fix: Don't parse tags inside words & URLs

[TheBlob42]

fixes #265

It is an easy solution but it should cover at least all cases I can think of 🤔
Is there a nicer way for "early returns" in F#?

[TheBlob42]

Some non-alphanumeric characters are allowed in URLs that could potentially appear before the # character (-, _, %, ., ;, = etc.). I personally don't know about any website that uses such URLs, but maybe someone knows a little more about it.

Here are some examples of made up valid URLs that my fix would not catch:

• www.demo-#demo.com
• www.demo_#demo.com
• www.demo.com?#demo

[artempyanykh]

Looks reasonable and will likely cover 99% of real-world cases. Thanks for the fix.

P.S. Early return is a bit of a pain in F#. You can emulate it using a computational expression and a guard, but this would be an overkill in this case.

[petobens]

Hi! So something like https://tableau.com/#/site/Sales/views/elastico/Elstico is expected not to be caught? Beyond theses cases (I have a few of those) things work much better now.

[TheBlob42]

Yes you are correct this is currently not caught. And with #268 now which allows / inside tags all of #/site/Sales/views/elastico/Elstico would be considered a tag 😅 to completely filter out URLs would require more work in regards to the parser