I have been doing some work with the excellent Moodle, an open source Learning Management System, and after doing an upgrade, started experiencing problems uploading SCORM objects. The specific errors I was getting were PCLZIP_ERR_BAD_FORMAT (-10) : Unable to find End of Central Dir Record signature, and An error occurs during package unzip (sic). After a lot of searching, it seemed a few people were having similar issues, and mostly it seems Internet Explorer was being blamed. Perhaps there is also a problem with Internet Explorer, however in my case, I ended up working out where Moodle stores the files on the server and inspected the folder. The file was corrupt, I couldn't download the zip file.

After trying to manually upload the files I discovered that the disk quota had been exceeded. After doing a cleanup of the disk, everything worked well.

Sometimes when searching for the cause of a problem some facts eclipse our abilities to be a little objective about the true nature of the problem. In my case it seemed clear that the problem occurred as soon as I upgraded, when in reality the cause was my making a backup on a full server. This could have been avoided in a few ways. Firstly I should keep an eye on the server quotas, that should have been a give away. However, I think the Moodle developers should have caught an IO exception when the disk was full, and reported back to the browser. This would have saved a lot of time. Having said that, I am very impressed with Moodle, this is just one of those habits that happen a lot in programming where such potential errors as stack overflow, out of memory, or IO errors are ignored with an optimistic hope that they should never occur. With this error, I assume something worse is happening. Because the page didn't crash with an error, I can only assume the error was swallowed, and thus vital helpful information was never presented on the web page.

As soon as I have caught up with this work I will look at the code and see if I can offer a bug fix.

Update: I have also experienced problems doing a course backup, with the same root cause, not enough disk space. When doing the course backup, the messages received are An error occurred while backing up course start and An error occurred while copying the zip file to the course directory. After looking at the code, it appears the Moodle error handling only looks for 4 types of IO errors. PHP 5 has 7 types, I believe it is the type 7 error that would cause this. Unfortunately the error trapping routine has a default handler for unknown error types.

---

A project I developed called for context sensitive dynamic help inside of Microsoft Word, based on the current cursor location. The idea was that if the cursor is on an image, then help information would be presented regarding the business rules pertaining to images. I.E. what size can the image be, policies regarding alternative text for web publishing etc. To achieve this there were two components. One was a Quality Assurance component to check for conformance to business rules. The other was the dynamic help system. I present a whittled down version of the help system here. The Quality Assurance component will be discussed in a future entry. There is everything necessary to implement this as a full fledged context sensitive dynamic help system. The help information is presented in a custom task pane, and uses HTML.

Figure 1 -Word 2007 showing the Context sensitive dynamic help system

Architecture

This solution uses three main components:

• The VSTO addin
• A MessageBroker which receives events from the VSTO addin
• The Help UserControl which consumes events from the MesageBroker

This centralised messaging architecture is beneficial when other components are introduced such as the QualityAssurance component referred to earlier. The VSTO Addin wouldn't need to be changed, the MessageBroker, the QualityAssurance component would subscribe as listeners for the events raised by the MessageBroker. Another advantage of this approach is that security and logging would be easily implemented in the one location.

So, when the Addin loads, the WindowSelectionChanged event is subscribed to by the addin.

this .Application.WindowSelectionChange+= new Microsoft.Office.Interop.Word.ApplicationEvents4_WindowSelectionChangeEventHandler(
   Application_WindowSelectionChange);

It would be possible for the Help UserControl to subscribe directly, but by doing it this way we can define a few rules around what events get triggered, and also cache the current style so that the event is not consumed by every registered listener every time the cursor moves even though the style hasn't changed.

void Application_WindowSelectionChange(Microsoft.Office.Interop.Word.Selection Sel) {
     Word.Style style = (Word.Style)Sel.get_Style();
     if (! style.NameLocal.Equals(_currentStyle)) {
        _currentStyle = style.NameLocal;
        _messageBroker.Publish(typeof(MessageEventArgs), Sel.Paragraphs[1], style);
     }
}

Note in this code how it is important to retrieve an instance of the style object and cast it to type Word.Style. This is COM Interop in Word. It would be possible to write an extension method to get around this, but I don't think its necessary for this example.

The constructor for the Help UserControl accepts an instance of the MessageBroker. Note: I should probably make this class a Singleton.

The Help UserControl receives event notifications when the current style has changed. It loads a web browser control and loads an HTML page named after the style. if no page exists for that style, it will display a default help page. There is a tabbed interface, with an Infoand aHelp tab. The Help tab has not been implemented, the idea was to provide links to examples.

