| read XML with IDL [message #71672] |
Wed, 14 July 2010 02:59  |
sh
Messages: 26
Registered: April 2010
|
Junior Member |
|
|
Hello together,
I already searched in the archive, but I didn't find a clear answer.
Every time you want to read .xml files, you have to write your own
parser (e.g. .edit xml_to_struct__define.pro) right?
But is there a possibilty to build up a struct automatically? E.g. the
input .xml file
<Solar_System>
<Planet NAME='Mercury'>
<Orbit UNITS='kilometers' TYPE='ulong64'>579100000</Orbit>
<Period UNITS='days' TYPE='float'>87.97</Period>
<Satellites TYPE='int'>0</Satellites>
</Planet>
...
</Solar_System>
output in IDL (structs within a "main" struct) the name of struct and
tags depends on the name of elements and the attributes
solar_system={solar_system,planet:{planet,orbit:
{orbit,units:'kilometers',type:'ulong64',value:'579100000'}, period:
{period,units:'days',type:'float',value:'87.97'},satellites:
{satellites,type:'int',value:'0'}}}
print, solar_system.planet.orbit
{ kilometers ulong64 579100000}
print, solar_system.planet.orbit.value
579100000
My questions are now: Should I write my own parser? Should I wait for
IDL8, maybe there are more adv. features for XML? Or does someone have
already such a routine?
thanks,
Sebastian
|
|
|
|
| Re: read XML with IDL [message #71746 is a reply to message #71672] |
Thu, 15 July 2010 10:13  |
penteado
Messages: 866
Registered: February 2018
|
Senior Member
Administrator |
|
|
On Jul 15, 1:57 pm, Karl <karl.w.schu...@gmail.com> wrote:
> You'd have to be careful here when defining the planet IDL structure.
> If you define it based on the first planet you encounter, you're going
> to have to redefine it later (big problem) or make your "schema
> discovery" code much more robust.
>
> I'm just trying to say that it isn't typical (I think) to write XML
> parsing code without integrating in some solid schema for the data.
> If you want to do it anyway, the XMLDOM class allows you the maximum
> flexibility in "discovering" your data on the fly.
Which is why it would be a lot easier with hashes instead of
structures, as the fields could be edited during the discovery
process. When the whole thing is done, if a structure is really
necessary, the final hash could be relatively easily converted to a
structure (how easily depends on which functions IDL 8 will have to
convert hashes to structures; this case would be a bit trickier due to
recursion).
|
|
|
|
| Re: read XML with IDL [message #71747 is a reply to message #71672] |
Thu, 15 July 2010 09:57 |
Karl[1]
Messages: 79
Registered: October 2005
|
Member |
|
|
On Jul 15, 8:05 am, Paulo Penteado <pp.pente...@gmail.com> wrote:
> On Jul 15, 5:41 am, sh <sebastian.h...@gmail.com> wrote:
>
>> But you have to overwrite some methods of the SAX parser to make use
>> of it for a certain xml file.
>> And I didn't want to write everytime a new xml_to_struct__define.pro!
>
>> Therefore I asked if there is a "general" solution to take the
>> elements/node/attributes names within a xml file to generate a
>> structur, like in my example above. Because in this case I have only
>> one parser who generates the output structure with a dependency on the
>> xml elements/node/attributes names and I can read every xml file!
>
> With IDL 8, it does not seem too difficult to write such a general
> parser to generate a hash. Or, for something more complicated to
> write, a class that inherited from the DOM class and presented a hash-
> like interface through overloaded [].
>
> Building a structure as the tree is parsed would always be awkward,
> and the result would be difficult to edit if it involved removing
> fields. But still there are situations where a structure would be more
> desirable as the end product, so that is one example of the need for a
> routine to create a structure from a hash, but I do not know if one
> will be provided.
One way to deal with an arbitrary XML file in IDL without doing
lexical-level parsing is to use IDLffXMLDOM. You can walk the entire
DOM tree, collecting elements and attributes, while maintaining
whatever relational information between them you need. When finished,
you can then generate the IDL structures to match the data
organization in the XML file. The important thing is that you don't
need to know the XML structure ahead of time in order to do this. I
think that it is possible to write a general routine for this that
would work for all XML input.
Just part of the algorithm would be:
Use XMLDOM methods to find all the element nodes. From this list find
all element nodes that do not have element nodes in them. These are
the "leaf" nodes. Then use IDL functions to make IDL structures for
each leaf element. The name of the struct is the name of the
element. Add element members for each attribute and the value of the
element.
For the non-leaf elements, you'll have to go back and build IDL
structs for these that contain structs for the leaf elements.
There are more details of course, but this looks like it could be
implemented fairly elegantly using recursion.
Anyway, all this strikes me as a bit odd. Most programs processing
XML input have a pretty good idea of the structure of the XML data
they are supposed to be reading. XML Schemas take this to the
extreme, where schema enforcement can be enabled to reject XML files
that do not conform to the schema. Finally, applications that read
XML data usually have some plan as to what they will do with the data
and won't know what to do with elements that they are not coded to
handle. But I guess that is the app's problem.
