Use Inkscape and XSLT to Create Cross-Platform Reports and Forms
I work for a health-care company developing application software. My colleagues and I are responsible for writing software to process health-care claims, manage work flow and make the company as efficient as possible. We recently decided to replace a piece of third-party software that took health-care claim data and overlaid it on standard HIPAA (Health Insurance Portability and Accountability Act) claim forms. The software would take the data and transpose it into PDF files that we stored on a large file server. Each PDF contained one claim on its proper form. We made the decision to replace the software because we needed something more agile. We wanted something that would create the claim image dynamically and not consume space on our servers.
Health-care claims are very intricate (Figure 1). Many boxes and boilerplate text have to be drawn. The conventional way to do this with a software application is to draw a series of lines using coordinates and lengths, and then lay the static and dynamic content on top of the newly drawn lines. The process of programming an application like this is long and tedious, not to mention error-prone. We wanted something that was easier to create and maintain. Our requirements were as follows:
We must be able to print high-quality versions of the claims.
Claims must be accessible from a Web browser.
The solution has to be programming language-independent. We use Python, PHP, Perl and Java. The images need to be created using any of these languages.
We must be able to convert the claim data and form into several different file formats, specifically PDF and PNG.
The entire solution must be platform-independent.
After reviewing the requirements, we looked at several different open- and closed-source options. None of them met all of our requirements, so we turned to creating our own solution. We tried scanning a blank claim form and using ImageMagick to put the claim data on it. This almost gave us what we wanted. The problem was that it was going to be tedious and redundant to create the solution in all of the required languages. Next, we turned to FOP (Formatting Objects Processor). This solution was closer to what we wanted. However, it would take too long create the claim forms. Plus, the solution was not really language-independent either (FOP is a Java library). We could have written wrappers for the FOP command-line interface, but we were convinced that there was still a better solution.
While exploring the FOP solution, we had the idea of using Scalable Vector Graphics (SVGs). Basically, we would take an SVG image of the claim form and make it into an XSLT (eXtensible Stylesheet Language Transformation), because the SVG format is a special XML format. Then, we would pull the claim data from our database and convert it into an XML string. Using any of our languages, we could then take the XSLT and the XML and create an SVG image of the claim. This solution met all of our requirements. It was language- and platform-independent. We could print the SVG images and embed them into Web pages. Furthermore, SVG images can be converted into different file formats easily. Another nice feature of this solution is the small file size of the SVG images. If we wanted to archive the images, they would take a fraction of the space the old solution did. Because SVG images are text, not compressed binary, the files can be compressed and save even more space.
One of the things that made the SVG solution so appealing was how easy it would be to create and maintain the master SVG image of the form. To do this, we would use Inkscape. Inkscape is an SVG-authoring tool that works on Linux, Mac OS X, Windows and other UNIX-like operating systems. Other SVG-authoring tools are available, but we chose Inkscape because it is open, and it is in the package manager for most Linux distributions.
The first thing we did to create the master SVG was open Inkscape and create a new US Letter size document. To keep things organized, we created four layers in the new document: scan, overlay, boilerplate and dynamic text. Using the scan layer, we imported a scan of a claim. Doing this allowed us to line up everything on the Inkscape stage without having to measure anything. After importing the image, we locked the layer so that it could not be modified accidentally. Actually, after we were finished with each layer on the SVG, we would lock it to ensure it was not tampered with.
Next, we used the overlay layer to trace all the lines and boxes from the original claim that we imported. This step was a little tricky. When the image we scanned was originally created, the lines were not spaced evenly for one reason or another. We decided to line up things correctly on our version. Fortunately, Inkscape has tools to do this automatically. By selecting all of the objects that needed to be spaced out (Shift-left-click) and using the Align and Distribute dialog (Object→Align and Distribute in the menu), Inkscape fixed the spacing issues. When finished, we had something that looked like Figure 2.
After drawing all of the lines, it was time to add all the boilerplate text. For this, we used the aptly named boilerplate layer. Before we got started, we decided to remove the scan layer completely, because we no longer needed it. To align the text properly, we used the Guides in Inkscape. Guides are exactly what the name suggests—guide lines that exist only inside of Inkscape for the purpose of aligning objects. To use a guide line, simply click the top or left-hand margin and drag the line into place. To get the most out of the guide lines, we enabled the Snap points to guides feature (File→Document Preferences→Guides). Doing this allowed us to place all of the text exactly in alignment. Figure 3 shows what the SVG looked like after this step.
Finally, we switched to the dynamic text layer and added placeholders where the claim data would be located. Again, we used the guides to align everything. For the text place holders, we used a single $ for each block of text. Then, to make life easier down the road, we renamed each of the dynamic text objects to something relevant. We did this by left-clicking on the object and going to Object→Object Properties in the menu. Figure 4 shows the final master SVG with the guide lines.
Creating the master SVG took about four full hours of work. I would venture to guess that it would have taken several days to do this programmatically.
- High-Availability Storage with HA-LVM
- DNSMasq, the Pint-Sized Super Dæmon!
- Localhost DNS Cache
- Real-Time Rogue Wireless Access Point Detection with the Raspberry Pi
- Days Between Dates: the Counting
- You're the Boss with UBOS
- The Usability of GNOME
- Linux for Astronomers
- Multitenant Sites
- PostgreSQL, the NoSQL Database