I just noticed an ommision of come character names in unicodedata module.
These are some CJK - Ideographs:

龼 (0x9fbc) - 鿋 (0x9fcb)
 (CJK Unified Ideographs [19968-40959] [0x4e00-0x9fff])

𪜀 (0x2a700) - 𫜴 (0x2b734)
(CJK Unified Ideographs Extension C [173824-177983] [0x2a700-0x2b73f])

𫝀 (0x2b740) - 𫠝 (0x2b81d)
 (CJK Unified Ideographs Extension D [177984-178207] [0x2b740-0x2b81f])

The names are probably to be generated - e.g. CJK UNIFIED IDEOGRAPH-2A700 ... etc.

(Tested with the recompiled unicodedata - using unicode 6.0; with the py 27 - builtin module (unidata_version: '5.2.0') only the first two ranges are relevant (as CJK Unified Ideographs Extension D is an adition of Unicode 6)

(Also there are the unprintable ASCII controls, surrogates and private use areas, where the missing names are probably ok.)


I tested with the following rather clumsy code:

# # # # # # # # # # # # # # # 
# wide_unichr = custom unichr emulating unicode ranges beyond FFFF on narrow python build
codepoints_missing_char_names = [[-2,-2],] # dummy
for i in xrange(0x10FFFF+1):
    if unicodedata.category(wide_unichr(i))[:1] != 'C' and unicodedata.name(wide_unichr(i), u"??noname??") == u"??noname??":
        if codepoints_missing_char_names[-1][1] == i-1:
            codepoints_missing_char_names[-1][1] = i
        else:
            codepoints_missing_char_names.append([i, i])

for first, last in codepoints_missing_char_names[1:]:
    print u"%s (%s) - %s (%s)" % (wide_unichr(first), hex(first), wide_unichr(last), hex(last),)
# # # # # # # # # # # # # # # # # # # # # # # # # # 

Unfortunately, I can't provide a fix, as unicodedata involves C code, where my knowledge is near zero.

vbr

Vlastimil Brom wrote:
> 
> New submission from Vlastimil Brom <vlastimil.brom@gmail.com>:
> 
> I just noticed an ommision of come character names in unicodedata module.
> These are some CJK - Ideographs:
> 
> 龼 (0x9fbc) - 鿋 (0x9fcb)
>  (CJK Unified Ideographs [19968-40959] [0x4e00-0x9fff])
> 
> 𪜀 (0x2a700) - 𫜴 (0x2b734)
> (CJK Unified Ideographs Extension C [173824-177983] [0x2a700-0x2b73f])
> 
> 𫝀 (0x2b740) - 𫠝 (0x2b81d)
>  (CJK Unified Ideographs Extension D [177984-178207] [0x2b740-0x2b81f])
> 
> The names are probably to be generated - e.g. CJK UNIFIED IDEOGRAPH-2A700 ... etc.

I don't think we should fill those rather big ranges with generated
names, unless there's a standard for this. There are quite a
few ranges in the Unicode database that are assigned, but don't
have a literal name associated with them.

Marc-Andre: Many of the characters you refer actually do have names assigned, even if the names don't appear in the Unicode character database. Instead, they are specified in section 4.8 of the Unicode standard, and unicodedata.c already implements that (it just wasn't updated when the ranges changed; I will look into this).

Martin v. Löwis wrote:
> 
> Martin v. Löwis <martin@v.loewis.de> added the comment:
> 
> Marc-Andre: Many of the characters you refer actually do have names assigned, even if the names don't appear in the Unicode character database. Instead, they are specified in section 4.8 of the Unicode standard, and unicodedata.c already implements that (it just wasn't updated when the ranges changed; I will look into this).

Thanks for pointing this out. I wasn't aware of there being a standard
for constructing names for CJK ideograph ranges.

For 3.2, this now fixed in r86681.

The patch for 3.1 is r86685. The patch for 2.7 is r86686.