he

he (for “HTML entities”) is a robust HTML entity encoder/decoder written in JavaScript. It supports all standardized named character references as per HTML, handles ambiguous ampersands and other edge cases just like a browser would, has an extensive test suite, and — contrary to many other JavaScript solutions — he handles astral Unicode symbols just fine. An online demo is available.

Installation

Via npm:

1npm install he

Via Bower:

1bower install he

Via Component:

1component install mathiasbynens/he

In a browser:

1<script src="he.js"></script>

In Node.js, io.js, Narwhal, and RingoJS:

1var he = require('he');

In Rhino:

1load('he.js');

Using an AMD loader like RequireJS:

1require(
2  {
3    'paths': {
4      'he': 'path/to/he'
5    }
6  },
7  ['he'],
8  function(he) {
9    console.log(he);
10  }
11);

API

he.version

A string representing the semantic version number.

he.encode(text, options)

This function takes a string of text and encodes (by default) any symbols that aren’t printable ASCII symbols and &, <, >, ", ', and `, replacing them with character references.

1he.encode('foo © bar ≠ baz 𝌆 qux');
2// → 'foo © bar ≠ baz 𝌆 qux'

As long as the input string contains allowed code points only, the return value of this function is always valid HTML. Any (invalid) code points that cannot be represented using a character reference in the input are not encoded:

1he.encode('foo \0 bar');
2// → 'foo \0 bar'

However, enabling the strict option causes invalid code points to throw an exception. With strict enabled, he.encode either throws (if the input contains invalid code points) or returns a string of valid HTML.

The options object is optional. It recognizes the following properties:

useNamedReferences

The default value for the useNamedReferences option is false. This means that encode() will not use any named character references (e.g. ©) in the output — hexadecimal escapes (e.g. ©) will be used instead. Set it to true to enable the use of named references.

Note that if compatibility with older browsers is a concern, this option should remain disabled.

1// Using the global default setting (defaults to `false`):
2he.encode('foo © bar ≠ baz 𝌆 qux');
3// → 'foo © bar ≠ baz 𝌆 qux'
4
5// Passing an `options` object to `encode`, to explicitly disallow named references:
6he.encode('foo © bar ≠ baz 𝌆 qux', {
7  'useNamedReferences': false
8});
9// → 'foo © bar ≠ baz 𝌆 qux'
10
11// Passing an `options` object to `encode`, to explicitly allow named references:
12he.encode('foo © bar ≠ baz 𝌆 qux', {
13  'useNamedReferences': true
14});
15// → 'foo © bar ≠ baz 𝌆 qux'

decimal

The default value for the decimal option is false. If the option is enabled, encode will generally use decimal escapes (e.g. ©) rather than hexadecimal escapes (e.g. ©). Beside of this replacement, the basic behavior remains the same when combined with other options. For example: if both options useNamedReferences and decimal are enabled, named references (e.g. ©) are used over decimal escapes. HTML entities without a named reference are encoded using decimal escapes.

1// Using the global default setting (defaults to `false`):
2he.encode('foo © bar ≠ baz 𝌆 qux');
3// → 'foo © bar ≠ baz 𝌆 qux'
4
5// Passing an `options` object to `encode`, to explicitly disable decimal escapes:
6he.encode('foo © bar ≠ baz 𝌆 qux', {
7  'decimal': false
8});
9// → 'foo © bar ≠ baz 𝌆 qux'
10
11// Passing an `options` object to `encode`, to explicitly enable decimal escapes:
12he.encode('foo © bar ≠ baz 𝌆 qux', {
13  'decimal': true
14});
15// → 'foo © bar ≠ baz 𝌆 qux'
16
17// Passing an `options` object to `encode`, to explicitly allow named references and decimal escapes:
18he.encode('foo © bar ≠ baz 𝌆 qux', {
19  'useNamedReferences': true,
20  'decimal': true
21});
22// → 'foo © bar ≠ baz 𝌆 qux'

encodeEverything

The default value for the encodeEverything option is false. This means that encode() will not use any character references for printable ASCII symbols that don’t need escaping. Set it to true to encode every symbol in the input string. When set to true, this option takes precedence over allowUnsafeSymbols (i.e. setting the latter to true in such a case has no effect).

1// Using the global default setting (defaults to `false`):
2he.encode('foo © bar ≠ baz 𝌆 qux');
3// → 'foo © bar ≠ baz 𝌆 qux'
4
5// Passing an `options` object to `encode`, to explicitly encode all symbols:
6he.encode('foo © bar ≠ baz 𝌆 qux', {
7  'encodeEverything': true
8});
9// → 'foo © bar ≠ baz 𝌆 qux'
10
11// This setting can be combined with the `useNamedReferences` option:
12he.encode('foo © bar ≠ baz 𝌆 qux', {
13  'encodeEverything': true,
14  'useNamedReferences': true
15});
16// → 'foo © bar ≠ baz 𝌆 qux'

strict

The default value for the strict option is false. This means that encode() will encode any HTML text content you feed it, even if it contains any symbols that cause parse errors. To throw an error when such invalid HTML is encountered, set the strict option to true. This option makes it possible to use he as part of HTML parsers and HTML validators.

1// Using the global default setting (defaults to `false`, i.e. error-tolerant mode):
2he.encode('\x01');
3// → ''
4
5// Passing an `options` object to `encode`, to explicitly enable error-tolerant mode:
6he.encode('\x01', {
7  'strict': false
8});
9// → ''
10
11// Passing an `options` object to `encode`, to explicitly enable strict mode:
12he.encode('\x01', {
13  'strict': true
14});
15// → Parse error

allowUnsafeSymbols

The default value for the allowUnsafeSymbols option is false. This means that characters that are unsafe for use in HTML content (&, <, >, ", ', and `) will be encoded. When set to true, only non-ASCII characters will be encoded. If the encodeEverything option is set to true, this option will be ignored.

