Initial Rust version.

.gitignore

@@ -14,4 +14,9 @@ slixmpp.egg-info/
 .DS_STORE
 .idea/
 .vscode/
-venv/
+venv/
+
+# Added by cargo
+
+/target
+/Cargo.lock

Cargo.toml

@@ -0,0 +1,13 @@
+[package]
+name = "slixmpp"
+version = "0.1.0"
+edition = "2021"
+
+# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+
+[dependencies]
+jid = "0.10"
+pyo3 = "0.21"
+
+[lib]
+crate-type = ["cdylib"]

slixmpp/jid.py

@@ -1,346 +1 @@
-
+from libslixmpp import JID, InvalidJID
-# slixmpp.jid
-# ~~~~~~~~~~~~~~~~~~~~~~~
-# This module allows for working with Jabber IDs (JIDs).
-# Part of Slixmpp: The Slick XMPP Library
-# :copyright: (c) 2011 Nathanael C. Fritz
-# :license: MIT, see LICENSE for more details
-from __future__ import annotations
-
-import re
-import socket
-
-from functools import lru_cache
-from typing import (
-    Optional,
-    Union,
-)
-
-from slixmpp.stringprep import nodeprep, resourceprep, idna, StringprepError
-
-HAVE_INET_PTON = hasattr(socket, 'inet_pton')
-
-#: The basic regex pattern that a JID must match in order to determine
-#: the local, domain, and resource parts. This regex does NOT do any
-#: validation, which requires application of nodeprep, resourceprep, etc.
-JID_PATTERN = re.compile(
-    "^(?:([^\"&'/:<>@]{1,1023})@)?([^/@]{1,1023})(?:/(.{1,1023}))?$"
-)
-
-#: The set of escape sequences for the characters not allowed by nodeprep.
-JID_ESCAPE_SEQUENCES = {'\\20', '\\22', '\\26', '\\27', '\\2f',
-                        '\\3a', '\\3c', '\\3e', '\\40', '\\5c'}
-
-
-# TODO: Find the best cache size for a standard usage.
-@lru_cache(maxsize=1024)
-def _parse_jid(data: str):
-    """
-    Parse string data into the node, domain, and resource
-    components of a JID, if possible.
-
-    :param string data: A string that is potentially a JID.
-
-    :raises InvalidJID:
-
-    :returns: tuple of the validated local, domain, and resource strings
-    """
-    match = JID_PATTERN.match(data)
-    if not match:
-        raise InvalidJID('JID could not be parsed')
-
-    (node, domain, resource) = match.groups()
-
-    node = _validate_node(node)
-    domain = _validate_domain(domain)
-    resource = _validate_resource(resource)
-
-    return node, domain, resource
-
-
-def _validate_node(node: Optional[str]):
-    """Validate the local, or username, portion of a JID.
-
-    :raises InvalidJID:
-
-    :returns: The local portion of a JID, as validated by nodeprep.
-    """
-    if node is None:
-        return ''
-
-    try:
-        node = nodeprep(node)
-    except StringprepError:
-        raise InvalidJID('Nodeprep failed')
-
-    if not node:
-        raise InvalidJID('Localpart must not be 0 bytes')
-    if len(node) > 1023:
-        raise InvalidJID('Localpart must be less than 1024 bytes')
-    return node
-
-
-def _validate_domain(domain: str):
-    """Validate the domain portion of a JID.
-
-    IP literal addresses are left as-is, if valid. Domain names
-    are stripped of any trailing label separators (`.`), and are
-    checked with the nameprep profile of stringprep. If the given
-    domain is actually a punyencoded version of a domain name, it
-    is converted back into its original Unicode form. Domains must
-    also not start or end with a dash (`-`).
-
-    :raises InvalidJID:
-
-    :returns: The validated domain name
-    """
-    ip_addr = False
-
-    # First, check if this is an IPv4 address
-    try:
-        socket.inet_aton(domain)
-        ip_addr = True
-    except socket.error:
-        pass
-
-    # Check if this is an IPv6 address
