How come unparsing invalid XML doesn't produce an error?

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

Hello DFDL community,

This is a follow-up to the thread that Christofer started last week about "fixed" values.

My input file consists of integers, on different lines. The first value in the file must be 44.

Using --Validate limited, I get an error message when the first value is not 44.

If the XML document is invalid (the first element does not have a value of 44), then I want unparsing to generate an error. I thought that Steve said the way to achieve this is to use:

dfdl:outputValueCalc="{ 44 }"

However, that does not result in an error. Why not?

Here is my DFDL schema:

<xs:element name="input">
    <xs:complexType>
        <xs:sequence dfdl:separator="%NL;" dfdl:separatorPosition="infix">
            <xs:element name="Fixed-Value" dfdl:occursCountKind="implicit"
                                  dfdl:outputValueCalc="{ 44 }">
                <xs:simpleType>
                    <xs:restriction base="xs:byte">
                        <xs:minInclusive value="44" />
                        <xs:maxInclusive value="44" />
                    </xs:restriction>
                </xs:simpleType>
            </xs:element>
            <xs:element name="num" type="xs:string" maxOccurs="unbounded"
                                  dfdl:occursCountKind="implicit" />
        </xs:sequence>
    </xs:complexType>
</xs:element>

Re: How come unparsing invalid XML doesn't produce an error?

["Sloane, Brandon" <bs...@tresys.com>]

dfdl:outputValueCalc is not used to validate the value in the input infoset, but rather to replace the value in the input infoset.


The intent of dfdl:outputValueCalc is for things such as length fields, where editing the infoset could change the desired value in a way that we want to be able to automatically compute during unparse. In such a use case, it would be undesirable to require that the value in the infoset matches the result of dfdl:outputValueCalc, as that would require manually computing the length field before unparsing, which is what we want to avoid.


Unfortunately, this means that the original value of the field is generally not available during unparse. In particular, during unparse, validation occurs on the augmented infoset, which is constructed by (among other things) replacing the logical values with the result of dfdl:outputValueCalc.


If you want to verify the contents of a field with dfdl:ouputValueCalc during unparse, you need to do so prior to running daffodil (eg. through normal xsd schema validation).


For fixed elements, such as in your example, an alternative would be to make the element hidden. That way there is no opportunity for it to become edited.


If you want to do the validation in Daffodil and keep the element visible, that is probably possible by defining it as a hidden field, and creating a second, visible, field that is populated using dfdl:inputValueCalc.



________________________________
From: Costello, Roger L. <co...@mitre.org>
Sent: Monday, May 13, 2019 9:57:37 AM
To: users@daffodil.apache.org
Subject: How come unparsing invalid XML doesn't produce an error?


Hello DFDL community,



This is a follow-up to the thread that Christofer started last week about “fixed” values.



My input file consists of integers, on different lines. The first value in the file must be 44.



Using --Validate limited, I get an error message when the first value is not 44.



If the XML document is invalid (the first element does not have a value of 44), then I want unparsing to generate an error. I thought that Steve said the way to achieve this is to use:



dfdl:outputValueCalc="{ 44 }"



However, that does not result in an error. Why not?



Here is my DFDL schema:



<xs:element name="input">
    <xs:complexType>
        <xs:sequence dfdl:separator="%NL;" dfdl:separatorPosition="infix">
            <xs:element name="Fixed-Value" dfdl:occursCountKind="implicit"
                                  dfdl:outputValueCalc="{ 44 }">
                <xs:simpleType>
                    <xs:restriction base="xs:byte">
                        <xs:minInclusive value="44" />
                        <xs:maxInclusive value="44" />
                    </xs:restriction>
                </xs:simpleType>
            </xs:element>
            <xs:element name="num" type="xs:string" maxOccurs="unbounded"
                                  dfdl:occursCountKind="implicit" />
        </xs:sequence>
    </xs:complexType>
</xs:element>