changeset 3914:4cb38c8312a1 default tip

plugin XEP-0384, xml_tools: avoid `getItems` timeout + fix empty node crash + parsing: - use `max_items` in `getItems` calls for bundles, as otherwise some pubsub service may return full nodes, which may be huge is `max_items=1` is not set on the node, possibly resulting in timeouts. - the plugin was crashing when TWOMEMO devices list node has no items at all. This is not the case anymore. - a naive parsing method has been implemented in `xml_tools` to replace the serialisation/deserialisation method. This should be more efficient and will avoid annoying `ns0:` prefixes in XML logs.
author Goffi <goffi@goffi.org>
date Sat, 24 Sep 2022 16:37:46 +0200
parents 944f51f9c2b4
children
files sat/plugins/plugin_xep_0384.py sat/tools/xml_tools.py
diffstat 2 files changed, 71 insertions(+), 17 deletions(-) [+]
line wrap: on
line diff
--- a/sat/plugins/plugin_xep_0384.py	Sat Sep 24 16:31:39 2022 +0200
+++ b/sat/plugins/plugin_xep_0384.py	Sat Sep 24 16:37:46 2022 +0200
@@ -419,7 +419,7 @@
     node = f"eu.siacs.conversations.axolotl.bundles:{device_id}"
 
     try:
-        items, __ = await xep_0060.getItems(client, jid.JID(bare_jid), node)
+        items, __ = await xep_0060.getItems(client, jid.JID(bare_jid), node, max_items=1)
     except Exception as e:
         raise omemo.BundleDownloadFailed(
             f"Bundle download failed for {bare_jid}: {device_id} under namespace"
@@ -432,7 +432,7 @@
             f" {namespace}: Unexpected number of items retrieved: {len(items)}."
         )
 
-    element = next(iter(domish_to_etree(cast(domish.Element, items[0]))), None)
+    element = next(iter(xml_tools.domish_elt_2_et_elt(cast(domish.Element, items[0]))), None)
     if element is None:
         raise omemo.BundleDownloadFailed(
             f"Bundle download failed for {bare_jid}: {device_id} under namespace"
@@ -476,7 +476,7 @@
                         client,
                         client.jid.userhostJID(),
                         node,
-                        etree_to_domish(element),
+                        xml_tools.et_elt_2_domish_elt(element),
                         item_id=str(bundle.device_id),
                         extra={
                             xep_0060.EXTRA_PUBLISH_OPTIONS: {
@@ -516,7 +516,7 @@
                         client,
                         client.jid.userhostJID(),
                         node,
-                        etree_to_domish(element),
+                        xml_tools.et_elt_2_domish_elt(element),
                         item_id=xep_0060.ID_SINGLETON,
                         extra={
                             xep_0060.EXTRA_PUBLISH_OPTIONS: { xep_0060.OPT_MAX_ITEMS: 1 },
@@ -563,7 +563,7 @@
                     )
 
                 element = next(
-                    iter(domish_to_etree(cast(domish.Element, items[0]))),
+                    iter(xml_tools.domish_elt_2_et_elt(cast(domish.Element, items[0]))),
                     None
                 )
                 if element is None:
@@ -650,7 +650,7 @@
                     client,
                     client.jid.userhostJID(),
                     node,
-                    etree_to_domish(element),
+                    xml_tools.et_elt_2_domish_elt(element),
                     item_id=xep_0060.ID_SINGLETON,
                     extra={
                         xep_0060.EXTRA_PUBLISH_OPTIONS: {
@@ -705,13 +705,15 @@
                     f" {namespace}"
                 ) from e
 
-            if len(items) != 1:
+            if len(items) == 0:
+                return {}
+            elif len(items) != 1:
                 raise omemo.DeviceListDownloadFailed(
                     f"Device list download failed for {bare_jid} under namespace"
                     f" {namespace}: Unexpected number of items retrieved: {len(items)}."
                 )
 
-            element = next(iter(domish_to_etree(cast(domish.Element, items[0]))), None)
+            element = next(iter(xml_tools.domish_elt_2_et_elt(cast(domish.Element, items[0]))), None)
             if element is None:
                 raise omemo.DeviceListDownloadFailed(
                     f"Device list download failed for {bare_jid} under namespace"
@@ -866,10 +868,10 @@
                 "timestamp": time.time()
             }))
 
-            message_data["xml"].addChild(etree_to_domish(element))
+            message_data["xml"].addChild(xml_tools.et_elt_2_domish_elt(element))
 
             try:
-                await client.send(message_data["xml"])
+                await client.a_send(message_data["xml"])
             except Exception as e:
                 raise omemo.MessageSendingFailed() from e
 
