Why my iPhone sucks: Screen Scraping the O2 Website

1
Why my iPhone sucksScreen Scraping the O2
Website

• Simon Lewis
• http//lewis.li - simon_at_lewis.li

2
Overview

• Observations about the iPhone
• Why Screen Scrape?
• MMS solution
• Conclusions

3
Good Points(Before the Bashing)

• Web Browser
• SSL IMAP Email Client
• Alarm Clock
• Timer

4
Bad Points
(Camera, Phone Interface, Text Message
Interface)

• EDGE
• Bluetooth
• NO 3G

Wet piece of string is better connected
5
MMS?
(maybe its not that popular)
ERROR
6
Mocking from O2
7
Message List Window
8
Beautiful Interface?
9
There must be a better Solution...
Problems

• No web service available
• http//o2.co.uk/m not optimised for iPhone

Opportunities

• HTTP interface can be exploited
• iPhone Mail app is very nice SSL IMAP

10
Screen Scraping?

• Programmatically accessing webpage content
• Parsing the contents of the page to extract the
  information you want
• Need a language suited to text parsing

Laziness - one of the principle virtues of a
programmer What tools are available?
Perl
11
CPAN(Comprehensive Perl Archive Network)
use ConfigIniFilesuse WWWMechanizeuse
GetoptLonguse MIMELiteuse YAML qw(LoadFile
DumpFile)use DataDumperuse CGI qw(standard)
12
System Overview
2
5
lewis.li
iPhone
4
1
3
o2.co.uk
13
Authentication
my mech WWWMechanize-gtnew( autocheck gt 1
) mech-gtget( site_url )mech-gtsubmit_form(
form_name gt 'fm', fields gt msisdn gt
user_name, pin gt user_password )
14
Inbox Listing
mech-gtcontent /From.?ltpgt(.?)lt\/pgt.?Subjec
t.?ltpgt(.?)lt\/pgt/s my _at_mms_message_links
mech-gtfind_all_links( url_regex gt
qr/showMessage/ ) my unique_mms_links map
_, 1 map _-gturl_abs
_at_mms_message_links
15
Iterating through MMS
for my option (_at_other_options)
mech-gtcurrent_form-gtvalue('selectedItem',
option) mech-gtsubmit() filename
"/tmp/mmsmessage_part" open fh, "gt filename"
or die "filename !" binmode fh print fh
mech-gtcontent close fh mms_details-gtfile
name mech-gtct() mech-gtback()
16
Creating the Email
my msg MIMELite-gtnew( From gt"sender",
To gt"mail_recipient", Subject
gt"subject", Type gt'multipart/mixed')
for my part (keys mms_details) ( my
nice_name part) s!/tmp/!! if
(mms_details-gtpart eq 'image/jpeg')
msg-gtattach(Type gt'image/jpeg',
Path gt part,
Filename gt"nice_name.jpeg",
Disposition gt 'attachment') msg-gtsend
17
Receiving MMS via Email
18
Conclusions

• CPAN had all the modules needed to automate
  interaction
• Messages on the O2 website expire. Email
  provides a good way to archive content
• WWWMechanize is very useful
• Radio buttons were the biggest challenge