formatlogs

#!/bin/bash

sed 's/<span class="user">/\n/g' ${1} > temp1


sed -n "/div class=\"thread\">${2}/=" temp1 > linenumbers

x=0

for line in `cat linenumbers`; do

tail -n +${line} temp1  > block${x}

l=`grep -n "<div class=\"thread\">" block${x} | cut -f1 -d: | head -2 | tail -n +2`

head -${l} block${x} > blocks${x}

x=$(($x+1))

done

tac blocks* > temp0

egrep "^(Patrick|Fatema)" temp0 > temp1

sed 's/<\/span><span class="meta">/\n/g' temp1 > temp2

sed 's/<\/span><\/div><\/div><p>/\n\n/g' temp2 > temp3

sed 's/<\/p><div class="message"><div class="message_header">/\n/g' temp3 > temp4

sed 's/&#039;/`/g' temp4 > temp5

sed 's/&quot;/"/g' temp5 > temp6

sed 's/<\/p><\/div><div class="thread">.*<div class="message"><div class="message_header">/\n/g' temp6 > temp7

sed 's/&lt;/</g' temp7 > fulltext

rm linenumbers
rm temp*
rm block*