Thread: Uppercase version of ß desired

On 2023-03-13 17:38:51 -0400, Celia McInnis wrote:
> I would be really happy if postgresql  had an upper case version of the ß
> german character. The wiki page
> https://en.wikipedia.org/wiki/%C3%9F
>
> indicates that the capital (U+1E9E ẞ LATIN CAPITAL LETTER SHARP S) was encoded
> by ISO 10646 in 2008.

The character is there, of course, and lower-casing it works as
expected:

hjp=> select 'ẞ', lower('ẞ');
╔══════════╤═══════╗
║ ?column? │ lower ║
╟──────────┼───────╢
║ ẞ        │ ß     ║
╚══════════╧═══════╝
(1 row)

But the 'ß' is a bit special as it is usually uppercased to 'SS'
(although 'ẞ' is now officially allowed, too).

Apparently your (and my) locale doesn't uppercase ß at all, which isn't
correct according to German spelling rules but was very common in the
last decades.

You can specify an alternate locale:

hjp=> select upper('ß');
╔═══════╗
║ upper ║
╟───────╢
║ ß     ║
╚═══════╝
(1 row)


hjp=> select upper('ß' collate "de-AT-x-icu");
╔═══════╗
║ upper ║
╟───────╢
║ SS    ║
╚═══════╝
(1 row)


The challenge now is to find a locale which uppercases ß to ẞ.

        hp

--
   _  | Peter J. Holzer    | Story must make more sense than reality.
|_|_) |                    |
| |   | hjp@hjp.at         |    -- Charles Stross, "Creative writing
__/   | http://www.hjp.at/ |       challenge!"

"Peter J. Holzer" <hjp-pgsql@hjp.at> writes:
> On 2023-03-13 17:38:51 -0400, Celia McInnis wrote:
>> I would be really happy if postgresql had an upper case version of the ß
>> german character.

> But the 'ß' is a bit special as it is usually uppercased to 'SS'
> (although 'ẞ' is now officially allowed, too).
> Apparently your (and my) locale doesn't uppercase ß at all, which isn't
> correct according to German spelling rules but was very common in the
> last decades.

Our code for libc locales doesn't support upcasing 'ß' to 'SS',
because it uses towlower() which can only manage
one-character-to-one-character transformations.  It should work for
upcasing to 'ẞ', but as you say, you need to find a locale that thinks
that should happen.

You might have better luck if you have a version of Postgres that
supports ICU and you can use an ICU locale.  That code path doesn't
appear to have any hard-wired assumption about how many characters
in convert to how many out.

            regards, tom lane

On Tue, 14 Mar 2023, Celia McInnis wrote:

>uc_alphabet = lc_alphabet.replace('ß', 'ẞ').upper()

That’s probably for the best. The uppercase Eszett was only added
to Unicode under the rule that the lowercase Eszett’s case rules
are kept unchanged, and the former’s considered normally only ever
typed manually.

Of course, the grammar rules about uppercasing ß have since changed,
but since there’s two valid ways, choosing is the application’s duty.

bye,
//mirabilos
--
15:41⎜<Lo-lan-do:#fusionforge> Somebody write a testsuite for helloworld :-)

> On Mar 13, 2023, at 5:38 PM, Celia McInnis <celia.mcinnis@gmail.com> wrote:
>
> HI:
>
> I would be really happy if postgresql  had an upper case version of the ß german character. The wiki page
> https://en.wikipedia.org/wiki/%C3%9F
>
> indicates that the capital (U+1E9E ẞ LATIN CAPITAL LETTER SHARP S) was encoded by ISO 10646 in 2008.
>
> BTW the reason that I'd like upper('ß') to give something different than 'ß'  is because I have written a simple
substitutionpuzzle for a large number of languages where I show the encrypted lower case words in upper case and the
successfulletter substitution submissions in lower case - so I need the upper and lower case versions of each letter to
bedifferent! 
>
> Thanks for any assistance! Maybe I can hack what I want in python (which is what I am using for the puzzle).

Hi Celia,
I ran into this too back when we were transitioning from Python 2 to 3 (2 behaved differently from 3). While
researchingit I discovered this Python issue which maybe sheds some additional light on the subject:
https://github.com/python/cpython/issues/74993

We ultimately found 90 characters that (under Python 3) grew longer when uppercased.

python -c "print([c for c in range(0x80, 0x22ff) if len(chr(c)) != len(chr(c).upper())])”


I hope this is at least interesting. :-)

Cheers
Philip