Also, processing an XML file in order to "learn" its schema is
problematic. You may not be able to discern the entire schema just by
reading one sample file. Taking the "planets" example, one might
suppose that a planet element may contain zero or more moon elements:
<planet name=Mercury>
</planet>
<planet name=Earth>
<moon name=Moon>
</moon>
</planet>
You'd have to be careful here when defining the planet IDL structure.
If you define it based on the first planet you encounter, you're going
to have to redefine it later (big problem) or make your "schema
discovery" code much more robust.
I'm just trying to say that it isn't typical (I think) to write XML
parsing code without integrating in some solid schema for the data.
If you want to do it anyway, the XMLDOM class allows you the maximum
flexibility in "discovering" your data on the fly.
|
|
|
|
| Re: read XML with IDL [message #71748 is a reply to message #71672] |
Thu, 15 July 2010 07:05 |
penteado
Messages: 866
Registered: February 2018
|
Senior Member
Administrator |
|
|
On Jul 15, 5:41 am, sh <sebastian.h...@gmail.com> wrote:
> But you have to overwrite some methods of the SAX parser to make use
> of it for a certain xml file.
> And I didn't want to write everytime a new xml_to_struct__define.pro!
>
> Therefore I asked if there is a "general" solution to take the
> elements/node/attributes names within a xml file to generate a
> structur, like in my example above. Because in this case I have only
> one parser who generates the output structure with a dependency on the
> xml elements/node/attributes names and I can read every xml file!
With IDL 8, it does not seem too difficult to write such a general
parser to generate a hash. Or, for something more complicated to
write, a class that inherited from the DOM class and presented a hash-
like interface through overloaded [].
Building a structure as the tree is parsed would always be awkward,
and the result would be difficult to edit if it involved removing
fields. But still there are situations where a structure would be more
desirable as the end product, so that is one example of the need for a
routine to create a structure from a hash, but I do not know if one
will be provided.
|
|
|
|
| Re: read XML with IDL [message #71750 is a reply to message #71672] |
Thu, 15 July 2010 01:41 |
sh
Messages: 26
Registered: April 2010
|
Junior Member |
|
|
On Jul 14, 9:16 pm, Karl <karl.w.schu...@gmail.com> wrote:
> On Jul 14, 7:58 am, kBob <krd...@gmail.com> wrote:
>
>
>
>
>
>> On Jul 14, 3:59 am, sh <sebastian.h...@gmail.com> wrote:
>
>>> Hello together,
>
>>> I already searched in the archive, but I didn't find a clear answer.
>
>>> Every time you want to read .xml files, you have to write your own
>>> parser (e.g. .edit xml_to_struct__define.pro) right?
>
>>> But is there a possibilty to build up a struct automatically? E.g. the
>>> input .xml file
>
>>> <Solar_System>
>>> <Planet NAME='Mercury'>
>>> <Orbit UNITS='kilometers' TYPE='ulong64'>579100000</Orbit>
>>> <Period UNITS='days' TYPE='float'>87.97</Period>
>>> <Satellites TYPE='int'>0</Satellites>
>>> </Planet>
>>> ...
>>> </Solar_System>
>
>>> output in IDL (structs within a "main" struct) the name of struct and
>>> tags depends on the name of elements and the attributes
>
>>> solar_system={solar_system,planet:{planet,orbit:
>>> {orbit,units:'kilometers',type:'ulong64',value:'579100000'}, period:
>>> {period,units:'days',type:'float',value:'87.97'},satellites:
>>> {satellites,type:'int',value:'0'}}}
>
>>> print, solar_system.planet.orbit
>>> { kilometers ulong64 579100000}
>
>>> print, solar_system.planet.orbit.value
>>> 579100000
>
>>> My questions are now: Should I write my own parser? Should I wait for
>>> IDL8, maybe there are more adv. features for XML? Or does someone have
>>> already such a routine?
>
>>> thanks,
>>> Sebastian
>
>> Have you looked into xml_to_struct__define.pro example provided with
>> the IDL distribution?
>
>> I would think you have, as ths Solar System XML is the example they
>> use to turn into a structure.
>
>> IDL Help provides a tutorial for this example code.
>
>> Kelly Dean
>> Fort Collins, CO
>
> IDLffXMLSAX
> and
> IDLffXMLDOM
>
> are two XML parsers that come with IDL. I think that the SAX parser
> would help here.
Yes I know these 2.
But you have to overwrite some methods of the SAX parser to make use
of it for a certain xml file.
And I didn't want to write everytime a new xml_to_struct__define.pro!
Therefore I asked if there is a "general" solution to take the
elements/node/attributes names within a xml file to generate a
structur, like in my example above. Because in this case I have only
one parser who generates the output structure with a dependency on the
xml elements/node/attributes names and I can read every xml file!
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
