XML Formatter

How to format XML

1. Paste one XML document or open one UTF-8 file up to 2 MiB. The tool supports XML 1.0 without a DOCTYPE; it does not open URLs or fetch schemas.
2. Choose Pretty with two spaces, four spaces or tabs, or deliberately choose Compact. Read the whitespace warning, then run the selected action.
3. Review the current checked output and any preview-truncation notice. Copy XML or download the complete UTF-8 file; edits and setting changes retire the previous result.

What the formatter checks and changes

Well-formedness checks document structure, legal characters and namespace bindings, including attributes that resolve to the same expanded name. It does not establish that an XML document follows a schema or means what an application expects.

Pretty replaces direct whitespace around child elements only in eligible containers. Compact removes the same whitespace without adding indentation. Without a schema, that whitespace cannot generally be proved insignificant, so neither operation promises semantic equivalence.

Subtrees rooted at elements with direct mixed text, direct CDATA or xml:space="preserve" keep their text layout, even across a descendant default reset. Whitespace-only leaves are retained. Comments, processing instructions, namespace declarations and child order are not discarded.

Output always uses an XML 1.0 UTF-8 declaration, no BOM and a final LF. Quote style, references, empty tags, line endings and processing-instruction separators can change. Escaping preserves parsed attribute controls and text carriage returns rather than accidentally changing them on the next parse.

Examples worth checking

The document <r> <a/> </r> gains line breaks and indentation in Pretty; Compact writes <r><a/></r> after the normalized declaration. Those layout changes alter character data even when the elements look unchanged.

In <p>Hello <b>world</b> !</p>, the spaces belong to mixed text. Its entire subtree keeps that text layout instead of splitting the sentence across new lines.

The attributes a:id and b:id conflict when both prefixes name the same namespace. An unprefixed id, however, is not put into an element's default namespace. Changing a prefix's spelling alone does not resolve an expanded-name duplicate.

Important limits

Only XML 1.0 and UTF-8 file input are supported. XML 1.1, other encoding declarations, every DOCTYPE and custom entities are rejected. Namespace URIs and stylesheet instructions are never opened.

This is not DTD or XSD validation, canonical XML, a signature checker, an XML-to-JSON converter or a repair tool. Do not substitute its normalized output for an original whose exact bytes or signature must be retained.

Input is limited to 2 MiB, depth 128, 100,000 records and 128 attributes per element; token limits also apply. Output is limited to 8 MiB. A near-line error marks where checking noticed a problem, not necessarily where the malformed token began.

The visible preview is capped, but current Copy and download contain the complete verified output. A cancelled, failed or timed-out operation has no downloadable partial result; resource limits are not a guarantee about every browser's memory use.

Local document processing

The tool processes your XML locally and does not upload or retain the source or result. Copy and download are explicit actions; downloaded files and clipboard contents remain under your browser and operating system's controls.