#!/bin/bash

# v0.1  : basic title extract
# v0.2  : added help arguments
# v0.3  : added version arguments
# v0.3.1: change !title to title in version info
# v0.4  : added case insensitive matching
# v0.5  : ignore SSL certificate errors
# v0.6  : switch to curl, add maximum filesize and time limits
# v0.7  : follow HTTP redirects, was broken since switch to curl
# v0.7.1: pretend to be Firefox, some broken websites reject cURL
# v0.8  : strip newlines and leading whitespace from "broken" titles
        : update user agent string
# v0.8.1: change name in version string to "blatitle"
# v0.9  : add support for gzip compression
# v0.9.1: fix support for servers that don't support HTTP HEAD
# v0.9.2: tidy up temporary files afterwards

OPTS='--location --insecure --silent --max-filesize 1048576 --max-time 10 -A "Mozilla/5.0 (X11; Linux x86_64; rv:36.0) Gecko/20100101 Firefox/36.0" -H "Accept-Encoding: gzip"'

read url

TIME=$(date +%s%N)

if [ -z "$TIME" ]
then
  echo "Timestamp error"
  exit 1
fi

if [ "$url" == "-h" ] || [ "$url" == "--help" ]
then
  echo "Usage: '!title <url>' where <url> is an http:// or https:// URL"
elif [ "$url" == "-v" ] || [ "$url" == "--version" ]
then
  echo "blatitle version 0.9.2"
else
  eval curl $OPTS -D /tmp/$TIME.header '$url' -o /tmp/$TIME.body.gz
  grep -q "Content-Encoding: gzip" /tmp/$TIME.header && gunzip /tmp/$TIME.body.gz || mv /tmp/$TIME.body.gz /tmp/$TIME.body
  sed -n -e 'H;${x;s!.*<head[^>]*>\(.*\)</head>.*!\1!I;tnext};b;:next;s!.*<title>\(.*\)</title>.*!\1!pI' /tmp/$TIME.body | sed ':a;N;$!ba;s/\n//g' | sed -e 's/^[ \t]*//'
  rm /tmp/$TIME.header /tmp/$TIME.body
fi