool = False) -> NormalizedLocationsType: