Obsolete MIME parameter in file content types

[hannes-ucsc]

There are many projects with files whose content-type property is described in the metadata using the now obsolete dcp-type MIME parameter. The use of this parameter was a quirk of the DSS, a component of the first incarnation of the DCP, dating from 2017. The parameter serves no purpose anymore and only complicates the processing of the content type by users of HCA. Given that there is an ecclectic mix of files that have the parameter in their content type, and those that don't, the presence of the parameter in some content types will surprise many users. Most would not have the acuity to parse the content type with a standards-compliant MIME parser so as to remove the suffix continuing to process it.

In a staging area this issue manifests as a value of the content_type property of file descriptors property ending in ; dcp-type=data.

https://docs.google.com/spreadsheets/d/1YeE7vzTcGmK2Z2EW61C1jJmvJ3tHS_z4Xv8hnKvnoHA/edit?usp=sharing

[hannes-ucsc]

https://humancellatlas.slack.com/archives/C01360XN04S/p1748393982910559

[amnonkhen]

The ; dcp-type: data suffix is added to the content type of a file in the code of hca-util (see commit), the command line tool contributors use to upload the data files to ingest’s upload area in aws s3. The tool only adds this suffix to files that have the ContentType attribute already set. I don’t know whether it has been designed like this intentionally. Looking at the contents of the upload area bucket, I notice that some files have the ContentType attribute set whilst others do not. That would explain why some files exported from ingest have this suffix.