I have a script which function like this for one file.

./script 0001g.log > output

for two or more files, like this

./script 0001g.log 0002g.log 0003g.log > output

The script take one special number from each input file and put it in one output file.

My question is I have 1000 input files, how can I do a loop to execute my script.

  • you're going to have to express many more of the details of what you're trying to accomplish in hopes of getting any assistance. to start with an obvious one, do the "log" files already exist? and if so, why doesn't this solve all your problems?: ./script *.log > output – Theophrastus Jun 21 '16 at 0:14
  • @Theophrastus the only reason I see not to use your solution is that *.log may be expanded over the ARG_MAX limit causing an Argument list too long error. That being said, your approach is the best solution unless OP really faces this issue. – Jedi Jun 21 '16 at 0:44
  • 2
    if it's just ARG_MAX holding you back, then 'find' should solve your problems (or find combined with xargs). something akin to: find . -type f -name '*.log' -exec ./script {} \; – Theophrastus Jun 21 '16 at 0:50
  • 1
    In regards to: The script take one special number from each input file and put it in one output file, I have to wonder what your script actually does. From the current description it sounds like you could be using a grep one-liner instead of a script. – Wildcard Jun 21 '16 at 3:03

You have a few possible solutions:


$ ./script *g.log >output

... and hope that *g.log doesn't expand to something that makes the command line too long. This is not very robust.

If your script doesn't depend on the number of files given to it, i.e., if output can just be appended to output for each input file, then this is another solution:

$ find ./ -type f -name "*g.log" | xargs ./script >output

A third solution would be to move the loop into the script itself:

for f in *g.log; do
  # old code using "$f" as file name

This does not have the problem with command line length restriction since it's in a script.

The invocation of the script would now be

$ ./script >output
| improve this answer | |


./script 0001g.log 0002g.log 0003g.log > output


./script 0001g.log > output
./script 0002g.log >> output
./script 0003g.log >> output

then you can use loop or

`seq -f '%04gg.log' 10` | script > output
| improve this answer | |
  • Have you tried this?  It looks like you have confused input with arguments. – Scott Jun 21 '16 at 4:33
  • This will try to use 0001g.log as a command name. Also, it will exceed maximum command length for large values. (was going to write "for large values of 10") – Kusalananda Jun 21 '16 at 8:26

If you like you can put the files into a directory


Then in your bash script you can try the following


# store path to files

# loop through files
for FILE in `ls $SOURCE*g.log`; do
    # do what you want to specific file
    echo $FILE
| improve this answer | |
  • The OP has both script and files in the same directory (and possibly other thing too, we don't know). Your for-loop will process these too. – Kusalananda Jun 21 '16 at 8:24
  • 1
    I've updated my answer to show how you can be a bit more precise with what files you are listing – TheLovelySausage Jun 21 '16 at 8:27
  • For the general case, if there was any file containing spaces or file globbing patterns in their file names, your loop would misbehave. Better to use for FILE in $SOURCE/*g.log. – Kusalananda Jun 21 '16 at 8:30
  • that's good advice – TheLovelySausage Jun 21 '16 at 8:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.