Yoshikoder

A new version of the Yoshikoder’s document converter is available from http://www.yoshikoder.org/ykconverter/.

The converter turns PDF, Word documents, and web pages, into UTF-8 encoded text for you to edit and save.

Visible changes from the previous version include:

• Online help
• Drag and drop documents onto the application window to convert them
• Preferences about where converted documents are saved

Philip Stone, pioneer of computer content analysis and author of the General Inquirer, is dead.

The Boston Globe has a short obituary.

There’s nothing more for me to say.

Colin Martindale’s Regressive Imagery dictionary is now available from the Yoshikoder resources page in English, French, German, Swedish and Portugese.

These are pretty much straight translations of the Wordstat files that Provalis Research makes available, except that I’ve ignored the exclusion lists that come with the English, French and Portugese versions.

Credit to Provalis for making it available electronically, and to the crowd of translators for making it available in multiple languages. This should happen a lot more often than it does.

If you go to all the work of making a dictionary, save it in an open format and make it available on the web. If you’re not sure how to do that, send it to me with your details, and I’ll put it on the resources page.

LingPipe looks like a pretty useful set of natural language processing components. The license is quirky, but seems broadly consistent with Yoshikoder’s GPL. Something for the weekend, I think.

Just realised I only opened the comments to those with a wordpress blog. That should now be fixed.

Will Gilbert, purveyor of the JarBundler ant task, tells me he wants to add the ability to specify help materials too. So I’ve sent him a little application showing how to use HelpHook and Info.plist values to get help working. Ask me if you want a copy, but I’m holding out for the new JarBundler task. Apparently, file associations are also on the agenda for next release. Can’t wait to play with it…

This is about getting yoshikoder, eclipse and subversion version control all working together. The following instructions should work for any code, not just mine, so they’re written generically.

I assume you’ve already figured out how to install and use eclipse. I also assume you know what you’re doing with subversion. If not, the rather excellent free Subversion Book is required reading. The task in this entry is to check out the code and build it without leaving the IDE.

First we need a subversion plugin for eclipse: I used these instructions. After eclipse has restarted, select Window > Preferences > Team > SVN. An unpleasant-looking error message appears, complaining about the lack of a javahl library. Cancel it and select the ‘JavaSVN (Pure Java)’ radio button. The plugin is now plugged in.

Next we need a subversion perspective. Clicking on the >> at the top right of the main window will offer ‘SVN Repository Exploring’ as an option. Choose it. The panel on the left of the main window should now be empty. Press the top right button on that panel that has tooltip ‘Add SVN Repository’ and enter the URL of the repository. With a bit of luck the repository folder structure appears as a tree. Navigate down to the code you want to check out, select the folder, right-click, and select ‘Checkout’.

From here, it’s like setting up a normal java project in eclipse. If the .project and .classpath files are also checked in (how rude, and yet how useful) then you’ll be offered the project fully formed.

By default, icons and text decorations indicate the versioned status of each of your files. As soon as you change a file and save it then an asterisk appears on its icon. To commit the changes, right-click and select ‘Commit’. From here on in, it’s regular version control stuff.

A little while ago I thought: wouldn’t it be nice to give OSX users help in the system HelpViewer? How hard could that be? Apple’s own documentation said that to launch html help pages in the system HelpViewer you just do the following:

• Drop a meta tag into the page you want to be the entry point
• Drop the folder of html help files into the resource bundle
• Jiggle the Info.plist values to tell the app how to find it

So, it’s all done in the bundle. A Help menu magically appears, and it all just goes.

Except that it doesn’t.

It used to work (1.3.1 and before), just as the old O’Reilly article suggests. These days you’ve got to fire up some native code.

The quickest way to do it, for those of us who don’t fancy objective-c much, is to embed Apple’s HelpHook code sample in our code. Helphook is a little piece of java for launching the HelpViewer via a JNI call to their native library.

Unlike the java, I have no idea whether this objective-c thingy will continue to work on Intel Macs, or even systems not exactly like my own.

Version 0.6.2 of the Yoshikoder is available from the home page.

Visible changes since 0.6.1 are:

• Pattern matching should be slightly more than twice as fast
• VBPro dictionary import now previews the file so you can specify the document encoding
• OSX users will find the same old help, but launched in HelpViewer

Rumour has it that all software should have an associated blog. This is the one for the Yoshikoder, a simple open-source multilingual content analysis program.

Development news will appear here rather than in comments to my version control system. Much more informative, that way…