I have only included stripped down versions of HTML pages for three styles:

• Heading 1
• Heading 2
• Heading 3

It is simply a matter of creating a page witht the style name, and an extension of .html and dropping it in the help folder. The code that loads the page looks like this:

private void LoadHelpIntoBrowser( string currentStyle) {
    String helpFilePath = CreateHelpFilePath(currentStyle);
     if (File.Exists(helpFilePath)) {
       webInfo.Navigate(helpFilePath);
    } else {
       LoadHelpHomePage();
    }
}

private void LoadHelpHomePage() {
   string p = Path.Combine(Path.GetDirectoryName(
   System.Reflection.Assembly.GetExecutingAssembly().CodeBase), "..\\..\\help" );
   String helpFilePath = CreateHelpFilePath( "WordJester" );
   webInfo.Navigate(helpFilePath);
}

private String CreateHelpFilePath(String style) {
   Uri baseUri = new Uri(
      Path.GetDirectoryName(System.Reflection.Assembly.GetExecutingAssembly().CodeBase));
    return String.Format( "{0}{1}..{1}..{1}help{1}{2}.html" , baseUri.LocalPath,
         Path.DirectorySeparatorChar, style);
}

The beauty of this approach is that you can add in as many styles as required, even non standard styles, and as long as a help file exists, it will be displayed. The file names will look like this:

Figure 2 - file names for the HYML pages based upon the word style name

Run the solution, and try creating some heading levels 1, 2 and 3, and move the cursor between them and check out the help in the custom task pane change. Let me know what you think of this application.

WordJester.zip (224.21 KB)

---

See part II of this post here.

By combining the Google Maps API with an IP address host service such as HostIP.info, it is possible to present a map showing the current visitors to your blog, and to display them on a map. This implementation shows how to do this for dasBlog. The result will look something like this:

Notice that the shadow on the marker for the united States is darker, indicating more hits for that area. Each of the marker icons can be clicked to give location details and how many hits for that location.

Client side or server side location determination

The location of an IP Address can be determined at the server side, or at the client side. Each has its advantages and disadvantages. A brief but not complete list of pros and cons is:

• Client side
  • Pros
    • No delay in serving the page as the processing is done after the page has been serves to the client.
    • Easy to implement
  • Cons
    • Requires a call to a web service on a different domain which will cause security concerns
    • Does not allow for users to be able to modify what is their perceived location
• Server side
  • Pros
    • More control as you are not dependent on an external web service
    • It is easier to cache results if required
  • Cons
    • More difficult to implement, especially if you want to keep to the architecture of the software you are developing for
    • You must maintain and update your own databse of IP address ranges and locations

The client side solution

This solution relies on a client side solution, taking advantage of the excellent Prototype JavaScript Framework. Further down the track I might possibly change this to a server side solution, but as it is only for an administrative page it is not so important. Note that there are security concerns with doing this client side. It is necessary top make an AJAX call to an external webservice, and this could be malicious, especially if a site were hacked. In order to view the example for this, you would have to add http://blog.focas.net.au to your trusted zones in Internet Explorer. If you do this, please make sure you remove it immediately afterwards. Usually I would only add a trusted site if I were using it frequently. The example page can be viewed at Visitors.aspx (Please scroll down to the bottom of the page as there are styling issues. The page may need refreshing)

I created a server page called Visitors.aspx, and a user control called VisitorsBox.ascx. These are directly based upon Referrers.aspx and ReferrersBox.ascx in the standard DasBlog source code.

How it works

C# code in the page creates JavaScript which is then registered with the RegisterClientScriptBlock method of the ClientScript object. The javascript requires some dynamic variables to be created and this is why it is in the page. Also the way dasBlog is set-up seems to make it difficult to add JavaScript to just one page without using this method. Atwo dimensional array is created which contains unique IP addresses and a count of hits for those addresses. THis is then processed on the client side The JavaScript code to create the map follows, it is well documented in the Google Maps API documentation, so i won't say too much about it.
     
if (GBrowserIsCompatible()) {
     map = new GMap2(document.getElementById('map_canvas'));
     map.setCenter(new GLatLng(37.4419, -122.1419), 1);
     var mapTypeControl = new GMapTypeControl();
     var topRight = new GControlPosition(G_ANCHOR_TOP_RIGHT, new GSize(10,10));
     var bottomRight = new GControlPosition(G_ANCHOR_BOTTOM_RIGHT, new GSize(10,10));
     map.addControl(mapTypeControl, topRight);
     map.addControl(new GSmallMapControl());

Now we need to loop thru the IP address array and call the hostip.info page to get the city/state/country information. There are different calls that can be made to return HTML or XML. I used the HTML version as it is a smaller response, and easy to parse with regular expressions. If there is no information for an IP address it will return (Unknown). Thed regular expression trims them out. This database seems to be maintained by one person, so consider donating to them.

