Instead of using dfdl:lengthKind="pattern", I think it might be better
to use a discriminator that fails when all the values in an entry are
zero (whatever zero means for each of the values), followed by an
element to consume those zero bytes. So something along these lines:
<xs:element name="Import_Directory_Table">
<xs:complexType>
<xs:sequence>
<xs:element ref="Import_Directory_Entry" maxOccurs="unbounded" />
<xs:element name="Terminating_Bytes" type="xs:hexBinary"
dfdl:lengthKind="explicit" dfdl:length="20">
<annotation>
<appinfo source="http://www.ogf.org/dfdl/";>
<dfdl:assert>
{ . eq xs:hexBinary("0000000000000000000000000000000000000000") }
</dfdl:assert>
</appinfo>
</annotation>
</xs:element>
</xs:sequence>
</xs:complexType>
</xs:element>
<xs:element name="Import_Directory_Entry">
<annotation>
<appinfo source="http://www.ogf.org/dfdl/";>
<dfdl:discriminator>
{
(./Import_Lookup_Table_RVA ne 0) and
(./Date_Time_Stamp ne xs:date("1970-01-01T00:00:00")) and
(./Forwarder_Change ne 0) and
(./Name_RVA ne 0) and
(./Import_Address_Table_RVA ne 0)
}
</dfdl:discriminator>
</appinfo>
</annotation>
<xs:complexType>
<xs:sequence>
<!-- your elements here -->
</xs:sequence>
</xs:complexType>
</xs:element>
So this will parse an unbounded number of Import_Directory_Entry's until
either the discriminator fails. This discriminator is designed to fail
when all of the values are zero. Note that this requires some thought
about what "zero" means for each of the values. I'm not sure of the type
of addressType, but for the dateTimeStamp, a zero value is represented
as the unix epoch. The above assumes address Type is an int, but that
maybe need to be adjusted.
Eventually that discriminator will fail when the values are all zero and
so it will backtrack that last terminating entry. We then consume those
20 bytes and assert that they were all zero. It might be reasonable to
put the Terminating_Bytes element in a hidden group since it's
effectively a constant. This would leave you with just the non-zero
entries in the infoset.
- Steve
On 2/4/19 6:00 PM, Costello, Roger L. wrote:
> Hello DFDL community,
>
> My input file contains a bunch of Import Directory Entries. The last Entry is
> empty (filled with null values). I figured that the way to handle this is
> with a
> lengthPattern (see below), but that results in this error message:
>
> [error] Schema Definition Error: Element element reference
> {}Import_Directory_Entry does not meet the requirements of Pattern-Based
> lengths
> and Scanability.
>
> What is the right way to express this? /Roger
>
> <xs:elementname="Import_Directory_Table">
> <xs:complexType>
> <xs:sequence>
> <xs:elementref="Import_Directory_Entry"maxOccurs="unbounded"
> dfdl:lengthKind="pattern"dfdl:lengthUnits="bytes"
> dfdl:lengthPattern=".+?(?=((\x00){20,20}|$))"/>
> </xs:sequence>
> </xs:complexType>
> </xs:element>
>
> <xs:elementname="Import_Directory_Entry">
> <xs:complexType>
> <xs:sequence>
> <xs:elementname="Import_Lookup_Table_RVA"type="addressType"/>
> <xs:elementname="Date_Time_Stamp"type="xs:dateTime"dfdl:length="4"
>
> dfdl:lengthKind="explicit"dfdl:binaryCalendarRep="binarySeconds"
>
> dfdl:lengthUnits="bytes"dfdl:binaryCalendarEpoch="1970-01-01T00:00:00"/>
> <xs:elementname="Forwarder_Chain"type="unsignedint32"/>
> <xs:elementname="Name_RVA"type="addressType"/>
> <xs:elementname="Import_Address_Table_RVA"type="addressType"/>
> </xs:sequence>
> </xs:complexType>
> </xs:element>
>
