test_site _pth file tests fail when stdlib path is very long

[itamaro]

Bug report

Bug description:

The getpath module tries to read certain _pth files during initialization).

This is tested in test_site with generated _pth files that include the stdlib path 200 times.

The getpath module disallows reading files over 32KB during initialization.

If the test suite runs from a very long base path, 200 repetitions of the stdlib path in the _pth file would be enough to exceed the 32KB limit.

To demonstrate, artificially increase the number of repetitions to some high number that would exceed 32KB, e.g. this patch:

diff --git a/Lib/test/test_site.py b/Lib/test/test_site.py
index 33d0975bda8..ba2f7d90a80 100644
--- a/Lib/test/test_site.py
+++ b/Lib/test/test_site.py
@@ -668,7 +668,7 @@ def test_underpth_nosite_file(self):
 exe_prefix = os.path.dirname(sys.executable)
 pth_lines = [
 'fake-path-name',
- *[libpath for _ in range(200)],
+ *[libpath for _ in range(5000)],
 '',
 '# comment',
 ]

and observe the test failing with the following error:

$ ./python.exe -m test test_site -v -m '*_pthFileTests.test_underpth_nosite_file'
== CPython 3.13.0a2+ (heads/main-dirty:b4b2cc10121, Jan 1 2024, 11:38:27) [Clang 15.0.0 (clang-1500.1.0.2.5)]
== macOS-14.2.1-arm64-arm-64bit little-endian
== Python build: debug
== cwd: /Users/itamaro/work/pyexe/main-dbg/build/test_python_worker_96404æ
== CPU count: 12
== encodings: locale=UTF-8 FS=utf-8
== resources: all test resources are disabled, use -u option to unskip tests

Using random seed: 4200167786
0:00:00 load avg: 8.99 Run 1 test sequentially
0:00:00 load avg: 8.99 [1/1] test_site
test_underpth_nosite_file (test.test_site._pthFileTests.test_underpth_nosite_file) ... Exception ignored in running getpath:
Traceback (most recent call last):
 File "<frozen getpath>", line 462, in <module>
MemoryError: cannot read file larger than 32KB during initialization
Fatal Python error: error evaluating path
Python runtime state: core initialized

Current thread 0x00000001e0e39000 (most recent call first):
 <no Python frame>
ERROR

======================================================================
ERROR: test_underpth_nosite_file (test.test_site._pthFileTests.test_underpth_nosite_file)
----------------------------------------------------------------------
Traceback (most recent call last):
 File "/Users/itamaro/work/cpython/Lib/test/test_site.py", line 683, in test_underpth_nosite_file
 output = subprocess.check_output([exe_file, '-c',
 ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
 'import sys; print("\\n".join(sys.path) if sys.flags.no_site else "")'
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 ], env=env, encoding='utf-8', errors='surrogateescape')
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 File "/Users/itamaro/work/cpython/Lib/subprocess.py", line 470, in check_output
 return run(*popenargs, stdout=PIPE, timeout=timeout, check=True,
 ~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 **kwargs).stdout
 ^^^^^^^^^
 File "/Users/itamaro/work/cpython/Lib/subprocess.py", line 575, in run
 raise CalledProcessError(retcode, process.args,
 output=stdout, stderr=stderr)
subprocess.CalledProcessError: Command '['/var/folders/ng/hmf622ys3wnc3c21prtx7nd80000gn/T/tmpkc8z54mu/python.exe', '-c', 'import sys; print("\\n".join(sys.path) if sys.flags.no_site else "")']' returned non-zero exit status 1.

----------------------------------------------------------------------
Ran 1 test in 0.032s

FAILED (errors=1)
test test_site failed
test_site failed (1 error)

== Tests result: FAILURE ==

1 test failed:
 test_site

Total duration: 150 ms
Total tests: run=1 (filtered)
Total test files: run=1/1 (filtered) failed=1
Result: FAILURE

CPython versions tested on:

3.12, 3.13, CPython main branch

Operating systems tested on:

Linux, macOS

Linked PRs

• gh-113628: Fix test_site test with long stdlib paths #113640
• [3.12] gh-113628: Fix test_site test with long stdlib paths (GH-113640) #113671
• [3.11] gh-113628: Fix test_site test with long stdlib paths (GH-113640) #113672

[itamaro]

A few ways to go about this:

1. Claim that to exceed 32KB the path needs to be so long that it's fine to fail in that case, and close the issue as unplanned. (while I acknowledge this requires quite a long path to hit in practice, I did run into this in our CI system at Meta, so it's not totally hypothetical)
2. Decrease the number of repetitions in test_site (e.g. from 200 to 20) to raise the threshold for the path length that would trigger this failure (this fixes it for me internally). I don't know what's the significance of the number of repetitions in these tests, and why 200 was chosen specifically.
3. Keep it at 200 repetitions, but add a check in the tests that if the generated _pth file exceeds 32KB, the test skips itself (e.g. raise some resource denied).
4. Some mix of 2 & 3 -- make the _pth generation "adaptive" - start with 200, but as long as the file exceeds 32KB try fewer repetitions (e.g. 100, 50, 20), and skip the test if even the fewest number of repetitions exceeds 32KB.

I'm happy to submit a PR for any of the 2-4 options.

[terryjreedy]

A test that fails with a legal stdlib path over about 160 (2**15//200 rounded down) chars is broken. I would replace 200 with min(200, 30000 // len(libpath)).