[va-resident]

Fixes #935

Change, cmd_check_bugreport.py (+49/−2) and modules/bugreport/base.py (+10/−3):

• DUMPSTATE_ENTRY_POINTS moves into modules/bugreport/base.py, next to the
  _get_dumpstate_file() that tries them in that order — so the entry points are named once
  rather than once per caller;
• _has_dumpstate(file_names) — any(fnmatch.filter(file_names, pattern) for pattern in DUMPSTATE_ENTRY_POINTS), the same matching _get_files_by_pattern() already uses;
• _nested_bugreport(outer) — only when they are absent: open the outer archive's .zip members
  in memory, names containing bugreport / dumpstate first, and take the first one that does
  carry an entry point. Logs which member was used.

The descent is unreachable for an archive that already has an entry point, which is all 30 normal
collections in the test set — including the 27 that carry some other zip member.

On placement: you asked on #935 where this code should live and passed the question to
@DonnchaC. This is my reading of it — the entry points next to the only code that consumes them,
the descent in the command that owns the archive — but nothing here depends on that choice. Say
where you would rather have it and I will move it.

Measured on 44 bug report collections, upstream vs patched — and re-run on the 14 wrappers
after the descent was simplified, with identical totals, so the numbers below come from the code
this PR ships:

set	upstream	patched
14 wrapper collections	0 artifacts and 0 records on 13; one tombstones.json with 4 records on the fourteenth; exit 0 throughout	260 artifacts, 586 638 records, 109 502 timeline rows, 94 alerts
30 normal collections, 14 brands	(baseline)	same artifacts, same records, same alerts — every file identical once record order is normalised

On the controls, 29 of 30 runs differ byte-for-byte in timeline.csv and tombstones.json and 2
of them in alerts.json; the contents are the same records in a different order, and two runs of
the same build differ the same way. Record order in those files is not deterministic in main —
worth a separate issue, and worth knowing before anyone uses a checksum to compare two runs.

Tests: tests/android/test_check_bugreport_wrapper.py, 2 cases — a wrapper is descended into,
and a plain archive that merely contains some other zip is left alone.

Applies to main@777db67 with no other change; pytest tests/android there goes from 99 passed /
1 skipped to 101 passed / 1 skipped. ruff check, ruff format --check and
mypy --platform linux (156 files) pass on the branch.

Related open work (re-checked 2026-09-22, none of these has landed):

• Run extraction modules in parallel #839 "Run extraction modules in parallel" rewrites modules/bugreport/base.py from
  _get_files_by_patterns() down (its hunk spans lines 66–226, which contains
  _get_dumpstate_file()) and also edits cmd_check_bugreport.py. If it lands first this PR needs
  a real rebase, not a context fix.
• Add Android bugreport output schema version #884 "Add Android bugreport output schema version" edits the same __init__ block of
  cmd_check_bugreport.py.
• Add versioned output schemas #885 "Add versioned output schemas" edits base.py's import block, immediately above where the
  new constant sits.
• My PR for A section's timing line truncates the section being parsed, silently #938 touches base.py's import block too. Whichever of the two you take first, the
  other is a trivial rebase — say the word and I will do it.
• Fix AndroidQF artifact discovery in root-level ZIP archives #933 "Fix AndroidQF artifact discovery in root-level ZIP archives" fixes a different
  archive-discovery defect — */packages.json not matching a root-level member — in
  modules/androidqf/base.py. No file and no logic in common; worth citing only as the
  AndroidQF-side analogue.
• No open issue or PR mentions a nested / wrapper bug report archive.

[va-resident]

Rebased onto main after #945 landed. The only conflict was in the import block of modules/bugreport/base.py: DUMPSTATE_ENTRY_POINTS now sits right after SECTION_DURATION. No other changes. Lint, mypy and tests pass.

[besendorf]

I merged this too early. Foind 2 issues:

The empty-success case remains. If the wrapper has no usable nested ZIP, from_zip() keeps the outer archive⁠�. The command can still exit 0 with an empty analysis, contrary to the expected behavior in #935⁠�. It should report that no bug report was found.

Descent can discard wrapper-only evidence. The same lines replace the outer ZIP and its file list entirely. The tombstone module can currently read tombstones from the wrapper; #935 reports four such records in one collection. If those files are absent from the inner ZIP, this change silently drops them. That case needs a regression test and a way to retain or explicitly account for those files.