-    if not ip_addr and HAVE_INET_PTON and domain[0] == '[' and domain[-1] == ']':
-        try:
-            ip = domain[1:-1]
-            socket.inet_pton(socket.AF_INET6, ip)
-            ip_addr = True
-        except (socket.error, ValueError):
-            pass
-
-    if not ip_addr:
-        # This is a domain name, which must be checked further
-
-        if domain and domain[-1] == '.':
-            domain = domain[:-1]
-
-        try:
-            domain = idna(domain)
-        except StringprepError:
-            raise InvalidJID(f'idna validation failed: {domain}')
-
-        if ':' in domain:
-            raise InvalidJID(f'Domain containing a port: {domain}')
-        for label in domain.split('.'):
-            if not label:
-                raise InvalidJID(f'Domain containing too many dots: {domain}')
-            if '-' in (label[0], label[-1]):
-                raise InvalidJID(f'Domain starting or ending with -: {domain}')
-
-    if not domain:
-        raise InvalidJID('Domain must not be 0 bytes')
-    if len(domain) > 1023:
-        raise InvalidJID('Domain must be less than 1024 bytes')
-
-    return domain
-
-
-def _validate_resource(resource: Optional[str]):
-    """Validate the resource portion of a JID.
-
-    :raises InvalidJID:
-
-    :returns: The local portion of a JID, as validated by resourceprep.
-    """
-    if resource is None:
-        return ''
-
-    try:
-        resource = resourceprep(resource)
-    except StringprepError:
-        raise InvalidJID('Resourceprep failed')
-
-    if not resource:
-        raise InvalidJID('Resource must not be 0 bytes')
-    if len(resource) > 1023:
-        raise InvalidJID('Resource must be less than 1024 bytes')
-    return resource
-
-
-def _format_jid(
-        local: Optional[str] = None,
-        domain: Optional[str] = None,
-        resource: Optional[str] = None,
-    ):
-    """Format the given JID components into a full or bare JID.
-
-    :param string local: Optional. The local portion of the JID.
-    :param string domain: Required. The domain name portion of the JID.
-    :param strin resource: Optional. The resource portion of the JID.
-
-    :return: A full or bare JID string.
-    """
-    if domain is None:
-        return ''
-    if local is not None:
-        result = local + '@' + domain
-    else:
-        result = domain
-    if resource is not None:
-        result += '/' + resource
-    return result
-
-
-class InvalidJID(ValueError):
-    """
-    Raised when attempting to create a JID that does not pass validation.
-
-    It can also be raised if modifying an existing JID in such a way as
-    to make it invalid, such trying to remove the domain from an existing
-    full JID while the local and resource portions still exist.
-    """
-
-
-class JID:
-
-    """
-    A representation of a Jabber ID, or JID.
-
-    Each JID may have three components: a user, a domain, and an optional
-    resource. For example: user@domain/resource
-
-    When a resource is not used, the JID is called a bare JID.
-    The JID is a full JID otherwise.
-
-    **JID Properties:**
-        :full: The string value of the full JID.
-        :jid: Alias for ``full``.
-        :bare: The string value of the bare JID.
-        :node: The node portion of the JID.
-        :user: Alias for ``node``.
-        :local: Alias for ``node``.
-        :username: Alias for ``node``.
-        :domain: The domain name portion of the JID.
-        :server: Alias for ``domain``.
-        :host: Alias for ``domain``.
-        :resource: The resource portion of the JID.
-
