If-statement in Shell Script

SeraphimNL · 01-10-2010, 04:40 AM

Hello LQ'er,

For a school assignment I have to write some sort of e-mail classification program. It's kinda fake since there are no mailservers involved, but only a collection of old e-mails. The commands I mostly needed to work with are: grep, sed, paste and other basic commands like that. Now I have to write a shell script that does the following: extract a e-mail subject, make a wordlist of that subject and check it it contains words that are classified as possible spam. This is what i've written so far:

Code:

      #!/bin/bash

      # spam words used for matching
      spamwords=output/only-spam.subject.all.lemm_stop.txt

      # files to examine
      FILES="data/lemm_stop/part*/*"

      #name of output file
      OUT="output/subspam-list.txt"
      rm -f $OUT  # remove output file (to prevent appending to an existing file) 
                  # -f: script will not produce an error message if file to remove does not exist

      for i in $FILES do
            egrep -wh '^Subject' | sed '/Subject: /d' | sed -r 's/[[:space:]]/\n/g' | sed -r 's/^[^[:alpha:]]*|[^[:alpha:]]*$//g' | egrep -wi '^[[:alpha:]]{3,}' | sed '/^$/d' > $i.temp
            grep -wf only-spam.subject.all.lemm_stop.txt $i.temp
            if filename == spam, echo $filename.subspam >> subspam.list.txt
      else echo $filename.subham >> subspam.list.txt
      rm $filename.temp

I am totally stuck in the part where i have to use an if-statement.

Code:

            grep -wf only-spam.subject.all.lemm_stop.txt $i.temp
            if filename == spam, echo $filename.subspam >> subspam.list.txt
      else echo $filename.subham >> subspam.list.txt
      rm $filename.temp

First I have to compare the created wordlist of the subject of the mail to the wordlist with classified spam words: only-spam.subject.all.lemm_stop.txt. How can I use the if-statement to make this work? How can I check for a match? I hope you can give me some advice!

Thanks in advance,

Seraphim

kofucii · 01-10-2010, 06:35 AM

Code:

#!/bin/bash

# spam words used for matching
      SPAMWORDS=`cat output/only-spam.subject.all.lemm_stop.txt`

      # files to examine
      FILES=`ls data/lemm_stop/part*/*`

      #name of output file
      OUT=output/subspam-list.txt
      rm -f $OUT  # remove output file (to prevent appending to an existing file) 
                  # -f: script will not produce an error message if file to remove does not exist

      for i in $FILES
do
            egrep -wh '^Subject' | sed '/Subject: /d' | sed -r 's/[[:space:]]/\n/g' | sed -r 's/^[^[:alpha:]]*|[^[:alpha:]]*$//g' | egrep -wi '^[[:alpha:]]{3,}' | sed '/^$/d' > $i.temp

for s in $SPAMWORDS
do            
  SPAM=`grep -wf $s $i.temp`
  if [ ! -z $SPAM ];then
    echo $i >> $OUT
    rm $i.temp
  fi
done

done

Try this out. Please consider, that there are two nested "for" loops, so the Big-o is O(N^2)

ozanbaba · 01-10-2010, 06:49 AM

for loop is wrong (actually it will run, but for will pass FILE to commands and difened variables used with $)

for $i in `ls $FILES` do # this will pass ls output one by one to the command

and using ` you can assign output of programs to variables easily.

example:

ls -d /home | grep -i ozan will output /home/ozan/ on my system
if i write this OZANHOME=`ls -d /home | grep -i ozan` OZAHOME will be /home/ozan/ (that slash would create some problems).

if statements are like this:

if [statement]; then
do something
else
do something else otherwise
fi

for statement part where you do tests: look at this, IBM wrote such nice document

chrism01 · 01-11-2010, 12:24 AM

There's good tutorials here
http://rute.2038bug.com/index.html.gz
http://tldp.org/LDP/Bash-Beginners-G...tml/index.html

GrapefruiTgirl · 01-11-2010, 12:30 AM

Plus, one is not limited to simply IF-THEN-ELSE:

Code:

if <something>; then
  <do stuff A>
elif <something else>; then
 <do different stuff B>
elif <try a third test>; then
 <do plan C>
else
 # none of the above tests were acted upon
 <do pretty much nothing, or plan D>
fi