Move deleted images to trash and clean metadata

[YDLuo-1]

Problem

The delete confirmation says that an image will be moved to the recycle bin, but Windows used os.remove, which permanently deleted it. After deletion, the code also removed the wrong fileStatedict key and only changed PPlabel in memory. fileState.txt, Label.txt, and Cache.cach could therefore retain references to a missing image, and Cache.cach could restore that stale entry on the next launch.

Changes

• move files to the recycle bin on Windows with the standard-library SHFileOperationW API
• use argument lists for Linux trash invocation and keep the existing Finder behavior on macOS
• stop without changing metadata when moving the file to trash fails
• remove the normalized image key from fileStatedict, PPlabel, and Cachelabel
• persist all three metadata files before refreshing the image list

No additional dependency is introduced.

Verification

• real Windows temporary-file check confirmed that the file is removed through the recycle-bin operation
• mock regression check confirmed successful deletion cleans and saves all three metadata stores
• mock failure check confirmed metadata remains unchanged when the file cannot be moved
• python -m py_compile PPOCRLabel.py
• git diff --check

[Copilot]

Pull request overview

This PR fixes cross-platform image deletion so that “delete” moves the image to the OS trash/recycle bin (instead of permanently removing it on Windows) and ensures related metadata stores are cleaned and persisted consistently to avoid stale entries on restart.

Changes:

• Adds a cross-platform moveFileToTrash() helper (Windows SHFileOperationW, Linux trash, macOS Finder via osascript).
• Stops deletion flow without mutating metadata if moving to trash fails.
• Removes the deleted image’s normalized key from fileStatedict, PPlabel, and Cachelabel, then saves all three stores before refreshing the image list.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[GreatV]

LGTM