-    :param string jid:
-        A string of the form ``'[user@]domain[/resource]'``.
-    :param bool bare:
-        If present, discard the provided resource.
-
-    :raises InvalidJID:
-    """
-
-    __slots__ = ('_node', '_domain', '_resource', '_bare', '_full')
-
-    def __init__(self, jid: Optional[Union[str, 'JID']] = None, bare: bool = False):
-        if not jid:
-            self._node = ''
-            self._domain = ''
-            self._resource = ''
-            self._bare = ''
-            self._full = ''
-            return
-        elif not isinstance(jid, JID):
-            node, domain, resource = _parse_jid(jid)
-            self._node = node
-            self._domain = domain
-            self._resource = resource if not bare else ''
-        else:
-            self._node = jid._node
-            self._domain = jid._domain
-            self._resource = jid._resource if not bare else ''
-        self._update_bare_full()
-
-    def _update_bare_full(self):
-        """Format the given JID into a bare and a full JID.
-        """
-        self._bare = (self._node + '@' + self._domain
-                      if self._node
-                      else self._domain)
-        self._full = (self._bare + '/' + self._resource
-                      if self._resource
-                      else self._bare)
-
-    @property
-    def bare(self) -> str:
-        return self._bare
-
-    @bare.setter
-    def bare(self, value: str):
-        node, domain, resource = _parse_jid(value)
-        assert not resource
-        self._node = node
-        self._domain = domain
-        self._update_bare_full()
-
-
-    @property
-    def node(self) -> str:
-        return self._node
-
-    @node.setter
-    def node(self, value: Optional[str]):
-        self._node = _validate_node(value)
-        self._update_bare_full()
-
-    @property
-    def domain(self) -> str:
-        return self._domain
-
-    @domain.setter
-    def domain(self, value: str):
-        self._domain = _validate_domain(value)
-        self._update_bare_full()
-
-    @property
-    def resource(self) -> str:
-        return self._resource
-
-    @resource.setter
-    def resource(self, value: Optional[str]):
-        self._resource = _validate_resource(value)
-        self._update_bare_full()
-
-    @property
-    def full(self) -> str:
-        return self._full
-
-    @full.setter
-    def full(self, value: str):
-        self._node, self._domain, self._resource = _parse_jid(value)
-        self._update_bare_full()
-
-    user = node
-    local = node
-    username = node
-
-    server = domain
-    host = domain
-
-    jid = full
-
-    def __str__(self):
-        """Use the full JID as the string value."""
-        return self._full
-
-    def __repr__(self):
-        """Use the full JID as the representation."""
-        return self._full
-
-    # pylint: disable=W0212
-    def __eq__(self, other):
-        """Two JIDs are equal if they have the same full JID value."""
-        if not isinstance(other, JID):
-            try:
-                other = JID(other)
-            except InvalidJID:
-                return NotImplemented
-
-        return (self._node == other._node and
-                self._domain == other._domain and
-                self._resource == other._resource)
-
-    def __ne__(self, other):
-        """Two JIDs are considered unequal if they are not equal."""
-        return not self == other
-
-    def __hash__(self):
-        """Hash a JID based on the string version of its full JID."""
-        return hash(self._full)

