[TYPO3-dev] String Handling in Extensions

[Steffen Kamper]

"Ries van Twisk" <typo3 at rvt.dds.nl> schrieb im Newsbeitrag 
news:mailman.139891.1176409925.21067.typo3-dev at lists.netfielders.de...
>
>>
>> "Ries van Twisk" <typo3 at rvt.dds.nl> schrieb im Newsbeitrag
>> news:mailman.139617.1176404290.21067.typo3-dev at lists.netfielders.de...
>>> hey Steffen,
>>>
>>> as soon as you work with other locales that can be multi-byte
>>> then you should use these functions.
>>> This is also true for values that come from posts/get's
>>> when the website was set to multu-byte like utf-8.
>>>
>>> Ries
>>>
>>
>> Hi Ries,
>>
>> you're right. So if the extension should work with all multibyte  langs. 
>> it
>> should only use this functions.
>> But i've not seen them used in extionsions before. Does this means  an 
>> ext
>> like tt_news gets in trouble on a chinese page ?
> Yes :)
>
> Personally I don't use them much myself...
> One solution what chinese people could do is
> map the mb version to the single byte versions
> and they are good again.. properly they have done that by default?!?!

How does that works ? I never heard of this possibility.

>
> Even core doesn't do it correct, since I had a template in Estonia 
> language
> that was not parsed correctly and it happend to be
> down the line that one of the string functions went 'haywire'.

so this should be reported to bt

>
> The good/bad thing is that errors don't happen to often thank
> to the smart guys of unicode.com that created utf-8.
>
utf8 is multibyte too, string function that work with byte fails.

> However since errors are hard to detect (can you read chinese??)
No :-) I'm happy that i didn't worked with utf16 or 32, utf8 is hard enough 
:-)


> it's also hard to debug and unfortunately PHP 'thinks' latin1.

php with ki ? (just kidding)

>
> So in short, extensions might get into trouble,
> they don't have to get in trouble.
>
> For example concate functions like . will goes likly ok.
> strlen can and will go wrong and returns the incorrect
> number of characters (strlen returns bytes).
>
> Other functions like strpos() might go wrong,
> but you can use strpos to find a chinese character
> if you like, but think binary!!!
>
for this the cs-library is expected - we should use it.

> typo3 tries to solve it with the locale class,
> I used it a couple of times where I knew
> it could go wrong.... But like you and many other
> we don't use them often enough!!!
>
>
> Ries
>

Yes, this shows this very clear. Maybe an article can help to spread the 
info to the devs.
I don't want to say, that we have to use it, but every dev should know about 
the problems behind.

vg  Steffen

>
>
>
>
>
>>
>> vg  Steffen
>>
>>> On Apr 12, 2007, at 1:36 PM, Steffen Kamper wrote:
>>>
>>>>
>>>> "Ingo Renner" <typo3 at ingo-renner.com> schrieb im Newsbeitrag
>>>> news:mailman.1.1176402192.13172.typo3-dev at lists.netfielders.de...
>>>>> Steffen Kamper wrote:
>>>>>
>>>>>> I looked in the core for an universal method, but didn't found a
>>>>>> strict
>>>>>> rule.
>>>>>> How should this be done ? mb-functions are not available on all
>>>>>> systems,
>>>>>> and there is a flag in installtool, iconv also, ...
>>>>>
>>>>> AFAIR there're some methods in csConv ...
>>>>>
>>>>> Ingo
>>>>
>>>> Hi Ingo,
>>>>
>>>> good hint, ty  - i didn't realized before.
>>>> There are functions in t3lib_cs like
>>>> substr
>>>> strlen
>>>> crop
>>>> etc.
>>>>
>>>> So every input string should be handled with tis methods to be sure.
>>>> Additional question: Are the form values has to be handled same  way ?
>>>>
>>>> vg Steffen
>>>>
>>>>
>>>> _______________________________________________
>>>> TYPO3-dev mailing list
>>>> TYPO3-dev at lists.netfielders.de
>>>> http://lists.netfielders.de/cgi-bin/mailman/listinfo/typo3-dev
>>>
>>> -- 
>>> Ries van Twisk
>>> Freelance Typo3 Developer
>>> email: ries at vantwisk.nl
>>> web:   http://www.rvantwisk.nl/
>>> skype: callto://r.vantwisk
>>>
>>>
>>>
>>>
>>
>>
>> _______________________________________________
>> TYPO3-dev mailing list
>> TYPO3-dev at lists.netfielders.de
>> http://lists.netfielders.de/cgi-bin/mailman/listinfo/typo3-dev
>
> -- 
> Ries van Twisk
> Freelance Typo3 Developer
> email: ries at vantwisk.nl
> web:   http://www.rvantwisk.nl/
> skype: callto://r.vantwisk
>
>
>
>