| Age | Commit message (Collapse) | Author |
|---|
|
- Replace the inaccurate "Go compiler can optimise away" claim in
parseContent's doc with an accurate explanation: callers that already
hold a string pass it directly, while []byte callers make a single
explicit conversion at the call site, keeping the conversion visible
rather than hidden inside the function (100 Go Mistakes #40).
- Add a comment in importBytes explaining why ctx is not threaded into
addAttachment or addTextEntry: both are synchronous, purely in-memory
operations with no I/O or blocking calls that could respect cancellation.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
|
|
(task q4)
Add 430 lines of new integration tests covering WriteBack edit round-trip,
Search action handlers (cat/export), ShredAllExported, ImportRecursive,
TestParsePickerAction (all 10 fzf keys), TestBuildPickerEntries with RowID
assertions, ImportForce, ensureRootGroup edge cases, and a captureStdout
helper. Add TTY-unavailability comment to FzfInteractive explaining why
Fzf/FzfInteractive/runFzfInteractive have no automated tests.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
|
|
k4+l4+m4)
Read-write KeePass backend implementing backend.Backend:
- WalkIndexes flattens groups into 'Group/Title' virtual entries
- LoadData returns formatted Password/User/URL/Notes for text entries
or raw bytes for binary attachments
- WriteBack hook enables edit round-trip through the kdbx file
- Add/Import/ImportRecursive/Remove with atomic tmp+rename save
- Binary attachments surface as virtual 'Group/Title/filename' entries;
Add to such a path creates/replaces the attachment on the parent entry
- AtomicSave exported for reuse by cli/kdbx_store.go
- parseContent is the tolerant inverse of formatContent
- Helpers EnsureGroup, UpsertEntryByTitle, SetEntryField, SplitDescriptionPath,
SanitizeRelativePath exported so cli/kdbx_store.go avoids duplication
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
|
|
The comment at the force=false import test said "should succeed and add
the entry" but the test actually verifies that a duplicate entry is
silently skipped (entry count must not change). Updated to say "should
skip silently because Work/Email already exists."
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
|
