Fix ExceptionGroup traceback filtering of pytest internals (#13380)

* init

* + author

* fix

* Simplify the code a bit

* Small refactor

* Import functions from `traceback` directly, to allow free use of `traceback` as a variable name.
* Extract `_filtered_traceback` into a function.
* Inline `_repr_exception_group_traceback` given it is used only in one place.
* Make a type alias for the type of `tbfilter`.

* action rerun

* + comment

* fix docs

* Revert "fix docs"

This reverts commit fc505cb.

* fix

---------

Co-authored-by: Bruno Oliveira <[email protected]>

Author: gesslerpd

AUTHORS

@@ -346,6 +346,7 @@ Pavel Karateev
 Pavel Zhukov
 Paweł Adamczak
 Pedro Algarvio
+Peter Gessler
 Petter Strandmark
 Philipp Loose
 Pierre Sassoulas

changelog/13380.improvement.rst

@@ -0,0 +1 @@
+Fix :class:`ExceptionGroup` traceback filtering to exclude pytest internals.

src/_pytest/_code/code.py

@@ -15,8 +15,10 @@
 from pathlib import Path
 import re
 import sys
-import traceback
+from traceback import extract_tb
+from traceback import format_exception
 from traceback import format_exception_only
+from traceback import FrameSummary
 from types import CodeType
 from types import FrameType
 from types import TracebackType
@@ -28,6 +30,7 @@
 from typing import Literal
 from typing import overload
 from typing import SupportsIndex
+from typing import TYPE_CHECKING
 from typing import TypeVar
 from typing import Union
 
@@ -208,10 +211,10 @@ def with_repr_style(
     def lineno(self) -> int:
         return self._rawentry.tb_lineno - 1
 
-    def get_python_framesummary(self) -> traceback.FrameSummary:
+    def get_python_framesummary(self) -> FrameSummary:
         # Python's built-in traceback module implements all the nitty gritty
         # details to get column numbers of out frames.
-        stack_summary = traceback.extract_tb(self._rawentry, limit=1)
+        stack_summary = extract_tb(self._rawentry, limit=1)
         return stack_summary[0]
 
     # Column and end line numbers introduced in python 3.11
@@ -694,8 +697,7 @@ def getrepr(
         showlocals: bool = False,
         style: TracebackStyle = "long",
         abspath: bool = False,
-        tbfilter: bool
-        | Callable[[ExceptionInfo[BaseException]], _pytest._code.code.Traceback] = True,
+        tbfilter: bool | Callable[[ExceptionInfo[BaseException]], Traceback] = True,
         funcargs: bool = False,
         truncate_locals: bool = True,
         truncate_args: bool = True,
@@ -742,7 +744,7 @@ def getrepr(
         if style == "native":
             return ReprExceptionInfo(
                 reprtraceback=ReprTracebackNative(
-                    traceback.format_exception(
+                    format_exception(
                         self.type,
                         self.value,
                         self.traceback[0]._rawentry if self.traceback else None,
@@ -851,6 +853,17 @@ def group_contains(
         return self._group_contains(self.value, expected_exception, match, depth)
 
 
+if TYPE_CHECKING:
+    from typing_extensions import TypeAlias
+
+    # Type alias for the `tbfilter` setting:
+    # bool: If True, it should be filtered using Traceback.filter()
+    # callable: A callable that takes an ExceptionInfo and returns the filtered traceback.
+    TracebackFilter: TypeAlias = Union[
+        bool, Callable[[ExceptionInfo[BaseException]], Traceback]
+    ]
+
+
 @dataclasses.dataclass
 class FormattedExcinfo:
     """Presenting information about failing Functions and Generators."""
@@ -862,7 +875,7 @@ class FormattedExcinfo:
     showlocals: bool = False
     style: TracebackStyle = "long"
     abspath: bool = True
-    tbfilter: bool | Callable[[ExceptionInfo[BaseException]], Traceback] = True
+    tbfilter: TracebackFilter = True
     funcargs: bool = False
     truncate_locals: bool = True
     truncate_args: bool = True
@@ -1100,11 +1113,7 @@ def _makepath(self, path: Path | str) -> str:
         return str(path)
 
     def repr_traceback(self, excinfo: ExceptionInfo[BaseException]) -> ReprTraceback:
-        traceback = excinfo.traceback
-        if callable(self.tbfilter):
-            traceback = self.tbfilter(excinfo)
-        elif self.tbfilter:
-            traceback = traceback.filter(excinfo)
+        traceback = filter_excinfo_traceback(self.tbfilter, excinfo)
 
         if isinstance(excinfo.value, RecursionError):
             traceback, extraline = self._truncate_recursive_traceback(traceback)
@@ -1178,14 +1187,15 @@ def repr_excinfo(self, excinfo: ExceptionInfo[BaseException]) -> ExceptionChainR
                 # Fall back to native traceback as a temporary workaround until
                 # full support for exception groups added to ExceptionInfo.
                 # See https://github.com/pytest-dev/pytest/issues/9159
+                reprtraceback: ReprTraceback | ReprTracebackNative
                 if isinstance(e, BaseExceptionGroup):
-                    reprtraceback: ReprTracebackNative | ReprTraceback = (
-                        ReprTracebackNative(
-                            traceback.format_exception(
-                                type(excinfo_.value),
-                                excinfo_.value,
-                                excinfo_.traceback[0]._rawentry,
-                            )
+                    # don't filter any sub-exceptions since they shouldn't have any internal frames
+                    traceback = filter_excinfo_traceback(self.tbfilter, excinfo)
+                    reprtraceback = ReprTracebackNative(
+                        format_exception(
+                            type(excinfo.value),
+                            excinfo.value,
+                            traceback[0]._rawentry,
                         )
                     )
                 else:
@@ -1194,9 +1204,7 @@ def repr_excinfo(self, excinfo: ExceptionInfo[BaseException]) -> ExceptionChainR
             else:
                 # Fallback to native repr if the exception doesn't have a traceback:
                 # ExceptionInfo objects require a full traceback to work.
-                reprtraceback = ReprTracebackNative(
-                    traceback.format_exception(type(e), e, None)
-                )
+                reprtraceback = ReprTracebackNative(format_exception(type(e), e, None))
                 reprcrash = None
             repr_chain += [(reprtraceback, reprcrash, descr)]
 
@@ -1545,3 +1553,15 @@ def filter_traceback(entry: TracebackEntry) -> bool:
         return False
 
     return True
+
+
+def filter_excinfo_traceback(
+    tbfilter: TracebackFilter, excinfo: ExceptionInfo[BaseException]
+) -> Traceback:
+    """Filter the exception traceback in ``excinfo`` according to ``tbfilter``."""
+    if callable(tbfilter):
+        return tbfilter(excinfo)
+    elif tbfilter:
+        return excinfo.traceback.filter(excinfo)
+    else:
+        return excinfo.traceback

testing/code/test_excinfo.py

@@ -1797,6 +1797,9 @@ def test():
             rf"FAILED test_excgroup.py::test - {pre_catch}BaseExceptionGroup: Oops \(2.*"
         )
     result.stdout.re_match_lines(match_lines)
+    # Check for traceback filtering of pytest internals.
+    result.stdout.no_fnmatch_line("*, line *, in pytest_pyfunc_call")
+    result.stdout.no_fnmatch_line("*, line *, in pytest_runtest_call")
 
 
 @pytest.mark.skipif(