Output format and encoding

The markup is re-serialized

The tool does not edit the text of a file: it parses the file, changes the nodes, and writes the whole tree again, as a browser serializes a document (the OuterHtml of the DOM). The content is the same, but the formatting of the markup is not preserved:

• the attribute values are written between double quotes: href='a.css' becomes href="a.css";
• the void elements have no slash: <link /> becomes <link>;
• the character references that do not need escaping are written as characters, and the ones that do are written in their standard form (&, <...);
• the elements that the parser adds are written, such as a missing tbody of a table;
• the whitespace between the elements stays, except where the parser moves it: the line break after the doctype is removed, and the whitespace after </body> moves into the body.

Replacing values shows a page before and after. To format the output, such as to indent or minify it, run your formatter after the tool. Serializing and formatting describes the serialization of WebPlatform.

Documents and fragments

A file that contains a doctype, or an html, head or body tag, is a document: it is written with its html, head and body elements, which the parser adds when one of them is missing. Any other file, such as a partial view or a template part, is a fragment: it is written without them, so a fragment stays a fragment.

Encoding

The tool reads each file with the encoding that the HTML parser finds, as a browser does: from the byte order mark, the <meta charset> element, or the content itself (see Parse bytes and streams). It writes the file back with the same encoding. A UTF-8 file keeps its byte order mark when it had one, and gets none otherwise.

When files are written

A file that is written is re-serialized as a whole, so even a file whose values did not change can get a different markup the first time.