Zombie driver fails when url contains "high bytes", non-ascii characters. The following example contains a valid Hungarian with accented characters.

https://hu.wikipedia.org/wiki/Műemlék

Desktop browsers and Mink Goutte driver translate the high bytes correctly:

https://hu.wikipedia.org/wiki/M%C5%B1eml%C3%A9k

Zombie driver sends string as-is to javascript, then bytes above 0x7f go wrong somewhere in Zombie:

https://hu.wikipedia.org/wiki/Mqeml\xe9k

It's a bit strange how characters are truncated:

• letter é becomes \xe9 that is character code in ISO-8859-1
• letter ű becomes q because this character does not exists in that code page

Characters that don't exist in ISO-8859-1 encoding are represented with regular letters, for example q, damage is irreversible.

Example shows that desktop browsers translate non-asci characters to percent-encoded bytes using their UTF-8 character codes:

• letter é becomes %C3%A9
• letter ű becomes %C5%B1

That's correct, web servers expect urls in this way.