cleanwvHtml.sh

 

i=$1


###
### ELIMINATE AND REPLACE WITH NOTHING
###



sed "s/ width=\"......\"//" $i > tmp.out
mv tmp.out $i
sed "s/ width=\".......\"//" $i > tmp.out
mv tmp.out $i

sed "s/ color: .....;//" $i > tmp.out
mv tmp.out $i

sed "s/ color: ....;//" $i > tmp.out
mv tmp.out $i

sed "s/ background-color: White;//" $i > tmp.out
mv tmp.out $i

sed "s/ rowspan=\"1\"//" $i > tmp.out
mv tmp.out $i
sed "s/ colspan=\"1\"//" $i > tmp.out
mv tmp.out $i

sed "s/ text=\".......\"//" $i > tmp.out
mv tmp.out $i

sed "s/ link=\".......\"//" $i > tmp.out
mv tmp.out $i

sed "s/ vlink=\".......\"//g" $i > tmp.out
mv tmp.out $i

sed "s/ cols="."//" $i > tmp.out
mv tmp.out $i
sed "s/ cols=\"..\"//" $i > tmp.out
mv tmp.out $i
sed "s/ cols=\"..\"//" $i > tmp.out
mv tmp.out $i

sed "s/ rows="."//g" $i > tmp.out
mv tmp.out $i
sed "s/ rows=\"..\"//" $i > tmp.out
mv tmp.out $i
sed "s/ rows=\"...\"//" $i > tmp.out
mv tmp.out $i

sed "s/ bgcolor=\".......\"//" $i > tmp.out
mv tmp.out $i
sed "s/ bgcolor=\".....\"//" $i > tmp.out
mv tmp.out $i

sed "s/ line-height: 4.166667mm;//" $i > tmp.out
mv tmp.out $i

sed "s/ padding: 0.00mm 0.00mm 0.00mm 0.00mm;//g" $i > tmp.out
mv tmp.out $i

sed "s/<li><p>/<li>/" $i > tmp.out
mv tmp.out $i

sed "s/ name=\"Normal\"//" $i > tmp.out
mv tmp.out $i

sed "s/ align=\"left\"//" $i > tmp.out
mv tmp.out $i

sed "s/text-indent: 0.00mm; //" $i > tmp.out
mv tmp.out $i

sed "s/text-align: left; //" $i > tmp.out
mv tmp.out $i


sed "s/ style=\"\"//" $i > tmp.out
mv tmp.out $i

sed "s/ style=\" \"//" $i > tmp.out
mv tmp.out $i

sed "s/ style=\"  \"//" $i > tmp.out
mv tmp.out $i







####
#### ELIMINATE EMPTY TAGS
####

sed "s/<b><\/b>//" $i > tmp.out
mv tmp.out $i


sed "s/<font color=\"Blue\"><u><\/u><\/font>//" $i > tmp.out
mv tmp.out $i


sed "s/<font color=\"....\"><\/font>//" $i > tmp.out
mv tmp.out $i

sed "s/<font color=\".....\"><\/font>//" $i > tmp.out
mv tmp.out $i


sed "s/<font color=\"....\"><b><u><\/u><\/b><\/font>//" $i > tmp.out
mv tmp.out $i

sed "s/<s>//" $i > tmp.out
mv tmp.out $i

sed "s/<\/s>//" $i > tmp.out
mv tmp.out $i

sed "s/<u>//" $i > tmp.out
mv tmp.out $i

sed "s/<\/u>//" $i > tmp.out
mv tmp.out $i





### DELETE CONSECUTIVE BLANK LINES EXCEPT THE FIRST ONE
### IN EACH SEQUENCE. DEL ALL BLANK LINES AT THE EOF.

sed "/^$/N;/\n$/D"  $i > tmp.out
mv tmp.out $i

sed "/^$/N;/\n$/D"  $i > tmp.out
mv tmp.out $i

sed "/^$/N;/\n$/D"  $i > tmp.out
mv tmp.out $i







###
### ADD STYLESHEET INFORMATION
###

sed "s/<body>$/<body><blockquote><blockquote><font class="grey2">/" $i > tmp.out
mv tmp.out $i

sed "s/^<\/head>/<META content=text\/css http-equiv=Content-Style-Type><LINK href=\"http:\/\/www.setgetweb.com\/lstyle.css\" rel=\"stylesheet\" type=\"text\/css\"><\/head>/" $i > tmp.out
mv tmp.out $i





###
### JOIN LINES WITH DONGLES AT BEGINNING
###

sed -e :a -e "$!N;s/\n<\/td>/<\/td>/;ta" -e "P;D" $i > $i.tmp
mv $i.tmp $i

sed -e :a -e "$!N;s/\n<\/td>/<\/td>/;ta" -e "P;D" $i > $i.tmp
mv $i.tmp $i

sed -e :a -e "$!N;s/\n<\/td>/<\/td>/;ta" -e "P;D" $i > $i.tmp
mv $i.tmp $i


sed -e :a -e "$!N;s/\n<\/p>/<\/p>/;ta" -e "P;D" $i > $i.tmp
   mv $i.tmp $i
sed -e :a -e "$!N;s/\n<\/p>/<\/p>/;ta" -e "P;D" $i > $i.tmp
   mv $i.tmp $i
sed -e :a -e "$!N;s/\n<\/p>/<\/p>/;ta" -e "P;D" $i > $i.tmp
   mv $i.tmp $i
sed -e :a -e "$!N;s/\n<\/p>/<\/p>/;ta" -e "P;D" $i > $i.tmp
   mv $i.tmp $i


sed -e :a -e "$!N;s/\n<\/li>/<\/li>/;ta" -e "P;D" $i > $i.tmp
   mv $i.tmp $i
sed -e :a -e "$!N;s/\n<\/li>/<\/li>/;ta" -e "P;D" $i > $i.tmp
   mv $i.tmp $i
sed -e :a -e "$!N;s/\n<\/li>/<\/li>/;ta" -e "P;D" $i > $i.tmp
   mv $i.tmp $i
sed -e :a -e "$!N;s/\n<\/li>/<\/li>/;ta" -e "P;D" $i > $i.tmp
   mv $i.tmp $i





###
### REMOVE ENTIRE LINES
###
grep -v "^<p>   </p>$" $i > tmp.out
mv tmp.out $i





###
### Search and replace
###

sed "s/<\/td>/<\/br><\/td>/" $i > tmp.out
mv tmp.out $i



###
### Enclose material in comments
###
sed "s/\[/<\!--/g" $i > tmp.out
mv tmp.out $i
sed "s/\]/-->/g" $i > tmp.out
mv tmp.out $i