PC Plus HelpDesk - issue 264

Paul Grosse This month, Paul Grosse gives you more insight into some of the topics dealt with in HelpDesk.

From the pages of HelpDesk, I look at:

  • Old virus infections;
  • World map clock;
  • Font types;
  • Global ad blocking;
  • Checking link status;
  • Non-ASCII websites;
  • Direct UTF-8 encoding; and,
  • Control your website.


Old virus infections

Anti-virus programs seem to spend a lot of time searching for viruses that are so old that they couldn't possibly still be around (could they?). There are thousands of them. You might, of course ask yourself if this is a waste of time?

Although there are many tens of thousands of viruses that are searched for that haven't seen light of day (or the inside of a computer) for years, there can always be a comeback.

Whilst we hear regular, although thankfully, infrequent reports of viruses infecting unlikely devices such as SatNav units, new hard drives and personal stereos, there was recently an infection of a batch of laptops for the German and Danish markets that were pre-installed with Microsoft Vista Home Premium and a copy of the 'Stoned.Angelina' virus.

The interesting thing about 'Stoned.Angelina' is that it is a boot sector virus that requires a DOS formatted floppy disk to propagate and, if that wasn't enough, it dates back to November 1994.

So, whilst it might seem like a waste of time, your scanner does need to know about historical viruses as well as new ones.

Stoned.Angelina might be pushing a Zimmer frame around but it still has life in it yet (but not as we know it).

World map clock

With a world map clock as your desktop wallpaper, you have a constant reminder of time passing by.

To get this, look for a program called 'kdeworld' by Matthias Hoelzer-Kleupfel. It is part of KDE and therefore should already be installed (or at least on the installation CDs/DVD).

You can run this as a windowed application by selecting the KDE menu, then 'Games', 'Toys', 'KWorldClock'. Right-clicking on the window will allow you to add times for particular places such as London for example.

If you want it as your desktop wallpaper, right-click on the desktop and select 'Configure Desktop...'.

Next, under 'Background', click on the 'Advanced Options' button and in the 'Background Program' frame, select 'kdeworld' from the list. Click on 'OK' and when you click on 'Apply', you will see it in all of its glory.

With this as your desktop wallpaper, there is no excuse for phoning people when they are in bed, although you can see that when I took the screenshot, I should have been in bed.

Font types

In Linux or BSD, all you need to do to install a font is to open up the Konqueror file browser and in the left pane, select the 'Services' tab and then expand the 'Fonts' line.

To install a font system-wide, click on the 'System' folder icon (right) and then drag your font from your desktop or another file browser and drop it there.

You will be prompted once for the root password and it will install. If you decide to install another (or a whole load of them by highlighting all you need and dropping them all in there), you do this within a certain time, you will not be prompted for the password again.

Linux and the BSDs accept a number of different types of fonts that include: Truetype (including UTF-8); PostScript Type 1; PCF Bitmap; OpenType; and, so on.

Global ad blocking

Increasingly, you might find yourself discovering a pop-up window when you close your browser. If this happens on many of the PCs on your home LAN, it is probably better to block it at the gateway using Squid and SquidGuard. You can do this by manually update the blacklists on an ad hoc basis.

For the uninitiated, SquidGuard is a web filter that works with the Squid web caching proxy. This runs on your gateway box, along with a firewall and filters out websites on the list. The lists include porn, drugs and so on.

You can get the latest lists from http://www.squidguard.org/blacklists.html but if you want to add to those on an ad hoc basis, all you need to do is to su to root, open /etc/squidguard.conf and look for the line that starts with 'dbhome'. Next, look for the 'DESTINATION CLASSES' line, below which you will find one for 'urllist'. Concatenate these two paths and open up a file browser there. Next, open up the 'urls' file and add any that you want, each on a separate line. If you need to block a whole domain, use the 'domains' file instead.

When you have finished, save the file and then, open a console and, as root again, restart Squid by typing...

/etc/init.d/squid restart

Now, your new configuration will run and the sites you have added will be blocked. One thing to consider is that there are plenty of proxy sites out there that are designed to preclude proxy filtering so, unless you include a fairly comprehensive list of these proxies, this is more about blocking nuisance ads than limiting adolescent browsing habits.

