sanitizeHtml throws TypeError on '&' symbol

[matejfalat]

To Reproduce

Step by step instructions to reproduce the behavior:

sanitizeHtml('<p>&</p>')
// or
sanitizeHtml('<p>&nbsp</p>')

Expected behavior

Not to crash.

Describe the bug

When the html text contains the ampersand symbol, the sanitizeHtml() is failing with:

Uncaught TypeError: Cannot read properties of undefined (reading '0')
    at Tokenizer.stateBeforeEntity (Tokenizer.js?6fbd:582:1)
    at Tokenizer.parse (Tokenizer.js?6fbd:818:1)
    at Tokenizer.write (Tokenizer.js?6fbd:158:1)
    at Parser.write (Parser.js?5804:459:1)
    at sanitizeHtml (index.js?5e22:578:1)
    at MaterialPreviewPage (MaterialPreviewPage.tsx?d2f2:41:55)

Details

React: 18.2.0,
Webpack: 5.75.0

Version of Node.js:
v18.13.0

Server Operating System:
Windows 11, WSL2, and Docker

Screenshots

[BoDonkey]

Hi @matejfalat,
I tested this on a mac using the repo tests and could not reproduce the error. Is this occurring only in the browser? I wonder if this is an htmlparser2 issue, rather than a sanitize-html issue since Tokenizer is part of that package. I guess we would need some minimal project set-up to replicate this error.
Cheers

[boutell]

Yes, what we would ask is that you contribute a PR with a failing unit test to this repo so we can see how this is possible in the context of this project and avoid any confusion with issues that might only exist in a larger project with parts that aren't actually dependencies of the module etc. htmlparser2 is a dependency so browser or no, a bug coming from that should be reproducible in a test.

[coolbueb]

probably because of missing ";" at the end

[NewEraCracker]

On Node.js CLI:

> const sanitizeHtml = require('sanitize-html');
undefined
> sanitizeHtml('<p>&</p>')
'<p>&amp;</p>'
> sanitizeHtml('<p>&nbsp</p>')
'<p> </p>'
>

Just tested this and it works good on both of these:

• With the patched version 2.7.3 (and htmlparser2 6.1.0);
• On another project using 2.13.0 (and htmlparser2 8.0.2).

Might be worth adding a test case. My two cents. Thank you.

PS: Found these related: fb55/htmlparser2#1426 & fb55/htmlparser2#1460