Fix: every TUI palette command taking a path was broken on Windows

The last two Windows failures were not POSIX-assuming tests. The tests were
right and the application was wrong.

`parse_palette_command` split its input with `shlex.split`, which defaults to
POSIX mode where a backslash is an ESCAPE character. On Windows that quietly
destroys every path a user types:

    C:\Users\me\Temp\exports  ->  C:UsersmeTempexports

So doc-export, export-field, db-export, db-import, parent-seed-backup and
totp-export all wrote to a mangled relative path — or nowhere — and then
reported success. A user asking the TUI to export their vault on Windows got
a success message and no file. The two failing tests were reporting exactly
that, from the one platform that runs them.

Both palettes now share `split_palette_args`, which uses non-POSIX splitting
on Windows and strips the quotes that mode leaves on a quoted token, so paths
with spaces still work. v3 imports it from v2 rather than keeping its own
copy, since two parsers that must agree will eventually not.

The `windows` argument is an explicit parameter rather than a bare os.name
check, so both branches are tested from either platform — including a test
that pins what POSIX mode does to a Windows path, so anyone "simplifying"
this back to shlex.split sees precisely what they are restoring. That
parameter is the actual fix for the class of bug: the reason this survived is
that the Windows branch could only ever run on Windows.

Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
Claude-Session: https://claude.ai/code/session_01Vtx94nXDmNgXEeThUZP4QN
This commit is contained in:
thePR0M3TH3ANandClaude Opus 5 committed 2026-08-20 11:37:17 -04:00
1 parent 6fcd94b51a
commit 81b0117c37
4 files changed
+99 -5

No files matched your search

+12 -2
View File
@@ -253,8 +253,18 @@ everything still open, in the order it should be tackled.
deliberately with the sync round-trip re-validated against a real
relay rather than folded into unrelated work.
(Found 2026-08-20 on the first CI run of this branch.)
- [ ] **Two TUI tests still fail on Windows, and only Windows ever runs
them.** Surfaced 2026-08-20 once the Tests workflow could report a
- [x] **Two TUI tests still fail on Windows, and only Windows ever runs
them.** RESOLVED — and they were right. The tests were not
POSIX-assuming; the APP was. `parse_palette_command` used
`shlex.split`, which defaults to POSIX mode where a backslash is
an escape character, so on Windows
`C:\Users\me\exports` became `C:Usersmeexports` and every
palette command taking a path — doc-export, export-field,
db-export, db-import, parent-seed-backup, totp-export — wrote to a
mangled relative path or nowhere, while telling the user it had
worked. Both palettes now share `split_palette_args`, which takes
an explicit `windows` flag so BOTH branches are tested from either
platform. Original wording kept below for the record. Surfaced 2026-08-20 once the Tests workflow could report a
Windows failure at all (see below). Four failed; two were literal
`/tmp/...` comparisons against paths the app echoes
platform-normalized, and those are fixed by deriving the expected
+29 -1
View File
@@ -2,6 +2,7 @@ from __future__ import annotations
import importlib.util
import json
import os
import shlex
import time
from pathlib import Path
@@ -30,7 +31,7 @@ def parse_palette_command(command: str) -> tuple[str, list[str]]:
if not raw:
raise ValueError("Palette: command required")
try:
parts = shlex.split(raw)
parts = split_palette_args(raw)
except ValueError as exc:
raise ValueError(f"Palette parse error: {exc}") from exc
if not parts:
@@ -38,6 +39,33 @@ def parse_palette_command(command: str) -> tuple[str, list[str]]:
return parts[0].lower(), parts[1:]
def split_palette_args(raw: str, *, windows: bool | None = None) -> list[str]:
"""Split a palette command line into tokens, preserving Windows paths.
`shlex.split` defaults to POSIX mode, where a backslash is an ESCAPE
character. On Windows that silently destroys every path a user types:
`C:\\Users\\me\\exports` comes back as `C:Usersmeexports`, so
doc-export, export-field, db-export, db-import and parent-seed-backup all
wrote to a mangled relative path — or nowhere — while reporting success.
Found by the Windows CI job once it could run these tests at all.
Non-POSIX mode keeps backslashes but also keeps the surrounding quotes on
a quoted token, so those are stripped back off. `windows` is a parameter
rather than a bare `os.name` check so both branches are testable from
either platform.
"""
on_windows = os.name == "nt" if windows is None else windows
if not on_windows:
return shlex.split(raw)
parts = shlex.split(raw, posix=False)
unquoted: list[str] = []
for part in parts:
if len(part) >= 2 and part[0] == part[-1] and part[0] in "\"'":
part = part[1:-1]
unquoted.append(part)
return unquoted
def pagination_window(
total_rows: int, page_size: int, page_index: int
) -> tuple[int, int, int, int]:
+5 -2
View File
@@ -48,10 +48,13 @@ class CommandProcessor:
self.app = app
def execute(self, raw: str) -> None:
import shlex
# Shared with the v2 palette so the two cannot drift: plain
# shlex.split eats backslashes, which destroys every Windows path a
# user types into a command that takes one.
from seedpass.tui_v2.app import split_palette_args
try:
parts = shlex.split(raw)
parts = split_palette_args(raw)
except Exception as e:
self.app.notify(f"Parse error: {e}", severity="error")
return
+53
View File
@@ -6,6 +6,7 @@ from seedpass.tui_v2.app import (
pagination_window,
parse_palette_command,
render_qr_ascii,
split_palette_args,
truncate_entry_for_display,
)
@@ -79,3 +80,55 @@ def test_pagination_window_large_vault_sizes(rows: int) -> None:
assert page == last_page
assert total_pages >= 1
assert 0 <= start <= end <= rows
class TestSplitPaletteArgs:
"""Both platform branches, tested from either platform.
`windows` is an explicit parameter rather than a bare `os.name` check
precisely so these can run everywhere. Without that, the Windows branch
would only ever be exercised on Windows — which is how the bug below
survived: shlex.split defaults to POSIX mode, where backslash is an ESCAPE
character, so every Windows path a user typed came back with its
separators eaten and doc-export, export-field, db-export, db-import and
parent-seed-backup wrote nowhere while reporting success.
"""
def test_windows_paths_keep_their_separators(self) -> None:
assert split_palette_args(
r"doc-export C:\Users\me\Temp\exports", windows=True
) == ["doc-export", r"C:\Users\me\Temp\exports"]
def test_posix_mode_would_have_eaten_them(self) -> None:
# The bug, pinned. If someone "simplifies" the helper back to a plain
# shlex.split, this is what Windows users get.
assert split_palette_args(
r"doc-export C:\Users\me\Temp\exports", windows=False
) == ["doc-export", "C:UsersmeTempexports"]
def test_quoted_windows_paths_survive_with_quotes_removed(self) -> None:
# Non-POSIX shlex keeps the quotes on the token; they have to come off
# or the path gains a literal quote character.
assert split_palette_args(
'db-export "C:\\Program Files\\out.enc"', windows=True
) == ["db-export", r"C:\Program Files\out.enc"]
def test_posix_behaviour_is_unchanged(self) -> None:
assert split_palette_args("doc-export /tmp/exports", windows=False) == [
"doc-export",
"/tmp/exports",
]
assert split_palette_args('db-export "/tmp/with space.enc"', windows=False) == [
"db-export",
"/tmp/with space.enc",
]
def test_quoting_still_groups_on_windows(self) -> None:
assert split_palette_args(
'link-add 42 related_to "note text"', windows=True
) == [
"link-add",
"42",
"related_to",
"note text",
]