src/lib.rs

@@ -0,0 +1,278 @@
+use pyo3::exceptions::{PyNotImplementedError, PyValueError};
+use pyo3::prelude::*;
+
+pyo3::create_exception!(py_jid, InvalidJID, PyValueError, "Raised when attempting to create a JID that does not pass validation.\n\nIt can also be raised if modifying an existing JID in such a way as\nto make it invalid, such trying to remove the domain from an existing\nfull JID while the local and resource portions still exist.");
+
+fn to_exc(err: jid::Error) -> PyErr {
+    InvalidJID::new_err(err.to_string())
+}
+
+/// A representation of a Jabber ID, or JID.
+///
+/// Each JID may have three components: a user, a domain, and an optional resource. For example:
+/// user@domain/resource
+///
+/// When a resource is not used, the JID is called a bare JID. The JID is a full JID otherwise.
+///
+/// Raises InvalidJID if the parser rejects it.
+#[pyclass(name = "JID", module = "slixmpp.jid")]
+struct PyJid {
+    jid: Option<jid::Jid>,
+}
+
+#[pymethods]
+impl PyJid {
+    #[new]
+    #[pyo3(signature = (jid=None, bare=false))]
+    fn new(jid: Option<&Bound<'_, PyAny>>, bare: bool) -> PyResult<Self> {
+        if let Some(jid) = jid {
+            if let Ok(py_jid) = jid.extract::<PyRef<PyJid>>() {
+                if bare {
+                    if let Some(py_jid) = &(*py_jid).jid {
+                        Ok(PyJid {
+                            jid: Some(jid::Jid::Bare(py_jid.to_bare())),
+                        })
+                    } else {
+                        Ok(PyJid { jid: None })
+                    }
+                } else {
+                    Ok(PyJid {
+                        jid: (*py_jid).jid.clone(),
+                    })
+                }
+            } else {
+                let jid: &str = jid.extract()?;
+                if jid.is_empty() {
+                    Ok(PyJid { jid: None })
+                } else {
+                    let mut jid = jid::Jid::new(jid).map_err(to_exc)?;
+                    if bare {
+                        jid = jid::Jid::Bare(jid.into_bare())
+                    }
+                    Ok(PyJid { jid: Some(jid) })
+                }
+            }
+        } else {
+            Ok(PyJid { jid: None })
+        }
+    }
+
+    /*
+    // TODO: implement or remove from the API
+    fn unescape() {
+    }
+    */
+
+    #[getter]
+    fn get_bare(&self) -> String {
+        match &self.jid {
+            None => String::new(),
+            Some(jid) => jid.to_bare().to_string(),
+        }
+    }
+
+    #[setter]
+    fn set_bare(&mut self, bare: &str) -> PyResult<()> {
+        let bare = jid::BareJid::new(bare).map_err(to_exc)?;
+        self.jid = Some(match &self.jid {
+            Some(jid::Jid::Bare(_)) | None => jid::Jid::Bare(bare),
+            Some(jid::Jid::Full(jid)) => jid::Jid::Full(bare.with_resource(&jid.resource())),
+        });
+        Ok(())
+    }
+
+    #[getter]
+    fn get_full(&self) -> String {
+        match &self.jid {
+            None => String::new(),
+            Some(jid) => jid.to_string(),
+        }
+    }
+
+    #[setter]
+    fn set_full(&mut self, full: &str) -> PyResult<()> {
+        // JID.full = 'domain' is acceptable in slixmpp.
+        self.jid = Some(jid::Jid::new(full).map_err(to_exc)?);
+        Ok(())
+    }
+
+    #[getter]
+    fn get_node(&self) -> String {
+        match &self.jid {
+            None => String::new(),
+            Some(jid) => jid
+                .node_str()
+                .map(ToString::to_string)
+                .unwrap_or_else(String::new),
+        }
+    }
+
+    #[setter]
+    fn set_node(&mut self, node: &str) -> PyResult<()> {
+        let node = jid::NodePart::new(node).map_err(to_exc)?;
+        self.jid = Some(match &self.jid {
+            Some(jid::Jid::Bare(jid)) => {
+                jid::Jid::Bare(jid::BareJid::from_parts(Some(&node), &jid.domain()))
+            }
+            Some(jid::Jid::Full(jid)) => jid::Jid::Full(jid::FullJid::from_parts(
+                Some(&node),
+                &jid.domain(),
+                &jid.resource(),
+            )),
+            None => Err(InvalidJID::new_err("JID.node must apply to a proper JID"))?,
+        });
+        Ok(())
+    }
+
+    #[getter]
+    fn get_domain(&self) -> String {
+        match &self.jid {
+            None => String::new(),
+            Some(jid) => jid.domain_str().to_string(),
+        }
+    }
+
+    #[setter]
+    fn set_domain(&mut self, domain: &str) -> PyResult<()> {
+        let domain = jid::DomainPart::new(domain).map_err(to_exc)?;
+        self.jid = Some(match &self.jid {
+            Some(jid::Jid::Bare(jid)) => {
+                jid::Jid::Bare(jid::BareJid::from_parts(jid.node().as_ref(), &domain))
