Schema integrity check is not performed

[OlivierCoen]

Hi!

When running a nf-core pipeline synced with nf-core/tools 3.4.1, I encountered the following issue:

...
WARN: The following invalid input values have been detected:

* --param1: value1
* --param2: value2
* .... any many other params (but not all of them)

although all these params were in the schema and their value had the proper type / pattern.

@awgymer made me notice that the reason was that nextflow_schema.json was malformed (it contained an extra bracked somewhere).

It raises a more general issue being that the check of the schema integrity is somehow too permissive.

Way to reproduce the error:

• Create a new pipeline with nf-core/tools 3.4.1
• Add an extra bracket somewhere in nextflow_schema.json
• Run nextflow run main.nf -profile test

Current output:

WARN: The following invalid input values have been detected:

* --config_profile_description: Minimal test dataset to check pipeline function
* --validate_params: true
* --custom_config_base: https://raw.githubusercontent.com/nf-core/configs/master
* --custom_config_version: master
* --fasta: s3://ngi-igenomes/igenomes//Saccharomyces_cerevisiae/Ensembl/R64-1-1/Sequence/WholeGenomeFasta/genome.fa
* --publish_dir_mode: copy
* --pipelines_testdata_base_path: https://raw.githubusercontent.com/nf-core/test-datasets/
* --input: https://raw.githubusercontent.com/nf-core/test-datasets/viralrecon/samplesheet/samplesheet_test_illumina_amplicon.csv
* --genome: R64-1-1
* --igenomes_base: s3://ngi-igenomes/igenomes/
* --config_profile_name: Test profile
* --trace_report_suffix: 2025-11-08_13-57-17
* --max_multiqc_email_size: 25.MB

Expected result:
The schema parser should report an error and quit.

Thanks a lot!

[awgymer]

Was there any additional output in the .nextflow.log from nf-schema for this run at all?

[awgymer]

A little bit of playing around suggests to me that this is an issue on two levels:

• nf-schema seems to parse the schema fine when doing the check for the schema version. This happens using schema = new JSONObject(schemaString) so it seems org.json.JSONObject is permissive here.
• jsonschema library underlying nf-schema also does not raise an error at all here and you can see my issue on that project here

[nvnieuwk]

Thanks for investigating @awgymer!

[awgymer]

See response from the linked issue (harrel56/json-schema#336):

JSON parsing behavior is fully dependent on the specific provider you use. Some of them provide means of configuring parsing process. In the provided example you use jackson (default) as the provider and you can pass your own ObjectMapper configuration to JacksonNode.Factory.

I think you can achieve your "strict" behavior by adding this configuration:

var mapper = new ObjectMapper().configure(DeserializationFeature.FAIL_ON_TRAILING_TOKENS, true);
var factory = new ValidatorFactory().withJsonNodeFactory(new JacksonNode.Factory(mapper));

[OlivierCoen]

Was there any additional output in the .nextflow.log from nf-schema for this run at all?

Hi! No, I did not see any additional log in .nextflow.log