Checking link status

If you have a website, you probably need to check that all of the site's links are okay. Even though you might, every now and then, get some spam that offers to check your website for you, there are there Linux-based programs that can do the job for free.

One such tool that will do the job perfectly. 'KLinkStatus' can be found in the KDE menu at 'Development', 'Web Development', 'KLinkStatus'.

To run the program, type the full URL of your site into the address bar and click on the play button in the toolbar. There are options for checking directories recursively, and if so, to what depth. You can also tell it not to check parent directories if you have given it a URL that is not the document root address. Perhaps, more importantly, you can instruct it not to check external links.

In addition to this, you can also use regular expressions to specify files that you don't want checking so, say, for instance, that you had a gallery of photographs, that all contained the string 'sa403', you could use the regex '*sa403*'.

Any page that throws something other than a large green tick in the status column deserves extra attention. Note that you can check more than one site at a time - just open up a new tab.

Non-ASCII websites

It doesn't matter if you've been asked to build a website in Punjabi or want to do some other work in another non-Roman text, the issues here apply to just about any non-European language. So, just to illustrate a point (and for the reason that I happen to be a bit more familiar with it than other texts), we'll use Punjabi (aka Panjabi*) as an example.

The issues are:

  • inputting text; and,
  • doing so correctly.

First of all, even though Punjabi has its own text (Gurmukhi), you might ask why can't it be translated into Roman text any way? The answer has a number of parts:

  • Why should someone have to learn another type of writing just so that they can communicate over the Internet?
  • Roman text, with its 26 letters, covers English quite well and we don't have to resort to using any accents. However, when you look at it carefully, you start to realise that there are, for native English speakers, a lot of hard-wired fiddles built into it and in some areas, the written language covers many ambiguities. Examples could include the 'ough' in 'Rough', 'Through', 'Though', 'Thorough', and so on. You get the idea.
  • Even though the Roman alphabet can cover English, there are some areas of the way you position your mouth and tongue that are not covered in everyday English speaking. One area is where you position your tongue when you say the letter 'D'. Another is whether you aspirate explosive consonants ('c' becomes 'ch' as in the Scottish word 'loch' and so on). In Punjabi, there are four versions of the letter 'D'. Two tongue positions and with/without aspiration. This area is covered by just the letter 'D' in English but it can alter meaning in Punjabi.
  • Gurmukhi is better at differentiating between sounds in areas that matter to the language that it is is optimised for than Roman letters are.

So, whilst it is possible, it won't mean as much as it could.

Inputting text

This (dh-i-s) is rather difficult and time consumingThere are 35 main letters in the Gurmukhi alphabet (the top line in the screenshot). Add in another six, an extra nine vowel sounds and half a dozen or so other signs and you are getting the idea.

Inputting this correctly needs knowledge of the important differences between the more confusing glyphs - you don't want to end up typing the wrong character, simply because you didn't understand that two characters that looked very similar are actually quite different..

Knowing how to type it is another matter and for this, you need to have the purpose of your typing established.

#### fruit teas. Hand-picked from Yorkshire hedgerowsSome fonts produce good characters for typesetting - the 'AmrLipi', and 'Anmol' families of fonts which are free on the web. Using them to produce PDFs or graphics doesn't represent a problem.

In the image on the right, the fonts are used graphically so it doesn't really matter how they were produced. Whether they were hand-written, typed using a ASCII-mapped font or a UTF-8 font is of no consequence because that data is not passed onto the user - only the shapes of the letters (here, the letters' outlines were extended on an underlying layer and flooded with white, then a similar process with a smaller amount of black, and then they were drop-shadowed. The letters were then filled with a gradient made from light and dark samples from the image itself). This was all done on The GIMP.

The words are actually English and say '#### fruit teas. Hand picked from Yorkshire hedgerows.'

However, some of these fonts map into the normal ASCII character range so using them in web pages makes search engines see the body text as garbage.

