Because I had so much trouble finding a good auto-linking implementation out in the wild, I decided to roll my own. It seemed that everything I found out there was either an implementation that didn't cover every case, or was just limited in one way or another.
So, this utility attempts to handle everything. It:
href
attribute inside anchor (<a>) tags (or any other tag/attribute),
and will not accidentally wrap the inner text of an anchor tag with a
new one (which would cause doubly-nested anchor tags).Hope that this utility helps you as well!
Full API Docs: http://gregjacobs.github.io/Autolinker.js/api/
Live Example: http://gregjacobs.github.io/Autolinker.js/examples/live-example/
See Upgrading from v3.x -> v4.x (Breaking Changes) at the bottom of this readme.
npm install autolinker --save
yarn add autolinker
bower install Autolinker.js --save
Simply clone this repository or download a zip of the project, and link to
either dist/Autolinker.js
or dist/Autolinker.min.js
with a script tag.
import Autolinker from 'autolinker';
const Autolinker = require('autolinker');
// note: npm wants an all-lowercase package name, but the utility is a class and
// should be aliased with a capital letter
<!-- 'Autolinker.js' or 'Autolinker.min.js' - non-minified is better for
debugging, minified is better for users' download time -->
<script src="https://github.com/gregjacobs/Autolinker.js/raw/master/path/to/autolinker/dist/Autolinker.min.js"></script>
Using the static link() method:
const linkedText = Autolinker.link(textToAutolink[, options]);
Using as a class:
const autolinker = new Autolinker([ options ]);
const linkedText = autolinker.link(textToAutoLink);
Note: if using the same options to autolink multiple pieces of html/text, it is slightly more efficient to create a single Autolinker instance, and run the link() method repeatedly (i.e. use the "class" form above).
const linkedText = Autolinker.link("Check out google.com");
// Produces: "Check out <a href="http://google.com" target="_blank" rel="noopener noreferrer">google.com</a>"
const linkedText = Autolinker.link("Check out google.com", {
newWindow: false
});
// Produces: "Check out <a href="http://google.com">google.com</a>"
The following are the options which may be specified for linking. These are specified by providing an Object as the second parameter to Autolinker.link(). These include:
newWindow : boolean
true
to have the links should open in a new window when clicked, false
otherwise. Defaults to true
.
urls : boolean/Object
true
to have URLs auto-linked, false
to skip auto-linking of URLs. Defaults
to true
.
This option also accepts an Object form with 3 properties to allow for
more customization of what exactly gets linked. All default to true
:
true
to match URLs found prefixed with a scheme,
i.e. http://google.com
, or other+scheme://google.com
, false
to
prevent these types of matches.true
to match URLs with known top level domains (.com, .net,
etc.) that are not prefixed with a scheme (i.e. 'http://'). Ex: google.com
,
asdf.org/?page=1
, etc. Set to false
to prevent these types of matches.true
to match IPv4 addresses. Ex: 192.168.0.1
.
false
to prevent these types of matches. Note that if the IP address had
a prefixed scheme (such as 'http://'), and schemeMatches
is true, it
will still be linked.Example usage: urls: { schemeMatches: true, tldMatches: false, ipV4Matches: true }
email : boolean
true
to have email addresses auto-linked, false
to skip auto-linking of
email addresses. Defaults to true
.
phone : boolean
true
to have phone numbers auto-linked, false
to skip auto-linking of
phone numbers. Defaults to true
.
mention : string
A string for the service name to have mentions (@username) auto-linked to. Supported
values at this time are 'twitter', 'soundcloud', 'instagram' and 'tiktok'. Pass false
to skip
auto-linking of mentions. Defaults to false
.
hashtag : boolean/string
A string for the service name to have hashtags auto-linked to. Supported
values at this time are 'twitter', 'facebook', 'instagram' and 'tiktok'. Pass false
to skip
auto-linking of hashtags. Defaults to false
.
stripPrefix : boolean
true
to have the 'http://'
(or 'https://'
) and/or the 'www.'
stripped from the beginning of displayed links, false
otherwise.
