Repository navigation
CoffeeWzRepacker V1.2.5
For optimal stability and performance, please use the latest version. Older versions may contain unresolved issues and performance limitations.
為確保最佳穩定性與效能,建議使用最新版本。舊版本可能存在未修正的問題與效能限制。
English Release Notes (Click to expand)
Added
- Added personal settings to control whether exported files should overwrite existing files
(supports batch operations: overwrite all / skip all) - Added batch node renaming functionality
(can be enabled or disabled in personal settings) - Added batch export support for
Icon/IconRawimages insideImg - Added
Ctrl + ZandCtrl + Shift + Zshortcuts for undo / redo - Added
AI TokenandAI Chatfunctionality- Users can now integrate external LLM APIs to assist with file editing
(supports bothAPI KeyandOAuthauthentication) - Currently supports:
ChatGPT (OpenAI)Gemini (Google)Grok (xAI)PixelLab
- AI-related features are still experimental; please report any issues encountered during usage
(PixelLabis currently the primary tested provider)
- Users can now integrate external LLM APIs to assist with file editing
- Added database functionality similar to
DB2, currently only applied to opened files- Supports map rendering
(Some older versions may experience rendering inconsistencies depending on resolution settings) - Allows exporting files into binary cache data for faster future loading
(Map rendering still requires the original files to be loaded)
- Supports map rendering
- Added automatic update checking
(can be enabled or disabled in personal settings) - Added
Ctrl + Sshortcut support for:WzsavingImgexportingCanvassaving
- Added support for reading
Lua
Fixed
- Fixed failure when saving
msfiles
(mscurrently does not support encryption conversion betweenv2 (Snow2)andv4 (ChaCha20)) - Fixed application freezing after resuming from sleep mode
- Fixed incorrect node highlighting after editing data through the bottom-right information panel
- Fixed video parsing failures
- Fixed save-related issues to prevent currently opened files from being closed while occupied
- Fixed potential save failures when batch-converting image compression formats
(BGRA-8888->DXT3) - Fixed occasional disappearance of loading animations when opening multiple files
- Fixed
Export All PNGs From Subtreenot functioning correctly on sub nodes - Fixed
sourcenodes failing to correctly resolve image resources - Fixed image resource resolution failures when consecutive
_inlink/_outlinkpaths were used inCanvas - Fixed incorrect total frame count calculations for frame-based animations
Improved
- Improved automatic detection of the current
Wzencryption format during saving - Improved file path restoration behavior after saving
- Improved resource preview reset logic to avoid unexpected resets after short idle periods
- Improved audio playback by switching to
bass.dlldecoder-based playback
(avoids exporting temporary audio files during playback) - Improved hotkey configuration workflow
- Improved image linking functionality, now supports cross-workspace image linking within the same tab
Changed
- Adjusted new
msfile creation to allow selecting:v2 (Snow2)v4 (ChaCha20)
- Adjusted
msfiles to support adding:- directories
Imgnodes
- Adjusted portions of the internal resource path structure
- Adjusted Keep the node tree collapsed when loading folders
Search System Refactor
- Completely refactored the search system
- New search logic now uses:
DFSBFS- tokenized search modes
- Added the
