AbstractRange - Web APIs 编辑

The AbstractRange abstract interface is the base class upon which all DOM range types are defined. A range is an object that indicates the start and end points of a section of content within the document.

As an abstract interface, you will not directly instantiate an object of type AbstractRange. Instead, you will use the Range or StaticRange interfaces. To understand the difference between those two interfaces, and how to choose which is appropriate for your needs.

  <div id="interfaceDiagram" style="display: inline-block; position: relative; width: 100%; padding-bottom: 11.666666666666666%; vertical-align: middle; overflow: hidden;"><svg style="display: inline-block; position: absolute; top: 0; left: 0;" viewbox="-50 0 600 70" preserveAspectRatio="xMinYMin meet"><a xlink:href="/wiki/en-US/docs/Web/API/AbstractRange" target="_top"><rect x="1" y="1" width="130" height="50" fill="#F4F7F8" stroke="#D4DDE4" stroke-width="2px" /><text  x="66" y="30" font-size="12px" font-family="Consolas,Monaco,Andale Mono,monospace" fill="#4D4E53" text-anchor="middle" alignment-baseline="middle">AbstractRange</text></a></svg></div>

  a:hover text { fill: #0095DD; pointer-events: all;}

Properties

collapsed Read only

A Boolean value which is true if the range is collapsed. A collapsed range is one whose start position and end position are the same, resulting in a zero-character-long range.

endContainer Read only

The DOM Node in which the end of the range, as specified by the endOffset property, is located.

endOffset Read only

An integer value indicating the offset, in characters, from the beginning of the node's contents to the beginning of the range represented by the range object. This value must be less than the length of the endContainer node.

startContainer Read only

The DOM Node in which the beginning of the range, as specified by the startOffset property, is located.

startOffset Read only

An integer value indicating the offset, in characters, from the beginning of the node's contents to the last character of the contents referred to  by the range object. This value must be less than the length of the node indicated in startContainer.

Methods

The AbstractRange interface offers no methods.

Usage notes

Range types

All ranges of content within a document are described using instances of interfaces based on AbstractRange. There are two such interfaces:

Range

The Range interface has been around for a long time and has only recently been redefined to be based upon AbstractRange as the need arose to define other forms of range data. Range provides methods that allow you to alter the range's endpoints, as well as methods to compare ranges, detect intersections beween ranges, and so forth.

StaticRange

A StaticRange is a basic range which cannot be changed once it's been created. Specifically, as the node tree mutates and changes, the range does not. This is useful when you need to specify a range that will only be used once, since it avoids the performance and resource impact of the more complex Range interface.

Contents of elements

When trying to access the contents of an element, keep in mind that the element itself is a node, but so is any text inside it. In order to set a range endpoint within the text of an element, be sure to find the text node inside the element:

let startElem = document.querySelector("p");
let endElem = startElem.querySelector("span");
let range = document.createRange();

range.setStart(startElem, 0);
range.setEnd(endElem, endElem.childNodes[0].length/2);
let contents = range.cloneContents();

document.body.appendChild(contents);

This example creates a new range, rng, and sets its starting point to the third child node of the first element whose class is elementclass. The end point is set to be the middle of the first child of the span, and then the range is used to copy the contents of the range.

Ranges and the hierarchy of the DOM

In order to define a range of characters within a document in a way that is able to span across zero or more node boundaries, and which is as resilient as possible to changes to the DOM, you can't specify the offset to the first and last characters in the HTML. There are a few good reasons for that.

First, after your page is loaded, the browser isn't thinking in terms of HTML. Once it's been loaded, the page is a tree of DOM Node objects, so you need to specify the beginning and ending locations of a range in terms of nodes and positions within nodes.

Second, in order to support the mutability of the DOM tree as much as possible, you need a way to represent positions relative to nodes in the tree, rather than global positions within the entire document. By defining points within the document as offsets within a given node, those positions remain consistent with the content even as nodes are added to, removed from, or moved around within the DOM tree—within reason. There are fairly obvious limitations (such as if a node is moved to be after the endpoint of a range, or if the content of a node is heavily altered), but it's far better than nothing.

Third, using node-relative positions to define the start and end positions will generally be easier to make perform well. Rather than having to negotiate the DOM figuring out what your global offset refers to, the user agent (browser) can instead go directly to the node indicated by the starting position and start from there, working its way forward until it reaches the given offset into the ending node.

