Add a Cython version of slixmpp.stringprep, using libidn.

This makes the validation of a JID a *lot* faster.

1 files changed, 71 insertions, 0 deletions

diff --git a/slixmpp/stringprep.pyx b/slixmpp/stringprep.pyx
new file mode 100644
index 00000000..e17c62c3
--- /dev/null
+++ b/slixmpp/stringprep.pyx
@@ -0,0 +1,71 @@
+# -*- coding: utf-8 -*-
+# cython: language_level = 3
+# distutils: libraries = idn
+"""
+    slixmpp.stringprep
+    ~~~~~~~~~~~~~~~~~~~~~~~
+
+    This module wraps libidn’s stringprep and idna functions using Cython.
+
+    Part of Slixmpp: The Slick XMPP Library
+
+    :copyright: (c) 2015 Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
+    :license: MIT, see LICENSE for more details
+"""
+
+from libc.stdlib cimport free
+
+
+# Those are Cython declarations for the C function we’ll be using.
+
+cdef extern from "stringprep.h" nogil:
+    int stringprep_profile(const char* in_, char** out, const char* profile, int flags)
+
+cdef extern from "idna.h" nogil:
+    int idna_to_ascii_8z(const char* in_, char** out, int flags)
+    int idna_to_unicode_8z8z(const char* in_, char** out, int flags)
+
+
+class StringprepError(Exception):
+    pass
+
+
+cdef str _stringprep(str in_, const char* profile):
+    """Python wrapper for libidn’s stringprep."""
+    cdef char* out
+    ret = stringprep_profile(in_.encode('utf-8'), &out, profile, 0)
+    if ret != 0:
+        raise StringprepError(ret)
+    unicode_out = out.decode('utf-8')
+    free(out)
+    return unicode_out
+
+def nodeprep(str node):
+    """The nodeprep profile of stringprep used to validate the local, or
+    username, portion of a JID."""
+    return _stringprep(node, 'Nodeprep')
+
+def resourceprep(str resource):
+    """The resourceprep profile of stringprep, which is used to validate the
+    resource portion of a JID."""
+    return _stringprep(resource, 'Resourceprep')
+
+def idna(str domain):
+    """The idna conversion functions, which are used to validate the domain
+    portion of a JID."""
+
+    cdef char* ascii_domain
+    cdef char* utf8_domain
+
+    ret = idna_to_ascii_8z(domain.encode('utf-8'), &ascii_domain, 0)
+    if ret != 0:
+        raise StringprepError(ret)
+
+    ret = idna_to_unicode_8z8z(ascii_domain, &utf8_domain, 0)
+    free(ascii_domain)
+    if ret != 0:
+        raise StringprepError(ret)
+
+    unicode_domain = utf8_domain.decode('utf-8')
+    free(utf8_domain)
+    return unicode_domain