From 4ff3d8eccf8a934c74e44578b44c56a1c4b59542 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Philipp=20H=C3=B6rist?= <philipp@hoerist.com>
Date: Sun, 17 Sep 2023 21:21:43 +0200
Subject: feat: JID: Add JID.from_iri()

---
 nbxmpp/protocol.py            | 63 +++++++++++++++++++++++++++++++------------
 nbxmpp/xmppiri.py             | 17 ++++++++++++
 test/unit/test_jid_parsing.py | 11 ++++++++
 3 files changed, 74 insertions(+), 17 deletions(-)

diff --git a/nbxmpp/protocol.py b/nbxmpp/protocol.py
index fe88984..0bcc0e9 100644
--- a/nbxmpp/protocol.py
+++ b/nbxmpp/protocol.py
@@ -39,6 +39,7 @@ from dataclasses import asdict
 from gi.repository import GLib
 
 import idna
+from nbxmpp.xmppiri import clean_iri
 from nbxmpp.xmppiri import escape_ifragment
 from nbxmpp.xmppiri import escape_inode
 from nbxmpp.xmppiri import escape_ires
@@ -491,6 +492,31 @@ def deprecation_warning(message):
     warnings.warn(message, DeprecationWarning)
 
 
+def split_jid_string(
+    jid_string: str
+) -> tuple[str | None, str, str | None]:
+
+    # https://tools.ietf.org/html/rfc7622#section-3.2
+
+    # Remove any portion from the first '/' character to the end of the
+    # string (if there is a '/' character present).
+
+    # Remove any portion from the beginning of the string to the first
+    # '@' character (if there is an '@' character present).
+
+    if jid_string.find('/') != -1:
+        rest, resourcepart = jid_string.split('/', 1)
+    else:
+        rest, resourcepart = jid_string, None
+
+    if rest.find('@') != -1:
+        localpart, domainpart = rest.split('@', 1)
+    else:
+        localpart, domainpart = None, rest
+
+    return localpart, domainpart, resourcepart
+
+
 @functools.lru_cache(maxsize=None)
 def validate_localpart(localpart: str) -> str:
     if not localpart or len(localpart.encode()) > 1023:
@@ -635,23 +661,7 @@ class JID:
     @classmethod
     @functools.lru_cache(maxsize=None)
     def from_string(cls, jid_string: str, force_bare: bool = False) -> JID:
-        # https://tools.ietf.org/html/rfc7622#section-3.2
-
-        # Remove any portion from the first '/' character to the end of the
-        # string (if there is a '/' character present).
-
-        # Remove any portion from the beginning of the string to the first
-        # '@' character (if there is an '@' character present).
-
-        if jid_string.find('/') != -1:
-            rest, resourcepart = jid_string.split('/', 1)
-        else:
-            rest, resourcepart = jid_string, None
-
-        if rest.find('@') != -1:
-            localpart, domainpart = rest.split('@', 1)
-        else:
-            localpart, domainpart = None, rest
+        localpart, domainpart, resourcepart = split_jid_string(jid_string)
 
         if force_bare:
             resourcepart = None
@@ -693,6 +703,25 @@ class JID:
                    domain=domainpart,
                    resource=None)
 
+    @classmethod
+    @functools.lru_cache(maxsize=None)
+    def from_iri(cls, iri_str: str, *, force_bare: bool = False) -> JID:
+        iri_str = clean_iri(iri_str)
+        localpart, domainpart, resourcepart = split_jid_string(iri_str)
+
+        if localpart is not None:
+            localpart = GLib.Uri.unescape_string(localpart)
+
+        if force_bare:
+            resourcepart = None
+
+        if resourcepart is not None:
+            resourcepart = GLib.Uri.unescape_string(resourcepart)
+
+        return cls(localpart=localpart,
+                   domain=domainpart,
+                   resource=resourcepart)
+
     def __str__(self) -> str:
         if self.localpart:
             jid = f'{self.localpart}@{self.domain}'
diff --git a/nbxmpp/xmppiri.py b/nbxmpp/xmppiri.py
index bc0035e..fd3c491 100644
--- a/nbxmpp/xmppiri.py
+++ b/nbxmpp/xmppiri.py
@@ -115,3 +115,20 @@ def escape_inode(node: str) -> str:
 
 def escape_ires(res: str) -> str:
     return _escape(res, rx_ires)
+
+
+def clean_iri(iri_str: str) -> str:
+    if not iri_str.startswith('xmpp:'):
+        raise ValueError('IRI must start with xmpp scheme')
+
+    iri_str = iri_str.removeprefix('xmpp:')
+
+    if iri_str.startswith('//'):
+        # Remove auth component
+        iri_str = iri_str.removeprefix('//')
+        iri_str = iri_str.split('/', maxsplit=1)[1]
+
+    # Remove query and fragment
+    iri_str = iri_str.split('?', maxsplit=1)[0]
+    iri_str = iri_str.split('#', maxsplit=1)[0]
+    return iri_str
diff --git a/test/unit/test_jid_parsing.py b/test/unit/test_jid_parsing.py
index 7199595..f118bf8 100644
--- a/test/unit/test_jid_parsing.py
+++ b/test/unit/test_jid_parsing.py
@@ -213,3 +213,14 @@ class JIDParsing(unittest.TestCase):
 
         jid = JID.from_user_input('call me "ishmael"@example.com')
         self.assertEqual(jid.to_iri(), 'xmpp:call%5C20me%5C20%5C22ishmael%5C22@example.com')
+
+    def test_iri_to_jid(self):
+        tests = [
+            ('nasty!#$%()*+,-.;=?[\\]^_`{|}~node@example.com', 'xmpp:nasty!%23$%25()*+,-.;=%3F%5B%5C%5D%5E_%60%7B%7C%7D~node@example.com'),
+            ('node@example.com/repulsive !#"$%&\'()*+,-./:;<=>?@[\\]^_`{|}~resource', 'xmpp:node@example.com/repulsive%20!%23%22$%25&\'()*+,-.%2F:;%3C=%3E%3F%40%5B%5C%5D%5E_%60%7B%7C%7D~resource'),
+            ('jiři@čechy.example/v Praze', 'xmpp:jiři@čechy.example/v%20Praze')
+        ]
+
+        for jid_string, iri_string in tests:
+            parsed_jid = JID.from_iri(iri_string)
+            self.assertEqual(str(parsed_jid), jid_string)
-- 
cgit v1.2.3