Help with Bash Script Please!!

[youngmoney07]

I need help to write a simple bash script. In a directory, there are 6 log files. Each file has a name and date, for example file1.Sep0918, file2.Sep0918, etc. For each file, I have to search for a key word, lets say the word "Error". I need to count the occurances of that word in each file and print them out in to file. The directory path should be an argument past to the script. The file output file should cut out the date from the filename and should resemble something like:

54 errors in file1
12 errors in file2
.
.
.

[Cabhan]

This sounds suspiciously like a homework assignment, which are not allowed per The Forum Rules.

I suggest that you read a few guides on Bash scripting:

Bash Guide for Beginners
Advanced Bash-Scripting Guide

[youngmoney07]

This is actually not a homework assignment. I am a beginner program and am trying to learn scripting. I would appreciate any help. I did not ask for exact code, even a description would be helpful.

Thanks

[ghostdog74]

to search for one file

egrep -o Error file | wc -l

i leave it to you to search the rest

[youngmoney07]

If i pass the script an argument thats the directory path, how do I get it to read each file on the path? Thats my biggest issue. There are several files in the directory, so to read each one, I thought using a for loop would work. So far, i've thought of...

for file in $1
do
<parse each file>
done

but all that does is make $file store the directory path.

[i92guboj]

If i pass the script an argument thats the directory path, how do I get it to read each file on the path? Thats my biggest issue. There are several files in the directory, so to read each one, I thought using a for loop would work. So far, i've thought of...

for file in $1
do
<parse each file>
done

but all that does is make $file store the directory path.

You'd rather do

for file in "$1"/*

Provided that $1 will hold the path.

If you need to process subdirs as well, you can do a recursive script, but I advise to use the command find instead. Read on the -exec option for find.

For more versatility, you could use this as well:

find /path/to/ -type f | while read file
do
  #whatever
done

[scm]

or

for file in $(find /path/to -type f)
do
    # whatever
done

[ghostdog74]

or

for file in $(find /path/to -type f)
do
    # whatever
done

files with spaces may be broken with this method. its better to use while read loop

[Cabhan]

That method can work with a tiny modification:

IFS=$'\012'

for file in $(find /path/to -type f)
do
    # whatever
done

This will tell the loop to only split on newlines, preventing spaces from messing with things. IFS stands for "Internal Field Separator".

[ghostdog74]

That method can work with a tiny modification:

IFS=$'\012'

for file in $(find /path/to -type f)
do
    # whatever
done

This will tell the loop to only split on newlines, preventing spaces from messing with things. IFS stands for "Internal Field Separator".

true. However, a while read loop enable one to work with fields as you work each line. IMO, it is more "flexible".