contains_xml method

CHANGELOG.md

@@ -10,6 +10,8 @@ and this project adheres to [Semantic Versioning](http://semver.org/spec/v2.0.0.
 - `Prompt` class `xml_tree` property
 ### Changed
 - `LLMModel` enum updated
+- `verify` parameter added to `contains_xml` method
+- Test system modified
 ## [0.9] - 2025-10-22
 ### Added
 - `Prompt` class `contains_xml` method

memor/message.py

@@ -232,9 +232,22 @@ def estimate_tokens(self, method: TokensEstimator = TokensEstimator.DEFAULT) ->
         """
         return method(self.render(render_format=RenderFormat.STRING, show_warning=False))
 
-    def contains_xml(self) -> bool:
-        """Check if the message contains any XML tags."""
-        return bool(re.search(XML_PATTERN, self.render(render_format=RenderFormat.STRING, show_warning=False)))
+    def contains_xml(self, verify: bool = False) -> bool:
+        """
+        Check if the message contains any XML tags.
+
+        :param verify: verify if the XML structure is well-formed
+        """
+        message = self.render(render_format=RenderFormat.STRING, show_warning=False)
+        wrapped = "<root>{message}</root>".format(message=message)
+        pattern_result = bool(re.search(XML_PATTERN, message))
+        if not verify:
+            return pattern_result
+        try:
+            _ = ElementTree.fromstring(wrapped)
+            return pattern_result
+        except Exception:
+            return False
 
     def _build_xml_tree(self) -> Dict[str, Any]:
         """Build XML tree."""

tests/test_prompt.py

@@ -898,6 +898,21 @@ def test_contains_xml1():
 
 
 def test_contains_xml2():
+    message = "How are you?"
+    response1 = Response(message="I am fine.", model=LLMModel.GPT_4, temperature=0.5, role=Role.USER, score=0.8)
+    response2 = Response(message="Thanks!", model=LLMModel.GPT_4, temperature=0.5, role=Role.USER, score=0.8)
+    template = PromptTemplate(content="{instruction}, {prompt[message]}", custom_map={"instruction": "Hi"})
+    prompt = Prompt(
+        message=message,
+        responses=[
+            response1,
+            response2],
+        role=Role.USER,
+        template=template)
+    assert not prompt.contains_xml(verify=True)
+
+
+def test_contains_xml3():
     message = "How are you?"
     response1 = Response(message="I am fine.", model=LLMModel.GPT_4, temperature=0.5, role=Role.USER, score=0.8)
     response2 = Response(message="Thanks!", model=LLMModel.GPT_4, temperature=0.5, role=Role.USER, score=0.8)
@@ -912,6 +927,51 @@ def test_contains_xml2():
     assert prompt.contains_xml()
 
 
+def test_contains_xml4():
+    message = "How are you?"
+    response1 = Response(message="I am fine.", model=LLMModel.GPT_4, temperature=0.5, role=Role.USER, score=0.8)
+    response2 = Response(message="Thanks!", model=LLMModel.GPT_4, temperature=0.5, role=Role.USER, score=0.8)
+    template = PromptTemplate(content="<inst>{instruction}<inst>, {prompt[message]}", custom_map={"instruction": "Hi"})
+    prompt = Prompt(
+        message=message,
+        responses=[
+            response1,
+            response2],
+        role=Role.USER,
+        template=template)
+    assert prompt.contains_xml()
+
+
+def test_contains_xml5():
+    message = "How are you?"
+    response1 = Response(message="I am fine.", model=LLMModel.GPT_4, temperature=0.5, role=Role.USER, score=0.8)
+    response2 = Response(message="Thanks!", model=LLMModel.GPT_4, temperature=0.5, role=Role.USER, score=0.8)
+    template = PromptTemplate(content="<inst>{instruction}<inst>, {prompt[message]}", custom_map={"instruction": "Hi"})
+    prompt = Prompt(
+        message=message,
+        responses=[
+            response1,
+            response2],
+        role=Role.USER,
+        template=template)
+    assert not prompt.contains_xml(verify=True)
+
+
+def test_contains_xml6():
+    message = "How are you?"
+    response1 = Response(message="I am fine.", model=LLMModel.GPT_4, temperature=0.5, role=Role.USER, score=0.8)
+    response2 = Response(message="Thanks!", model=LLMModel.GPT_4, temperature=0.5, role=Role.USER, score=0.8)
+    template = PromptTemplate(content="<inst>{instruction}</inst>, {prompt[message]}", custom_map={"instruction": "Hi"})
+    prompt = Prompt(
+        message=message,
+        responses=[
+            response1,
+            response2],
+        role=Role.USER,
+        template=template)
+    assert prompt.contains_xml(verify=True)
+
+
 def test_equality1():
     message = "Hello, how are you?"
     response1 = Response(message="I am fine.", model=LLMModel.GPT_4, temperature=0.5, role=Role.USER, score=0.8)

tests/test_response.py

@@ -825,10 +825,30 @@ def test_contains_xml1():
 
 
 def test_contains_xml2():
+    response = Response(message="I am fine.")
+    assert not response.contains_xml(verify=True)
+
+
+def test_contains_xml3():
     response = Response(message="I am fine. <note>test</note>")
     assert response.contains_xml()
 
 
+def test_contains_xml4():
+    response = Response(message="I am fine. <note>test<note>")
+    assert response.contains_xml()
+
+
+def test_contains_xml5():
+    response = Response(message="I am fine. <note>test<note>")
+    assert not response.contains_xml(verify=True)
+
+
+def test_contains_xml6():
+    response = Response(message="I am fine. <note>test</note>")
+    assert response.contains_xml(verify=True)
+
+
 def test_equality1():
     response1 = Response(message="I am fine.", model=LLMModel.GPT_4, temperature=0.5, role=Role.USER, score=0.8)
     response2 = response1.copy()