fix: strip ANSI escape sequences from report messages

src/pytest_reportlog/plugin.py

@@ -1,10 +1,13 @@
 import json
+import re
 from typing import Dict, Any, TextIO
 
 from _pytest.pathlib import Path
 
 import pytest
 
+_ANSI_ESCAPE_SEQUENCE_RE = re.compile(r"\x1b\[[0-?]*[ -/]*[@-~]")
+
 
 def pytest_addoption(parser):
     group = parser.getgroup("terminal reporting", "report-log plugin options")
@@ -69,10 +72,12 @@ def close(self):
             self._file = None
 
     def _write_json_data(self, data):
+        data = _strip_ansi_escape_sequences(data)
         try:
             json_data = json.dumps(data)
         except TypeError:
             data = cleanup_unserializable(data)
+            data = _strip_ansi_escape_sequences(data)
             json_data = json.dumps(data)
         self._file.write(json_data + "\n")
         self._file.flush()
@@ -148,3 +153,21 @@ def cleanup_unserializable(d: Dict[str, Any]) -> Dict[str, Any]:
             v = str(v)
         result[k] = v
     return result
+
+
+def _strip_ansi_escape_sequences(value: Any) -> Any:
+    """
+    Return a new value with ANSI escape sequences removed from any strings.
+
+    Report log output is intended to be machine-readable, and ANSI codes can
+    appear in pytest's formatted output (e.g. colored diffs).
+    """
+    if isinstance(value, str):
+        return _ANSI_ESCAPE_SEQUENCE_RE.sub("", value)
+    if isinstance(value, dict):
+        return {k: _strip_ansi_escape_sequences(v) for k, v in value.items()}
+    if isinstance(value, list):
+        return [_strip_ansi_escape_sequences(v) for v in value]
+    if isinstance(value, tuple):
+        return tuple(_strip_ansi_escape_sequences(v) for v in value)
+    return value

tests/test_reportlog.py

@@ -184,6 +184,7 @@ def __str__(self):
 
 def test_subtest(pytester, tmp_path):
     """Regression test for #90."""
+    pytest.importorskip("pytest_subtests")
     pytester.makepyfile("""
         def test_foo(subtests):
             with subtests.test():
@@ -196,3 +197,36 @@ def test_foo(subtests):
     for line in lines:
         data = json.loads(line)
         assert "$report_type" in data
+
+
+def test_strips_ansi_escape_sequences(testdir, tmp_path):
+    testdir.makeconftest(r"""
+def pytest_report_to_serializable(config, report):
+    data = {
+        "$report_type": "TestReport",
+        "nodeid": report.nodeid,
+        "outcome": report.outcome,
+    }
+
+    if getattr(report, "when", None) == "call":
+        data["longreprtext"] = "\x1b[31mRED\x1b[0m"
+        data["sections"] = [("Captured stdout call", "\x1b[32mGREEN\x1b[0m")]
+
+    return data
+""")
+    testdir.makepyfile("""
+        def test_fail():
+            assert 0
+    """)
+
+    log_file = tmp_path / "log.json"
+    result = testdir.runpytest("--report-log", str(log_file))
+    assert result.ret == pytest.ExitCode.TESTS_FAILED
+
+    json_objs = [json.loads(x) for x in log_file.read_text().splitlines()]
+    report = next(obj for obj in json_objs if obj.get("longreprtext"))
+
+    assert report["longreprtext"] == "RED"
+    assert "\x1b" not in report["longreprtext"]
+    assert report["sections"][0][1] == "GREEN"
+    assert "\x1b" not in report["sections"][0][1]