soft hyphens and generateAlias

[ghost]

Issue by @micha-nerdlichter
August 2nd, 2016, 09:53 GMT

Following the arguments in #2709 shouldn't generateAlias be using the shy-free title? I see no sense in replacing the shy entities with - e.g. the page title Schiff[-]fahrts[-]markt[-]bericht generates the alias schiff-fahrts-markt-bericht.

[ghost]

Comment by @leofeyer
August 2nd, 2016, 15:15 GMT

It actually does not on my system. Can this be reproduced in the online demo?

[ghost]

Comment by @Toflar
August 2nd, 2016, 20:14 GMT

I read the source code and it should really not happen. Moreover, I couldn't reproduce it in the latest 3.5 version which you should obviously be using before reporting a bug :)

[ghost]

Comment by @ADoebeling
August 2nd, 2016, 20:25 GMT

I also can't reproduce this issue.

[ghost]

Comment by @micha-nerdlichter
August 3rd, 2016, 10:33 GMT

Oh, I tried this in Contao 4.2.1 and started this issue in 3.5 Sorry, my bad. I'll check this in 3.5. But shouldn't this be fixed in 4.2.1 too?

[leofeyer]

It does not need to be fixed in Contao 3.5, because the issue does not occur there. However, I can reproduce it in Contao 4.2.

[leofeyer]

I have debugged the issue and the different behavior is caused by the Patchwork\Utf8::toAscii() method, which converts soft-hyphens to regular hyphens. In Contao 3, we are using utf8_romanize(), which leaves the soft-hyphens untouched, so they are replaced further down the code.

I would fix this by adding the following line to the standardize() function:

$strString = str_replace('­', '', $strString);

@contao/developers Any objections?

[Toflar]

I am not happy with any of these methods anyway so it's okay for me.

[aschempp]

Yeah, same here. Please also add the following to get the original behavior back…

$strString = str_replace(['ä', 'Ä', 'ö', 'Ö', 'ü', 'Ü'], ['ae', 'AE', 'oe', 'OE', 'ue', 'UE'], $strString)

[leofeyer]

That's only a small subset of "the original behavior", because it fixes the German characters only.

[leofeyer]

Also, Ä should be converted to Ae instead of AE, because you write Aepfel and not AEpfel. 😄

[aschempp]

Also, Ä should be converted to Ae instead of AE, because you write Aepfel and not AEpfel. 😄

You might be right about that. Except if the whole word is uppercase… dunno what the previous behavior was.

[leofeyer]

https://github.com/contao/core/blob/master/system/helper/utf8_lookup.php#L183

[leofeyer]

If you look at the romanize sub-array, you see what we would actually have to convert to restore the previous behavior. Maybe we should not have switched to the Utf8 library.

[leofeyer]

I have created a PR here: #569