Back

XML to Crystal

Generate Crystal classes/structs automatically from an XML document.

100% Client-Side Local Execution
Class Name:
Input Data
1 2 3 4 5
Transformed Result
1 2 3 4 5 6 7 8 9

What is XML to Crystal?

This free online tool turns a sample XML document into a ready-to-use class with JSON::Serializable for Crystal, so you can use this xml to crystal struct generator tool to go straight from raw XML to a typed class without writing the boilerplate yourself. It is built around Crystal's actual idiom — a Crystal class — rather than a generic one-size-fits-all class, and pairs naturally with XML.parse for real-world (de)serialization. Use it for convert xml to crystal class or any time you need xml to crystal lang struct, whether you're scaffolding a new Crystal project or mapping an existing API response.

Technical Execution

Paste or upload a sample XML document and the tool walks its structure field by field, inferring a concrete Crystal type for each value — mapping basic XML types to `String`, `Int32`, `Bool`, and `Float64` as appropriate. For nested data, it maps nested XML elements to nested Crystal classes and repeated elements to `Array(T)`, since Crystal's standard library parses XML as a DOM tree rather than deserializing directly into typed objects. On the naming side, it keeps snake_case keys as-is to match Crystal's idiomatic naming, and adds a `@[JSON::Field(key: "originalKey")]` annotation when a key doesn't already follow Crystal convention. For optional data, fields that appear as `null`, or are missing from some sample objects, are typed as a nilable union (`String?`, `Int32?`) so Crystal's static type checker allows them to hold `nil`.

Zero Data Storage Guarantee

Everything runs client-side in your browser: the sample XML you paste in — which may contain real field names, endpoint shapes, or structure from a private API or project — is parsed and converted entirely on your device and is never uploaded to any server. Nothing you enter is logged, stored, or transmitted anywhere, so you can safely paste a real payload from an internal or unreleased API to generate an accurate Crystal class. Closing or refreshing the tab clears everything, since no copy of your data ever left your machine in the first place.

How to Use XML to Crystal (Step-by-Step Guide)

  1. Copy a representative sample of your XML (the more fields and nested objects it includes, the more complete the generated class will be).
  2. Paste it into the input panel of the XML to Crystal tool, or click "Upload" to load it from a xml file.
  3. Review the auto-detected root name, and rename it if you want the generated Crystal class to use a specific class or type name instead of the default.
  4. Click "Generate" to produce the Crystal class, which appears instantly in the output panel with syntax highlighting.
  5. Click "Copy" to copy the generated code to your clipboard, or "Download" to save it as a `.cr` file.
  6. Paste the result into your Crystal project and adjust field names or types if your data has edge cases the sample didn't cover.

Frequently Asked Questions (FAQ)

Q: How does this tool handle nested objects and arrays when converting XML to Crystal?
A: The tool maps nested XML elements to nested Crystal classes and repeated elements to `Array(T)`, since Crystal's standard library parses XML as a DOM tree rather than deserializing directly into typed objects That keeps the generated code organized the way a Crystal developer would structure it by hand, instead of flattening everything into one giant class.
Q: What happens to fields that are nullable or missing from some records?
A: For optional data, fields that appear as `null`, or are missing from some sample objects, are typed as a nilable union (`String?`, `Int32?`) so Crystal's static type checker allows them to hold `nil`. If your sample only has one example object, fields that could be null in other records might not be detected as optional — paste a few varied samples (or an array of objects) for the most accurate result.
Q: Is this better than writing the class by hand?
A: For anything beyond a trivial, flat payload, yes — writing Crystal's `JSON::Serializable` classes by hand for nested JSON means working out every nilable union type yourself, since Crystal's compiler will reject a type declaration that's wrong at compile time. The generator produces the same result in seconds and gives you a correct starting point to refine, rather than a blank page.
Q: Is my XML data safe to paste into this tool, especially if it's from a private API?
A: Yes. Everything runs client-side in your browser: the sample XML you paste in — which may contain real field names, endpoint shapes, or structure from a private API or project — is parsed and converted entirely on your device and is never uploaded to any server. Nothing you enter is logged, stored, or transmitted anywhere, so you can safely paste a real payload from an internal or unreleased API to generate an accurate Crystal class. Closing or refreshing the tab clears everything, since no copy of your data ever left your machine in the first place.
Q: Can I use this as a xml to crystal lang struct?
A: Yes — the tool is commonly used as a xml to crystal lang struct, and works the same way whether you're converting a one-off response for a prototype or generating model code for a production project.
Q: Does the tool rename fields to match Crystal naming conventions?
A: Yes. It keeps snake_case keys as-is to match Crystal's idiomatic naming, and adds a `@[JSON::Field(key: "originalKey")]` annotation when a key doesn't already follow Crystal convention, so the generated code reads naturally in Crystal while still deserializing the original XML correctly.
Q: What Crystal library does the generated code work with?
A: The output is written to pair cleanly with XML.parse, the most common choice for Crystal developers handling XML, so you can drop it into an existing project without changing your serialization setup.
Q: Does this tool work for deeply nested or large XML samples?
A: Yes. The tool recurses through every level of nesting it finds and generates a complete set of linked class definitions, not just the top level — though for very large samples, trimming it down to one representative record per array is usually enough to get an accurate result.
Report an issue with this tool