Re: Arabic alphabetical lists (xslt 1.0)

Note: I used floor($num-1 div string-length($arabletters)), not just the div.
-- Patricia


On 30 August 2013 13:45, Wolfgang Laun <wolfgang.laun@xxxxxxxxx> wrote:
> Here's a viable solution, as confirmed by Patricia. For XSLT 2.0:
> s/div/idiv/ (and use a function):
>
> <xsl:stylesheet version="1.0"
>     xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
>     xmlns:xsd="http://www.w3.org/2001/XMLSchema">
>
> <xsl:variable name="arabletters"
> select="'&#x0627;&#x0628;&#x062A;&#x062B;&#x062C;&#x062D;&#x062E;&#x062F;&#x0630;&#x0631;&#x0632;&#x0633;&#x0634;&#x0635;&#x0636;&#x0637;&#x0638;&#x0639;&#x063A;&#x0641;&#x0642;&#x0643;&#x0644;&#x0645;&#x0646;&#x0647;&#x0648;&#x064A;'"/>
>
> <xsl:template name="hijai-order">
>   <xsl:param name="number"/>
>   <xsl:variable name="num-1" select="$number - 1"/>
>   <xsl:variable name="unit" select="$num-1 mod string-length($arabletters)"/>
>   <!--xsl:value-of select="$unit"/-->
>   <xsl:variable name="tens" select="$num-1 div string-length($arabletters)"/>
>   <xsl:if test="$tens != 0">
>     <xsl:call-template name="hijai-order">
>        <xsl:with-param name="number" select="$tens"/>
>     </xsl:call-template>
>   </xsl:if>
>   <xsl:value-of select="substring($arabletters,$unit+1,1)"/>
> </xsl:template>
>
> <!-- test with a sequence of <item>s  -->
> <xsl:template match="row">
>   <xsl:variable name="pos" select="count(preceding-sibling::row)+1" />
>   <item>
>     <xsl:value-of><xsl:call-template name="hijai-order">
> <xsl:with-param name="number" select="$pos"/>
> </xsl:call-template>  <xsl:value-of select="$pos"/></xsl:value-of>
>   </item>
> </xsl:template>
>
> </xsl:stylesheet>
>
>
>
> On 28/08/2013, Wolfgang Laun <wolfgang.laun@xxxxxxxxx> wrote:
>> Another deviation in the list on ask.metafilter.com (as well as in the one
>> I
>> can produce with my KSLT 2.0 processor) is that it simply iterates a range
>> of Unicode code points; this appears to be the sort order used in
>> modern Arabic dictionaries. If Wikipedia is correct, this is based on
>> 28 Unicode letters from the codepage at U+0600, i.e.: 0627 - 0628,
>> 062A - 063A, 0641 - 0648, 064A. Note the gaps.
>>
>> -W
>>
>>
>> On 28/08/2013, Wolfgang Laun <wolfgang.laun@xxxxxxxxx> wrote:
>>> After a look at Unicode code table U+0600..U+06FF I suspect that XSLT
>>> processors rely on the underlying Java for producing the arabic
>>> alphabet in order to get a set of symbols for creating the "numbering"
>>> letter combinations.
>>>
>>> An XSLT 2.0 processor (Saxon) produces 36 single letter numbers, which
>>> isn't right either: it contains the teh marbutah and  U+063B through
>>> U+063F, and the tatweel, none of which is, I think, correct.
>>>
>>> I'm afraid you may have to roll your own, code something to create the
>>> correct sequence of letters from an integer. If you need help with
>>> this, there's others on this list better equipped to do it in XSLT
>>> 1.0.
>>>
>>> HTH
>>> -W
>>>
>>>
>>> On 28/08/2013, Patricia Piolon <skyanth@xxxxxxxxx> wrote:
>>>> Hi!
>>>>
>>>> I've been trying to ask a question here a number of times but my
>>>> message keeps getting bounced, presumably because of the special
>>>> characters?
>>>>
>>>> My question is about xslt-generated alphabetical lists with Arabic
>>>> characters, also posted here:
>>>> http://ask.metafilter.com/247029/Please-tell-me-about-Arabic-alphabetical-lists.
>>>> Could you please read the question in the link and reply here?
>>>>
>>>> Sorry for the convoluted way of asking this question; I'm not going to
>>>> spend any more time trying to find out why this listbot hates my mail
>>>> so much, so this is my last attempt.
>>>>
>>>> -- Patricia