UNIX for Dummies Questions & Answers

Are the sizes of the "date only" file always smaller than the other files? So, can we not use size of file for the criterion? Then we do not have to worry about all possible file contents.

Yeah the date only file is smaller but the other file sizes may vary.

Can it be done by sed or awk command.

get head -1 and then check if the record is like YYYYMMDDHH:MM:SS or get only first 8 characters from the first record and see if it is YYYYMMDD ?

error correction in red
Let's assume your "empty files" are never larger than 20 bytes. You pick the number.

Code:

filesize=$( cat filename | wc -c)
if [ $filesize -gt 20 ] ; then
   echo 'good file'
else
   echo 'empty file'
fi
# the above stuff can be made into a one liner:
 [ $( cat filename | wc -c) -gt 20 ] && echo 'good file' || echo 'empty file'

Be sure to leave the spaces in those commands.

Note: the use cat is intentional, we just want a number.

wc -l < filename would be equivalent...

Yes, it can be done by: "sed", "awk", "grep" and probably some dozen other utilities, but that is not the point. The point is, you - you - have to come up with a definition of what constitutes an "empty file" versus an "other file". We can help you put that definition into code, but we can't do the definition for you, because we do not know your input as well as you do.

Let's see: Your empty files have only one lines, whereas other files have several lines, yes? (I don't really know this, i just deduce from your samples and might as well be wrong. Bottom line is, you will have to decide for yourself, ultimately.) This could be tested with

Code:

wc -l /path/to/your.file

If this gives a value greater than 1 then it is an "other" file, otherwise it might be both.

Next: if it is an empty file it has a time stamp of a certain form at the beginning of the line, so we can match it:

Code:

grep -q "^20[0-9]\{7\}:[0-9]\{2\}:[0-9]\{2\}$" /path/to/your.file

This will return 0 if the expression (your sample timestamp as a regexp) is true and your file is an empty file, otherwise it will return "$? > 0" and your file is an other file.

Couple these checks and in all likelihood you will do it correctly provided that your samples were significant for your input data. If there are other things distinguishing one class of files from the other than work some checks for these into your script so that you raise the chance of correct answers.

Was that so difficult?

I hope this helps.

bakunin

The reason for wc -c is that some examples seemed to have more than one line...
if that is wrong the wc -l approach is better.

@eskay: wc -l counts lines, wc -c counts bytes, in a given file

Perhaps consider using "touch" to see if the first line has a valid date...

Code:

if touch -t $(head -c 13 abc.txt|cut -c1-10,12-13) abc.txt
then 
     echo empty
else
     echo not empty 
fi