init

#!/usr/bin/env bash
mkdir data
cd data && wget --recursive -np -l 800 'https://oa.cc.iitk.ac.in/Oa/Jsp/OAServices/IITk_SrchStudRoll.jsp?selstudrol=1&selstuddep=&selstudnam='

#copy files to data.

for i in data/oa.cc.iitk.ac.in/Oa/Jsp/OAServices/*numtxt*
do
    cp $i data/
done
#rm -rf data/oa.cc.iitk.ac.in

cd data
ls | while read file; do
     new=$( echo $file | cut -d "=" -f 3 | cut -d "&" -f 1 )
     new="$new"
     mv "$file" "$new"
done

for file in *
do
    rollno=$(cat $file | grep -iPA1 "<b>Roll no: </b>" | sed "s/[[:space:]]\+/ /g" | tail -n 1)
    name=$(cat $file | grep -iPA1 "<b>Name: </b>" | sed "s/[[:space:]]\+/ /g" | tail -n 1)
    program=$(cat $file | grep -iPA1 "<b>Program: </b>" | sed "s/[[:space:]]\+/ /g" | tail -n 1)
    dept=$(cat $file | grep -iPA1 "<b>Department: </b>" | sed "s/[[:space:]]\+/ /g" | tail -n 1)
    hostel=$(cat $file | grep -iPA1 "<b>Hostel Info: </b>" | sed "s/[[:space:]]\+/ /g" | tail -n 1)
    username=$(cat $file | grep -iPo "mailto:[a-zA-z]+@iitk.ac.in" | cut -d ":" -f2 | cut -d "@" -f1)
    gender=$(cat $file | grep -iPA1 "<b> Gender:</b>" | sed "s/[[:space:]]\+/ /g" | tail -n 1)
    output="$rollno | $name | $program | $dept | $hostel | $username | $gender"
    echo $output > $file.dat
    rm -f $file
done