4
\$\begingroup\$

I inherited a script at work the other day. I know very little about the command line in general but I'm not entirely new to programming. I am using this as an opportunity to learn... Very little of the original code is left (so if it's all wrong, it's my fault now). I was asked to modify it retain sub-directories so I chose to use tar.

The directories being archived are huge; maybe not crazy huge but definitely big. It should be ran often enough where there aren't too many files to process. Hopefully. I'm not sure how concerned I should be about that yet. It can be assumed that more than once instance of this script will never be running at the same time.

function archive() { log_dir=$1 # /home/u123/initial_files/SessLogs age=$2 # age of the files to process archive=$3 # base of the archive directory exclude=$4 # exclude_list of directories/files base_dir=${log_dir%/*} # /home/u123/initial_files ftype=${log_dir##*/} # SessLogs date_dir=$(make_date_string $age) # make the new folder in the archive with the date arch_dir=$archive/$ftype/$date_dir [ ! -d $archive/$ftype ] && ( mkdir $archive/$ftype ) [ ! -d $arch_dir ] && ( mkdir $arch_dir ) outfile=$arch_dir/archivelog.txt printf "$(date)""\n\n" >>$outfile printf "%s\n" "processing $ftype files..." >>$outfile printf "%s\n" " - type = $ftype" >>$outfile printf "%s\n" " - age = $age days" >>$outfile printf "%s\n" " - log_dir = $log_dir" >>$outfile printf "%s\n" " - archive = $archive" >>$outfile printf "%s\n" " - exclude = $exclude" >>$outfile printf "\n">>$outfile # test for valid age if [ $age -lt 1 ]; then printf "\n%s\n" "age parameter is invalid" >>$outfile return 0 # test for existence of log file elif [ ! -d $log_dir ]; then printf "\n%s\n" "log directory cannot be found $log_dir" >>$outfile return 0 # test for existence of exclude list elif [ ! -f $exclude ]; then printf "\n%s\n" "exclude list cannot be found $exclude" >>$outfile return 0 fi # i hate to run the full find twice since these are very large directories... idk. files=$( find "$log_dir" -type f -mtime +"$age" | grep -v -f "$exclude" | wc -l ) # no files to move... if [[ "$files" -eq "0" ]] ; then #printf "find ""$log_dir"" -type f -mtime +"$age" | grep -v -f "$exclude" | wc -l" >>$outfile printf \n"%s\n" "there are no files to update for $log_dir" >>$outfile return 0 fi # save this for after tar is done # get directories that have old files in them and give me a unique list sorted in reverse cleanup=( $( find "$log_dir" -type f -mtime +"$age" | grep -v -f "$exclude" | sed 's/\(.*\)\/.*/\1/' | sort -ur ) ) # shove the files in a .tar into the archive directory for this date find "$log_dir" -type f -mtime +"$age" | grep -v -f "$exclude" | sort -r \ | tar vciPhf "$arch_dir"/archive.tar --remove-files --same-owner --atime-preserve --files-from - 1>&2>>$outfile if [ "$?" != "0" ]; then printf "\n%s\n" "process stopped for $ftype because of errors with .tar" >> $outfile return 0 fi # unzip the tar file in the archive directory sc=$( grep -o "/" <<< "$log_dir" | wc -l ) tar xvp -C "$arch_dir" -f "$arch_dir"/archive.tar --strip-components="$sc" 1>&2>>$outfile printf "\n\n" >>$outfile # if no error: rm "$arch_dir"/archive.tar if [ "$?" = "0" ]; then for dir in "${cleanup[@]}"; do ct=$( find $dir | wc -l ) if [ "$ct" -eq 1 ]; then printf "rmdir ""$dir""\n" >>$outfile rmdir "$dir" fi done printf "\n\n%s\n" "archive successful" >>$outfile rm "$arch_dir"/archive.tar return 1 fi return 0 } # this is entirely from the original code function make_date_string() { local -ix age=$1 calc_time=$(expr $age \* 86400) date_calc=`eval "perl -e 'print scalar localtime( time - $calc_time ) . \"\n\";'"` final_date=`date +%Y-%m-%d --date="$date_calc"` echo $ftype"_"$final_date # creates date string } # a settings file will be read in... it's just this way for testing right now archive '/home/u123/initial_files/SrcFiles' 7 '/home/u123/ArchiveFiles' '/home/u123/exclude_list' 
\$\endgroup\$
7
  • \$\begingroup\$Welcome to Code Review. Please indicate what exactly you did inherit (it is not subject to review per CR charter), and what is your contribution to be reviewed.\$\endgroup\$
    – vnp
    CommentedOct 25, 2014 at 23:29
  • \$\begingroup\$I pretty much stated that (I thought clearly) in the post. The only thing left over is the make_date_string function.\$\endgroup\$CommentedOct 25, 2014 at 23:31
  • \$\begingroup\$the original file was the date thing and a parameter form of (not verbatim) find ./stuff -type f -mtime +N -exec mv {} \; It needed to use an exclude list and keep sub-directory structure. And logging, which I'm doing now.\$\endgroup\$CommentedOct 25, 2014 at 23:37
  • 1
    \$\begingroup\$@vnp I disagree. OP is clearly responsible for maintaining all of the code.\$\endgroup\$CommentedOct 26, 2014 at 2:36
  • \$\begingroup\$I mean... I keep debating on whether I should post a portion of the original to prove everything is mine but that date part. But I really don't feel comfortable posting someone else's work for any reason. With comments and whitespace removed, including the date stuff, Egyptian brackets - the original function was 12 lines long and used global variables.\$\endgroup\$CommentedOct 26, 2014 at 5:30

