Como classificar e filtrar imagens por hora de modificação e visualizá-las no nsxiv no FreeBSD

Você terá que passar por xargs corretamente para fazer a classificação; então, algo como

find . -maxdepth 1 -type f \
    \( -iregex '.+\.jpe?g$' -o -iregex '.+\.png$' \) \
    -newermt 2023-10-01 \
    -print0 \
    \|
      xargs -0 \
        stat -n -f '%Y %N\0' \
    \|
      sort -z \
    \|
      cut -z -d ' ' -f2-  \
    \|
      nsxiv -0 -

We use the "decorate – sort – undecorate" trick that you would normally find in languages with tuples. Here, we need to separate our sort key (the time in seconds since epoch, %Y in stat) from our value (the file name) with a space. And then cut that back out. A bit awkward!

If it is important for you to be able to make it work on a vanilla FreeBSD system there is another answer with find/xargs/cut/stat. If you are able to install additional utilities there is a cleaner answer using zsh as well.

But if you are used to GNUism and Linux then it is very easy to live a comfortable life on FreeBSD if you install the GNU findutils.

pkg install findutils

In the same vein people have got bitten by BSD ls, grep, make, sed and awk. But their GNU counterparts are easily available as well as gnuls, gnugrep, gmake, gsed and gawk. Are you going all in then get coreutils

They can all coexist with the BSD versions on your system. You will then typically prefix the command with a "g" for the GNU version: gnuls, gfind, gxargs, gmake, gsed and gawk.

And if you hate to prefix your beloved gnu variants then see "How to make GNU grep the default in FreeBSD?"

If you get used to the BSD variants instead then they are available on Linux as well. See bsdutils

You could do:

zsh +o caseglob -c '
  autoload after
  print -rN -- *.(jp(e|)g|png)(.e[after 2023-10-01]om)' | nsxiv -0 -

For portability + reliability, perl is likely your best bet:

perl -MTime::HiRes=lstat -MPOSIX -MFile::Glob=:nocase -e '
  $start = mktime(0,0,0, 1,10-1,2023-1900); # 2023-10-01
  for (<*.{jpg,jpeg,png}>) {
    push @files, [$_, $s[9]] if @s = lstat and -f _ and $s[9] >= $start;
  }
  print "$_->[0]\0" for sort {$b->[1] <=> $a->[1]} @files
  ' | nsxiv -0 -

Note that your -exec ls -t --zero -- {} + won't work properly at sorting the files by modification time if the list of files is big enough that find needs to invoke ls several times.

I have found a solution that works in FreeBSD's /bin/sh. I have tried it in FreeBSD 13.2. It seems to properly handle filenames that contain backslashes and whitespace (spaces, tabs, and newlines), including leading and trailing whitespace.

find . -maxdepth 1 -type f \
        \( -iregex '.\{1,\}\.jpe\{0,1\}g$' -o -iregex '.\{1,\}\.png$' \) \
        -newermt 2023-10-01 \
        -exec ls -l -B -D%s {} + \
    | cut -w -f6- | sort -r | cut -f2- \
    | while IFS="$(printf '\n')" read -r line; do
        printf '%b\0' "$line"
      done \
    | nsxiv -0 -

How it works:

• ls -l -B -D%s
  • -B displays non-printable characters in octal notation. This is important for handling filenames that contain newlines and tabs.
  • -D%s formats file modification time in seconds since the Epoch. This will be useful for sorting by modification time.
• cut -w -f6-: remove all columns from the output of ls -l ... except the columns that contain file modification times and filenames. -w is necessary to make consecutive spaces count as one delimiter.
• sort -r: sort files by modification time from the most recently modified to the oldest.
• cut -f2-: remove the column of file modification times, leaving only the (newline-separated) filenames.
• The last part: read the newline-separated filenames one by one and use printf's %b format string to convert the octal notation in the filenames back to the actual characters (i.e. including newline and tabs). Separate the filenames using the NULL character (\0), pipe the NULL-separated filenames into nsxiv -0 -.

Additional notes:

• cut -w -f6- | sort -r | cut -f2- cannot be replaced by ls's -t option. As user "Stéphane Chazelas" wrote in another answer on this webpage:

  [find ... -exec ls -t ... -- {} +] won't work properly at sorting the files by modification time if the list of files is big enough that find needs to invoke ls several times.

• find ... \( -iregex '.\{1,\}\.jpe\{0,1\}g$' -o -iregex '.\{1,\}\.png$' \) could be simplified by using extended regular expressions. e.g. find -E ... -iregex '.+\.(jpe?g|png)$'.
• In the unusual case where there are space character(s) in the owner or group names, it might be necessary to add the -n option to ls, so that ls outputs user ids and group ids instead of names.