fixes buggy extraction of XML attributes

refinery/units/formats/__init__.py

@@ -342,12 +342,11 @@ def rank_attribute(attribute: str):
             def walk(node: XMLNodeBase):
                 candidates = [
                     candidate for candidate, count in Counter(
-                        key
-                        for child in node.children
-                        for key, val in child.attributes.items()
-                        if re.fullmatch(R'[-\s\w+,.;@(){}]{2,64}', nval(val))
+                        key for child in node.children for key, val in child.attributes.items()
+                        if len(val) in range(2, 65) and re.fullmatch(R'[-\s\w+,.;@()]+', nval(val))
                     ).items()
-                    if count == len(node.children)
+                    if count == len(node.children) == len(
+                        {child.attributes[candidate] for child in node.children})
                 ]
                 if not candidates:
                     attr = None

refinery/units/formats/xml.py

@@ -1,7 +1,7 @@
 #!/usr/bin/env python3
 # -*- coding: utf-8 -*-
 from refinery.lib.structures import MemoryFile
-from refinery.lib.meta import metavars
+from refinery.lib.meta import metavars, is_valid_variable_name
 from refinery.lib import xml
 from refinery.units.sinks.ppxml import ppxml
 from refinery.units.formats import XMLToPathExtractorUnit, UnpackResult
@@ -23,7 +23,17 @@ def extract(node: xml.XMLNode = node):
                 with MemoryFile() as stream:
                     node.write(stream)
                     return bytes(stream.getbuffer() | ppxml)
-            yield UnpackResult('/'.join(parts), extract, **node.attributes)
+
+            attributes = {
+                self._normalize_key(k): self._normalize_val(v)
+                for k, v in node.attributes.items()
+            }
+
+            if not all(is_valid_variable_name(k) for k in attributes):
+                attributes = {F'_{k}': v for k, v in attributes.items()}
+
+            yield UnpackResult('/'.join(parts), extract, **attributes)
+
             for child in node.children:
                 yield from walk(child, *parts, path(child))
 

test/units/formats/test_xml.py

@@ -30,3 +30,13 @@ class data:
         unit = self.load('WLANProfile/SSIDConfig')
         test = str(data | unit)
         self.assertIn('33344333', test)
+
+    def test_lure_document(self):
+        data = self.download_sample('e6daa00e095948acfc176d71c5bf667a0403e5259653ea5ac8950aee13180ae0')
+        data = data | self.ldu('xt', 'settings.xml') | bytes
+
+        pipe = self.load_pipeline('xtxml w.settings/w.rsids/w.rsidRoot [| eat val ]')
+        self.assertEqual(data | pipe | str, '00BA5B2D')
+
+        pipe = self.load_pipeline('xtxml docVars/10* [| eat val ]| hex | wshenc | carve -dn5 string [| dedup | pop k | swap k | hex | xor var:k ]| xtp url')
+        self.assertEqual(data | pipe | str, 'http'':/''/trust-certificate''.net')