[gmmcosta15]

Description

• Feature
• Bug Fix
• Code Refactor
• Documentation

Printing a file from a USB drive now copies it into ~/printer_data/gcodes/ and prints the local copy, so removing the drive mid-print no longer kills the job. The copy is atomic and only starts printing once Moonraker (and Happy Hare's preprocessor) has finished with it. Reprinting the same USB file reuses the existing local copy instead of making a new one.

Changes

• New lib/utils/usb_import.py: validates the USB path (mounted drive, no symlink or .. escape, .gcode only), checks free space, copies via a fsynced temp file in printer_data/.usb_import/ and os.replace, runs off the UI thread as UsbImportTask
• Reuse: an index in .usb_import/index.json maps each USB file (name, size, head and tail hash) to its processed local copy; byte-identical same-name files are reused too, a different file with the same name gets a _n suffix
• printTab.start_print routes USB files through the import; a new copy prints after Moonraker's create_file for it, or after a 120 s timeout
• Import errors show a notification and never start a print
• Unit tests for the import module and the print-start flow

Motivation

Prints started from a USB drive read the file from the drive for the whole job, so pulling it or a loose connection ended the print. Happy Hare also rewrites new gcodes, so printing before it finishes left Klipper reading a deleted file.