[SOLVED] recursively copy specific files using a list

atjurhs · 04-10-2017, 12:23 PM

hi guys,

I need to copy a set of specific files that exist in a directory structure with sub-directories. my initial attempt was to use

Code:

find . -name "*.h5" -exec cp {} ~/home/tmp ";"

this copied all of files that are of type .h5 but some of these .h5 files are really really big BIG, on the order of several gigs, and i don't need/want them.

so i need to find a way to copy specific files that all end in .h5 but also have different names such as:

Code:

.123_yellow.h5
.123_yellow.csv
123_Chevrolet.h5
abcdefg.log
abc_red.log
abc_red.h5
windows_123.h5
     /subdirectory
          .green.h5
          .purple.h5
          mac_ford.h5
          linux_white.h5

where in this directory structure i only want to keep those files with a "color" as part their name. so i would first make a separate file that has a listing of the files i need:

Code:

.123_yellow.h5
.green.h5
.purple.h5
abc_red.h5
linux_white.h5

then execute a command that reads the list and copies only the files i need and that are in the list to the ~/home/tmp

is this doable or way to involved for an awk or bash or python, btw i don't know python or perl

can you help or not?

thanks!!!

Todd

Turbocapitalist · 04-10-2017, 12:37 PM

Since you've started with find you can continue with that just as well as with other options. Options in find are connected by a logical AND unless specified otherwise. If you use a logical OR then you'll need to group parts using parenthesis. And to keep the shell from interpreting the parenthesis, they need to be escaped or quoted:

Code:

find . -name '*.h5' \( -name '*red*' -o -name '*green*' \) -print

When you have what you want, swap out the -print for -exec as you had.

atjurhs · 04-10-2017, 02:09 PM

cool!

that does look much simpler than i had thought this do be!

i have a long list of "color" files but this solution is easy and brute force (i like brute force) will get it done

rtmistler · 04-10-2017, 02:28 PM

First, you could move them versus copy them.

Second, that first option does seem unlikely, you may have a database of files which you wish to retain and then reference. Therefore my way of thinking would be that you would be creating a custom new file or set of files and perhaps taking that off system? Thus it might be going to a temporarily mounted file system like a thumb stick? Otherwise if you just need the reference names as part of your find command, then you could output just the list of found files and their locations and keep that versus copy the files all the time. All depends how much, how often, and how large the sub-selected files happen to be.

atjurhs · 04-10-2017, 02:59 PM

Quote:

Originally Posted by rtmistler

First, you could move them versus copy them.

Second, that first option does seem unlikely, you may have a database of files which you wish to retain and then reference. Therefore my way of thinking would be that you would be creating a custom new file or set of files and perhaps taking that off system? Thus it might be going to a temporarily mounted file system like a thumb stick? Otherwise if you just need the reference names as part of your find command, then you could output just the list of found files and their locations and keep that versus copy the files all the time. All depends how much, how often, and how large the sub-selected files happen to be.

thanks for the thought!

but i can not do a mv as there are other users that also need access to the files

allend · 04-10-2017, 04:49 PM

If the concern is the size of the files, you could consider using the -size option with find.

Code:

find . -name '*.h5' -size -<your size specification> -print

syg00 · 04-10-2017, 07:15 PM

Do you need to copy them at all - would a link suffice ?.