* [Client] Expose custom zone management in the deck editor
Wires the state layer into every editor surface that shows deck zones.
- Deck dock: context menu on zones gains New/Rename/Delete/Change
board actions, with per-zone submenus for adding cards.
- Card database dock and visual database display gain an add-to-zone
submenu listing custom zones per board plus a create-zone entry.
- All prompt call sites pass validateNewZoneName so duplicates and
reserved names are rejected inline before Ok unlocks.
- Rename reuses the same dialog in name-only mode, keeping one
validation contract for every zone-name entry point.
- Change board marks the current board instead of offering a no-op,
and the state layer refuses moves onto boards holding a same-named
zone from imported decks.
* [DeckEditor] Address custom-zone menu and export review feedback
* [DeckLoader] Keep the sideboard marker and block ordering when exporting nested zones
- saveToStream_DeckZone threads the owning board zone name down to the card
writer, so cards in a custom zone under the sideboard keep their SB:
prefix instead of being re-imported into the maindeck
- nested sub-zones are collected during the loop and written after the
parent zone's own header and cards, so they no longer read as part of the
zone printed before them
* [DeckEditor] Fix move-to-zone menu use-after-free and per-zone enabled state
- resolve the card name/provider/collector number before createNewCustomZone
rebuilds the model tree, then re-find the refreshed index via findCard and
move it (mirrors the decrementCard re-find pattern)
- the enabled test now compares the card's own zone (nearest custom-zone
ancestor, else its board), matching moveCardToZone's lookup, so moving a
card out of a custom zone back to the board root is offered and the card's
own zone is disabled
---------
Co-authored-by: Lukas Brübach <Bruebach.Lukas@bdosecurity.de>
* Fix#6952: prevent deck loss when saving to a full disk
DeckLoader::saveToFile() opened the target with QFile in WriteOnly mode,
which truncates the existing file to 0 bytes the moment it is opened. The
serializers (DeckList::saveToFile_Native/_Plain) always return true and the
result of flush() was ignored, so a write that failed part-way -- e.g.
because the disk was full -- left a 0-byte file behind yet was still
reported (and logged) as a successful save. The same truncate-then-write
pattern in updateLastLoadedTimestamp() could destroy a deck on load.
Switch both paths to QSaveFile, which writes to a temporary file and only
atomically replaces the target if commit() succeeds. On any write or flush
failure commit() returns false, the original deck is left untouched, and
the failure is logged instead of being reported as success.
* Use QSaveFile in convertToCockatriceFormat() too
convertToCockatriceFormat() had the same data-loss pattern: QFile WriteOnly
truncated the .cod, saveToFile_Native() always returns true, and the original
file was then removed unconditionally -- so a full disk during conversion wrote
a 0-byte .cod and then deleted the source deck.
Switch to QSaveFile (write + atomic commit), remove the original only after a
successful commit, and move the format check ahead of the file open so an
already-Cockatrice or unsupported deck never truncates or deletes anything.
Raised in review by ZeldaZach.
* [DeckList] Refactor load from plaintext to take normalizer as param
* update usages
* weaken unit test
* weaken unit test more
* revert unit test
* move CardNameNormalizer to libcockatrice_card
* update unit test
* formatting
* remove helpers
* create getZoneNodes method
* replace direct calls to getRoot and forEachCard
* remove more non-const uses of forEachCard
* make node getter return const lists
* one more usage
* address comment
* address comment again
* fix hash
* fix hashes (for real this time)
* [DeckLoader] Refactor last load info into struct
* Use constant
* [[nodiscard]]
* do discard, I guess.
---------
Co-authored-by: Brübach, Lukas <lukas.bruebach@student.fhws.de>
* [DeckList] Disable copy constructor
Took 1 hour 44 minutes
Took 1 minute
# Commit time for manual adjustment:
# Took 28 seconds
Took 33 seconds
* Revert member to pointer.
Took 19 minutes
* Revert pulling up setters/getters now that getDeckList is no longer const.
Took 6 minutes
* Revert more.
Took 2 minutes
* One more fix.
Took 1 minute
* Update cockatrice/src/interface/deck_loader/deck_loader.cpp
Co-authored-by: RickyRister <42636155+RickyRister@users.noreply.github.com>
---------
Co-authored-by: Lukas Brübach <Bruebach.Lukas@bdosecurity.de>
Co-authored-by: RickyRister <42636155+RickyRister@users.noreply.github.com>
* Initialize all deck list member variables in constructor.
Took 10 minutes
* Revert "Initialize all deck list member variables in constructor."
This reverts commit fba2455808.
* setParent
Took 1 hour 2 minutes
---------
Co-authored-by: Lukas Brübach <Bruebach.Lukas@bdosecurity.de>
* A DeckLoader is not a DeckList.
Took 2 hours 39 minutes
* Explicitly initialize base class in copy constructor?
Took 3 minutes
---------
Co-authored-by: Lukas Brübach <Bruebach.Lukas@bdosecurity.de>
* Deck loader is a gui class.
Took 31 minutes
Took 3 minutes
* Deck Loader is responsible for printing.
Took 8 minutes
Took 2 seconds
* Style proxy.
Took 14 minutes
Took 6 minutes
Took 1 minute
* Don't need to include QBrush anymore.
Took 3 minutes
Took 7 seconds
* Includes for printer.
Took 5 minutes
* Nuke getDeckList()
Took 9 minutes
* Adjust to rebase.
Took 35 seconds
* Lint.
Took 3 minutes
* Braces for one line return statements.
Took 13 minutes
Took 50 seconds
* Enum for model columns.
Took 9 minutes
* One more single line if.
Took 1 minute
* Another style lint on a sunday night
Took 5 minutes
* Move enum to namespace.
Took 3 minutes
* Fix a critical blocker.
Took 5 minutes
* Update docs.
Took 3 minutes
* Doxygen and namespace enums.
Took 2 minutes
Took 15 seconds
* Adjust to namespace.
Took 4 minutes
Took 1 minute
---------
Co-authored-by: Lukas Brübach <Bruebach.Lukas@bdosecurity.de>
