fontoxpath

A minimalistic XPath 3.1 and XQuery 3.1 engine for (XML) nodes with XQuery Update Facility 3.0 support.

Demo page

How to use

Querying XML

evaluateXPath(xpathExpression, contextNode, domFacade, variables, returnType, options);

The following are convenience functions for a specific returnType.

evaluateXPathToArray(xpathExpression, contextNode, domFacade, variables, options);
evaluateXPathToAsyncIterator(xpathExpression, contextNode, domFacade, variables, options);
evaluateXPathToBoolean(xpathExpression, contextNode, domFacade, variables, options);
evaluateXPathToFirstNode(xpathExpression, contextNode, domFacade, variables, options);
evaluateXPathToMap(xpathExpression, contextNode, domFacade, variables, options);
evaluateXPathToNodes(xpathExpression, contextNode, domFacade, variables, options);
evaluateXPathToNumber(xpathExpression, contextNode, domFacade, variables, options);
evaluateXPathToNumbers(xpathExpression, contextNode, domFacade, variables, options);
evaluateXPathToString(xpathExpression, contextNode, domFacade, variables, options);
evaluateXPathToStrings(xpathExpression, contextNode, domFacade, variables, options);

• xpathExpression <String> The query to evaluate.
• contextNode <Node> The node in which context the xpathExpression will be evaluated. Defaults to null.
• domFacade <IDomFacade> An IDomFacade implementation which will be used for querying the DOM. Defaults to an implementation which uses properties and methods on the contextNode as described in the DOM spec.
• variables <Object> The properties of variables are available variables within the xpathExpression. Defaults to an empty Object. Can only be used to set variables in the global namespace.
• returnType <number> Determines the type of the result. Defaults to evaluateXPath.ANY_TYPE. Possible values:
  • evaluateXPath.ALL_RESULTS_TYPE Returns the result of the query, can be anything depending on the query. This will always be an array, and the result can be mixed: contain both nodes and strings for example.
  • evaluateXPath.NUMBER_TYPE Resolve to a number, like count((1,2,3)) resolves to 3.
  • evaluateXPath.STRING_TYPE Resolve to a string, like //someElement[1] resolves to the text content of the first someElement.
  • evaluateXPath.BOOLEAN_TYPE Resolves to a boolean true or false, uses the effective boolean value to determine the result. count(1) resolves to true, count(()) resolves to false.
  • evaluateXPath.NODES_TYPE Resolve to all nodes Node[] the XPath resolves to. Returns nodes in the order the XPath would. Meaning (//a, //b) resolves to all A nodes, followed by all B nodes. //*[self::a or self::b] resolves to A and B nodes in document order.
  • evaluateXPath.FIRST_NODE_TYPE Resolves to the first Node node.NODES_TYPE would have resolved to.
  • evaluateXPath.STRINGS_TYPE Resolve to an array of strings string[].
  • evaluateXPath.MAP_TYPE Resolve to an Object, as a map.
  • evaluateXPath.ARRAY_TYPE Resolve to an array [].
  • evaluateXPath.ASYNC_ITERATOR_TYPE
  • evaluateXPath.NUMBERS_TYPE Resolve to an array of numbers number[].
  • evaluateXPath.ANY_TYPE Returns the result of the query, can be anything depending on the query. Note that the return type is determined dynamically, not statically: XPaths returning empty sequences will return empty arrays and not null, like one might expect. This is deprecated, use evaluateXPath.ALL_RESULTS_TYPE instead, since that is more predictable.
• options <Object> Options used to modify the behavior. The following options are available:
  • namespaceResolver <function(string):string?> By default, the namespaces in scope of the context item (if it is a node) are used. This is fine for most queries if you can assume how your XML uses prefixes. Use this function to override those namespaces to remove that assumption. This function will be called with a prefix (the empty string for the default namespaceURI) and should return a namespaceURI (or null for the null namespace).
  • nodesFactory INodesFactory A INodesFactory implementation which will be used for creating nodes.
  • language string The query language to use. Defaults to evaluateXPath.XPATH_3_1_LANGUAGE. Possible values:
    • evaluateXPath.XPATH_3_1_LANGUAGE Evaluate xpathExpression according the XPath spec.
    • evaluateXPath.XQUERY_3_1_LANGUAGE Evaluate xpathExpression according the XQuery spec.
  • moduleImports <Object<string, string>
  • debug <boolean> If a debug trace should be tracked, see debugging for more information.
  • logger <Object> Object with functions used to override the standard logger.
    • trace: <function(string):void> The logger for the trace() function. The argument is the string of the original message.
  • defaultFunctionNamespaceURI <string> To modify or change the default function namespaceURI. Defaults to http://www.w3.org/2005/xpath-functions. Defining the default function namespaceURI in the xpath expression overwrites this option.
  • functionNameResolver <({prefix, localName}, arity) => {namespaceURI, localName}> To influence the function name resolving algorithm. Useful to extend the protected namespaces, such as the fn namespace.

