XOM API for Processing XML with Java

[Michael Kay <mike AT saxonica.com>]

There is of course no universal definition of what constitutes "lexical" 
information.

However, the use cases for retaining line/column information are perhaps 
stronger than for the things like CDATA sections and entity references that 
we often classify under this heading. Almost anything that can be described 
as "validation" - whether it's XSD-like validation, or semantic validation 
such as an XSLT processor performs on XSLT input, needs to give the user 
feedback in terms of line/column locations, and a tree model that preserves 
line and column information is almost essential for such applications.

Michael Kay
Saxonica


> On 31 May 2016, at 19:31, Elliotte Rusty Harold <elharo AT ibiblio.org> wrote:
> 
> No, Xerces/SAX throws away a lot of lexical information; and XOM throws
> away more. General purpose parsers are not adequate for the rare use cases
> such as editors that actually need to preserve the lexical information.
> 
> On Tue, May 31, 2016 at 8:04 AM, <nu.xom AT io7m.com> wrote:
> 
>> Hello.
>> 
>> I'm writing a program that accepts a XOM-produced AST and performs
>> various operations on it.
>> 
>> There are various properties of the AST that cannot be expressed by a
>> RELAX-NG or XSD schema, so there are inevitably errors that need to be
>> raised when the AST is invalid in some way. In order to allow those
>> errors to be acceptable to users, I need to be able to show them
>> line and column numbers indication the locations of the errors.
>> 
>> The problem: XOM appears to discard all lexical information (if any is
>> even provided by the underlying parser). Is there any way I can arrange
>> for lexical information to be preserved in any way? I'm open to any
>> suggestion whatsoever (even populating an IdentityHashMap that maps XOM
>> produced nodes to lexical information would be fine).
>> 
>> I would assume that the solution would first involve getting SAX or
>> whatever is doing the parsing to supply that information (which it
>> presumably does at some point, because many of the exception types have
>> line and column information). I would then somehow need to intercept
>> that information before XOM consumes it, and then somehow associate the
>> produced XOM nodes with the information I intercepted.
>> 
>> M
>> _______________________________________________
>> XOM-interest mailing list
>> XOM-interest AT lists.ibiblio.org
>> http://lists.ibiblio.org/mailman/listinfo/xom-interest
>> 
> 
> 
> 
> -- 
> Elliotte Rusty Harold
> elharo AT ibiblio.org
> _______________________________________________
> XOM-interest mailing list
> XOM-interest AT lists.ibiblio.org
> http://lists.ibiblio.org/mailman/listinfo/xom-interest