web-html-stream
v2.0.11
Published
Streaming HTML element matching, processing & extraction, using web streams.
Downloads
18
Readme
web-html-stream 
Efficient streaming element matching and processing for HTML5 DOM serialized HTML. Works with Web Streams as returned by fetch.
Usage
const htmlStream = require('web-html-stream');
/**
* @param {object} node, a DOM node like object.
* @return {object} Anything really; return values are accumulated in an
* array.
*/
function handler(node, ctx) {
// Do something with the node
return node;
}
const testDoc = "<html><body><div>"
+ "<test-element foo='bar'>foo</test-element>"
+ "</div></body>";
const inputStream = new ReadableStream({
start(controller) {
controller.enqueue(testDoc);
controller.close();
}
});
// Create a matcher to handle some elements, using CSS syntax. To avoid
// shipping a CSS parser to clients, CSS selectors are only supported in node.
var reader = new htmlStream.HTMLTransformReader(inputStream, {
transforms: [
{ selector: 'test-element[foo="bar"]', handler: handler },
{ selector: 'foo-bar', handler: handler },
],
ctx: { hello: 'world' }
});
// Create the same matcher using more verbose selector objects. These are
// especially useful when processing dynamic values, as this avoids the need to
// escape special chars in CSS selectors.
reader = new htmlStream.HTMLTransformReader(inputStream, {
transforms: [{
selector: {
nodeName: 'test-element',
attributes: [['foo', '=', 'bar']]
},
handler: handler,
// Optional: Request node.innerHTML / outerHTML as `ReadableStream`
// instances. Only available in rule objects.
stream: false
}],
ctx: { hello: 'world' }
});
// Read matches
reader.read()
.then(res => {
console.log(res);
return reader.read();
})
// {
// done: false,
// value: [
// "<html><body><div>",
// {
// "nodeName": "test-element",
// "attributes": {
// "foo": "bar"
// },
// "outerHTML": "<test-element foo='bar'>foo</test-element>",
// "innerHTML": "foo"
// },
// "</div></body>"
// ]
// }
.then(res => console.log);
// { done: true, value: undefined }Performance
Using the Barack Obama
article (1.5mb HTML, part of npm test):
web-html-streammatch & replace all 32<figure>elements: 1.95msweb-html-streammatch & replace all links: 14.98msweb-html-streammatch & replace a specific link (a[href="./Riverdale,_Chicago"]): 2.24msweb-html-streammatch & replace references section (ol[typeof="mw:Extension/references"]): 3.7mslibxmlDOM parse: 26.3mslibxmlDOM round-trip: 50.8mshtmlparser2DOM parse: 66.8mshtmlparser2DOM round-trip: 99.7mshtmlparser2SAX parse: 70.6msdominoDOM parse: 225.8msdominoDOM round-trip: 248.6ms
Using a smaller (1.1mb) version of the same page:
- SAX parse via libxmljs (node) and no-op handlers: 64ms
- XML DOM parse via libxmljs (node): 16ms
- XPATH match for ID (ex:
dom.find('//*[@id = "mw123"]')) : 15ms - XPATH match for class (ex:
dom.find("//*[contains(concat(' ', normalize-space(@class), ' '), ' interlanguage-link ')]"): 34ms
- XPATH match for ID (ex:
- HTML5 DOM parse via Mozilla's html5ever: 32ms
- full round-trip with serialization: 60ms
- HTML5 DOM parse via domino (node): 220ms
Syntactical requirements
web-html-stream gets much of its efficiency from leveraging the syntactic
regularity of HTML5 and
XMLSerializer
DOM serialization.
Detailed requirements (all true for HTML5 and XMLSerializer output):
- Well-formed DOM: Handled tags are balanced.
- Quoted attributes: All attribute values are quoted using single or double quotes.