@@ -1493,7 +1495,6 @@
             encrypted.
         @return: Whether to continue the message received flow.
         """
-
         muc_plaintext_cache_key: Optional[MUCPlaintextCacheKey] = None
 
         sender_jid = jid.JID(message_elt["from"])
@@ -1583,7 +1584,7 @@
         if twomemo_encrypted_elt is not None:
             try:
                 message = twomemo.etree.parse_message(
-                    domish_to_etree(twomemo_encrypted_elt),
+                    xml_tools.domish_elt_2_et_elt(twomemo_encrypted_elt),
                     sender_bare_jid
                 )
             except (ValueError, XMLSchemaValidationError):
@@ -1597,7 +1598,7 @@
         if oldmemo_encrypted_elt is not None:
             try:
                 message = await oldmemo.etree.parse_message(
-                    domish_to_etree(oldmemo_encrypted_elt),
+                    xml_tools.domish_elt_2_et_elt(oldmemo_encrypted_elt),
                     sender_bare_jid,
                     client.jid.userhost(),
                     session_manager
@@ -1999,11 +2000,11 @@
 
         if namespace == twomemo.twomemo.NAMESPACE:
             # Add the encrypted element
-            stanza.addChild(etree_to_domish(twomemo.etree.serialize_message(message)))
+            stanza.addChild(xml_tools.et_elt_2_domish_elt(twomemo.etree.serialize_message(message)))
 
         if namespace == oldmemo.oldmemo.NAMESPACE:
             # Add the encrypted element
-            stanza.addChild(etree_to_domish(oldmemo.etree.serialize_message(message)))
+            stanza.addChild(xml_tools.et_elt_2_domish_elt(oldmemo.etree.serialize_message(message)))
 
         if muc_plaintext_cache_key is not None:
             self.__muc_plaintext_cache[muc_plaintext_cache_key] = plaintext
@@ -2031,7 +2032,7 @@
             log.debug("Ignoring empty device list update.")
             return
 
-        item_elt = domish_to_etree(item)
+        item_elt = xml_tools.domish_elt_2_et_elt(item)
 
         device_list: Dict[int, Optional[str]] = {}
         namespace: Optional[str] = None
--- a/sat/tools/xml_tools.py	Sat Sep 24 16:31:39 2022 +0200
+++ b/sat/tools/xml_tools.py	Sat Sep 24 16:37:46 2022 +0200
@@ -18,7 +18,7 @@
 
 
 import re
-from typing import Optional
+from typing import Optional, Tuple
 import html.entities
 from collections import OrderedDict
 from xml.dom import minidom, NotFoundErr
@@ -30,6 +30,7 @@
 from sat.core.i18n import _
 from sat.core.constants import Const as C
 from sat.core.log import getLogger
+import xml.etree.ElementTree as ET
 
 
 log = getLogger(__name__)
@@ -1957,3 +1958,55 @@
 def ppElt(elt):
     """Pretty print a domish.Element"""
     print(pFmtElt(elt))
+
+
+# ElementTree
+
+def et_get_namespace_and_name(et_elt: ET.Element) -> Tuple[Optional[str], str]:
+    """Retrieve element namespace and name from ElementTree element
+
+    @param et_elt: ElementTree element
+    @return: namespace and name of the element
+        if not namespace if specified, None is returned
+    """
+    name = et_elt.tag
+    if not name:
+        raise ValueError("no name set in ET element")
+    elif name[0] != "{":
+        return None, name
+    end_idx = name.find("}")
+    if end_idx == -1:
+        raise ValueError("Invalid ET name")
+    return name[1:end_idx], name[end_idx+1:]
+
+
+def et_elt_2_domish_elt(et_elt: ET.Element) -> domish.Element:
+    """Convert ElementTree element to Twisted's domish.Element
+
+    Note: this is a naive implementation, adapted to XMPP, and some content are ignored
+        (attributes namespaces, tail)
+    """
+    namespace, name = et_get_namespace_and_name(et_elt)
+    elt = domish.Element((namespace, name), attribs=et_elt.attrib)
+    if et_elt.text:
+        elt.addContent(et_elt.text)
+    for child in et_elt:
+        elt.addChild(et_elt_2_domish_elt(child))
+    return elt
+
+
+def domish_elt_2_et_elt(elt: domish.Element) -> ET.Element:
+    """Convert Twisted's domish.Element to ElementTree equivalent
+
+    Note: this is a naive implementation, adapter to XMPP, and some text content may be
+        missing (content put after a tag, i.e. what would go to the "tail" attribute of ET
+        Element)
+    """
+    tag = f"{{{elt.uri}}}{elt.name}" if elt.uri else elt.name
+    et_elt = ET.Element(tag, attrib=elt.attributes)
+    content = str(elt)
+    if content:
+        et_elt.text = str(elt)
+    for child in elt.elements():
+        et_elt.append(domish_elt_2_et_elt(child))
+    return et_elt