Re: XTRE1160: The fragment identifier {62;n} is not a

Actually, in the case of a data URI you can know the media type, by
parsing the URI, if you want to.

2008/11/19 Michael Kay <mike@xxxxxxxxxxxx>:
> Saxon isn't 100% conformant in its handling of fragment identifiers, though
> it's a pretty fuzzy area and I'm not sure it really affects the outcome in
> this case. The reality is that when URIs are processed using the JAXP
> URIResolver concept, Saxon has no knowledge of a media type, so it cannot
> interpret a fragment identifier in a way that depends on the media type, and
> it therefore treats it as an NCName irrespective of the actual media type.
>
> But the root cause of the problem is that the "#" in the URI, if it is to be
> interpreted as an ordinary character, should have been escaped as %23.
> Except that in this case, the error is deeper, in that "&amp;#62;" should
> have been "&#62;".
>
> Michael Kay
> http://www.saxonica.com/
>
>> -----Original Message-----
>> From: Andrew Welch [mailto:andrew.j.welch@xxxxxxxxx]
>> Sent: 19 November 2008 11:54
>> To: xsl-list@xxxxxxxxxxxxxxxxxxxxxx
>> Subject:  XTRE1160: The fragment identifier {62;n} is
>> not a valid NCName
>>
>> Possibly Saxon specific issue here, or maybe something related to the
>> data: scheme:
>>
>> I'm calling doc() passing in content from an RSS feed which
>> attempts to be tidied using tagsoup, but this particular value:
>>
>> <xsl:value-of select="doc('data:,More Th&amp;#62;n')"/>
>>
>> causes this exception:
>>
>> Error at xsl:value-of on line 30 of rss-to-xhtml-text-plus-image.xsl:
>>   XTRE1160: The fragment identifier {62;n} is not a valid NCName
>>
>> net.sf.saxon.trans.XPathException: The fragment identifier
>> {62;n} is not a valid NCName
>>         at net.sf.saxon.functions.Document.makeDoc(Document.java:257)
>>         at net.sf.saxon.functions.Doc.doc(Doc.java:155)
>>         at net.sf.saxon.functions.Doc.evaluateItem(Doc.java:129)
>>
>> it doesn't even make it to the URIResolver
>>
>> The value is "More Th>n" double escaped - are there any
>> restrictions on the "data" scheme, or should this be fine?
>> Shouldn't it at least make it to the custom URIResolver?
>>
>> A correct, single-escaped value works fine:
>>
>> <xsl:value-of select="doc('data:,More Th&#62;n')"/>
>>
>> Running that my debug output shows what goes in and out of tagsoup:
>>
>> ** in uri resolver: data:,More Th>n
>> More Th&gt;n
>>
>>
>>
>> Any ideas?
>>
>>
>> --
>> Andrew Welch
>> http://andrewjwelch.com
>> Kernow: http://kernowforsaxon.sf.net/