XPath Tester for HTML and XML

The example contains three fictional product cards. The expression //article[@class="product"]/h2/text() returns Desk Lamp, Travel Mug and Notebook. Each result includes its location in the parsed tree. Changing the input clears the previous result, so the displayed output always belongs to the last completed query.

//article[@class="product"]/h2/text()

HTML mode parses the supplied markup as an HTML5 tree. It normalizes the document structure, including implied elements, and evaluates unnamespaced element names. Scripts are not executed and linked resources are not fetched. Paste the actual response body when debugging a response-based scraper; a live browser DOM can contain elements that were added later by JavaScript.

XML mode preserves namespace declarations. In the XML example, the default namespace is https://example.com/catalog. Map it to a query prefix:

{"p":"https://example.com/catalog"}

Then evaluate //p:item/p:name/text(). The prefix is a local query label; the namespace URI is what identifies the elements. DTD and ENTITY declarations are not accepted.

Expression	Result in the HTML example
//article/h2/text()	Three product names
//article/@data-id	Three ID attributes
count(//article)	One numeric value: 3
boolean(//article[@data-id='p999'])	One boolean value: false
//article/h3	Valid expression, zero matches

Use Node markup when you need to inspect the selected element and its attributes. Use Text values when you need the extracted content. Numeric, string and boolean expressions produce a scalar value rather than a node-match count. The displayed node locations describe the parsed tree; adapt namespace prefixes before reusing a location as a query.

The workbench uses the browser's XPath 1.0 evaluation API. XPath 2.0 and 3.1 functions are outside its scope. HTML/XML input is limited to 200,000 characters and expressions to 2,000 characters. The result view and JSON export contain at most 500 node matches; the total count is shown separately.

If a selector works here but fails in a library, check that the library received the same document, uses the same namespace map and supports the same XPath version. ElementTree's abbreviated XPath support, for example, is different from a full XPath 1.0 engine.