Example

const {
    evaluateXPath,
    evaluateXPathToBoolean,
    evaluateXPathToString,
    evaluateXPathToFirstNode,
    evaluateXPathToNumber,
} = require('fontoxpath');

const documentNode = new DOMParser().parseFromString('<xml/>', 'text/xml');

console.log(evaluateXPathToBoolean('/xml => exists()', documentNode));
// Outputs: true

console.log(evaluateXPathToString('$foo', null, null, { foo: 'bar' }));
// Outputs: "bar"

// We pass the documentNode so the default INodesFactory can be used.
console.log(
    evaluateXPathToFirstNode('<foo>bar</foo>', documentNode, null, null, {
        language: evaluateXPath.XQUERY_3_1_LANGUAGE,
    }).outerHTML
);
// Outputs: "<foo>bar</foo>"

// We pass the Math namespaceURI for the pi() function to be used
console.log(
    evaluateXPathToNumber(
        'pi()',
        documentNode,
        undefined,
        {},
        {
            language: evaluateXPath.XQUERY_3_1_LANGUAGE,
            defaultFunctionNamespaceURI: 'http://www.w3.org/2005/xpath-functions/math',
        }
    )
);
// Outputs: Math.PI (3.14...)

Creating typed values

When having to pass JavaScript values as variables to an evaluateXPath call you can create a typed value of it to ensure it will be used as that specific type.

If you do not do this and instead pass a plain JavaScript value as variable it will get converted automatically into a type which fits but you will not be able to control the exact type.

const integerValueFactory = createTypedValueFactory('xs:integer');
const integerValue = integerValueFactory(123, domFacade);

// Will return true as we specified it to be an xs:integer
evaluateXPathToBoolean('$value instance of xs:integer', null, null, {
    value: typedValue,
}),

// Will return false as JavaScript numbers are by default converted to an xs:double
evaluateXPathToBoolean('$value instance of xs:integer', null, null, {
    value: 123,
}),

Debugging

FontoXPath can output a basic trace for an error if the debug option is set to true. This is disabled by default because of performance reasons.

evaluateXPathToBoolean(`
if (true()) then
  zero-or-one((1, 2))
else
  (1, 2, 3)
`, null, null, null, {debug: true});

// Throws:
1: if (true()) then
2:   zero-or-one((1, 2))
     ^^^^^^^^^^^^^^^^^^^
3: else
4:   (1, 2, 3)

Error: FORG0003: The argument passed to fn:zero-or-one contained more than one item.
  at <functionCallExpr>:2:3 - 2:22
  at <ifThenElseExpr>:1:1 - 4:12

Besides errors, the fn:trace function can be used to output information to the developer console.

Performance

FontoXPath can use the Performance API to provide some insight in the speed of XPaths. To use it, first give FontoXPath an implementation of the Performance interface:

import { profiler } from 'fontoxpath';

profiler.setPerformanceImplementation(window.performance); // or global.performance or self.performance, depending on you surroundings

// And start profiling all XPath / XQuery usage

profiler.startProfiling();

At some point, you may want to get a summary of all evaluated XPaths:

const summary = profiler.getPerformanceSummary();

This summary contains an array of XPaths, their execution times, their total runtime and their average runtime. Starting a performance profile will also output measurements on the timeline of the performance profiler of the browser.