Defaults to true
.
This option also accepts an Object form with 2 properties to allow for
more customization of what exactly is prevented from being displayed.
Both default to true
:
true
to prevent the scheme part of a URL match
from being displayed to the user. Example: 'http://google.com'
will be displayed as 'google.com'
. false
to not strip the
scheme. NOTE: Only an 'http://'
or 'https://'
scheme will be
removed, so as not to remove a potentially dangerous scheme (such
as 'file://'
or 'javascript:'
).true
to prevent the 'www.'
part of a URL match
from being displayed to the user. Ex: 'www.google.com'
will be
displayed as 'google.com'
. false
to not strip the 'www'
.stripTrailingSlash : boolean
true
to remove the trailing slash from URL matches, false
to keep
the trailing slash. Example when true
: http://google.com/
will be
displayed as http://google.com
. Defaults to true
.
truncate : number/Object
A number for how many characters long URLs/emails/Twitter handles/Twitter
hashtags should be truncated to inside the text of a link. If the match is
over the number of characters, it will be truncated to this length by
replacing the end of the string with a two period ellipsis ('..').
Example: a url like 'http://www.yahoo.com/some/long/path/to/a/file' truncated to 25 characters may look like this: 'yahoo.com/some/long/pat..'
In the object form, both length
and location
may be specified to perform
truncation. Available options for location
are: 'end' (default), 'middle',
or 'smart'. Example usage:
truncate: { length: 32, location: 'middle' }
The 'smart' truncation option is for URLs where the algorithm attempts to strip out unnecessary parts of the URL (such as the 'www.', then URL scheme, hash, etc.) before trying to find a good point to insert the ellipsis if it is still too long. For details, see source code of: TruncateSmart
className : string
A CSS class name to add to the generated anchor tags. This class will be added
to all links, as well as this class plus "url"/"email"/"phone"/"hashtag"/"mention"
suffixes for styling url/email/phone/hashtag/mention links differently.
The name of the hashtag/mention service is also added as a CSS class for those types of matches.
For example, if this config is provided as "my-link", then:
decodePercentEncoding: boolean
true
to decode percent-encoded characters in URL matches, false
to keep
the percent-encoded characters.
Example when true
: https://en.wikipedia.org/wiki/San_Jos%C3%A9
will
be displayed as https://en.wikipedia.org/wiki/San_José
.
Defaults to true
.
replaceFn : Function
A function to use to programmatically make replacements of matches in the
input string, one at a time. See the section
Custom Replacement Function for
more details.
sanitizeHtml : boolean
true
to HTML-encode the start and end brackets of existing HTML tags found
in the input string. This will escape <
and >
characters to <
and
>
, respectively.
Setting this to true
will prevent XSS (Cross-site Scripting) attacks,
but will remove the significance of existing HTML tags in the input string. If
you would like to maintain the significance of existing HTML tags while also
making the output HTML string safe, leave this option as false
and use a
tool like https://github.com/cure53/DOMPurify (or others) on the input string
before running Autolinker.
Defaults to false
.
For example, if you wanted to disable links from opening in new windows, you could do:
const linkedText = Autolinker.link("Check out google.com", {
newWindow: false
});
// Produces: "Check out <a href="http://google.com">google.com</a>"
And if you wanted to truncate the length of URLs (while also not opening in a new window), you could do:
const linkedText = Autolinker.link("http://www.yahoo.com/some/long/path/to/a/file", {
truncate: 25,
newWindow: false
});
// Produces: "<a href="http://www.yahoo.com/some/long/path/to/a/file">yahoo.com/some/long/pat..</a>"
One could update an entire DOM element that has unlinked text to auto-link them as such:
const myTextEl = document.getElementById('text');
myTextEl.innerHTML = Autolinker.link(myTextEl.innerHTML);
Using the same pre-configured Autolinker instance in multiple locations of a codebase (usually by dependency injection):
const autolinker = new Autolinker({ newWindow: false, truncate: 25 });
//...
autolinker.link("Check out http://www.yahoo.com/some/long/path/to/a/file");
// Produces: "Check out <a href="http://www.yahoo.com/some/long/path/to/a/file">yahoo.com/some/long/pat..</a>"
//...
autolinker.link( "Go to www.google.com" );
// Produces: "Go to <a href="http://www.google.com">google.com</a>"
If you're just interested in retrieving the list of Matches without producing a transformed string, you can use the parse() method.
