Arrays become repeated tags, and the array-ness disappears

JSON arrays are a core data structure, but XML has no native equivalent. This converter handles an array by repeating the same element once per item — no wrapper element is introduced. So {"fields": ["math", "cs"]} produces two sibling tags:

<root>
  <fields>math</fields>
  <fields>cs</fields>
</root>

The consequence shows up at the edge case. A one-element array collapses to a single tag that is byte-identical to the same key holding a plain value — both {"fields": ["math"]} and {"fields": "math"} emit exactly <fields>math</fields>. Nothing in the XML records which one you started with, so a converter reading it back has to guess, and it will guess "plain value". This is the classic reason JSON → XML → JSON is not a safe round trip: the trip is lossless right up until a list happens to have one item in it, which is precisely the case your test data is least likely to cover.

Invalid key names silently become the tag "item"

XML element names must start with a letter or underscore and may then contain only letters, digits, underscores, dots and hyphens. A JSON key that breaks that rule — one starting with a digit, containing a space, or holding a character like @ — is replaced with the literal tag name item, and the original key is discarded. There is no warning. {"2024 total": 15000} becomes:

<root>
  <item>15000</item>
</root>

The label is simply gone, and nothing in the output hints that it ever existed. Keys of this shape are common in exported data — year-prefixed columns, keys with spaces from spreadsheet headers, @-prefixed metadata keys from other XML converters — so it is worth scanning your keys before converting rather than discovering the loss later. Renaming 2024 total to total_2024 first preserves the meaning.

No attributes, everything is an element

XML supports both elements and attributes; JSON has no concept of an attribute, so this converter never produces one. Every value becomes an element, and primitives become that element's text content, escaped for ampersand, less-than, greater-than and double-quote. Null values and empty objects both become self-closing tags — {"a": null, "b": {}} gives <a /> and <b />, another pair that the XML cannot tell apart afterwards.

The output always opens with the declaration <?xml version="1.0" encoding="UTF-8"?> and wraps everything in a single element literally named root, because XML permits exactly one root element and JSON supplies no name for it. Nesting is indented two spaces per level. Invalid input produces an error beginning Invalid JSON: followed by the parser's own message.

Both formats nest, so this conversion keeps the structure — unlike flattening either one into a table, where the nesting has nowhere to go. The XML to JSON converter goes back the other way.