Shell Programming and Scripting

Bash/scripting newbie here - I feel this might be a trivial problem, but I'm not sure how to tackle it. I've got a folder of a year's worth of files, with some random number of files generated every day of the year (but at least one per day). I'm writing a script to automatically grab the file with the earliest daily timestamp and copy it to another folder. For example, if three files were generated on Jan 1st at 01:15, 01:30, and 01:50, I want to copy the one timestamped 01:15 into another folder and move on to Jan 2nd.

I've figured out the very basic bash commands to list files in a directory and copy them, but not how to read those filenames & their timestamps into a variable so I can perform operations on them, or how to go back from the variable to actually copy the files (in my other programming life, mostly using Java, file I/O is enough of a general mess that I'm making a foray into scripting ...). Any suggestions?

Thanks!
-Goose

When you do

Code:

ls -lrt  *

does the order of the files come out matching the order of the timestamps i.e.,
file_1:15
file_1:30

and so on. It should if the timestamps in the filename mean anything. Please show an example ls -lrt output for a few days worth of files.

Do you want to copy the files with earliest creation date or with earliest modification date?
Can you post one line of ls -l output?

Here's the ls -l output for 10 files over 6 days. I'm just interested in the timestamp in the filename, which doesn't correspond much if at all to the file modification date. So I guess I need to figure out some way of parsing out the date (ie, 2009-01-01) and the time (ie, 12-09-44-58) and comparing them - and I think I have examples of how to do that in other scripting languages (R & perl). But I'm still not sure how to go from the output of something like that back to telling the file system which actual file to copy to a different directory.

Thanks!

Code:

ls -l | awk '!a[$6]++' | while read line; do cp $(echo "$line" | awk '{print $8}') /target/dir; done

I ended up using

Code:

date=$(echo $n | cut -f 2-4 -d '_')

to extract the date from the filename, storing it as the previous date, then comparing the next file date to it to determine if they differed. It'll be a while before I can come up with nice one-liners like pseudocoder's, but I'm pretty happy to have my first running bash script! The archives on here helped a lot.

Code:

awk '{split($NF,a,"_");if(a[2]!=$6){print "cp",$NF,"/dir/"}}' file | sh