command line - How do I find the most recent timestamp in directory names in a bash script? - Ask Ubuntu

i have timestamped directories named prefix may vary, , timestamp in handy form yyyymmdd, followed varying suffixes denote chronological order (not in sane fashion). unfortunately, title part of name can include _ character, used field delimiter.

for example:

/collect/mydir_20161102_0a /collect/mydir_20161102_0b /collect/mydir_20161102_0c /collect/mydir_20161102_1 /collect/mydir_20161102_2 /collect/other_dir_20161103_0a /collect/other_dir_20161103_0b /collect/mydir_20161104_0a /collect/mydir_20161104_0b /collect/mydir_20161104_0c /collect/mydir_20161104_1 /collect/mydir_20161104_2 

the order of names displayed here chronological order of creation, including suffixes, 0a comes before 0b, , 0s come before 1. there should not incidence of different title prefix same timestamp.

the directories may have been changed or modified since creation, use of ctime , mtime out.

i need output string containing title , timestamp, or without preceding path mydir_20161104 or /collect/mydir_20161104, must come recent directory. search should not recurse through directory levels.

i avoid parsing ls too!

might not pretty, , i'm not handling filenames newlines in them:

find collect/ -mindepth 1 |    awk -f_ '{print $(nf-1),$nf,$0}' |    sort -v |    sed -r 's/^([^ ]* ){2}//' 

so:

1. listing files find
2. adding last 2 _-delimited fields in filename start of using awk
3. sorting (-v - version sort - can handle fields 0a , 1)
4. removing added part sed

it made safe filenames containing valid character, i'd have replace awk sed that.

my output:

$ find collect/ -mindepth 1 | awk -f_ '{print $(nf-1),$nf,$0}' | sort -v | sed -r 's/^([^ ]* ){2}//'     collect/mydir_20161102_0a collect/mydir_20161102_0b collect/mydir_20161102_0c collect/mydir_20161102_1 collect/mydir_20161102_2 collect/other_dir_20161103_0a collect/other_dir_20161103_0b collect/mydir_20161104_0a collect/mydir_20161104_0b collect/mydir_20161104_0c collect/mydir_20161104_1 collect/mydir_20161104_2 

of course, parsing ls. ;)

if need title , timestamp without suffix, reverse sort (sort -vr) , modify last sed to:

sed -r 's:.*/::;s/_[^_]*$//;q' 

so:

$ find collect/ -mindepth 1 | awk -f_ '{print $(nf-1),$nf,$0}' | sort -rv | sed -r 's:.*/::;s/_[^_]*$//;q' mydir_20161104 

and version can handle filenames newlines:

find collect/ -mindepth 1 -print0 | sed -rz 's/(.*)(_[^_]*)(_[^_]*)$/\2\3 &/' | sort -zrv | sed -zr 's:.*/::;s/_[^_]*$//;q' 

this uses \0-delimited lines throughout (-print0 in find, -z in sed , sort). awk replaced equivalent sed command.