[Moin-devel] case insensitive title index suggestion

[Magnus Lyckå]

At 13:12 2003-05-29 +0200, Karolina Lindqvist wrote:
>I have pages with start both with and upper case initial and a lower case.
>That causes the title index to have two groups for each letter, one for the
>upper case version and one for the lower case. A nice feature would to make
>the title index case-insensitive.
>
>As a stopgap, I changed the sort() call in _macro_TitleIndex() to use the
>following caseinsensitive_sort() function, which I found on the net, and it
>(+ one letter = letter.upper()) does the job
>
>def caseinsensitive_sort(list):
>     tuplesList = [(x.upper(), x) for x in list]
>     tuplesList.sort()
>     return [x[1] for x in tuplesList]
>
>This is supposedly faster than using a compare function to sort().

Yes, using l.sort(some_python_function) is slow, since it the fast
C-based sort function has to call the Python function over and over
again. But your sort won't do the right thing, even it its better
than a plain sort.

 >>> l = ['Hej', 'allan', 'åke', 'Östen', 'Ärling']

 >>> l.sort()
 >>> for n in l: print n,
...
  Hej allan Ärling Östen åke
 >>> l.sort(lambda x, y: cmp(x.upper(), y.upper()))
 >>> for n in l: print n,
...
  allan Hej Ärling åke Östen
 >>>

At least 'a' came before 'H' now, but 'å' should come before 'Ä'
in Sweden.

 >>> import locale
 >>> locale.setlocale(locale.LC_ALL, '')
'Swedish_Sweden.1252'
 >>> l.sort(locale.strcoll)
 >>> for n in l: print n,
...
allan Hej åke Ärling Östen

The schwartzian transform is still useful, but the locale aware version
would be (untested):

import locale
locale.setlocale(locale.LC_ALL, '')
...
def locale_aware_sort(list):
     tuplesList = [(locale.strxfrm(x), x) for x in list]
     tuplesList.sort()
     return [x[1] for x in tuplesList]


--
Magnus Lycka (It's really Lyckå), magnus at thinkware.se
Thinkware AB, Sweden, www.thinkware.se
I code Python ~ The shortest path from thought to working program