#-- DESCRIPTION:
#--   Note: the script has really been superceded by the scripts 'login-geocities.sh'
#     and 'upload-geocities.sh'  
#
#--   A description of the process of the script is as
#--   follows:
#--    1) Get the Yahoo! Geocities Login Form from http://geocities.yahoo.com
#--    2) Put the user account information into the form
#--       (user-name & password)
#--    3) Reformat the 'input' fields in the form so that
#--       they can be used by the curl -d switch
#--       (i.e. strip html tags, url encode fields, remove
#--        newline characters etc)
#--    4) Post (submit) this information to the Form's target
#--       and save any cookies and headers which the server
#--       sends. Also follow any redirects which the server
#--       specifies.
#--    5) Get the final Yahoo! Mail home-page location by looking at
#--       the final redirection instruction sent by the server
#--
#--
#-- NOTES:
#--   This appears to be succesfully logging in to the geocities server.
#     (august 2003)
#
#--   The script needs to overcome several devices used
#--   by yahoo.
#--     a) The use of cookies
#--     b) The use of automatic redirects of the web-client
#--        (using the LOCATION http header field)
#--     c) The use of ssl to encrypt the logon process
#--     d) The use of hidden fields with random numbers within
#--        the Yahoo! Mail Log-in Form to discourage automatic logins
#--
#--   These things can all be overcome because of the
#--   capabilities of the 'curl' non-interactive web client.
#--
#--   This script has only been used within the 'cygwin'
#--   unix emulating bash shell running on WinMe.
#
#-- Revisions:
#--   may 2002, I created this script,
#--   august 2003, Barcelona, I am revising this script in order to adapt it
#--    to uploading a file to a free geocities account. Also I would like to 
#--    rewrite this as a java application, (even an applet, although that is a little
#--    optimistic) so that it could be used from an internet cafe or from some
#--    other somewhat 'hostile' location. 
#-- AUTHOR: mj bishop (matth3wbishop@yahoo.com)
#--
#--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*

sTargetUrl=""              #-- The initial target of the login form                                                           
sYahooMailHomeUrl=""       #-- The final url of the Yahoo! Mail home page
sYahooMailHostName=""      #-- Just the host-name of the above.
sInitialUrl="http://www.geocities.com"
curl -L http://www.geocities.com > login-form.html
#-- Extract the form fields from the html
#--

sTargetUrl=$(\
   expand login-form.html | \
   sed -n '/<form.*name=login_form/s/.*action= *\([^ ]\+\) .*/\1/p')

   
#-- The code below wont work if the tags are upper case, unless sed 
#-- is working in a case insensitive way, which is unlikely
echo $sTargetUrl;
expand login-form.html | \
  sed -n '/<form.*name=login_form/,/<\/form>/p' | \
  sed '/<input.*name=/!d' | \
  sed "s/.*\(<input[^>]\+>\).*/\1/g" | \
  sed 's/.*name="\([^"]*\)".*value="\([^"]*\)".*/\1=\2/g' | \
  sed 's/.*name="\([^"]*\)".*value=[ ]*\([^ >]\+\)[ >].*/\1=\2/g' | \
  sed 's/.*name=[ ]*\([^ >]\+\)[ >].*value="\([^"]*\)".*/\1=\2/g' | \
  sed "s/.*name=[ ]*\([^ >]\+\)[ >].*/\1=/g" | \
  sed "s/login=.*/login=matth3wbishop/g" | \
  sed "s/passwd=.*/passwd=bacibaci/g"  | \
  sed -e 's/ *$//g' -e 's/^ *//g' | sed 's/ /+/g' | \
  sed 's/$/\&/g' | tr -d '\012' > form-data.txt 

 sManagerUrl="http://geocities.yahoo.com/filemanager"