For example:
const matches = Autolinker.parse("Hello google.com, I am asdf@asdf.com", {
urls: true,
email: true
});
console.log(matches.length); // 2
console.log(matches[0].type); // 'url'
console.log(matches[0].getUrl()); // 'google.com'
console.log(matches[1].type); // 'email'
console.log(matches[1].getEmail()); // 'asdf@asdf.com'
A custom replacement function (replaceFn) may be provided to replace url/email/phone/mention/hashtag matches on an individual basis, based on the return from this function.
const input = "..."; // string with URLs, Email Addresses, Mentions (Twitter, Instagram), and Hashtags
const linkedText = Autolinker.link(input, {
replaceFn : function(match) {
console.log("href = ", match.getAnchorHref());
console.log("text = ", match.getAnchorText());
switch(match.type) {
case 'url':
console.log("url: ", match.getUrl());
return true; // let Autolinker perform its normal anchor tag replacement
case 'email':
const email = match.getEmail();
console.log("email: ", email);
if(email === "my@own.address") {
return false; // don't auto-link this particular email address; leave as-is
} else {
return; // no return value will have Autolinker perform its normal anchor tag replacement (same as returning `true`)
}
case 'phone':
console.log("Phone Number: ", match.getPhoneNumber());
return '<a href="http://newplace.to.link.phone.numbers.to/">' + match.getPhoneNumber() + '</a>';
case 'mention':
console.log("Mention: ", match.getMention());
console.log("Mention Service Name: ", match.getServiceName());
return '<a href="http://newplace.to.link.mention.handles.to/">' + match.getMention() + '</a>';
case 'hashtag':
console.log("Hashtag: ", match.getHashtag());
return '<a href="http://newplace.to.link.hashtag.handles.to/">' + match.getHashtag() + '</a>';
}
}
} );
const input = "..."; // string with URLs, Email Addresses, Mentions (Twitter, Instagram), and Hashtags
const linkedText = Autolinker.link( input, {
replaceFn : function( match ) {
console.log("href = ", match.getAnchorHref());
console.log("text = ", match.getAnchorText());
const tag = match.buildTag(); // returns an `Autolinker.HtmlTag` instance for an <a> tag
tag.setAttr('rel', 'nofollow'); // adds a 'rel' attribute
tag.addClass('external-link'); // adds a CSS class
tag.setInnerHtml('Click here!'); // sets the inner html for the anchor tag
return tag;
}
} );
The replaceFn
is provided one argument:
A replacement of the match is made based on the return value of the function. The following return values may be provided:
undefined
), or true
(boolean): Delegate back to
Autolinker to replace the match as it normally would.false
(boolean): Do not replace the current match at all - leave as-is.The full API docs for Autolinker may be referenced at: http://gregjacobs.github.io/Autolinker.js/api/
http://gregjacobs.github.io/Autolinker.js/examples/
urls.wwwMatches
config has been removed. A www.
prefix is now treated
like any other subdomain of a top level domain (TLD) match (such as
'subdomain.google.com'). Match.getType()
should be replaced with Match.type
. This allows for
TypeScript type narrowing of Match
objects returned by the parse()
method or inside the replaceFn
.Matcher
classes have been removed in favor of a single finite state
machine parser, greatly improving the performance of Autolinker but removing
some of the customizability of the old regular expressions. Will address this customizability in a future release.Autolinker.AnchorTagBuilder
, Autolinker.HtmlTag
, and Autolinker.match.*
references have been removed. These shouldn't be needed as public APIs, but
please raise a GitHub issue if these are for some reason needed.HtmlParser
class has been removed in favor of an internal parseHtml()
function which replaces the old regexp-based implementation with a state
machine parser that is guaranteed to run in linear time. If you were using
the HtmlParser
class directly, I recommend switching to htmlparser2, which implements the HTML semantics
better. The internal parseHtml()
function that Autolinker now uses is
fairly geared towards Autolinker's purposes, and may not be useful in a
general HTML parsing sense.If you are still on v0.x, first follow the instructions in the Upgrading from v0.x -> v1.x section below.