          var myAjax; for (var x=0;x<addresses.length;x++)
{ myAjax = new
Ajax.Request('http://api.hostip.info/get_html.php?ip='
+ addresses[x][0] ,  { method: 'get', onSuccess: function(originalRequest)
{ var result=originalRequest.responseText; var tokens=result.split('\n'); var count=addresses[x][1]; var country=/\:[^\(]+/i.exec(tokens[0])[0].substring(1).strip(); var city=/\:[^\(]+/i.exec(tokens[1])[0].substring(1).strip(); var location=(city.length>0)?
city + ',':'';
location+=country; varmessage=location + ':
' + addresses[x][1] + '
visitor'; if (addresses[x][1]
> 1) message+='s'; 

 Now we can call the GClientGeocoder object to turn the city/state/country information into a latitude/longtitude combination. The GClientGeocoder's getLatLng object expects a method to be passed in to handle a callback. In this code, it is used to create the marker on the map which can be clicked to view more information.

             var geocoder = new
GClientGeocoder();    geocoder.getLatLng(location, function(point)
{    if (point)
{       var marker = new
GMarker(point);        map.addOverlay(marker);
   GEvent.addListener(marker, 'click', function()
{    marker.openInfoWindow(message);
    });  }  }); 

This javascript is created in the C# code using a StringBuilder object, and is then added to the page using the RegisterClientScriptBlock method of the ClientScript object.

Page.ClientScript.RegisterClientScriptBlock(this.GetType(), "ipaddressdata", sb.ToString(), true);

Download

The download code contains the Visitors.aspx and Visitors.ascx page and component. These should be dropped into the newtelligence.DasBlog.Web folder of your dasBlog source code. Before going any further you will need to change the Google Maps key in VisitorsBox.ascx.cs file.

          Page.ClientScript.RegisterClientScriptBlock(this.GetType(), "googlemaps", @"<script
src='http://maps.google.com/maps?file=api&amp;v=2&amp;key=YOUR
KEY GOES HERE'"+"
type='text/javascript'></script>");
        

When you have compiled the code, drop those two files into the root of your website, and the newtelligence.DasBlog.Web.dll into the bin folder of your website. Remember to change the trust settings for your blog, and enjoy seeing who is visiting your site. Note that the Google Maps JavaScript and Prototype JavaScripts are served up from the Google servers so you don't need to download them.

Visitors.zip (10.57 KB)

---

Wikipediaise - What is it?

Wikipediaise is a Visual Tools for Office addin (VSTO) developed in Microsoft Visual Studio 2008 as an addin for Microsoft Word. It is written in C#.  It  was designed to hyperlink acronyms and jargon  to Wikipedia.

I do a lot of technical documentation for my work, and the IT industry being what it is, the documents end up with a ridiculous number of acronyms. To make life easier, we usually put an abbreviation section at the top of the document, but this is a time consuming process to go thru every time, so I automated it. Additionally I added another method which will seek out the first occurrence of an abbreviation or acronym, and hyperlink it. First I will describe how this works, then how to use and customize the functionality.

Initially Wikipedia was used as the reference point, as it is an excellent reference point for technical information. After a while it became clear that many acronyms were better documented elsewhere, or in internal company documents, so I added the ability to use alternative reference sources.

Note that although I refer to acronyms, the addin is good for jargon and technical terms as well.

The following images show a before and after shot of a simple document, additionally it shows the document with an acronym table inserted at the top.

Figure 1 - Before shot of a Word document with acronyms and jargon to be hyperlinked

Figure 2 - The same document after it has been hyperlinked

Figure 3 - The same document, hyperlinked, and with an acronym table inserted at the beginning

How it works

The application comes with an embedded XML file with a set of pre-defined acronyms. This serves as an example only. The application will look in the %mydocuments% folder for a file called wikipediaise.dic. If this file exists it will override the embedded file, so the application can be customized for most requirements.

Format of the XML file

There are two elements available in the wikipediaise.dic file shown below.

Table 1 - Elements available in wikipediaise.dic

Element	Description	Comment
excludeStyle	Lists a Word style to be excluded from the process	This could be a built in style or a user defined style. If a word is in this style it will not be hyperlinked.
entry	Contains a mandatory key term that will be searched for. Optional attributes will be described later.	This text will be searched for in a case sensitive manner. If the term is found in the middle of a word, it will still be matched. For this reason, position longer superset acronyms earlier e.g. place https before http

The excludeStyle element has no attributes, so just looks like this

Figure 4 - excludeStyle element example

The entry element has attributes, these are described below.

Attribute name	Mandatory	Description	Comment
key	Yes	The term that will be searched for and hyperlinked.	Case sensitive. position longer superset acronyms earlier e.g. place https before http
wikipediaEntry	No	This attribute is only used for entries in Wikipedia where the page name is not the same as the attribute.  E.g. the entry in Wikipedia for Apache has a page name of Apache_HTTP_Server
description	No	This will be used as a tooltip when a hyperlink is created in Word. It will also be used in the acronym table if that feature is used,
url	No	This is an alternative URL if Wikipedia is not to be the source of reference.

Table 2 - entry element attributes

Focas.NET.wikipediaise.zip (21.4 KB)

---