Modifying XML

To modify XML you can use XQuery Update Facility 3.0 as following

evaluateUpdatingExpressionSync(xpathExpression, contextNode, domFacade, variables, options);

The arguments are the same as evaluateXPath. This returns an Object, the object has a xdmValue and pendingUpdateList. The xdmValue is the result of query as if it was run using evaluateXPath with evaluateXPath.ANY_TYPE as returnType. The pendingUpdateList is an <Object[]> in which each entry represents an update primitive where the type identifies the update primitive.

The pending update list can be executed using

executePendingUpdateList(pendingUpdateList, domFacade, nodesFactory, documentWriter);

• pendingUpdateList <Object[]> The pending update list returned by evaluateUpdatingExpression.
• domFacade <IDomFacade> See evaluateXPath. The default will use nodes from the pendingUpdateList.
• nodesFactory INodesFactory A INodesFactory implementation which will be used for creating nodes. Defaults to an implementation which uses properties and methods of nodes from the pendingUpdateList.
• documentWriter <IDocumentWriter> An IDocumentWriter implementation which will be used for modifying a DOM. Defaults to an implementation which uses properties and methods of nodes from the pendingUpdateList.

Example

const { evaluateUpdatingExpression, executePendingUpdateList } = require('fontoxpath');
const documentNode = new DOMParser().parseFromString('<xml/>', 'text/xml');

const result = evaluateUpdatingExpressionSync('replace node /xml with <foo/>', documentNode)

executePendingUpdateList(result.pendingUpdateList);
console.log(documentNode.documentElement.outerHTML);
// Outputs: "<foo/>";

An example of using XQUF with XQuery modules:

registerXQueryModule(`
module namespace my-custom-namespace = "my-custom-uri";
(:~
    Insert attribute somewhere
    ~:)
declare %public %updating function my-custom-namespace:do-something ($ele as element()) as xs:boolean {
    if ($ele/@done) then false() else
    (insert node
    attribute done {"true"}
    into $ele, true())
};
`);
// At some point:
const contextNode = null;
const pendingUpdatesAndXdmValue = evaluateUpdatingExpressionSync(
    'ns:do-something(.)',
    contextNode,
    null,
    null,
    { moduleImports: { ns: 'my-custom-uri' } }
);

console.log(pendingUpdatesAndXdmValue.xdmValue); // this is true or false, see function

executePendingUpdateList(pendingUpdatesAndXdmValue.pendingUpdateList, null, null, null);

// At this point the context node will have its attribute set

Global functions

To register custom functions. They are registered globally.

registerCustomXPathFunction(name, signature, returnType, callback);

• name {namespaceURI: string, localName: string} The function name.
• signature string[] The arguments of the function.
• returnType string The return type of the function.
• callback function The function itself.

Example:

const fontoxpath = require('fontoxpath');

// Register a function called 'there' in the 'hello' namespace:
fontoxpath.registerCustomXPathFunction(
    { namespaceURI: 'hello', localName: 'there' },
    ['xs:string'],
    'xs:string',
    (_, str) => `Hello there, ${str}`
);

// and call it, using the BracedUriLiteral syntax (Q{})
const out = fontoxpath.evaluateXPathToString('Q{hello}there("General Kenobi")');

// Or by using a prefix instead:
const URI_BY_PREFIX = { hi: 'hello' };
const out2 = fontoxpath.evaluateXPathToString('hi:there("General Kenobi")', null, null, null, {
    namespaceResolver: (prefix) => URI_BY_PREFIX[prefix],
});

Including modules

Use the registerXQueryModule function to register an XQuery module. Registered modules will be globally available, but will have to be imported before they can be used.

Example:

const fontoxpath = require('fontoxpath');

fontoxpath.registerXQueryModule(`
    module namespace test = "https://www.example.org/test1";

    declare %public function test:hello($a) {
        "Hello " || $a
    };
`);

// Import the module using the XQuery way:
fontoxpath.evaluateXPathToString(
    `
    import module namespace test = "https://www.example.org/test1";
    (: Invoke the test:hello function :)
    test:hello('there')
    `,
    null,
    null,
    null,
    { language: fontoxpath.evaluateXPath.XQUERY_3_1_LANGUAGE }
);

// Or by using the moduleImports API, which can be used in XPath contexts as well
fontoxpath.evaluateXPathToString(
    `
    (: Invoke the test:hello function :)
    test:hello('there')
    `,
    null,
    null,
    null,
    { moduleImports: { test: 'https://www.example.org/test1' } }
);

Typescript

We support TypeScript; and expose a minimal Node type. You can use generic types to get the type of the DOM implementation you are using without having to cast it.

const myNodes = evaluateXPathToNodes<slimdom.Node>('<foo>bar</foo>', null, null, null, {
    language: evaluateXPath.XQUERY_3_1_LANGUAGE,
});

// Type of myNodes is: slimdom.Node[] .

