Cross-Browser Document Capture System
Name
Marti Kaljuve
Abstract
A web page is seldom displayed in the exact same manner in different browser and operating system combinations. There are several reasons for different rendering outcomes: interpretation of web standards by the browser, the browser's rendering engine, available fonts in the operating system, plugins installed in the browser, screen resolution etc. Neglecting to consider these differences as a web designer may lead to webpage layout issues that result in lost customers.
Web designers might consider it common practice to test webpages on several browsers to eliminate cross-browser layout issues. Experiments show that finding visual differences is a dull and cumbersome task for people. Knowing this, another member working at Browserbite has created an algorithm that has proved to be much faster and more accurate at finding layout issues compared to humans. The algorithm works by comparing a baseline (oracle in software testing terms) webpage in image form to other image captures of the same webpage in different browsers, finding differences in layout and position that a human might consider erroneous.
This thesis concentrates on the problem of creating the input to the aforementioned algorithm. A selective overview of existing solutions and services for webpage capture and automation is given, measuring their performance where possible. A list of requirements are established for a cross-platform capture solution to be commercialized. A fast and cross-platform method of capturing full webpages is then introduced, and an overview of a scalable Software-as-a-Service system implemented for cross-browser and cross-platform capture in several virtual and physical machines asynchronously is given.
Graduation Thesis language
English
Graduation Thesis type
Master - Software Engineering
Supervisor(s)
Marlon Dumas, Kaspar Loog
Defence year
2013