script to count the no. of a particular word

[devhariprasad]

I am trying to write a script to count the no. of occurences of a word (say "swinstall") in the files in a particular directory.

I tried doing this using "wc" and "grep" commands but was unsuccessful.

Please help me out.

[Steven E. Protter]

Shalom,

Post the code you are using please.

grep swinstall * | wc -l

That will get you close.

SEP

[Yogeeraj_1]

hi.

if you have one occurence per line,
you can also try:

awk '/swinstall/{n++}; END {print n+0}' *.txt

hope this helps too!
kind regards
yogeeraj

[devhariprasad]

grep swinstall * | wc -l

This will not serve the purpose. Suppose if there are 3 words by name swinstall in a line that will be counted as one. I want those 3 words in a single line to the counted as 3.

Below is the code that I used

for i in *
do

count=`grep -c swinstall $i`

echo $i "has " $count "swinstalls"

done

This code also has the same problem as the code you have given

[devhariprasad]

Hi Yogeeraj,

Suppose the following are the contents of the file
---------------------------------------------
swinstall
swinstall use
i am called swinstall
swinstall is called swinstall everywhere else
---------------------------------------------

You awk script will count the no. of occurences of swinstall as 4, whereas the answer I am expecting is 5. The last line has 2 occurences of swinstall. This is counted as just one.

[Wouter Jagers]

I would try this:

# sed -n 's/\(\\)/\n\1\n/gp' myfile | grep '^swinstall$' | wc -l

Cheers,
Wout

[Wouter Jagers]

Sorry, seeing that line I think I should clarify :-)

The sed command will just look for any occurence of 'swinstall' and put each occurence on a line of its own (adding linebreaks before and after). Then, it becomes possible to grep for lines which hold only this word and count them.

Cheers,
Wout

[Wouter Jagers]

Poured it into a little script to check a complete directory, as per your original question. Note that the directory should -only- contain text files ;-)

I guess it's a start, whatever your ultimate plan is.



#!/bin/sh
# Will count the number of occurences of MYSTRING in the files within MYDIR
#
MYDIR="/root/tmp"
MYSTRING="swinstall"

for i in `ls $MYDIR`
do
echo $i has `sed -n "s/\(\<${MYSTRING}\>\)/\n\1\n/gp" $i | grep "^${MYSTRING}$" | wc -l` occurences of $MYSTRING
done

exit 0

[James R. Ferguson]

Hi:

To count the number of occurancs of a pattern ("word") you can use:

# perl -lne 'BEGIN{$count=0};$count++ while (m/\bswinstall\b/ig);END{print $count}' file

This counts the total number of occurances of the string "swinstall" in 'file'.

You can total the count for any number of files by passing multilple file arguments.

Regards!

...JRF...

[john korterman]

Hi,

try the attached script for a start: $1="string to search for", $2="file to search".
The count is faked, but it may produce the correct result.

regards,
John K.