XML Validator

Check whether an XML document is well formed, see the first error with its line and column, and format it without changing its text.

Overview

A well formed document obeys the grammar of XML 1.0: every tag is closed by a matching one, attribute values are quoted, an ampersand starts a reference, and there is exactly one root element. That is what this tool checks. Whether a document is also valid, whether it matches a schema such as a DTD, an XSD or RELAX NG, is a further question that needs the schema, and this tool does not ask for one.

Web interface

Paste a document into the input; it is checked on every keystroke. The verdict under the input either says Well formed XML, with the size, the number of elements and attributes and the depth, or names the error and quotes its line with a caret under the column.

The formatted output takes an indent of 2, 4 or Tab, or Minified, and three switches: Keep comments, on by default, Sort attributes and Collapse whitespace. Structure shows the element tree, the counts and the namespaces the document declares. The dice in the corner of the input loads a random example, well formed or with a typical mistake.

Errors

Only the first error is reported, since XML gives a parser no reliable way to carry on after it. Lines and columns are counted from 1, in characters as a reader counts them.

The browser's own parser is asked as well, as a backstop. The tool's scanner accepts a handful of documents the browser rejects, around namespace URIs; for those the browser's message is shown, without the boilerplate of its error page. A test holds the scanner to never reject what the browser accepts.

Formatted output

Formatting changes the whitespace between elements and nothing else:

• text is written as it stands, including spaces at its start and end and no-break spaces;
• an element with mixed content, such as <p>Hello <b>world</b>!</p>, is written in one piece, since new lines between its parts would change its text;
• everything under xml:space="preserve" is written exactly as it came;
• CDATA sections, processing instructions and the DOCTYPE are kept.

The output is the document as the parser read it, so a character reference such as é comes back as é, and an entity declared in the DOCTYPE comes back as its value. Checked on 1,440 formatted random documents with mixed content, CDATA, no-break spaces and xml:space, in every indent and in Chrome, Safari's engine and Firefox: the text of each came back unchanged. Collapse whitespace is the one switch that changes text on purpose: it turns runs of spaces and line breaks into one space, outside xml:space="preserve".

Limits

Entities declared in a DOCTYPE are not read by the scanner, so once a DOCTYPE is present any entity name is accepted. Schemas are not checked, see Overview.

A document of 2.7 MB was checked and formatted in about half a second in Chrome. The check runs on every keystroke, so very large documents make typing in the input slow; paste them instead.

No API

Privacy

Nothing leaves the browser. The document is parsed on the page, nothing is stored, and the tool works with the network disconnected.