[SOLVED] How to safely iterate glob with time modified order?

Issue

Currently, I am doing this in a bash script.

for i in $(ls -tr *.tar); do
# something
done

When I run shellcheck, I got a (correct) warning

for i in $(ls -tr *.tar); do
         ^-------------^ SC2045 (warning): Iterating over ls output is fragile. Use globs.

OK, but what to do here? I cannot use globs, or at least I don't know how.

I will add I need to keep the order as ls -tr does, that is, ordered by time modified (and reversed, newest at the bottom).

I have seen some ways to do this, some of them on Stack Overflow (for example In a bash script how to loop through files sorted by date ), doing find/sort/awk combo, but that all seem even more fragile than just using ls.

Is there a safe way to do this? (I am using Linux with "standard" coreutils etc)

Solution

Largely, this is covered by BashFAQ #3; the below extracts a relevant subset.

---

On a GNU system, find's -printf can print timestamp metadata.

Using NUL-delimited output and sort -z ensures correct operation with all possible filenames, including ones with literal newlines included (the NUL is the only character that cannot possibly exist in a path across all UNIXy systems).

while IFS= read -r -d : time && IFS= read -r -d '' file; do
  echo "Read ${file} with timestamp ${time}"
done < <(find "$dir" -mindepth 1 -maxdepth 1 -type f -printf '%T@:%p\0' | sort -zn)

The use of -printf ensures a well-defined output format, and is thus (assuming, again, availability of GNU tools) not at all fragile.

If you want to reverse the order and list newest-to-oldest, add the -r argument to sort.

Answered By - Charles Duffy

Answer Checked By - David Goodson (WPSolving Volunteer)