Traverse, copy & transform file names

Question

My issue: I need to copy log files, all with identical names, but stored 1-deep in subdirectories, and change their names.

I'm copying from the a folder called Logfiles which contains a bunch of folders:

W3SVC114, W3SVC1507562355, W3SVC350179472, etc.

Each of these folders will contain a file named: u_ex[YYMMDD].log, where [YYMMDD] is a datestamp.

I need all of these logfiles in one folder. Since all names are identical I would like to add the directory name to the resulting file name, for example:

W3SVC1507562355/u_ex150407.log becomes W3SVC1507562355_u_ex150407.log

How do I make all of these happen? I'm a bit out of my depth here... :-/

Answer 1

Zsh comes with a function called zmv that makes it easy to move or copy files and apply pattern-based transformations on the name. Put this in your .zshrc (or run it on the command line):

autoload -U zmv
alias zcp='zmv -C'
alias zln='zmv -L'

Then your copy-with-renaming can be done with any of the following equivalent commands:

zcp 'Logfiles/(*)/(*.log)' 'destination_directory/${1}_$2'
zcp -w 'Logfiles/*/*.log' 'destination_directory/${1}_${2}.log'
zcp -W 'Logfiles/*/*.log' 'destination_directory/*_*.log'

If you don't want to install zsh, you can do the same thing with a loop in any shell:

cd Logfiles
for x in */*.log; do
  cp -- "$x" "destination_directory/${x%/*}_${x##*/}"
done

Answer 2

You can use following script to copy files from one location to another with modified name.

Note: In following script we have hard-coded two values.

1. /Logfiles/ :- change Logfiles name to valid folder name from which you have to copy files.

2. /tmp/ :- It is a directory under which you want to copy files. Change this directory name according to your requirements.

   #!/bin/bash
   
   find /Logfiles/ -maxdepth 1  -type f |  #find all files located under folder Logfiles with maxdepth one.
   
   while read FILEDIR                          #Read all files line by line.
   
   do
   
      DIR="${FILEDIR%/*}"                   # Get the folder name its inside
   #       echo $DIR
   
      FILE="${FILEDIR/*\/}"                 # Get the plain file name.
   #    echo $FILE
   
      NEWFILE="${DIR}"_"${FILE}"           # set new filename.
   #       echo $NEWFILE
   
      echo -e "coping $FILE from $DIR in /tmp with name $NEWFILE \n"
   
   
     cp $DIR/$FILE /tmp/$NEWFILE     #copy file from old location to new location with modified name.           
   
   
   done
   

Answer 3

With pax:

pax -'rws|/|_|' -- */u_ex150407.log target/dir

pax is the standard file-archiver specified by POSIX. Unfortunately - though it is required in the Linux Standard Base and has been for a few version increments of same - for whatever reason, many distribtions do not package it with the default installation as both standards require. I consider this a shame, because it is a very versatile tool.

Above I tell pax to go into -read and -write mode (which translates to a copy) and - when naming its output files - to -substitute the first / in any of its named copy file parameters with a _. You can basically do standard BRE sed syntax there (except that literal newlines in the left-hand-side are permissible here), but there's no need for anything elaborate in this case.

Anyway, that's all you need. If you don't have it installed for some reason, I recommend you get it and give pax a try.

Answer 4

Plain POSIX shell answer:

#!/bin/sh
for i in */*.log
do mv "$i" "${i//\//_}"
done

moves every file matching */*.log to the same name with / substituted by _.