Strip ANSI sequences before writing junitxml text

changelog/12365.bugfix.rst

@@ -0,0 +1,3 @@
+The ``--junitxml`` report no longer includes ANSI escape codes from
+Pygments-highlighted assertion diffs in failure messages or details, or from
+captured output written to ``<system-out>`` and ``<system-err>``.

src/_pytest/junitxml.py

@@ -43,6 +43,15 @@
 _JunitDurationReport = Literal["total", "call"]
 _JunitFamily = Literal["legacy", "xunit1", "xunit2"]
 
+# SGR (Select Graphic Rendition) sequences pre-baked into assertion diffs by Pygments
+# when color is enabled. These are invalid in XML and must be stripped before writing.
+_ANSI_ESCAPE_RE = re.compile(r"\x1b\[[\d;]*m")
+
+
+def strip_ansi_escapes(text: str) -> str:
+    """Remove ANSI SGR escape sequences from *text*."""
+    return _ANSI_ESCAPE_RE.sub("", text)
+
 
 def bin_xml_escape(arg: object) -> str:
     r"""Visually escape invalid XML characters.
@@ -163,6 +172,8 @@ def to_xml(self) -> ET.Element:
 
     def _add_simple(self, tag: str, message: str, data: str | None = None) -> None:
         node = ET.Element(tag, message=message)
+        if data is not None:
+            data = strip_ansi_escapes(data)
         node.text = bin_xml_escape(data)
         self.append(node)
 
@@ -194,7 +205,7 @@ def _prepare_content(self, content: str, header: str) -> str:
 
     def _write_content(self, report: TestReport, content: str, jheader: str) -> None:
         tag = ET.Element(jheader)
-        tag.text = bin_xml_escape(content)
+        tag.text = bin_xml_escape(strip_ansi_escapes(content))
         self.append(tag)
 
     def append_pass(self, report: TestReport) -> None:
@@ -213,7 +224,7 @@ def append_failure(self, report: TestReport) -> None:
                 message = reprcrash.message
             else:
                 message = str(report.longrepr)
-            message = bin_xml_escape(message)
+            message = bin_xml_escape(strip_ansi_escapes(message))
             self._add_simple("failure", message, str(report.longrepr))
 
     def append_collect_error(self, report: CollectReport) -> None:
@@ -236,7 +247,9 @@ def append_error(self, report: TestReport) -> None:
             msg = f'failed on teardown with "{reason}"'
         else:
             msg = f'failed on setup with "{reason}"'
-        self._add_simple("error", bin_xml_escape(msg), str(report.longrepr))
+        self._add_simple(
+            "error", bin_xml_escape(strip_ansi_escapes(msg)), str(report.longrepr)
+        )
 
     def append_skipped(self, report: TestReport) -> None:
         if hasattr(report, "wasxfail"):

testing/test_junitxml.py

@@ -1772,8 +1772,28 @@ def test_esc(my_setup):
     _, dom = run_and_parse()
     node = dom.get_first_by_tag("testcase")
     snode = node.get_first_by_tag("error")
-    assert "#x1B[31mred#x1B[m" in snode["message"]
-    assert "#x1B[31mred#x1B[m" in snode.text
+    assert "error: red" in snode["message"]
+    assert "#x1B" not in snode["message"]
+    assert "error: red" in snode.text
+    assert "#x1B" not in snode.text
+
+
+def test_ansi_escapes_stripped_from_captured_output(pytester: Pytester) -> None:
+    pytester.makepyfile(
+        """
+        import sys
+
+        def test_color():
+            sys.stdout.write("out: \033[31mred\033[0m")
+            sys.stderr.write("err: \033[32mgreen\033[0m")
+    """
+    )
+    xmlf = pytester.path.joinpath("junit.xml")
+    pytester.runpytest(f"--junitxml={xmlf}", "-o", "junit_logging=all")
+    text = xmlf.read_text(encoding="utf-8")
+    assert "#x1B" not in text
+    assert "out: red" in text
+    assert "err: green" in text
 
 
 @parametrize_families
@@ -1860,6 +1880,27 @@ def test_no_message_quiet(pytester: Pytester) -> None:
     result.stdout.no_fnmatch_line("* generated xml file: *")
 
 
+def test_junitxml_strips_ansi_from_failure_details(
+    pytester: Pytester, monkeypatch: MonkeyPatch
+) -> None:
+    """ANSI escapes from Pygments-highlighted diffs must not appear in junitxml (#12365)."""
+    pytest.importorskip("pygments")
+    monkeypatch.setenv("PY_COLORS", "1")
+    pytester.makepyfile(
+        """
+        def test_foo():
+            assert [1, 2, 3] == [3, 2, 1]
+        """
+    )
+    xml_path = pytester.path.joinpath("junit.xml")
+    result = pytester.runpytest(f"--junitxml={xml_path}", "-vv")
+    assert result.ret == 1
+    xml_text = xml_path.read_text(encoding="utf-8")
+    assert "\x1b[" not in xml_text
+    assert "#x1B" not in xml_text
+    assert "assert [1, 2, 3] == [3, 2, 1]" in xml_text
+
+
 @pytest.mark.parametrize(
     ("name", "value"),
     [