For the Internet, you need to use a Unicode font. Saab ( http://guca.sourceforge.net/typography/fonts/saab/ ) is one such font and it is free. Just drop it into your fonts directory or use the font installer. On Windows and Linux, add another language to your keyboard (Input Languages) and when you activate it, the keyboard maps to the relevant utf-8 range.

Doing so correctly

Windows lets you use the Tablet PC Input Panel displaying the key mappings but using this is rather difficult and time consuming because the display is too small. On the right, you can see part of it at actual size.

You can input text using this interface by clicking on the keys. It works reasonably well apart from two things:

  • It is too small; and,
  • The mouse is too slow.
This is an area of it blown up three times. You can see that many of the characters are difficult to read.

Also, the area of the keys is quite small, making it difficult to do with a mouse. It is a bit better with a tracker ball but still slow.

There is also plenty of opportunity for confusion. On the right, in the red circle is Thatha, the letter than sounds like 'th'. You can compare this with 'Nana' ('N', inset) and see that there are only two pixels different (or, looking at it another way, one of the pixels has moved by one square).
Here is another example - one that is even more difficult to see. In the circle is Thatha (Th - yes, there are two, along with four characters that would be described in the Roman alphabet as 'D', two 'T's and so on - think tongue position) and inset is Baba (B). Again, only one pixel moved and it is not obvious if you don't know what to look for or, if your eyes aren't too perfect.

* The number five in Punjabi is 'Panj' - the Punjab gets its name from five rivers and, as you might have guessed, aab is Punjabi for water. Without an explicit vowel, the short 'u' and 'a' are virtually indistinguishable and unless you were listening out for them, you would think that it was a 'u'. Modern texts tend to have Panjab and Panjabi instead of the 'u' version of the spelling.

Direct UTF-8 encoding

ASCII codes render all right in browsers by default but if you want to display text using an alternative set of glyphs, you run into some problems.

The standard AnmoLipi, AmrLipi and GurbaniAkhar fonts map to ASCII codes used by the western alphabet and don't need any special coding but you do need the fonts installed specified in the browser's page otherwise, the browser will assume that it is just normal text and display it using Roman characters.

This is because the address space collides with other languages.

You can see this with the example on the right (the source will be protected).

In this case the machine that the browser was running on didn't have GurbaniWebThick on it so 'Thathaa' came out as an uppercase 'Q'.

Also, if your web pages need to be indexable by search engines and display correctly, you need to use UFT-8 encoding which gives the characters their own character space.

There are two ways of doing this:

  • Use HTML character codes to specify characters within specific Unicode ranges. These start with '&' and end with a semicolon - the Gurmukhi character 'Phapha' (ਫ) has the code 'ਫ' for example. However, for long texts, this is utterly impractical;
  • Type the characters into the web page directly. The Gurmukhi keyboard layout has all of the vowel sounds on the left hand similar to the Dvorak keyboard so typing is faster than you would think.

One problem with using typed characters (ie, they don't use the '&#nnnn;' codes) is that on some browsers, the title text is not represented in the intended text.

So, to solve this, you need to include the following line in the <head> section of the page...

<meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />
Now, you can type directly into the web pages, mix different languages and even get your page title to display properly.
And here it is in the title bar on a KDE-based UNIX system.
In addition, alt and title text display correctly.
If you click here, you can load a keyboard mapping page that has mixed English and Gurmukhi scripts - the latter being both in the direct and the escaped formats.

You can use that web page as a keyboard layout and, if you want to see it bigger, just make the browser text bigger. One thing you might find is that the Rara (J) has the other Rara (AltGr [ ) on the web page on some systems and Iri is a shift-B. This is because the phonetic layout is not entirely standardised yet.

All you have to do is to open up notepad or KWrite and press each key to check that the mapping is correct. Below the keyboard layout on that web page are the mappings for all of the characters in the Gurmukhi range, along with their HTML escape sequences so, if you want to modify the layout, you can use those sequences - all of the hard work has already been done for you by laying it out. If you need to change any, it should be no more than one or two characters.

