Merged SBOMs contain duplicates

[alexthemark]

Hello, and thanks for the awesome CLI -- it's helping me merge up a bunch of SBOMs. However, I noticed that my SBOMs have duplicate components in them.

I think this would be fixed by the latest version of the dotnet library -- specifically, due to https://github.com/CycloneDX/cyclonedx-dotnet-library/pull/216.

Is there a plan to upgrade this library and make a new release?

Thanks!

[ataraxus]

I would also be interested in this behaviour and possible workarounds...

[andreas-hilti]

CycloneDX-cli version 0.25.0 should now contain the latest cyclonedx-dotnet-library (specifically version 6.0.0). Thus, the behavior w.r.t. duplicates should be improved.

[pepcots]

Is it solved? I'm still having duplicates with merge command...

[OscheibeSymbio]

I am running into an issue where the result of 2 merged SBOMs create a invalid result. (Error message: "Validation failed at line number 14 and position 7: There is a duplicate key sequence '[email protected]' for the 'http://cyclonedx.org/schema/bom/1.5:bom-ref' key or unique identity constraint.")

Here are the repro steps:

I have two input files "sbom1-metadata-component.xml" and "sbom2-anything.xml" from which I have removed as much information from as possible. Both of these files are still being reported as valid and the resulting error message is the same as the full SBOMs.

sbom1-metadata-component.xml

<?xml version="1.0" encoding="utf-8"?>
<bom xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:xsd="http://www.w3.org/2001/XMLSchema" serialNumber="urn:uuid:a1f4c1c1-d3b6-4db4-b061-41223a62cb53" version="1" xmlns="http://cyclonedx.org/schema/bom/1.5">
  <metadata>
	<component type="application" bom-ref="[email protected]">
      <name>will-cause-issues</name>
    </component>
  </metadata>
  <components>
	<component type="library" bom-ref="pkg:nuget/[email protected]">
      <name>NotRelevant</name>
    </component>
  </components>
</bom>

Note here the component within the metadata region. The NotRelevant component needs to be there so the result contains anything at all but the content of NotRelevant is not relevant.

sbom1-metadata-component.xml

<?xml version="1.0" encoding="utf-8"?>
<bom xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:xsd="http://www.w3.org/2001/XMLSchema" serialNumber="urn:uuid:a1f4c1c1-d3b6-4db4-b061-41223a62cb53" version="1" xmlns="http://cyclonedx.org/schema/bom/1.5">
</bom>

I just used an empty SBOM to execute the merge command with, which still caused the issue. My normal SBOM is much bigger.

Using the docker image to merge these two files creates a result that will not pass the "validate" command.

docker run -v ${PWD}:/sbom cyclonedx/cyclonedx-cli validate --input-file sbom/sbom1-metadata-component.xml
docker run -v ${PWD}:/sbom cyclonedx/cyclonedx-cli validate --input-file sbom/sbom2-anything.xml
docker run -v ${PWD}:/sbom cyclonedx/cyclonedx-cli merge --input-files sbom/sbom1-metadata-component.xml sbom/sbom2-anything.xml --output-file sbom/merged-sbom.xml
docker run -v ${PWD}:/sbom cyclonedx/cyclonedx-cli validate --input-file sbom/merged-sbom.xml --input-version v1_5

The result contains two component references. One within the metadata region and one within the components region containing the same bom-ref. Error message: "Validation failed at line number 14 and position 7: There is a duplicate key sequence '[email protected]' for the 'http://cyclonedx.org/schema/bom/1.5:bom-ref' key or unique identity constraint."

Is this behavior relevant to this issue here? This invalid SBOM prevents me from doing further analysis.

merged-sbom.xml

<?xml version="1.0" encoding="utf-8"?>
<bom xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:xsd="http://www.w3.org/2001/XMLSchema" serialNumber="urn:uuid:ad5e9f70-3191-4486-878c-00bb9019de92" version="1" xmlns="http://cyclonedx.org/schema/bom/1.5">
  <metadata>
    <component type="application" bom-ref="[email protected]">
      <name>will-cause-issues</name>
    </component>
  </metadata>
  <components>
    <component type="library" bom-ref="pkg:nuget/[email protected]">
      <name>NotRelevant</name>
    </component>
    <component type="application" bom-ref="[email protected]">
      <name>will-cause-issues</name>
    </component>
  </components>
</bom>

I have attached all files to this comment, but their content is the same as described above. sbom-repro.zip

Thank you for providing us with such a powerful open source tool and helping your users.

[andreas-hilti]

I think it is similar to this issue: https://github.com/CycloneDX/cyclonedx-cli/issues/364 in the sense that in the FlatMerge, it is not completely clear what should happen to the components in the metadata. The root cause is again here: https://github.com/CycloneDX/cyclonedx-dotnet-library/blob/57972c202d267366954599a948445196cedd0dda/src/CycloneDX.Utils/Merge.cs#L84-L88 together with this behavior here https://github.com/CycloneDX/cyclonedx-cli/blob/03b8019b24e847b6fdc91822eae2e9a220d525fa/src/cyclonedx/Commands/MergeCommand.cs#L92-L97 that if the metadata is not provided, the first non-null is used. Then the same component appears in the metadata and inside components.

[andreas-hilti]

As a workaround, you can for now specify the name using the "--name" option (or similar metadata).