Problem on reading file with "***" in the beginning of each line

abc00001 · 10-16-2012, 07:13 AM

I am using the following code to read line by line from a file.

while read line
do
echo $line
done <　logfile.out

The code is very common of course it works well in most situation. However, since the log file that I am going to read contains "***" at the beginning of each line. The while loop above will not work as expected. For example the log file contains,

*** Backup completed
*** Backup failed

When trying to read it using the code listed above, the output will become,

gconfd-root kde-root ksocket-root orbit-root ssh-gKQZBa3854 temp.out test.sh test.txt VMwareDnD vmware-root Backup completed

gconfd-root kde-root ksocket-root orbit-root ssh-gKQZBa3854 temp.out test.sh test.txt VMwareDnD vmware-root Backup failed

Is there any way to read the file line by line properly? As the log is generated by a vendor software, the format cannot be changed.

Thanks a lot for help.

druuna · 10-16-2012, 07:19 AM

Try this:

Code:

while read line
do
echo "$line"
done <　logfile.out

Mind the double quotes....

EDIT: Using set -f turns off globbing and in this case your original solution (no double quotes) works.

NevemTeve · 10-16-2012, 09:04 AM

try this:

Code:

cat logfile.out

grail · 10-16-2012, 10:19 AM

Any asterisk not quoted will be expanded by the shell, hence all files and folders present in the directory you are running it from will be displayed along with your string.

druuna has provided multiple solutions to help.

abc00001 · 10-16-2012, 09:28 PM

Thanks for everyone's reply. I am deeply appreciated.

By the way, the double quote does not help on solving the issue. Thus, I do it in a indirect way. Since I just only concern about the content of the line but not that three "***". I just remove them before read the line content. Below is my stupid solution.

Example of the log file content:
*** (warn) backup failed
*** (info) backup successful
*** (warn) backup did not start

The code to read this file.
#!/bin/bash
LOGFILE=/tmp/log.out
TEMPFILE=/tmp/temp.out

cat $LOGFILE | sed -r 's/^.{3}//' > $TEMPFILE

while read line
do
echo $line
Perform something on the content...
fi
done < "$TEMPFILE"

rm $TEMPFILE
exit 0

NevemTeve · 10-16-2012, 11:35 PM

1. Please use [code] and [/code] tags
2. What do you mean by 'double quote does not help'? What did you try, what happened, what should have happened?

druuna · 10-17-2012, 01:55 AM

Quote:

Originally Posted by abc00001

By the way, the double quote does not help on solving the issue. Thus, I do it in a indirect way. Since I just only concern about the content of the line but not that three "***". I just remove them before read the line content.

Although your solution seems to do what you want, it can be simplified.

I'm also curious why the double quotes do not work, but maybe this will help:

Code:

#!/bin/bash

LOGFILE="infile"

while read FIRST SECOND REST
do
#  echo "$FIRST"
  echo "$SECOND"
  echo "$REST"
done < "$LOGFILE"

exit 0

When using the example data given in post #5, this will be the output:

Code:

$ ./tets.sh
(warn)
backup failed
(info)
backup successful
(warn)
backup did not start

The while read construct can use more then one variable. In the above code snippet it uses three (FIRST SECOND REST) variables to read a line. The first part [the ***] will be stored in FIRST, the backup status [(xyz)] will be stored in SECOND and the rest of the line in REST.

The # echo "$FIRST" part is there to show you that something can be done with it (even though you mention not needing it).

grail · 10-17-2012, 10:01 AM

The other thing to consider with your second solution is, what if a line does not start with "***"? You may remove valuable information.

Also, +1 to being curious how the double quotes did not work as I have been unable to reproduce what you have shown when using quotes?