<body>
<h3>SRT File
</h3>
<p>This file is used to specify the character sorting order and which
characters are the 'same' for the purposes of searching. All accented versions
of a character will have the same code.
</p>
<p>The first section has three byte records that have the following meaning
</p>
<table>
<tr>
<th>Byte number
</th>
<th>Description
</th>
</tr>
<tr>
<td>1
</td>
<td>The low nibble is 1 for letters, 2 for digits and zero for everything
else.
</td>
</tr>
<tr>
<td>2
</td>
<td>This is the sort order for the unaccented form of the character.
So a, à, á, ä and å would all have the same code here. The next byte allows
you to put them in order.
</td>
</tr>
<tr>
<td>3
</td>
<td>Upper nibble is 2 for uppercase letters, 1 for lowercase letters.
It can also have the values 3 or 4 for unknown reasons.
<p>The lower nibble is a number that can be added to byte 2 to get the full
sorting order. This sorts the accented versions of the characters
</p></td>
</tr>
</table>
<p>The next section has two byte records with an unknown function.
</p>
<h3>Unknowns
</h3>
At the time of writing Jan 2010 certain things are not known.
<ul>
<li>If it can deal with character pairs that sort as one.
</li>
<li>What the second section is for.
</li>
<li>Sorting for scandinavian languages where the accented characters come at
the end of the alphabet. Perhaps you just treat them as separate letters and
not as a base + accent, or perhaps there is a way of specifying it.
</li>
</ul>
</body>