 #-- The line below submits the form data gathered above and collects the 
 #-- cookies which the geocities server sends back in the file 'headers.txt'
 curl -D headers.txt -d @form-data.txt -L $sTargetUrl 
 curl -b headers.txt $sManagerUrl > other-side.html
 #curl -b headers.txt geocities.yahoo.com  > other-side.html

expand other-side.html | \
  tr [A-Z] [a-z] | \
  sed -n '/<form.*name=dispfiles/,/<\/form>/p' | \
  sed '/<input.*name=/!d' | \
  sed "s/.*\(<input[^>]\+>\).*/\1/g" | \
  sed 's/.*name="\([^"]*\)".*value="\([^"]*\)".*/\1=\2/g' | \
  sed 's/.*name="\([^"]*\)".*value=[ ]*\([^ >]\+\)[ >].*/\1=\2/g' | \
  sed 's/.*name=[ ]*\([^ >]\+\)[ >].*value="\([^"]*\)".*/\1=\2/g' | \
  sed "s/.*name=[ ]*\([^ >]\+\)[ >].*/\1=/g" 

 #-- Curl syntax
 # directory=&op-uploadtodir=upload+files
 # -F/--form <name=content> Specify HTTP POST data (H)
 # cat other-side.html | grep -Ei "(input|form)"


 curl -b headers.txt  $sManagerUrl/upload > other-side.html
 #sNewTargetUrl=$(\
 #  curl -D headers.txt -d @form-data.txt -L $sTargetUrl | \
 #  sed -n '/url=/s#.*url=\(http://[^"]\+\)".*#\1#gp')

 #echo "sNewTargetUrl=$sNewTargetUrl"
 #curl -D headers.txt -d @form-data.txt -L $sTargetUrl > other-side.html
 #curl -b headers.txt -L $sNewTargetUrl > other-side2.html 


exit 1;

#-- A user progress message
#--
echo "
  --------------------------------------------------
  Extracting the HTML login form from the Yahoo! Geocities 
  site [http://mail.yahoo.com]

  a) Putting account (user-name & password) into the login form
  b) Retrieving the target url of the login form
  c)
  --------------------------------------------------
     "

if [ "x" = "y" ]
then
  echo "
    ===========================
    = CONTENTS OF $sInitialUrl
    ==========================="
  wget -O- $sInitialUrl 
fi

#-- some greps have a -P switch and some an -E switch, for 
#-- extended regular expressions

curl -L $sInitialUrl | sed -n '/<form.*name=login_form/,/<\/form>/p'  | \
  grep -E '(<form|<input|<\/form>)' > t001.txt

#exit 1;
#-- used for debugging
if [ "x" = "x" ]
then
  echo "
    ===========================
    = CONTENTS OF t001.txt 
    ==========================="
  cat t001.txt
fi
#--
#-- Place the particular account details (user-name & password)
#-- within the Geocities login form
#--

echo '<html><body>' > login-form.html
sed '/name="\?login"\?/s/value=""/value="matth3wbishop"/g' t001.txt > t000.txt
sed '/name="passwd"/s/type="password"/& value="bacibaci"/g' t000.txt >> login-form.html
echo '</body></html>' >> login-form.html

if [ "x" = "x" ]
then
  echo "
    ===========================
    = CONTENTS OF login-form.html
    ==========================="
    
  cat login-form.html
fi

#exit 1;

sTargetUrl=$(sed -n '/<form /s/.*action="\([^"]*\)".*/\1/p' login-form.html)
echo "-->>sTargetUrl=$sTargetUrl"