The codebase has been converted to TypeScript, and uses ES6 exports. You can
now use the import
statement to pull in the Autolinker
class and related
entities such as Match
:
// ES6/TypeScript/Webpack
import Autolinker, { Match } from 'autolinker';
The require()
interface is still supported as well for Node.js:
// Node.js
const Autolinker = require('autolinker');
You will no longer need the @types/autolinker
package as this package now
exports its own types
You will no longer be able to override the regular expressions in the
Matcher
classes by assigning to the prototype (for instance, something like
PhoneMatcher.prototype.regex = ...
). This is due to how TypeScript creates
properties for class instances in the constructor rather than on prototypes.
The idea of providing your own regular expression for these classes is a
brittle notion anyway, as the Matcher
classes rely on capturing groups in
the RegExp being in the right place, or even multiple capturing groups for
the same piece of information to support a different format. These capturing
groups and associated code are subject to change as the regular expression
needs to be updated, and will not involve a major version release of
Autolinker.
In the future you will be able to override the default Matcher
classes
entirely to provide your own implementation, but please raise an issue (or
+1 an issue) if you think the library should support a currently-unsupported
format.
twitter
option removed, replaced with mention
(which accepts 'twitter',
'instagram' and 'soundcloud' values)twitter
option) now defaults to
being turned off. Previously, Twitter handle matching was on by
default.replaceFn
option now called with just one argument: the Match
object (previously was called with two arguments: autolinker
and
match
)replaceFn
) TwitterMatch
replaced with
MentionMatch
, and MentionMatch.getType()
now returns 'mention'
instead of 'twitter'
replaceFn
) TwitterMatch.getTwitterHandle()
->
MentionMatch.getMention()
Pull requests definitely welcome. To setup the project, make sure you have Node.js installed. Then open up a command prompt and type the following:
cd Autolinker.js # where you cloned the project
npm install
To run the tests:
npm run test
npm run test
command to testFor this you will need Ruby installed (note: Ruby comes pre-installed on MacOS), with the JSDuck gem.
See https://github.com/senchalabs/jsduck#getting-it for installation instructions on Windows/Mac/Linux
JSDuck is used to build the project's API/documentation site. See [Documentation Generator Notes](#Documentation Generator Notes) for more info.
Run:
yarn serve
Then open your browser to: http://localhost:8080/docs/examples/index.html
You should be able to make a change to source files, and refresh the page to see the changes.
Note: If anyone wants to submit a PR converting gulp watch
to webpack
with
the live development server, that would be much appreciated :)
This project uses JSDuck for its documentation generation, which produces the page at http://gregjacobs.github.io/Autolinker.js.
Unfortunately, JSDuck is a very old project that is no longer maintained. As
such, it doesn't support TypeScript or anything from ES6 (the class
keyword,
arrow functions, etc). However, I have yet to find a better documentation
generator that creates such a useful API site. (Suggestions for a new one are
welcome though - please raise an issue.)
Since ES6 is not supported, we must generate the documentation from the ES5 output. As such, a few precautions must be taken care of to make sure the documentation comes out right:
@cfg
documentation tags must exist above a class property that has a
default value, or else it won't end up in the ES5 output. For example:
// Will correctly end up in the ES5 output
/**
* @cfg {String} title
*/
readonly title: string = '';
// Will *not* end up in ES5 output, and thus, won't end up in the generated
// documentation
/**
* @cfg {String} title
*/
readonly title: string;
@constructor
tag must be replaced with @method constructor
See Releases