UNIX for Dummies Questions & Answers

No, just the order they happen to be in the filesystem, which may have nothing to do with folder or file names or even their time of creation.

You'll need to forgive my lack of knowledge. Only been doing this for a few months.

My understanding of

Code:

touch -t $(date -d "2012-07-24 + $i day" "+%Y%m%d%H%M%S") filename

touch updates the date +1 for each loop. However, I can't figure out the logic to include a loop that updates the 48 daily files by 30 mins for each day as well.

Thanks
Cludgie

you didn't give him the time-of-day, only the date...

Code:

#!/bin/bash
for i in {1..14}; do
        for j in {1..48}; do
                ((h = (j-1) / 2, m = (j-1) % 2 * 30))
                ts=$(date +'%Y%m%d%H%M' -d "2012-07-24 + $i day $h:$m")
                echo touch -t "$ts" "Day$i/File$j/BlockedNodes.csv"
        done
done

remove the echo if it looks correct.

Code:

touch -d 201207250000 Day1/File1/BlockedNodes.csv
touch -d 201207250030 Day1/File2/BlockedNodes.csv
...
touch -d 201207252330 Day1/File48/BlockedNodes.csv
touch -d 201207260000 Day2/File1/BlockedNodes.csv

Yes, you can include the option(s) to find that allow you to sort alphabetically, numerically, chronologically, etc. Example, you may be looking for files between 45 days and 90 days and you want them sorted chronologically;

Code:

find . -mtime +45 -mtime -90 -type f -exec ls -lart {} \;

(assuming you are already in the cwd and you do not have too many files that forces you to use xargs)

The 'ls -lart' will sort chronologically (date) in ascending order.

From the list that is generated you can execute the 'touch' command, or the 'cp' command, or the 'mv' command, or the 'tr' command, etc. Using the example from above;

Code:

find . -mtime +45 -mtime -90 -type f -exec ls -rt {} \; | while read line
do
<your commands>
done

Find type files that are between 45 and 90 days old (that would be mtime - when the file was last modified) and sort them by date in ascending order (if you leave off the 'la' from 'ls -lart' you are left with 'ls -rt' - you will not see the perms and dates but it is still sorting). The resulting files are then piped through a while loop and each line is read individually and that allows you to act on each line depending on some condition, i.e., the date. The <your commands> can be your touch command, acting on each line that is produced from the output of the find command.

As I mentioned above, if you have too many files you will need to use the xargs with your find command.

The find command by itself will do just that, find what you are looking for. Options added to the find command allow you to find files/directories based on some conditional, the "-mtime" or "-type" for example. Using the "-depth" option with find will allow you to find . files (hidden) as well.

The find command can be very simple or it can be very complex; you could have it sort, convert upper to lower, reverse the file names, etc. It's what you include with the find command that allows you to be choosy.

Almost always, there is more than 1 way to do things in UNIX/Linux. You may find another way to do what you want that we did not list here. And that is usually good, you are thinking instead of just copying what we said.

You may choose to use a perl 1 liner to identify and select the files you want.

Code:

find . -depth | perl -lne 'print if /1st date/ && /2nd date/ &&

and so on ...

with find, and bash regular expression matching:

Code:

#!/bin/bash

startday='2012-07-24'
re='/Day([0-9]+)/File([0-9]+)/'

find . -type f -name '*.csv' -print0 |
while read -rd '' file; do
        [[ $file =~ $re ]] || continue
        day=${BASH_REMATCH[1]}
        time=${BASH_REMATCH[2]}
        ((
                hr  = (time - 1) / 2,
                min = (time - 1) % 2 * 30
        ))
        ts=$(date -d "$startday + $day day $hr:$min" +%Y%m%d%H%M)
        echo touch -t "$ts" "$file"
done

Thanks very much NeutronScott, that's brilliant. This seems to have the logic I'm looking for, only it's returning "touch: invalid date format `201207292000'." Yet, if I enter in this date for 1 file, it accepts it fine. Any thoughts?

Thanks again
Cludgie

---------- Post updated at 12:24 AM ---------- Previous update was at 12:10 AM ----------

Thank you also Raggmopp. I should refresh my page more often.

My only challenge to using the mtime, is that they were all files were modified in the same day, so would have lost their original creation and modified dates. Would that affect it?

Thanks
Cludgie

Creation date - take that with a grain of salt. A file will have the creation date (if you want to call it that) if the file has never been modified. Example, create a file with something in it. Now, leave that file alone for some time. Somebody else will do 'ls -lart' and see a date/time stamp. To that somebody else that is the modified time. You created the file and you know when, so you know that is also the create date.

Something I did notice a couple of weeks ago. I am running Ubuntu 12.04 on my desktop which has a newer version of the 'stat' command. In the output of the stat command is a field called 'Birth Date:' Hmmm, could this be?

In my mind, the date/time stamp on a file is the modified time. This is the date/time stamp you will see associated with a file when you do 'ls -la'. It is this date/time stamp that is sorted when you run the 'ls -lart' command.

Here is something that may be useful with your task.

Show files which are modified after the specified file. The following find command displays all the files that are created/modified after ordinary_file.

Note the 'ls -lrt' : the files (and dirs if there are any) are being sorted by date/time, Oldest to youngest.

Code:

# find -newer ordinary_file

Looks like you're using a combination of proposed solutions. Good job!

---------- Post updated at 06:20 PM ---------- Previous update was at 06:14 PM ----------

Creation date - take that with a grain of salt. A file will have the creation date (if you want to call it that) if the file has never been modified. Example, create a file with something in it. Now, leave that file alone for some time. Somebody else will do 'ls -lart' and see a date/time stamp. To that somebody else that is the modified time. You created the file and you know when, so you know that is also the create date.

Something I did notice a couple of weeks ago. I am running Ubuntu 12.04 on my desktop which has a newer version of the 'stat' command. In the output of the stat command is a field called 'Birth Date:' Hmmm, could this be?

In my mind, the date/time stamp on a file is the modified time. This is the date/time stamp you will see associated with a file when you do 'ls -la'. It is this date/time stamp that is sorted when you run the 'ls -lart' command.

Here is something that may be useful with your task.

Show files which are modified after the specified file. The following find command displays all the files that are created/modified after ordinary_file.

Note the 'ls -lrt' : the files (and dirs if there are any) are being sorted by date/time, Oldest to youngest.

Code:

# find -newer ordinary_file

Looks like you're using a combination of proposed solutions. Good job!