To illustrate this, consider the HTML below:

<div class="container">
  <div class="header">
    <img src="" class="sitelogo">
    <h1>The Ultimate Website</h1>
  </div>
  <article>
    <section class="entry" id="entry1">
      <h2>Section 1: An interesting thing...</h2>
      <p>A <em>very</em> interesting thing happened on the way to the forum...</p>
      <aside class="callout">
        <h2>Aside</h2>
        <p>An interesting aside to share with you...</p>
      </aside>
    </section>
  </article>
  <pre id="log"></pre>
</div>

After loading the HTML and constructing the DOM representation of the document, the resulting DOM tree looks like this:

In this diagram, the nodes representing HTML elements are shown in green. Eah row beneath them shows the next layer of depth into the DOM tree. Blue nodes are text nodes, containing the text that gets shown onscreen. Each element's contents are linked below it in the tree, potentially spawning a series of branches below as elements include other elements and text nodes.

If you want to create a range that incorporates the contents of the <p> element whose contents are "A <em>very</em> interesting thing happened on the way to the forum...", you can do so like this:

let pRange = document.createRange();
pRange.selectNodeContents(document.querySelector("#entry1 p"));

Since we wish to select the entire contents of the <p> element, including its descendants, this works perfectly.

If we wish to instead copy the text "An interesting thing..." from the <section>'s heading (an <h2> element) through the end of the letters "ve" in the <em> within the paragraph below it, the following code would work:

let r = document.createRange();
let startNode = document.querySelector("section h2").childNodes[0];
r.setStart(startNode, 11);

let endNode = document.querySelector("#entry1 p em").childNodes[0];
r.setEnd(endNode, 2);

let fragment = r.cloneContents();

Here an interesting problem arises—we are capturing content from multiple nodes located at different levels of the DOM hierarchy, and then only part of one of them. What should the result look like?

As it turns out, the DOM specification fortunately addresses this exact issue. For example, in this case, we're calling cloneContents() on the range to create a new DocumentFragment object providing a DOM subtree which replicates the contents of the specfied range. To do this, cloneContents() constructs all the nodes needed to preserve the structure of the indicated range, but no more than necessary.

In this example, the start of the specified range is found within the text node below the section's heading, which means that the new DocumentFragment will need to contain an <h2> and, below it, a text node.

The range's end is located below the p element, so that will be needed within the new fragment. So will the text node containing the word "A", since that's included in the range. Finally, an <em> and a text node below it will be added below the <p> as well.

The contents of the text nodes are then determined by the offsets into those text nodes given when calling setStart() and setEnd(). Given the offset of 11 into the heading's text, that node will contain "An interesting thing...". Similarly, the last text node will contain "ve", given the request for the first two characters of the ending node.

The resulting document fragment looks like this:

Notice especially that the contents of this fragment are all below the shared common parent of the topmost nodes within it. The parent <section> is not needed to replicate the cloned content, so it is isn't included.

Example

Consider this simple HTML fragment of HTML.

<p><strong>This</strong> is a paragraph.</p>

Imagine using a Range to extract the word "paragraph" from this. The code to do that looks like the following:

let paraNode = document.querySelector("p");
let paraTextNode = paraNode.childNodes[1];

let range = document.createRange();
range.setStart(paraTextNode, 6);
range.setEnd(paraTextNode, paraTextNode.length-1);

let fragment = range.cloneContents();
document.body.appendChild(fragment);

First we get references to the paragraph node itelf as well as to the second child node within the paragraph. The first child is the <strong> element. The second child is the text node " is a paragraph.".

With the text node reference in hand, we create a new Range object by calling createRange() on the Document itself. We set the starting position of the range to the sixth character of the text node's string, and the end position to the length of the text node's string minus one. This sets the range to encompass the word "paragraph".

We then finish up by calling cloneContents() on the Range to create a new DocumentFragment object which contains the portion of the document encompassed by the range. After that, we use appendChild() to add that fragment at the end of the document's body, as obtained from document.body.

The result looks like this:

Specifications

Browser compatibility

BCD tables only load in the browser

The compatibility table on this page is generated from structured data. If you'd like to contribute to the data, please check out https://github.com/mdn/browser-compat-data and send us a pull request.