Upgrade to 2.6.0, add IDNA functions to library

ada_url/__init__.py

@@ -1,6 +1,9 @@
 from ada_url.ada_adapter import (
     URL,
     check_url,
+    idna,
+    idna_to_ascii,
+    idna_to_unicode,
     join_url,
     normalize_url,
     parse_url,
@@ -10,6 +13,9 @@
 __all__ = [
     'URL',
     'check_url',
+    'idna',
+    'idna_to_ascii',
+    'idna_to_unicode',
     'join_url',
     'normalize_url',
     'parse_url',

ada_url/ada_adapter.py

@@ -18,10 +18,10 @@
 SET_ATTRIBUTES = frozenset(URL_ATTRIBUTES)
 
 
-def _get_urlobj(constructor, *args):
-    urlobj = constructor(*args)
+def _get_obj(constructor, destructor, *args):
+    obj = constructor(*args)
 
-    return ffi.gc(urlobj, lib.ada_free)
+    return ffi.gc(obj, destructor)
 
 
 def _get_str(x):
@@ -79,11 +79,14 @@ def __init__(self, url, base=None):
         url_bytes = url.encode('utf-8')
 
         if base is None:
-            self.urlobj = _get_urlobj(lib.ada_parse, url_bytes, len(url_bytes))
+            self.urlobj = _get_obj(
+                lib.ada_parse, lib.ada_free, url_bytes, len(url_bytes)
+            )
         else:
             base_bytes = base.encode('utf-8')
-            self.urlobj = _get_urlobj(
+            self.urlobj = _get_obj(
                 lib.ada_parse_with_base,
+                lib.ada_free,
                 url_bytes,
                 len(url_bytes),
                 base_bytes,
@@ -168,7 +171,7 @@ def check_url(s):
     except Exception:
         return False
 
-    urlobj = _get_urlobj(lib.ada_parse, s_bytes, len(s_bytes))
+    urlobj = _get_obj(lib.ada_parse, lib.ada_free, s_bytes, len(s_bytes))
     return lib.ada_is_valid(urlobj)
 
 
@@ -191,8 +194,13 @@ def join_url(base_url, s):
     except Exception:
         raise ValueError('Invalid URL') from None
 
-    urlobj = _get_urlobj(
-        lib.ada_parse_with_base, s_bytes, len(s_bytes), base_bytes, len(base_bytes)
+    urlobj = _get_obj(
+        lib.ada_parse_with_base,
+        lib.ada_free,
+        s_bytes,
+        len(s_bytes),
+        base_bytes,
+        len(base_bytes),
     )
     if not lib.ada_is_valid(urlobj):
         raise ValueError('Invalid URL') from None
@@ -258,7 +266,7 @@ def parse_url(s, attributes=PARSE_ATTRIBUTES):
         raise ValueError('Invalid URL') from None
 
     ret = {}
-    urlobj = _get_urlobj(lib.ada_parse, s_bytes, len(s_bytes))
+    urlobj = _get_obj(lib.ada_parse, lib.ada_free, s_bytes, len(s_bytes))
     if not lib.ada_is_valid(urlobj):
         raise ValueError('Invalid URL') from None
 
@@ -295,7 +303,7 @@ def replace_url(s, **kwargs):
     except Exception:
         raise ValueError('Invalid URL') from None
 
-    urlobj = _get_urlobj(lib.ada_parse, s_bytes, len(s_bytes))
+    urlobj = _get_obj(lib.ada_parse, lib.ada_free, s_bytes, len(s_bytes))
     if not lib.ada_is_valid(urlobj):
         raise ValueError('Invalid URL') from None
 
@@ -315,3 +323,51 @@ def replace_url(s, **kwargs):
             raise ValueError(f'Invalid value for {attr}') from None
 
     return _get_str(lib.ada_get_href(urlobj))
+
+
+class idna:
+    """Process international domains according to the UTS #46 standard.
+
+    :func:`idna.encode` implements the UTS #46 ``ToASCII`` operation.
+    Its output is a Python ``bytes`` object.
+    It is also available as :func:`idna_to_ascii`.
+
+    .. code-block:: python
+
+        >>> from ada_url import idna
+        >>> idna.encode('meßagefactory.ca')
+        b'xn--meagefactory-m9a.ca'
+
+    :func:`idna.decode` implements the UTS #46 ``ToUnicode`` operation.
+    Its oputput is a Python ``str`` object.
+    It is also available as :func:`idna_to_unicode`.
+
+    .. code-block:: python
+
+        >>> from ada_url import idna
+        >>> idna.decode('xn--meagefactory-m9a.ca')
+        'meßagefactory.ca'
+
+    Both functions accept either ``str`` or ``bytes`` objects as input.
+    """
+
+    @staticmethod
+    def decode(s):
+        if isinstance(s, str):
+            s = s.encode('ascii')
+
+        data = _get_obj(lib.ada_idna_to_unicode, lib.ada_free_owned_string, s, len(s))
+        return _get_str(data)
+
+    @staticmethod
+    def encode(s):
+        if isinstance(s, str):
+            s = s.encode('utf-8')
+
+        val = _get_obj(lib.ada_idna_to_ascii, lib.ada_free_owned_string, s, len(s))
+        return ffi.string(val.data, val.length) if val.length else b''
+
+
+idna_to_unicode = idna.decode
+
+idna_to_ascii = idna.encode

ada_url/ada_c.h

@@ -97,4 +97,8 @@ bool ada_has_search(ada_url result);
 // returns a pointer to the internal url_aggregator::url_components
 const ada_url_components* ada_get_components(ada_url result);
 
+// idna methods
+ada_owned_string ada_idna_to_unicode(const char* input, size_t length);
+ada_owned_string ada_idna_to_ascii(const char* input, size_t length);
+
 #endif  // ADA_C_H

setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = ada-url
-version = 1.0.1
+version = 1.1.0
 description = 'URL parser and manipulator based on the WHAT WG URL standard'
 long_description = file: README.rst
 long_description_content_type = text/x-rst

tests/test_ada_url.py

@@ -3,6 +3,9 @@
 from ada_url import (
     URL,
     check_url,
+    idna,
+    idna_to_ascii,
+    idna_to_unicode,
     join_url,
     normalize_url,
     parse_url,
@@ -290,3 +293,16 @@ def test_replace_url_error(self):
             with self.subTest(s=s, kwargs=kwargs):
                 with self.assertRaises(ValueError):
                     replace_url(s, **kwargs)
+
+    def test_idna_decode(self):
+        self.assertEqual(idna.decode('xn--meagefactory-m9a.ca'), 'meßagefactory.ca')
+        self.assertEqual(
+            idna_to_unicode(b'xn--meagefactory-m9a.ca'), 'meßagefactory.ca'
+        )
+
+    def test_idna_encode(self):
+        self.assertEqual(idna.encode('meßagefactory.ca'), b'xn--meagefactory-m9a.ca')
+        self.assertEqual(
+            idna_to_ascii('meßagefactory.ca'.encode('utf-8')),
+            b'xn--meagefactory-m9a.ca',
+        )