**Describe the bug** After the patches of #722, I manually found a simple bypass of the HTML entity regex, and ran the fuzzer again to find another HTML corruption. **To Reproduce** 1. The browser decodes `:` to `:` forming a `javascript:` protocol URL (entities can be prefixed with any amount of `0`'s). ```md [Click me](javascript:alert(origin)) ``` ```html Click me ``` 2. Weird combination of syntax again resulting in raw HTML being output: ```md `)>`) ``` ```html <x> ``` **Expected behavior** Safe HTML is output. Maybe use a proper HTML entity decoder to solve the first. **Debug info** Version of library being used: https://github.com/trentm/python-markdown2/commit/ba92f30e9101f485bfd6b2a3178908ee4a82da9f (latest `master` at time of writing)
Activity
Fix zero-padded colon entities passing _safe_href check
Add the tables case from trentm#726
Fix XSS from code spans in safe mode