Compiling queries to JavaScript for better execution performance

:warning: Warning: this functionality considered experimental. :warning:

FontoXPath supports compiling a small but useful subset of XPath 3.1 to pure JavaScript code. Query execution performance benefits from this: execution speed can be 2 to 7 times higher than when using evaluateXPath, according to our benchmarks.

Two API's provide this functionality:

• compileXPathToJavaScript Compiles a query and its return type to JavaScript code. This result should be evaluated to a function, for example with new Function.
• executeJavaScriptCompiledXPath Evaluates a to a function evaluated compiled query (see the example below) and applies it to the given context node, returning its resulting value.

Supported functionality

Here is a list of supported functionality so you can determine if compiling to JavaScript is suitable for your project. These functionalities are supported:

• Absolute and relative path expressions, including an arbitrary amount of steps.
• child, self, parent and attribute axes.
• NodeTests: NameTest, ElementTest, Wildcard and TextTest.
• Predicates (the [ and ] in /xml[child::title]).
• Logical operators (and and or).
• Compares (compare string to string and node to string).
• Return types evaluateXPath.NODES_TYPE, evaluateXPath.BOOLEAN_TYPE, evaluateXPath.FIRST_NODE_TYPE, evaluateXPath.STRING, evaluateXPath.ANY.

Functions, XQuery and other more advanced features are not supported (yet).

Example usage:

import {
    compileXPathToJavaScript,
    CompiledXPathFunction,
    evaluateXPath,
    executeJavaScriptCompiledXPath,
} from 'fontoxpath';

const documentNode = new DOMParser().parseFromString('<p>Beep beep.</p>', 'text/xml');

const compiledXPathResult = compileXPathToJavaScript(
    '/child::p/text()',
    evaluateXPath.BOOLEAN_TYPE
);
if (compiledXPathResult.isAstAccepted === true) {
    // Query is compiled succesfully, it can be evaluated.
    const evalFunction = new Function(compiledXPathResult.code) as CompiledXPathFunction;

    console.log(executeJavaScriptCompiledXPath(evalFunction, documentNode));
    // Outputs: true
} else {
    // Not supported by JS codegen (yet).
}

Ideas to improve the example to better fit your project:

• If a query could not be compiled to JavaScript, fall back on the stable evaluateXPath function.
• Add caching so compiling and new Function does not have happen more than once per unique query.
• Store compiled code to disk.

Features

Note that this engine assumes XPath 1.0 compatibility mode turned off.

Not all XPath 3.1 functions are implemented yet. We accept pull requests for missing features. A full list of supported queries can be found on the playground. Select the 'Report on which functions are implemented' example to get a full dynamic report!

The following features are unavailable at this moment, but will be implemented at some point in time (and even sooner if you can help!):

• Some DateTime related functions
• Collation related functions (fn:compare#3)
• Some other miscellaneous functions
• XML parsing
• The treat as operator
• Some parts of FLWOR expressions

For all available features, see the unit tests, or just try it out on the Demo page.

Extensions to the spec

FontoXPath implements a single function that is public API: fontoxpath:version() as xs:string. It resides in the 'http://fontoxml.com/fontoxpath' namespace. Call it to check what version of FontoXPath you are running.

Compatibility

This engine is pretty DOM-agnostic, it has a good track record with the browser DOM implementations and slimdom.js. There are a number of known issues with other DOM implementations such as xmldom because it does not follow the DOM spec on some features including namespaces.

When using namespaces in general, be sure to not use the HTML DOM since it does not always implement namespaces how you'd expect!

Contribution

If you have any questions on how to use FontoXPath, or if you are running into problems, just file a github issue! If you are looking to contribute, we have a Contribution Guide that should help you in getting your development environment set up.