Metadata Updates in Dataverse

gistfile1.md

Purpose

In preparing to add File Level metadata to Dataverse these are notes on several places where experience has shown that the API could be improved.

Adding an Email value to a Dataset's Metadata

Expected code (java)

newDatasetVersion.addEmailContact("[email protected]");

Current code (Dataverse)

• high barrier of entry to collaborators
• time-consuming
• hard to maintain

 DatasetFieldType emailDatasetFieldType = datasetFieldService.findByNameOpt(DatasetFieldConstant.datasetContact);
    DatasetField emailDatasetField = DatasetField.createNewEmptyDatasetField(emailDatasetFieldType, datasetVersion);
    for (DatasetField childField : emailDatasetField.getDatasetFieldCompoundValues().get(0).getChildDatasetFields()) {
        if (DatasetFieldConstant.datasetContactEmail.equals(childField.getDatasetFieldType().getName())) {
            childField.getSingleValue().setValue(user.getDisplayInfo().getEmailAddress());
        }
    }
    datasetVersion.getDatasetFields().add(emailDatasetField);

Current Native API: Adding JSON via Dataverse

This example shows the JSON required for adding two authors and their affilations

Expected JSON (not Dataverse)

{
"author": [
    {
      "authorName": "Smith, Robert",
      "authorAffiliation": "The Smiths",
    },
    {
      "authorName": "Kew, Susie",
      "authorAffiliation": "Creedence Clearwater Revival",
    }
  ]
 }

This is the JSON expected by Dataverse

This is what the JSON looks like for adding two authors and their affiliations.

No one uses this because of the complexity

• Note: This was written as a favor to the Dataverse team to help them save test data when the DB was dropped

{
            "typeName": "author",
            "multiple": true,
            "typeClass": "compound",
            "value": [
              {
                "authorName": {
                  "typeName": "authorName",
                  "multiple": false,
                  "typeClass": "primitive",
                  "value": "Smith, Robert"
                },
                "authorAffiliation": {
                  "typeName": "authorAffiliation",
                  "multiple": false,
                  "typeClass": "primitive",
                  "value": "The Smiths"
                }
              },
              {
                "authorName": {
                  "typeName": "authorName",
                  "multiple": false,
                  "typeClass": "primitive",
                  "value": "Kew, Susie"
                },
                "authorAffiliation": {
                  "typeName": "authorAffiliation",
                  "multiple": false,
                  "typeClass": "primitive",
                  "value": "Creedence Clearwater Revival"
                }
              }]

References:

• sample data:https://github.com/IQSS/dataverse/blob/develop/scripts/api/data/dataset-create-new.json)
• parsing program for JSON -> Relational DB: https://github.com/IQSS/dataverse/blob/28758208bef3fb713c903450067a47c0814655b1/src/main/java/edu/harvard/iq/dataverse/util/json/JsonParser.java

Native API: JSON output

Java code to create JSON output

• reference: https://github.com/IQSS/dataverse/blob/master/src/main/java/edu/harvard/iq/dataverse/util/json/JsonPrinter.java

• Not so bad for metadata. But if stored as JSON, this wouldn't be necessary

• Also, the output is confusing--more data than user would want--it's the same as this JSON:

  • https://github.com/IQSS/dataverse/blob/develop/scripts/api/data/dataset-create-new.json