An alternative is that you take a real keyboard and, using a 0.5mm - 0.7mm permanent marker (the type you would use to write on CDRs), write the characters on them yourself.

You can do this fairly easily...

  • Make sure that the keys map to the characters you want by pressing them in both the shift and the normal state (also the AltGr modifier key if you are going to use that as well);
  • Unplug the keyboard and then;
  • Hold down the key so that it isn't going to move and then write on it.
  • If you are in any doubt, check it again.

The keyboard in the photograph is used on a Windows and UNIX system so the Bindi and Tippi are reversed and there are the extra shift number patterns for the haha, wawa, (yaya ?) rara at the bottom and the adhikk. On Windows, the top row, Ikk, Doh, Tinn, Charr and so on are accessed with the [Ctrl] key modifier (the normal Roman numbers appear without this modifier key) whereas, on KDE, Ikk, Doh, Tinn ... to ... Atth, Naun and Sifar are accessible without modifiers (if you want the Roman versions, use the number pad instead).

Using a keyboard like this speeds up typing quite substantially and soon you can learn to touch-type Gurmukhi. That's something to put on your CV.

If you really need to make sure that you know the characters off by heart, you can always produce some flash cards similar to those in the photograph on the right so that you can practice identifying them out of sequence.

Control your website

If you are always updating your website, it might well become a bit of a style mess.

If this goes on for a while, you might decide that you want to redesign it all so that the main features of the site are consistent.

If you run Apache and (optionally) can write simple Perl scripts, it is quite easy - the page on the right is entirely the product of a Perl script that runs from a stub file called 'index.html'. In this case, the menu is across the bottom because the graph takes up the full width of smaller displays.

Anybody looking at the address line would think that it was running from the document root but in fact it is running from the cgi-bin.

In many cases, you don't have to write any Perl at all. All of this can make design changes and page updates easy.

For those who haven't done this, editing web pages is very easy when you have your own web server because you can edit files on the server - no uploading - just use KWrite (or NotePad if you are using Windows) or similar to edit your page, press [Ctrl][S] to save the file and then press [F5] on the browser to refresh the page. If you want to change the value of a variable - say a cellpadding from 2 to 3, you can do so very quickly. You can do this with plain html files just in a browser, without using a server but this is limited if you want to try out something a bit more advanced.

You can make life simpler still so that maintaining a site and keeping consistency is very easy. In fact, if you do it the right way, you can re-brand your site just by editing a few, strategically written files.

The obvious way of doing this is to use tables - Cascading Style Sheets (CSSs) aren't yet mature enough to be cross-browser /platform consistent and many messy pages with inconsistent rendering and poor content control appear from poor use of CSSs.

In the screenshot, you can see how this table-based page consists of a header (red), a menu down the left hand side (blue) and then the page content (green). Using SSIs allows you to create pages or page-creating scripts for the header and menu sections - you might even want to add a breadcrumb for more intricate sites. There is nothing stopping you from adding a categorised section down the right of the page as well as a footer. Simple, complex - the choice is yours.

The menu cell in this case is 'valign="top"' and 'align="center"' with the menu itself in a nested table within that. Doing this allows it to remain centred over the background image when changing text size in the browser.

The body content is kept a respectable distance from the menu simply by using '<blockquote>' tags to bracket the text. A layout like this will survive reasonable text and window re-sizing without breaking - you often see sites that are packed into an 800 or less pixel width block regardless of the text size or that of the browser (something that doesn't look too good on a monitor with 1280x1024 resolution or larger).

You can design your page like this, using whatever WYSIWYG designer or text editor you like and then, look inside the html code and see what is going to be consistent between pages and delegate the job of presenting that to some files.

By delegating the consistent web parts to other files in this way, a new page can be created from a simple, minimal 'template', re-using the web parts so that you don't have to re-invent the wheel each time you create a new page or have to edit all of the pages each time you want to change something that is common to them all.

You can see in the diagram on the right how the master design (1) has been divided up into parts (2) that need to be general and those that need to be specific.

Tables work best with designs like these because they hold everything in its place. They also have the flexibility to survive resizes (remember that if someone has poor eyesight, they might have a large monitor with a large font size and having a site taking up half the screen-width with an illegibly small font isn't going to work very well).

