ABLA: an algorithm for repairing structure-based locators through attribute annotations

Abstract

The growth of the web has been unstoppable in the last decade, which leads to an increasing demand for extracting information from it. Apart from the need to extract information, this growth also has brought the necessity to adapt web pages to user requirements, create annotations or test web applications. Due to the evolution of web pages, the complexity of the implementation of these techniques has increased. Being able to test, annotate, adapt and extract information from web pages correctly and efficiently has become a primary task. In order to perform all these tasks, it is mandatory to have the best mechanisms to effectively and unequivocally locate the desired elements throughout the web page life cycle, especially when a web page evolves. Different mechanisms are used to find web nodes. These mechanisms, called locators, are prone to fail over time owing to changes on websites. Many authors improve life expectancy of locators developing algorithms that use different types of locators. Some others have created algorithms that regenerate locators by saving extra information from the previous structure of the website. These algorithms extend the useful life of locators but their computational and storage cost is much higher. To avoid these problems, we have designed an algorithm that employs an attribute system embedded in the HTML code. The algorithm is able to regenerate the locators based on these attributes every time a single change takes place in a given element attribute. The evaluation of the proposal shows a much lower computational cost than in previous works.