[dasc] Persist Only Required Fields, Schema Defaults, and User Data

[kozmaadrian]

Summary

The form should save only schema dedined mandatory fields, schema defined defaults or user-entered data only.

---

General Functionality

The current implementation uses the JSON Schema + JSON data to dynamically generate the editor UI. There are two cases:

• Creating a new document: In this case the data is auto-generated with default empty values.
• Authoring an existing document: In this case the editor will get already saved data loaded.

A field can have a value from one of:

• Default from schema — Value from the schema default keyword
• User-entered — Value typed or selected by the user
• Auto-generated — Value produced by the data generator (e.g. empty string, empty array) for UI rendering

---

Problem

Currently the form stores the complete JSON data—not only the user-entered data but the auto-generated default values as well.

The generated values are used to ensure the UI can render the corresponding input with some default values. However, all these generated values are not needed to be saved all the time.

---

Expected Behavior

JSON Schema supports the option to mark a field as required. This feature adds the restriction that the property marked required must be present in the JSON data, but does not enforce anything about its value.

The editor should store the JSON data having (in priority order):

1. P1 — All data properties that are marked as required
2. P2 — Default values defined by the schema
3. P3 — If validation is defined in the schema (e.g. format, etc.) → only valid values are stored
4. P4 — User-entered data — but not auto-generated default values

---

Additional benefit: JSON to HTML Output

In this way, saving the JSON as HTML would also remove a lot of empty unnecessary blocks.

[mhaack]

To add on the current problem, we have 3! extra challenges here with "Currently the form stores the complete JSON data—not only the user-entered data but the auto-generated default values as well."

1. this leads to the JSON > HTML converting it to lots of empty tables, as the converter only sees the JSON data and not the schema
2. on JSON > HTML (without schema info) we can not distinguish if an empty filed, empty object, empty array should be there (because required by the schema) or should not be there (as it was there before but the used made the fields empty)
3. on reading a clean HTML, with no empty tables, no empty rows, the HTML > JSON can only reconstruct the part which is stored, hence if can lead to invalid JSON if required fields are missing.