Copy files with certain extension from many nested sub-directories to a single directory and append to each copied file the name of the directory

Question

I have a parent directory with thousands of directories within sub-directories and so on. In some of the sub-directories I have few image files.

I want to copy (rsync is preferred) all of the bmp and jpg files from the sub-directories to a single new directory without any sub-directories. But at the same time I want to append the name of the LAST directory where the file comes from to the copied file.

├── example1
│   ├── fdfd
│   │   ├── db_files
│   │   │   ├── asdasd.zip
│   │   │   ├── dfdfsta.zip
│   │   │   ├── csf_4545.zip
│   │   │   ├── 45455.zip
│   │   │   ├── 4544.zip
│   │   │   ├── 45545.zip
│   │   │   ├── gfdfgd.zip
│   │   │   ├── retert.zip
│   │   │   ├── vcxvxc.zip
│   │   │   ├── vcxcvcx.zip
│   │   │   ├── asdr3.zip
│   │   │   ├── tetst.zip
│   │   │   ├── testeh.bmp
│   │   │   └── testst43.zip
│   │   ├── TEST_FILE.bmp
│   │   └── hfexport.csv
│   └── testFOLDER2018-05
│       ├── Databasef.txt
│       ├── Folder_Backup_2014-01-05.7z
│       ├── full_Export.pdf
│       ├── f-to-7000.csv
│       ├── f-to-505.csv
│       ├── f-to-600.csv
│       ├── f-to-960.bmp
│       ├── g-to-1000.jpg
│       └── 7000.csv
├── example3
│   ├── img_copy
│   │   ├── automation
│   │   │   ├── f.sh
│   │   │   ├── fff.sh
│   │   │   └── test.bat
│   │   ├── all.bmp
│   │   ├── h23.txt
│   │   ├── old2_copy.jpg

In the example above, the copied files will be as follows:

db_files_testeh.bmp
fdfd_TEST_FILE.bmp
testFOLDER2018-05_f-to-960.bmp
testFOLDER2018-05_g-to-1000.jpg
img_copy_all.bmp
img_copy_old2_copy.jpg

How can I achieve this?

I use the following commands to copy only bmp and jpg files to a new directory.

find -regex '.*\.\(bmp\|jpg\)' | tee /home/tmp/matches.txt
while read -r LINE; do rsync -vh --append-verify --chmod=ugo=rwX --progress --stats "$LINE" /mnt/storage3/backups/NEW_DIR/ONLY_IMAGES; done < /home/tmp/matches.txt

Answer 1

find . -type f '(' -name '*.bmp' -o -name '*.jpg' ')' -exec sh -c '
    for pathname do
        newname="${pathname%/*}_${pathname##*/}"  # a/b/c/d.bmp --> a/b/c_d.bmp
        newname="target/${newname##*/}"           # a/b/c_d.bmp --> target/c_d.bmp

        printf "Would move %s to %s\n" "$pathname" "$newname"
        # mv -i "$pathname" "$newname"
    done' sh {} +

This finds all regular files whose filenames end with either .jpg or .bmp. For batches of these files, a short shell script is executed.

The shell script extracts the directory pathname of each file, along with its filename and concatenates the two with an underscore in-between. This forms the new name. The new name is then trimmed of it's initial directories, leaving only the filename, and a target directory path is appended to this.

The final mv is commented out for safety. Run this once with it commented out to see that it does the correct thing.

Related:

• Understanding the -exec option of `find`