#--
#-- Reformat the HTML form fields into a format which
#-- is usable by the curl -d switch. i.e Name=Value
#-- Ensure that each field is 'url encoded'.
#--
cat login-form.html | grep '<input' \
  | sed 's/.*name="\([.a-zA-Z0-9 ]*\)".*value="\([.a-zA-Z0-9 ]*\)".*/\1=\2/g' \
  | sed 's/.*name="\([.a-zA-Z0-9 ]*\)".*value=\([0-9]*\).*/\1=\2/g' \
  | expand | sed -e 's/ *$//g' -e 's/^ *//g' | sed 's/ /+/g' \
  | sed 's/$/\&/g' | tr -d '\012' > formData.txt


if [ "x" = "x" ]
then
  echo "
    ===========================
    = CONTENTS OF formData.txt 
    ==========================="
    
  cat formData.txt
fi
#--
#--
echo "
  --------------------------------------------------
  Submitting (post method) the Yahoo! Mail Login form
    -- Saving cookies and other header into 'headers.txt'
    -- Automatically following any redirects
  --------------------------------------------------
     "
#curl -D headers.txt -d @formData.txt -L $sTargetUrl > ymail.html
curl -D headers.txt -d @formData.txt -L $sTargetUrl > ymail.html


if [ "x" = "x" ]
then
  echo "
    ===========================
    = CONTENTS OF ymail.html 
    ==========================="
    
  cat ymail.html
fi

exit 1;

sNewLocation=$(cat ymail.html | sed -n '/url=/s#.*url=\(http://[^"]\+\)".*#\1#gp')
echo "==> sNewLocation=$sNewLocation"

curl -D headers.txt -d @formData.txt -L $sNewLocation 
exit 1;

sYahooMailHomeUrl=$(sed -n '/Location:/s/Location: *//p' headers.txt)
echo "sYahooMailHomeUrl = $sYahooMailHomeUrl"

sYahooMailHostName=$(echo $sYahooMailHomeUrl | sed 's/\.com\/.*/\.com/g')
echo "sYahooMailHostName = $sYahooMailHostName"

sInboxUrl=$( grep -i 'check mail' ymail.html | sed 's/.*href="\([^"]*\)".*/\1/g')
echo "sInboxUrl = $sInboxUrl"


#--
#-- Go to the Yahoo! Mail Inbox
#--
echo "
  --------------------------------------------------
  Following the link to the Yahoo! Mail inbox
  using previously set cookies
  --------------------------------------------------
     "
  
curl -b headers.txt ${sYahooMailHostName}${sInboxUrl} > ymail.html

# links -dump ymail.html | tr -s ' ' | less

#-- It is not working to manually redirect as the lines below
#-- attempt to do. Maybe have to redirect twice.
# sYahooMailHomeUrl=$(sed -n '/Location:/s/Location: *//gp' headers.txt)
# echo "sYahooMailHomeUrl = $sYahooMailHomeUrl"
# curl -i -b headers.txt -d @formData.txt $sYahooMailHomeUrl | less

# ---------------------------------------------
#  511  (echo '<html><body>'; curl mail.yahoo.com | sed -n '/<form /,/<\/form>/p' | grep -P '(<form|<input|<\/form>)' | sed -e '/name="login"/s/value=""/value="matth3wbishop"/g' -e '/name="passwd"/s/type="password"/& value="bacibaci"/g'; echo '</body></html>') > login-form.html
#  543  cat login-form.html | grep '<input' | sed 's/.*name="\([.a-zA-Z0-9 ]*\)".*value="\([.a-zA-Z0-9 ]*\)".*/\1=\2/g' | sed 's/.*name="\([.a-zA-Z0-9 ]*\)".*value=\([0-9]*\).*/\1=\2/g'
#  556  cat login-form.html | grep '<input' | sed 's/.*name="\([.a-zA-Z0-9 ]*\)".*value="\([.a-zA-Z0-9 ]*\)".*/\1=\2/g' | sed 's/.*name="\([.a-zA-Z0-9 ]*\)".*value=\([0-9]*\).*/\1=\2/g' | expand | sed -e 's/ *$//g' -e 's/^ *//g' | sed 's/ /+/g' | sed 's/$/\&/g' | tr -d '\012'
#  557  history 2 >> getYMail.txt
#  580  curl -D - -d @qsy.txt -L https://login.yahoo.com/config/login?90ep5ie0i4i3q | less
#  581  history 2 >> getYMail.txt