+            }
+            Some(jid::Jid::Full(jid)) => jid::Jid::Full(jid::FullJid::from_parts(
+                jid.node().as_ref(),
+                &domain,
+                &jid.resource(),
+            )),
+            None => jid::Jid::Bare(jid::BareJid::from_parts(None, &domain)),
+        });
+        Ok(())
+    }
+
+    #[getter]
+    fn get_resource(&self) -> String {
+        match &self.jid {
+            None => String::new(),
+            Some(jid) => jid
+                .resource_str()
+                .map(ToString::to_string)
+                .unwrap_or_else(String::new),
+        }
+    }
+
+    #[setter]
+    fn set_resource(&mut self, resource: &str) -> PyResult<()> {
+        let resource = jid::ResourcePart::new(resource).map_err(to_exc)?;
+        self.jid = Some(match &self.jid {
+            Some(jid::Jid::Bare(jid)) => jid::Jid::Full(jid.with_resource(&resource)),
+            Some(jid::Jid::Full(jid)) => jid::Jid::Full(jid::FullJid::from_parts(
+                jid.node().as_ref(),
+                &jid.domain(),
+                &resource,
+            )),
+            None => Err(InvalidJID::new_err(
+                "JID.resource must apply to a proper JID",
+            ))?,
+        });
+        Ok(())
+    }
+
+    /// Use the full JID as the string value.
+    fn __str__(&self) -> String {
+        match &self.jid {
+            None => String::new(),
+            Some(jid) => jid.to_string(),
+        }
+    }
+
+    /// Use the full JID as the representation.
+    fn __repr__(&self) -> String {
+        match &self.jid {
+            None => String::new(),
+            Some(jid) => jid.to_string(),
+        }
+    }
+
+    /// Two JIDs are equal if they have the same full JID value.
+    fn __richcmp__(&self, other: &Bound<'_, PyAny>, op: pyo3::basic::CompareOp) -> PyResult<bool> {
+        let other = if let Ok(other) = other.extract::<PyRef<PyJid>>() {
+            other
+        } else if other.is_none() {
+            Bound::new(other.py(), PyJid::new(None, false)?)?.borrow()
+        } else {
+            Bound::new(other.py(), PyJid::new(Some(other), false)?)?.borrow()
+        };
+        match (&self.jid, &other.jid) {
+            (None, None) => Ok(true),
+            (Some(jid), Some(other)) => match op {
+                pyo3::basic::CompareOp::Eq => Ok(jid == other),
+                pyo3::basic::CompareOp::Ne => Ok(jid != other),
+                _ => Err(PyNotImplementedError::new_err(
+                    "Only == and != are implemented",
+                )),
+            },
+            _ => Ok(false),
+        }
+    }
+
+    /// Hash a JID based on the string version of its full JID.
+    fn __hash__(&self) -> isize {
+        if let Some(jid) = &self.jid {
+            // Use the same algorithm as the Python JID.
+            let string = jid.to_string();
+            unsafe { pyo3::ffi::_Py_HashBytes(string.as_ptr() as *const _, string.len() as isize) }
+        } else {
+            0
+        }
+    }
+
+    // Aliases
+
+    #[getter]
+    fn get_user(&self) -> String {
+        self.get_node()
+    }
+
+    #[setter]
+    fn set_user(&mut self, user: &str) -> PyResult<()> {
+        self.set_node(user)
+    }
+
+    #[getter]
+    fn get_server(&self) -> String {
+        self.get_domain()
+    }
+
+    #[setter]
+    fn set_server(&mut self, server: &str) -> PyResult<()> {
+        self.set_domain(server)
+    }
+
+    #[getter]
+    fn get_host(&self) -> String {
+        self.get_domain()
+    }
+
+    #[setter]
+    fn set_host(&mut self, host: &str) -> PyResult<()> {
+        self.set_domain(host)
+    }
+
+    #[getter]
+    fn get_jid(&self) -> String {
+        self.get_full()
+    }
+
+    #[setter]
+    fn set_jid(&mut self, jid: &str) -> PyResult<()> {
+        self.set_full(jid)
+    }
+}
+
+#[pymodule]
+#[pyo3(name = "libslixmpp")]
+fn py_jid(py: Python<'_>, m: &Bound<'_, PyModule>) -> PyResult<()> {
+    m.add_class::<PyJid>()?;
+    m.add("InvalidJID", py.get_type_bound::<InvalidJID>())?;
+    Ok(())
+}

tests/test_jid.py

@@ -3,7 +3,6 @@ from __future__ import unicode_literals
 import unittest
 from slixmpp.test import SlixTest
 from slixmpp import JID, InvalidJID
-from slixmpp.jid import nodeprep
 
 
 class TestJIDClass(SlixTest):
@@ -279,9 +278,5 @@ class TestJIDClass(SlixTest):
         #self.assertRaises(InvalidJID, JID, '%s@example.com' % '\\20foo2')
         #self.assertRaises(InvalidJID, JID, '%s@example.com' % 'bar2\\20')
 
-    def testNodePrepIdemptotent(self):
-        node = 'ᴹᴵᴷᴬᴱᴸ'
-        self.assertEqual(nodeprep(node), nodeprep(nodeprep(node)))
-
 
 suite = unittest.TestLoader().loadTestsFromTestCase(TestJIDClass)