#--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*
#-- SCRIPT: autoseekjob.sh
#-- 
#-- DESCRIPTION:
#--  This file contains some command lines which automatically
#--  submit an application on the
#--   http://www.seek.com.au  Australian Jobs site..
#--  The 'Seek' web-site has been set up in a manner that
#--  encourages the users of the site to submit job
#--  applications through the site rather than by directly
#--  emailing the advertiser of the job. This is probably in order
#--  to ensure that the job advertiser is aware that the
#--  job-seeker has found the advertisement on the 'Seek' site.
#--
#-- To this end, many of the job advertisements on the site do
#-- not contain any email address for the advertiser. Instead
#-- an 'Apply now' button/ link is provided on the Job-Ad
#-- page which provides access to an online application
#-- form. Once this form is filled out, its contents are
#-- (apparently) emailed by some 'Seek' daemon to the
#-- advertiser.
#--
#-- The 'Seek' site has a number of mechanisms to
#-- ensure that the user must access the application form
#-- through the use of a traditional interactive browser
#-- (using point-and-click), namely;
#--
#-- a) There are a number of automatic redirects sent
#-- to the client browser to
#-- disguise the ultimate destination url (as opposed to
#-- the url displayed as the form's/hyperlink's target
#-- in the page-source html.
#--  For example:
#--  http://it.seek.com.au/showjob.asp?[job-id-number]
#--     takes the web-client to a page containing
#--     frames where the desired job-as url is 
#--  http://it.seek.com.au/users/joblist.asp?AdID=[job-id-number]
#--
#-- b)
#-- A cookie is sent to the web-client when the client
#-- first views the job application. This cookie then
#-- appears to be used to ensure that the web-client
#-- viewing the application form has previously also
#-- viewed the Job-ad.
#--
#-- c)
#-- The form in the job-application page contains a
#-- hidden element with some kind of time-string value.
#--
#-- The script below is an attempt to by-pass the
#-- measures which the 'Seek' site developers have
#-- used to disallow the non-interactive submission
#-- of job-applications through the 'Seek' site.
#--
#-- The purpose of the script is to obviate the
#-- (slightly tedious) necessity of manually filling
#-- out the job application form on the 'Seek' site
#-- and therefore to allow the 'batch' or multiple
#-- submission of Job applications.
#--
#--


#-- DATES:
#--   may 02
#-- NOTES:
#--   The crucial tool used below is 'curl'.
#--   Other tools that I attempted to use were wget and lynx.
#--   The two key features that I needed in a non-interactive
#--   web-client (to complete the task) were the capability
#--   to send cookie data back to the web-server (in order to 
#--   fool the server into believing that a cookie had 
#--   previously been sent to the web-client) and the capability
#--   to 'post' data to the web-server.
#-- 
#--   The lynx text browser contains the ability to post data
#--   with the --post_data command line switch, but I was unable
#--   to discover how to send cookie data back to the server 
#--   with lynx. (The --cookie_file switch did not do what I was
#--   hoping). The 'wget' tool is capable of sending cookie data
#--   but did not seem to be able to post data.
#--
#--   The 'curl' tool contains both capabilities. 
#--   This script was a late-night 'folly' spurred on by my annoyance
#--   at having to continually fill out the 'Seek' job application
#--   form with the same information, over and over again
#--   (especially in a job-market atmosphere where most of the 
#--   job application seem to be ignored anyway)
#-- AUTHOR: matthew bishop
#-- STATUS: dev  
#--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*

#  This line doesnt work because there is no cookie file (hed.txt)
#    curl -d txtAdID=1933148 'http://it.seek.com.au/users/JobApply.asp?txtAdID1933148&' | less

#-- Test if the job-id number is still valid
#-- (That the advertisement has not been removed from the
#-- 'Seek' site.

if [ -n "$1" ];
then
  jobid=$1
else
  #-- testing
  jobid=1933149
fi

url1=http://it.seek.com.au/users/joblist.asp?AdID=$jobid
url2=http://it.seek.com.au/showjob.asp?jobid=$jobid
url3=http://it.seek.com.au/users/JobApply.asp?txtAdID$jobid'&'
# echo url3=$url3
# return;


removed=$(wget -nd -O - -o t.log $url1 | grep -i 'recently removed this position' | wc -l)
removed=$(echo $removed | sed 's/ //g')

if  [ $removed -eq 1 ]; then
  echo "'Seek' id-number=$jobid (removed)"
  return 1
fi


#-- Capture the cookie which is sent to the web-client by the
#-- 'showjob' page
     curl -D seek-cookie.txt $url2

#-- Use curl to post data and retrieve the on-line job-app form 
     curl -b seek-cookie.txt -d txtAdID=$jobid $url3 

#-- to do:
#--   sed the app form to insert good values (name,phone etc)