Cool, that helps.

This is specific to UTF-8 decoding, but I sort of wanted to have generic
bitwise operations. I was actually surprised those aren't in the base
operators but whatever.

Cheers,

E.
On 2/16/11 9:15 AM, "Chris Maloney" <voldrani@xxxxxxxxx> wrote:

> [Sorry, hit the "Send" button by mistake.]
>
> I thought of some more optimizations on the way to work this morning.
> First of all, you can simplify your expressions:
>
>   ....
>     <xsl:variable name="byte1Bit7"
>                   select="if ($byte1 > 127) then 1 else 0"
>                   as="xs:integer"/>
>     <xsl:variable name="byte1Bit6"
>                   select="if ($byte1 mod 128 > 63) then 1 else 0"
>                   as="xs:integer"/>
>     <xsl:variable name="byte1Bit5"
>                   select="if ($byte1 mod 64 > 31) then 1 else 0"
>                   as="xs:integer"/>
>     <xsl:variable name="byte1Bit4"
>                   select="if ($byte1 mod 32 > 15) then 1 else 0"
>                   as="xs:integer"/>
>     <xsl:variable name="byte1Bit3"
>                   select="if ($byte1 mod 16 > 7) then 1 else 0"
>                   as="xs:integer"/>
>     <xsl:variable name="byte1Bit2"
>                   select="if ($byte1 mod 8 > 3) then 1 else 0"
>                   as="xs:integer"/>
>     <xsl:variable name="byte1Bit1"
>                   select="if ($byte1 mod 4 > 1) then 1 else 0"
>                   as="xs:integer"/>
>     <xsl:variable name="byte1Bit0"
>                   select="$byte1 mod 2"
>                   as="xs:integer"/>
>     ...
>
> Then, it occurred to me that if you're doing bitwise-and specifically
> for utf-8 conversion, then you can optimize this quite a bit.  The
> ANDing needed by utf-8 conversion is never to a random bit pattern,
> but instead to fixed bit-masks, where the bits are always grouped
> together.  So, for example, you could make a specific function to AND
> with 0x3F:
>
>   <xsl:function name="relpath:bitwiseAnd3F" as="xs:integer">
>       <xsl:param name="arg" as="xs:integer"/>
>       <xsl:variable name='result'
>                     as='xs:integer'
>                     select="$arg mod 64"/>
>       <xsl:sequence select="$result"/>
>   </xsl:function>
>
> and its complement to AND with 0xC0:
>
>   <xsl:function name="relpath:bitwiseAndC0" as="xs:integer">
>       <xsl:param name="byte" as="xs:integer"/>
>       <xsl:variable name='result'
>                     as='xs:integer'
>                     select="($byte - $byte mod 64) / 64"/>
>       <xsl:sequence select="$result"/>
>   </xsl:function>
>
> Cheers!
>
>
> On Tue, Feb 15, 2011 at 5:41 PM, Eliot Kimber <ekimber@xxxxxxxxxxxx> wrote:
>> I have implemented UTF-8 URI decoding in XSLT 2. It required that I
>> implement left-shift and bitwise AND operations. Left-shift is easy but
>> bitwise AND is a little more involved. My solution is below, but I'm
>> wondering if there's a more elegant and/or efficient way to do this?
>>
>> My approach was based on info from Ken Holman's site that shows how to
>> calculate each bit of the byte. I do that and then sum the result of
>> multiplying the AND of each bit pair times the appropriate power of two.
>>
>>  <xsl:function name="relpath:bitwiseAnd" as="xs:integer">
>>    <xsl:param name="byte1" as="xs:integer"/>
>>    <xsl:param name="byte2" as="xs:integer"/>
>>
>>    <xsl:variable name="byte1Bit7" select="if (($byte1 mod 256) - ($byte1
>> mod 128) > 0) then 1 else 0" as="xs:integer"/>
>>    <xsl:variable name="byte1Bit6" select="if (($byte1 mod 128) - ($byte1
>> mod 64) > 0) then 1 else 0" as="xs:integer"/>
>>    <xsl:variable name="byte1Bit5" select="if (($byte1 mod 64)  - ($byte1
>> mod 32) > 0) then 1 else 0" as="xs:integer"/>
>>    <xsl:variable name="byte1Bit4" select="if (($byte1 mod 32)  - ($byte1
>> mod 16) > 0) then 1 else 0" as="xs:integer"/>
>>    <xsl:variable name="byte1Bit3" select="if (($byte1 mod 16)  - ($byte1
>> mod 8) > 0) then 1 else 0" as="xs:integer"/>
>>    <xsl:variable name="byte1Bit2" select="if (($byte1 mod 8)   - ($byte1
>> mod 4) > 0) then 1 else 0" as="xs:integer"/>
>>    <xsl:variable name="byte1Bit1" select="if (($byte1 mod 4)   - ($byte1
>> mod 2) > 0) then 1 else 0" as="xs:integer"/>
>>    <xsl:variable name="byte1Bit0" select="if (($byte1 mod 2)  > 0) then 1
>> else 0" as="xs:integer"/>
>>
>>    <xsl:variable name="byte2Bit7" select="if (($byte2 mod 256) - ($byte2
>> mod 128) > 0) then 1 else 0" as="xs:integer"/>
>>    <xsl:variable name="byte2Bit6" select="if (($byte2 mod 128) - ($byte2
>> mod 64) > 0) then 1 else 0" as="xs:integer"/>
>>    <xsl:variable name="byte2Bit5" select="if (($byte2 mod 64)  - ($byte2
>> mod 32) > 0) then 1 else 0" as="xs:integer"/>
>>    <xsl:variable name="byte2Bit4" select="if (($byte2 mod 32)  - ($byte2
>> mod 16) > 0) then 1 else 0" as="xs:integer"/>
>>    <xsl:variable name="byte2Bit3" select="if (($byte2 mod 16)  - ($byte2
>> mod 8) > 0) then 1 else 0" as="xs:integer"/>
>>    <xsl:variable name="byte2Bit2" select="if (($byte2 mod 8)   - ($byte2
>> mod 4) > 0) then 1 else 0" as="xs:integer"/>
>>    <xsl:variable name="byte2Bit1" select="if (($byte2 mod 4)   - ($byte2
>> mod 2) > 0) then 1 else 0" as="xs:integer"/>
>>    <xsl:variable name="byte2Bit0" select="if (($byte2 mod 2)  > 0) then 1
>> else 0" as="xs:integer"/>
>>
>>    <xsl:variable name="result" as="xs:integer"
>>      select="
>>      (128 * $byte2Bit7 * $byte1Bit7) +
>>      (64 * $byte2Bit6 * $byte1Bit6) +
>>      (32 * $byte2Bit5 * $byte1Bit5) +
>>      (16 * $byte2Bit4 * $byte1Bit4) +
>>      (8* $byte2Bit3 * $byte1Bit3) +
>>      (4 * $byte2Bit2 * $byte1Bit2) +
>>      (2 * $byte2Bit1 * $byte1Bit1) +
>>      (1 * $byte2Bit0 * $byte1Bit0)
>>      "
>>    />
>>    <xsl:sequence select="$result"/>
>>  </xsl:function>
>>
>> Cheers,
>>
>> E.
>> --
>> Eliot Kimber
>> Senior Solutions Architect
>> "Bringing Strategy, Content, and Technology Together"
>> Main: 512.554.9368
>> www.reallysi.com
>> www.rsuitecms.com
>

--
Eliot Kimber
Senior Solutions Architect
"Bringing Strategy, Content, and Technology Together"
Main: 512.554.9368
www.reallysi.com
www.rsuitecms.com