How could I add the containing directory as a prefix to a copied file name? - linux

Question

The issue: I have bunch of files split across multiple directories, which all have the same name (input.txt).
What I am after: I want to firstly copy all of these to a new directory, while adding the containing directory as a suffux to avoid confusion between them and prevent overwriting. This is the basis of what I am trying to do:
cp -nr /foo/bar/*/input.txt /new/path/
Where do I go from here?
To respond to the comments below, if my file structure in /old/directory contains folders:
/old/directory/1/input.txt
/old/directory/2/input.txt
/old/directory/3/input.txt
This is an example of my desired output:
/new/directory/ should contain:
1input.txt
2input.txt
3input.txt
Thanks

Answer 1

This will do the trick and also handle any directories that might have spaces in their names (or any other odd characters).
#!/bin/bash
tld=./old/directory
newpath=./new/directory
while IFS= read -r -d $'\0' file; do
tmp="${file#*${tld}/}"
echo cp "$file" "$newpath/${tmp//\//}"
done < <(find "$tld" -type f -name "input.txt" -print0)
Proof of Concept
$ tree ./old/directory/
./old/directory/
├── 1
│   └── input.txt
├── 2
│   └── input.txt
└── 3
├── 3a
│   └── input.txt
└── input.txt
4 directories, 4 files
$ ./mvinput.sh
cp ./old/directory/3/input.txt ./new/directory/3input.txt
cp ./old/directory/3/3a/input.txt ./new/directory/33ainput.txt
cp ./old/directory/1/input.txt ./new/directory/1input.txt
cp ./old/directory/2/input.txt ./new/directory/2input.txt

Answer 2

Well, the tough news is that there's not an obvious manner of doing this in one line - not one that isn't nonsensically difficult to understand anyway. There may be a way to do it with rsync, and I'm sure that someone smarter than I could do it in awk, but in my opinion, you're better off making a script, or even writing a custom binary that does this for you.
find . -name input.txt | while read line
do
cp "$line" /new/path/`echo $line | cut -d '/' -f3- | sed 's/\//_/'`
done
Note that you'll probably have to change the -f3- part of the cut command in order to select which directory name you want to start your suffix as.

Answer 3

One approach is to use an array to save the files, also since / is not allowed on file names an alternative is to change it to something else, like say an underscore.
#!/usr/bin/env bash
##: Just in case there are no files/directories the * glob will not expand by itself.
shopt -s nullglob
files=(foo/bar/*/input.txt)
for file in "${files[#]}"; do
new_file=${file//\//_} ##: Replace all /'s with an _ by Parameter Expansion
echo cp -v "$file" new/path/"$new_file"
done
As per OP's request here is the new answer.
#!/usr/bin/env bash
shopt -s nullglob
##: Although files=(/old/directory/{1..3}/input.txt)
##: can be a replacement, no need for nullglob
files=(/old/directory/*/input.txt)
for file in "${files[#]}"; do
tmp=${file%[0-9]*}
new_file=${file#*$tmp}
echo cp -v "$file" new/path/"${new_file//\//}"
done
Another option is to split the fields using / as the delimiter.
#!/usr/bin/env bash
##: Do not expand a literal glob * if there are no files/directories
shopt -s nullglob
##: Save the whole path and files in an array.
files=(/old/directory/*/input.txt)
for file in "${files[#]}"; do
IFS='/' read -ra path <<< "$file" ##: split via / and save in an array
tmp=${path[#]:(-2)} ##: Remain only the last 2, e.g. 1 input.txt
new_file=${tmp// } ##: Remove the space, becomes 1input.txt
echo cp -v "$file" new/path/"$new_file"
done
Remove the echo if you think that the output is correct.
It is easy to understand which directory the file came from just replace the under scores with a /