fix(trash): strip only trailing trash suffix in to_origin_dir_path

[Siraryansingh]

What this PR does / why we need it:

Fixes to_origin_dir_path() and to_origin_file_path() in trestle/common/trash.py so they strip only the trailing __bk and .bk suffix instead of splitting on all occurrences of __bk or .bk inside directory and file names.

Previously, directory names containing __bk (e.g. alpha__bkbeta) were incorrectly truncated to alpha during recovery.

Which issue(s) this PR fixes:

Fixes #2210

How was this tested?:

• Added test_to_origin_dir_path_with_bk_in_name unit test in tests/trestle/utils/trash_test.py
• Ran all 17 trash unit tests via pytest (17 passed in 0.04s)
• Linted with ruff (all checks passed)

[degenaro]

@Siraryansingh Thx for this PR. Are you on our slack channel #oscal-compass-trestle-agileauthoring-c2p? To join you need a LF ID, you can get one here https://identity.linuxfoundation.org/.

Below is Bob's review. Please fix the DCO and missing test issues, thx!

=====

PR #2341 Review — fix(trash): strip only trailing trash suffix in to_origin_dir_path

Author: Aryan Singh (@Siraryansingh)
Base: develop
Files changed: trestle/common/trash.py (+7 / -4), tests/trestle/utils/trash_test.py (+10 / 0)
Issue fixed: #2210

---

🚫 Blocker — DCO not correctly signed

The commit 6e1a56a contains:

Signed-off-by: Aryan Singh <aryansingh.as1012@gmail.com>

However, the commit was authored under the GitHub identity siraryansingh2005@gmail.com. The DCO requires that the email in the Signed-off-by trailer exactly matches the email associated with the commit author identity. Because these two addresses differ, the DCO check fails.

Resolution: The author must amend the commit so that the Signed-off-by email matches the author email on record, then force-push:

git commit --amend --signoff
# verify the Signed-off-by email matches your git user.email, then:
git push --force-with-lease

---

Summary

This PR fixes a real bug in trestle/common/trash.py: both to_origin_dir_path() and to_origin_file_path() used str.split() to remove the __bk / .bk trash suffixes. That approach silently truncates any path component or filename that contains those strings in the middle — for example, a directory named alpha__bkbeta would be recovered as alpha.

---

Fix Quality ✅ Correct

The change is minimal and precise.

Before (same pattern in both functions):

parts = item.split(TRESTLE_TRASH_DIR_EXT)
origin_path_parts.append(parts[0])

str.split() splits on every occurrence, so alpha__bkbeta__bk → ['alpha', 'beta', ''] → 'alpha'.

After:

if item.endswith(TRESTLE_TRASH_DIR_EXT):
    item = item[:-len(TRESTLE_TRASH_DIR_EXT)]
origin_path_parts.append(item)

This correctly strips only the trailing suffix. alpha__bkbeta__bk → alpha__bkbeta. ✅

The fix is applied symmetrically to both to_origin_dir_path() (directory parts) and to_origin_file_path() (filename).

---

Test Coverage ✅ Adequate — one minor gap

• test_to_origin_dir_path_with_bk_in_name correctly exercises to_origin_dir_path() with a directory whose name contains __bk in the middle.
• All 17 trash unit tests pass locally.

Minor gap (non-blocking): There is no analogous regression test for to_origin_file_path() with a filename that contains .bk in the middle (e.g. report.bk2.md). A suggested addition:

def test_to_origin_file_path_with_bk_in_name(tmp_path: pathlib.Path) -> None:
    """Test to_origin_file_path handles file names containing .bk."""
    test_utils.ensure_trestle_config_dir(tmp_path)
    (tmp_path / trash.TRESTLE_TRASH_DIR).mkdir(exist_ok=True, parents=True)
    bk_file = tmp_path / 'report.bk2.md'
    trash_file_path = trash.to_trash_file_path(bk_file)
    origin_file_path = trash.to_origin_file_path(trash_file_path)
    assert bk_file.resolve() == origin_file_path.resolve()

---

Round-trip Correctness ✅

to_trash_dir_path() appends the suffix to the full relative path string exactly once. The fixed to_origin_dir_path() now strips it from the end exactly once. The forward/reverse round-trip is mathematically consistent.

---

Code Style ✅

item[:-len(TRESTLE_TRASH_DIR_EXT)] is idiomatic Python. An equally readable alternative is item.removesuffix(TRESTLE_TRASH_DIR_EXT) (Python 3.9+), but the current form is acceptable.

---

Overall Verdict

Concern	Status
DCO signature mismatch	🚫 Blocker
Logic correctness	✅ Correct
Test for directory case	✅ Present
Test for file case with .bk in name	⚠️ Missing (non-blocking suggestion)
Code style	✅ Fine

The fix itself is correct and ready to merge once the DCO issue is resolved.

[degenaro]

Also, some tips. Run locally:

make develop
make code-format
make code-lint
make mdformat
make test

[degenaro]

Some additional recommendations:

• Use the constant in the existing dir test — replace the hardcoded 'alpha__bkbeta' string with f'alpha{trash.TRESTLE_TRASH_DIR_EXT}beta'.

• Add a trailing-marker dir test — cover the alpha__bk → alpha__bk__bk → alpha__bk round-trip (the case PR fix: preserve __bk names in trash restore #2211 has but fix(trash): strip only trailing trash suffix in to_origin_dir_path #2341 is missing).

• Add a file-name embedded-marker test — to_origin_file_path is fixed by this PR but has no regression test for a file like report.bkfinal.md.