Ctrl + Fshortcut, which opens the advanced search settings - Unified node search behavior
- Implemented
Matchlogic support - Moved search processing to background threads to prevent severe UI freezing
- Added optional deep search support
- Deep search supports up to 2 nested levels
- Otherwise, only expanded nodes will be searched
中文更新說明 (點擊展開)
新增
- 新增個人設定,現可透過設定決定匯出時是否複寫檔案(支援批次的全部覆寫 / 全部略過)
- 新增批量重新命名節點名稱功能 (可於個人設定中開啟或關閉)
- 新增批次匯出
Img內Icon/IconRaw圖示功能 - 新增
Ctrl + Z及Ctrl + Shift + Z復原 / 重做的快捷鍵 - 新增
AI Token以及AI Chat功能- 使用者現在可以串接各 LLM API 來協助編輯檔案 (提供
API Key及OAuth兩種模式) - 目前支援
ChatGPT (OpenAI)Gemini(Google)Grok (xAI)PixelLab
- AI 功能目前仍屬實驗性功能,若使用過程中遇到問題請協助回報 (目前僅實測
PixelLab)
- 使用者現在可以串接各 LLM API 來協助編輯檔案 (提供
- 新增類似
DB2的資料庫功能,目前僅對當前開啟檔案生效- 支援地圖渲染
P.S. 部分舊版本可能因解析度設定不同而產生顯示異常 - 支援將檔案匯出為二進制快取資料,方便後續快速讀取 (地圖渲染依然需要載入對應檔案)
- 支援地圖渲染
- 新增自動檢查更新功能 (可於個人設定中決定是否啟用)
- 新增
Ctrl + S快捷鍵支援:Wz保存Img匯出Canvas保存
- 新增支援讀取
Lua
修正
- 修正
ms檔案保存失敗的問題 (ms檔案暫不支援在v2 (Snow2)與v4 (ChaCha20)間進行加密格式轉換) - 修正從睡眠模式喚醒後造成軟體沒有回應的問題
- 修正透過右下角資訊窗編輯資料後,節點未正確
HighLight的問題 - 修正影片解析失敗問題
- 修正存檔時相關錯誤,避免檔案被占用時提前關閉目前檔案
- 修正批次轉換圖片壓縮格式時 (
BGRA-8888->DXT3) 可能造成無法正常儲存問題 - 修正載入多個檔案時,載入動畫偶發性消失問題
- 修正針對
Sub節點執行匯出子樹全部PNG功能時未正確生效問題 - 修正
source節點無法正確取得圖片資源問題 - 修正
Canvas解析路徑連續為_inlink/_outlink時無法正確取得圖資問題 - 修正連續幀動畫總幀數計算錯誤問題
優化
- 優化存檔時自動帶入當前
Wz解密格式 - 優化存檔後重新帶入檔案路徑
- 優化重置資源顯示邏輯,避免短時間未操作導致預覽資源被重置
- 優化音訊播放邏輯,改為依賴
bass.dll解碼器處理 (避免播放音訊時匯出暫存檔) - 優化快捷鍵設定方式
- 優化比對模式下的補圖功能,現已支援跨工作區補圖
- 優化透過資料夾載入時檔案採用虛擬資料夾分類
調整
- 調整新建
ms檔案時,可選擇v2 (Snow2)v4 (ChaCha20)
- 調整
ms檔案可支援新增目錄及Img節點 - 調整部份資源路徑結構
- 調整資料夾載入時,節點樹保持收合狀態
搜尋系統重構
- 全面重構搜尋系統
- 全新搜尋邏輯採用
DFSBFSToken化搜尋模式
- 新增
Ctrl + F快捷鍵,該快捷可觸發進階搜尋設定 - 統一節點搜尋邏輯
- 實作
Match邏輯 - 搜尋功能移動到背景執行序處理,避免嚴重卡死 UI 介面
- 新增可選深度搜尋功能
- 深度搜尋最多支援 2 層節點
- 否則僅搜尋目前已展開節點
Important Notes / 注意事項
- New features have undergone basic testing, but undiscovered issues may still exist.
新功能已完成基礎測試,但仍可能存在未發現的問題。 - Please report any issues on the project page if encountered.
若發現任何異常,請於頁面回報 Issue。
開發者說明 (點擊展開)
首先,感謝所有持續使用 CoffeeWzRepacker 的使用者,也很抱歉這次版本更新間隔較長。
由於本次更新加入了部分較為複雜的功能,因此花費了較多時間進行調試與穩定性驗證,同時也深入修復了一些較細部的問題。
目前各項功能仍可能存在部分未發現的問題,若各位在使用過程中遇到 Bug,歡迎透過 Issue 頁面回報。若能附上影片、圖片或重現步驟,將有助於我更快速地進行排查與修正。
本軟體仍將盡可能維持免費使用,也歡迎各位提出功能建議與改善方向。
另外,由於目前整體功能與架構已逐漸趨於穩定,未來若無重大問題,版本更新頻率可能不會像早期開發階段那麼頻繁。不過當有新版本發布時,現在也可透過更新檢查功能進行確認。
最後,如果你願意支持本專案的持續開發,也非常感謝你的支持。
Developer Notes (Click to expand)
First of all, thank you to everyone who continues to use CoffeeWzRepacker, and I apologize for the longer delay between this release and the previous update.
This version introduced several more complex features, which required additional time for debugging, stability testing, and resolving a number of smaller issues internally.
There may still be undiscovered bugs or edge cases within certain features. If you encounter any issues while using the software, please report them through the Issues page. Providing screenshots, videos, or reproduction steps will greatly help with investigation and fixing.
This software will continue to remain free to use whenever possible, and feature suggestions or improvement ideas are always welcome.
As the overall architecture and feature set are becoming more stable, future updates may not be released as frequently unless major issues or significant new features are introduced. However, newly released versions can now also be checked through the built-in update checker.
Lastly, if you would like to support the continued development of this project, your support would be greatly appreciated.