Best way to convert a 4-byte, little endian integer to a symbolic value?

["Costello, Roger L." <co...@mitre.org>]

I have a 4-byte, little endian integer.

Each integer denotes a shape.

For example, the integer 31 denotes the shape MultiPatch.

So, if the input file contains this value (hex for integer 31):

	1F000000

Then I want the XML to be this:

	<shapetype>MultiPatch</shapetype>

Of course, this input (lowercase F):

	1f000000

should produce the same result:

	<shapetype>MultiPatch</shapetype>

What is the best way to express this in DFDL?

/Roger

Re: Best way to convert a 4-byte, little endian integer to a symbolic value?

[Steve Lawrence <sl...@apache.org>]

Right now, you'd need to do something like this:

<xs:element name="shapeTypeValue" type="xs:int"
dfdl:lengthKind="explicit" dfdl:length="4" dfdl:length="bytes"
dfdl:byteOrder="littleEndian" dfdl:representation="binary" />
<xs:element name="shapeType" type="xs:string" dfdl:inputValueCalc="{
if (../shapeTypeValue eq 31) then "MultiPatch"
else if (../shapeTypeValue eq 32) then "ShapeType2"
else if (../shapeTypeValue eq 33) then "ShapeType3"
else if (../shapeTypeValue eq 34) then "ShapeType4"
...
"}
/>

You might even want to put the shapeTypeValue in a group and reference
it with dfdl:hiddenGroupRef so the physical value isn't in the infoset,
in which case the shapeTypeValue element needs a dfdl:outputValueCalc
property for unparsing, which would look something like this:

dfdl:outputValueCalc="{
if (../shapeType eq "MultiPatch") then 31
else if (../shapeType eq "ShapeType2") then 32
else if (../shapeType eq "ShapeType3") then 33
else if (../shapeType eq "ShapeType4") then 34
...
}"

The main problem with this is is does not scale very well. If there are
a lot of shapetypes, this expression could be huge and slow. And
unfortunately, we see this pattern alot--so often in fact that Mike B.
has written a proposal to support this exact thing. The proposal is here:

https://cwiki.apache.org/confluence/display/DAFFODIL/Proposal%3A+Features+to+Support+Table-Lookup

It isn't implemented yet so if you take a look at it and you think it
won't meet your needs, let us know. We're definitely open to suggestions.

- Steve


On 05/30/2018 06:29 AM, Costello, Roger L. wrote:
> I have a 4-byte, little endian integer.
> 
> Each integer denotes a shape.
> 
> For example, the integer 31 denotes the shape MultiPatch.
> 
> So, if the input file contains this value (hex for integer 31):
> 
> 	1F000000
> 
> Then I want the XML to be this:
> 
> 	<shapetype>MultiPatch</shapetype>
> 
> Of course, this input (lowercase F):
> 
> 	1f000000
> 
> should produce the same result:
> 
> 	<shapetype>MultiPatch</shapetype>
> 
> What is the best way to express this in DFDL?
> 
> /Roger
>