1 Answer 1

2
\$\begingroup\$

You can simplify this:

arch_dir=$archive/$ftype/$date_dir [ ! -d $archive/$ftype ] && ( mkdir $archive/$ftype ) [ ! -d $arch_dir ] && ( mkdir $arch_dir ) 

to this:

[ -d $arch_dir ] || mkdir -p $arch_dir 

That is:

  • The -p flag of mkdir will make it create all intermediary directories as necessary
  • No need to put (...) around the mkdir
  • Using ... || instead of ! ... && is shorter, and perhaps even more natural too

printf "$(date)""\n\n" >>$outfile 

First of all, unless you have a good reason, don't use a $(...) subshell to capture the output of a command just to print it. Use the output directly:

date >>$outfile 

Secondly, if a simple echo without parameters is enough, I would use that instead of a printf. If you need any of the flags of echo, then it's better to use printf, as it's more portable. For printing a blank line, I'd use echo.


Redirecting many lines to $outfile like this feels a bit repetitive:

printf "%s\n" "processing $ftype files..." >>$outfile printf "%s\n" " - type = $ftype" >>$outfile printf "%s\n" " - age = $age days" >>$outfile printf "%s\n" " - log_dir = $log_dir" >>$outfile printf "%s\n" " - archive = $archive" >>$outfile printf "%s\n" " - exclude = $exclude" >>$outfile printf "\n">>$outfile 

A better way would be:

{ printf "%s\n" "processing $ftype files..." printf "%s\n" " - type = $ftype" printf "%s\n" " - age = $age days" printf "%s\n" " - log_dir = $log_dir" printf "%s\n" " - archive = $archive" printf "%s\n" " - exclude = $exclude" printf "\n" } >> $outfile 

As mentioned earlier, an even better way would be rewriting these with simple echo, for example:

echo "processing $ftype files..." 

Finally, since you have many lines appending to $outfile, it might be best to create a helper function for it.


if [[ "$files" -eq "0" ]] ; then 

When using [[ ... ]], you don't need to quote variables on the left side. You also don't need to quote barewords like "0". So this is enough:

if [[ $files == 0 ]] ; then 

This probably doesn't do what you expect:

tar xvp -C "$arch_dir" -f "$arch_dir"/archive.tar --strip-components="$sc" 1>&2>>$outfile printf "\n\n" >>$outfile # if no error: rm "$arch_dir"/archive.tar if [ "$?" = "0" ]; then 

$? is the exit code of the last command. In this case the printf, and I'm guessing that you're in fact really interested in the tar instead. If you want to act on the exit code of the tar, then you can either use the || operator right after it, or save the exit code, for example:

tar xvp -C "$arch_dir" -f "$arch_dir"/archive.tar --strip-components="$sc" 1>&2>>$outfile tar_exit=$? printf "\n\n" >>$outfile # if no error: rm "$arch_dir"/archive.tar if [ $tar_exit = 0 ]; then 

And no need to quote $? as it's always a number, never blank, never contains spaces.


Copy-paste your code to http://www.shellcheck.net/# and it will point out a couple of more things that you should fix.

\$\endgroup\$
4
  • \$\begingroup\$thanks for that link...! Ok, so... a lot of things said don't use echo because it is inconsistent (which I was surprised by). Also, when I write things, my brain can't handle the fact a naked command (like without backticks or - don't kill me - the jquery looking notation) is syntactically correct and will be executed.\$\endgroup\$CommentedOct 26, 2014 at 13:34
  • \$\begingroup\$i knew $? was something for the last command... but for some unknown reason I didn't consider printf a real command... in my head. I would have never caught that.\$\endgroup\$CommentedOct 26, 2014 at 13:49
  • \$\begingroup\$About echo versus printf, as I explained above, if all you need is printf 'something\n' then use the simpler echo 'something', if you need anything more sophisticated then use printf\$\endgroup\$
    – janos
    CommentedOct 26, 2014 at 14:13
  • 1
    \$\begingroup\$yeah... I've worked with php before. Seems about the same. I came across several things like this: unix.stackexchange.com/q/65803/82330 while searching commands and stuff. It doesn't really matter; I would have to assume none of the issues listed at that link would affect the (simple) things I am writing right now. All the servers I use have the same rehl version anyway.\$\endgroup\$CommentedOct 26, 2014 at 14:53

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.