1he.encode('foo © and & ampersand', {
2  'allowUnsafeSymbols': true
3});
4// → 'foo © and & ampersand'

Overriding default encode options globally

The global default setting can be overridden by modifying the he.encode.options object. This saves you from passing in an options object for every call to encode if you want to use the non-default setting.

1// Read the global default setting:
2he.encode.options.useNamedReferences;
3// → `false` by default
4
5// Override the global default setting:
6he.encode.options.useNamedReferences = true;
7
8// Using the global default setting, which is now `true`:
9he.encode('foo © bar ≠ baz 𝌆 qux');
10// → 'foo © bar ≠ baz 𝌆 qux'

he.decode(html, options)

This function takes a string of HTML and decodes any named and numerical character references in it using the algorithm described in section 12.2.4.69 of the HTML spec.

1he.decode('foo © bar ≠ baz 𝌆 qux');
2// → 'foo © bar ≠ baz 𝌆 qux'

The options object is optional. It recognizes the following properties:

isAttributeValue

The default value for the isAttributeValue option is false. This means that decode() will decode the string as if it were used in a text context in an HTML document. HTML has different rules for parsing character references in attribute values — set this option to true to treat the input string as if it were used as an attribute value.

1// Using the global default setting (defaults to `false`, i.e. HTML text context):
2he.decode('foo&ampbar');
3// → 'foo&bar'
4
5// Passing an `options` object to `decode`, to explicitly assume an HTML text context:
6he.decode('foo&ampbar', {
7  'isAttributeValue': false
8});
9// → 'foo&bar'
10
11// Passing an `options` object to `decode`, to explicitly assume an HTML attribute value context:
12he.decode('foo&ampbar', {
13  'isAttributeValue': true
14});
15// → 'foo&ampbar'

strict

The default value for the strict option is false. This means that decode() will decode any HTML text content you feed it, even if it contains any entities that cause parse errors. To throw an error when such invalid HTML is encountered, set the strict option to true. This option makes it possible to use he as part of HTML parsers and HTML validators.

1// Using the global default setting (defaults to `false`, i.e. error-tolerant mode):
2he.decode('foo&ampbar');
3// → 'foo&bar'
4
5// Passing an `options` object to `decode`, to explicitly enable error-tolerant mode:
6he.decode('foo&ampbar', {
7  'strict': false
8});
9// → 'foo&bar'
10
11// Passing an `options` object to `decode`, to explicitly enable strict mode:
12he.decode('foo&ampbar', {
13  'strict': true
14});
15// → Parse error

Overriding default decode options globally

The global default settings for the decode function can be overridden by modifying the he.decode.options object. This saves you from passing in an options object for every call to decode if you want to use a non-default setting.

1// Read the global default setting:
2he.decode.options.isAttributeValue;
3// → `false` by default
4
5// Override the global default setting:
6he.decode.options.isAttributeValue = true;
7
8// Using the global default setting, which is now `true`:
9he.decode('foo&ampbar');
10// → 'foo&ampbar'

he.escape(text)

This function takes a string of text and escapes it for use in text contexts in XML or HTML documents. Only the following characters are escaped: &, <, >, ", ', and `.

1he.escape('<img src=\'x\' onerror="prompt(1)">');
2// → '&lt;img src=&#x27;x&#x27; onerror=&quot;prompt(1)&quot;&gt;'

he.unescape(html, options)

he.unescape is an alias for he.decode. It takes a string of HTML and decodes any named and numerical character references in it.

Using the he binary

To use the he binary in your shell, simply install he globally using npm:

1npm install -g he

After that you will be able to encode/decode HTML entities from the command line:

1$ he --encode 'föo ♥ bår 𝌆 baz'
2föo ♥ bår 𝌆 baz
3
4$ he --encode --use-named-refs 'föo ♥ bår 𝌆 baz'
5föo ♥ bår 𝌆 baz
6
7$ he --decode 'föo ♥ bår 𝌆 baz'
8föo ♥ bår 𝌆 baz

Read a local text file, encode it for use in an HTML text context, and save the result to a new file:

1$ he --encode < foo.txt > foo-escaped.html

Or do the same with an online text file:

1$ curl -sL "http://git.io/HnfEaw" | he --encode > escaped.html

Or, the opposite — read a local file containing a snippet of HTML in a text context, decode it back to plain text, and save the result to a new file:

1$ he --decode < foo-escaped.html > foo.txt

Or do the same with an online HTML snippet:

1$ curl -sL "http://git.io/HnfEaw" | he --decode > decoded.txt

See he --help for the full list of options.

Support

he has been tested in at least:

• Chrome 27-50
• Firefox 3-45
• Safari 4-9
• Opera 10-12, 15–37
• IE 6–11
• Edge
• Narwhal 0.3.2
• Node.js v0.10, v0.12, v4, v5
• PhantomJS 1.9.0
• Rhino 1.7RC4
• RingoJS 0.8-0.11

Unit tests & code coverage

After cloning this repository, run npm install to install the dependencies needed for he development and testing. You may want to install Istanbul globally using npm install istanbul -g.

Once that’s done, you can run the unit tests in Node using npm test or node tests/tests.js. To run the tests in Rhino, Ringo, Narwhal, and web browsers as well, use grunt test.

To generate the code coverage report, use grunt cover.

Acknowledgements

Thanks to Simon Pieters (@zcorpan) for the many suggestions.

Author

Mathias Bynens

License

he is available under the MIT license.