Image Markup Tool blog

The previous post detailed a bug I thought I'd found when running IMT on Linux under Wine. However, after extensive testing, I've been unable to reproduce it. I think that it was actually caused by the propagation of a file which was created with the previous version, prior to my last couple of fixes, into three more files. Once the <facsimile> and <text> elements are out of order, they probably stay that way, but unless a file was corrupted in the first place, I don't think it can happen. So I'll shelve this for the moment, but keep my eye on the markup files created on Linux for any more signs of this.

If you do New / With Same Categories, then choose an image, then save; then you add annotations; you end up with a file in which the <facsimile> element is wrongly positioned after the <text> element. I've already added code to ensure that <facsimile> should always be inserted right after <teiHeader>, but I haven't checked the code which writes <text> to make sure that it always places text as the last child of <TEI>. I need to do that. The problem may only show up on Linux (haven't tested on Windows yet).

It took me a while, but I've finally completed the rewrite of this code. It was worth doing to get a more robust highlighting setup, and it's still very fast, even on large files.

I've been unhappy with the XML syntax highlighter limitations, so I've decided to start again with a more rigorous approach. This is a nice scale of algorithm to get more familiar with C++ syntax too. Made a good start -- processing instructions, doctype declarations and basic open and close tags are already working, and I'm starting in on the attributes.

Found another problem related to the limitations in Wine's implementation of version info routines. Fixed it and released a new version. As I work with the IMT under Wine on the ColDesp project, I may find a couple more of these annoyances. In the long run, this is good, because the app is effectively now supported and working well on Linux.

Did a new release of IMT (1.8.1.6) after I discovered that WINE fails to implement some of the Windows version info routines, meaning that two attributes in the <appInfo> tag end up empty, and thus invalid. Also did a fix for a bug I'm unable to reproduce, and no-one has reported, but which somehow caused me to end up with a file in which the <text> element was before the <facsimile> element -- also resulting in an invalid file.

I have half-way-decent syntax highlighting working now, using a kind of hybrid of regular expression matches, start-end matches, and testing the existing format of text which has already been highlighted. I think this will do for the moment. The only problems arise when attribute+value structures are split across multiple lines. I think we can live with that for the moment.

In the process, I discovered how to convert text documents from UTF-8 using QTextCodec. This is simple enough, but I'll ultimately need to figure out how to detect encodings. I'll do a bit of research on this.

I have all the non-interactive multi-line highlights working (highlights that are discrete, such that if you're inside one, you're not inside the others). I'm just starting on attributes, which are more complicated because they interact with open tags.

Implemented proper regexps for the W3C NameStartChar and NameChar components. The only limitation is characters above \xFFFF, which won't fit into the \xhhhh format for QRegExp. Also fixed a couple of bugs in the existing implementation, by making the attribute matching minimal (= not greedy), and allowing spaces between attribute names, equals signs, and attribute values.

We finally have a French interface file for the Image Markup Tool, kindly provided by Marjorie Burghart (EHESS, France). I've built a new version of the application (although no other changes have been made), and added the appropriate credits and acknowledgements in the Help file and on the website.