Refactor IRI to URI

2 files changed, 24 insertions, 64 deletions

diff --git a/openid/test/test_xri.py b/openid/test/test_xri.py
index a5f0bfa..341472e 100644
--- a/openid/test/test_xri.py
+++ b/openid/test/test_xri.py
@@ -32,19 +32,10 @@ class XriTransformationTestCase(TestCase):
     def test_to_iri_normal(self):
         self.assertEqual(xri.toIRINormal('@example'), 'xri://@example')
 
-    try:
-        unichr(0x10000)
-    except ValueError:
-        # bleh narrow python build
-        def test_iri_to_url(self):
-            s = u'l\xa1m'
-            expected = 'l%C2%A1m'
-            self.assertEqual(xri.iriToURI(s), expected)
-    else:
-        def test_iri_to_url(self):
-            s = u'l\xa1m\U00101010n'
-            expected = 'l%C2%A1m%F4%81%80%90n'
-            self.assertEqual(xri.iriToURI(s), expected)
+    def test_iri_to_url(self):
+        s = u'l\xa1m\U00101010n'
+        expected = 'l%C2%A1m%F4%81%80%90n'
+        self.assertEqual(xri.iriToURI(s), expected)
 
 
 class CanonicalIDTest(TestCase):
diff --git a/openid/yadis/xri.py b/openid/yadis/xri.py
index 60e0675..73ad798 100644
--- a/openid/yadis/xri.py
+++ b/openid/yadis/xri.py
@@ -4,53 +4,15 @@
 @see: XRI Syntax v2.0 at the
       U{OASIS XRI Technical Committee<http://www.oasis-open.org/committees/tc_home.php?wg_abbrev=xri>}
 """
-
 import re
+import warnings
+from urllib import quote
 
-XRI_AUTHORITIES = ['!', '=', '@', '+', '$', '(']
+import six
 
-try:
-    unichr(0x10000)
-except ValueError:
-    # narrow python build
-    UCSCHAR = [
-        (0xA0, 0xD7FF),
-        (0xF900, 0xFDCF),
-        (0xFDF0, 0xFFEF),
-    ]
-
-    IPRIVATE = [
-        (0xE000, 0xF8FF),
-    ]
-else:
-    UCSCHAR = [
-        (0xA0, 0xD7FF),
-        (0xF900, 0xFDCF),
-        (0xFDF0, 0xFFEF),
-        (0x10000, 0x1FFFD),
-        (0x20000, 0x2FFFD),
-        (0x30000, 0x3FFFD),
-        (0x40000, 0x4FFFD),
-        (0x50000, 0x5FFFD),
-        (0x60000, 0x6FFFD),
-        (0x70000, 0x7FFFD),
-        (0x80000, 0x8FFFD),
-        (0x90000, 0x9FFFD),
-        (0xA0000, 0xAFFFD),
-        (0xB0000, 0xBFFFD),
-        (0xC0000, 0xCFFFD),
-        (0xD0000, 0xDFFFD),
-        (0xE1000, 0xEFFFD),
-    ]
-
-    IPRIVATE = [
-        (0xE000, 0xF8FF),
-        (0xF0000, 0xFFFFD),
-        (0x100000, 0x10FFFD),
-    ]
-
-
-_escapeme_re = re.compile('[%s]' % ''.join(u'%s-%s' % (unichr(m_n[0]), unichr(m_n[1])) for m_n in UCSCHAR + IPRIVATE))
+from openid.urinorm import GEN_DELIMS, PERCENT_ENCODING_CHARACTER, SUB_DELIMS
+
+XRI_AUTHORITIES = ['!', '=', '@', '+', '$', '(']
 
 
 def identifierScheme(identifier):
@@ -96,15 +58,22 @@ def toURINormal(xri):
     return iriToURI(toIRINormal(xri))
 
 
-def _percentEscapeUnicode(char_match):
-    c = char_match.group()
-    return ''.join(['%%%X' % (ord(octet),) for octet in c.encode('utf-8')])
+def iriToURI(iri):
+    """Transform an IRI to a URI by escaping unicode.
 
+    According to RFC 3987, section 3.1, "Mapping of IRIs to URIs"
 
-def iriToURI(iri):
-    """Transform an IRI to a URI by escaping unicode."""
-    # According to RFC 3987, section 3.1, "Mapping of IRIs to URIs"
-    return _escapeme_re.sub(_percentEscapeUnicode, iri)
+    @type iri: six.text_type, six.binary_type deprecated.
+    @rtype: six.text_type
+    """
+    # Transform the input to the binary string. `quote` doesn't quote correctly unicode strings.
+    if isinstance(iri, six.text_type):
+        iri = iri.encode('utf-8')
+    else:
+        assert isinstance(iri, six.binary_type)
+        warnings.warn("Binary input for iriToURI is deprecated. Use text input instead.", DeprecationWarning)
+
+    return quote(iri, (GEN_DELIMS + SUB_DELIMS + PERCENT_ENCODING_CHARACTER).encode('utf-8')).decode('utf-8')
 
 
 def providerIsAuthoritative(providerID, canonicalID):