Tables are laid out in rows from top left to bottom right so here, the top left (pitcher leaf) would be part of a table that never changed. It might have a fragment of code just for a single cell, saved as a file in its own right. The code for the next cell (top right) would be in the page file (there is nothing stopping you from having all but the name in that cell as part of the top left file).

Next, you would have a file for the menu cell that runs down the left side. You could have this as static or, if you can write your own Perl, you can make it look at one of the environment variables and find out the name of the page it is serving. With that, you can make the link to the current page dead.

Finally, you have the body content. You can, of course, delegate whatever parts you want to any number of files and even get Perl to process file content such as colours or graphics according to what is requested on the browser's address line.

The index.html stub file you can see in the screen shot shows two of the types of server-side includes (SSIs) that you can use with Apache.
  • The first one, '#include file=' specifies a static page fragment to paste into the page; and,
  • The second, '#exec cgi=' specifies a CGI script which can be any type of executable (binaries or scripts). The output of the CGI script is included in the same way as the include file above, replacing all of the include tag so that the end user would never know that anything like this had occurred.

Note that these page fragments can include other page fragments.

Now that we know how to include web part files, we can build up a given page that has all of the consistency saved elsewhere, thus allowing us to keep the consistent content to a minimum (this reduces the chance of messing it up as well - something that is important).

So, this is what your mypage.shtml file might look like...

    <!--#include file="heading.htf" -->
    <title>Paul's Home Page</title>
  <body topmargin=0 leftmargin=0>
    <!--#include file="topleft.htf" -->
    <p align="center"><font size="+2"
       color="#7f7fcf"><b>Paul's Home
       Page.</b></font><br><font size="-1"><i>Talk
       about food and odd-looking carnivorous
    <!--#include file="menuleft.htf" -->
    <!-- body content starts here -->

    <!-- body content ends here -->
    <!--#include file="footer.htf" -->

...where 'heading.htf', 'topleft.htf', 'menuleft.htf' and 'footer.htf' are static html fragments (note, you don't need to have any knowledge of Perl or anything else to write static web parts) that make up the missing bits. These fragments don't have to have whole numbers of cells in them, you can see that the footer.htf file would have the end of the table definition (ie, </table> tag). By saving your page like this, you end up with consistently laid out pages that you can tweak to your heart's content.

If you decide that you want to add another item to the menu (or take one out), you just have to edit 'menuleft.htf' and all of the pages will be affected in the same way. The pages don't have to be called '*.htf', you can have any file name you like (or put them anywhere you like within the document root's file structure).

If you want to make your menu (or any other part) interactive with the page that is loading it (for example, you want the menus to change for a given page), you can take out that file and turn it into a cgi script (best done in Perl) and change the line in the page from...

<!--#include file="menuleft.htf" -->


<!--#exec cgi="/cgi-bin/menuleft.cgi" -->

Note that you are taking the static data file out of the document root, turning it into an executable and putting it into the cgi-bin directory. This is purely down to common sense.

You should never put executable files in the document root because if somebody managed to place a file there and it was allowed to execute, it could do serious damage. All sensibly configured web servers do not allow executables to run outside the cgi-bin directory.

Note that you cannot download the contents of the cgi-bin, you can only retrieve the output of the programs in there when they are run.

So, how do we get our web pages to have their contents examined by the server and any directives acted upon, instead of merely copied to the client machine?

SSIs need the web server to know when to look inside an HTML file to see if there is anything to process.

  • On UNIX servers, the executable bit can be used ('XBit hack') and the file extensions can all be the same (this allows you the flexibility of being able to change the status of a web page without having to change its extension and therefore every link that leads to it).
  • On Windows, you need to use a different file extension - '.shtml' is normally used although it is completely configurable on Apache.

With your page-top, breadcrumb, menu and any other bits sorted out and stored as separate files, you can fill your page's stub file with the content from your old site knowing that if you want to change a menu, you can do so with complete consistency.

Back to PC Plus Archive Index Page