Printing books from HTML and CSS: metrics, formatters and results

In this document we analyzed the possibility to use CSS as a solid successor of XSL-FO for producing books in PDF format. We collected three test suites: a combination of crafted documents, some book excerpts from the publisher “Il Mulino” and some complete books from the public domain. We also devised a set of metrics to evaluate the quality of the produced PDFs. We tested the most used formatters to generate PDFs so as to get an overview of the market status in terms of support for CSS Paged Media. While book excerpts and crafted short PDFs were easily checkable manually, for the larger books we recurred to a software we developed to automatically scan the pages and check for errors. Most of the commercial PDF formatters’ output was up to expectations. Given enough time we hope that the gap between the open source offering and the commercial one will be reduced enough to consider them comparable. Overall, we were able to demonstrate how CSS Paged Media supports all of the requirements and output capabilities of XSL-FO either directly or, in a few cases, with manual intervention on the HTML/CSS code. Specifically, XSL-FO is designed with the concept of page sequences, lacking in CSS which instead works on individual pages. This different paradigm resulted in the inability to insert conditional spaces and blank pages at predefined positions. Thus, the overall process cannot be fully automated yet.