[comparefamily] [agd] trouble with five-digit code points

[frankrolf]

I noticed the comparefamily report for some code points in Source Serif was suspicious:

	Warning. Font uses working name 'u1F16A' rather than Adobe final name 'u1F160'. SourceSerif4-Regular
	Warning. Font uses working name 'u1F16B' rather than Adobe final name 'u1F160'. SourceSerif4-Regular

The code points in question are

• U+1F16A RAISED MC SIGN 🅪
• U+1F16B RAISED MD SIGN 🅫

comparefamily suggests the final name (and hence the code point) should be U+1F160 NEGATIVE CIRCLED LATIN CAPITAL LETTER Q 🅠 – wrong, of course.

When investigating, I noticed that the method for getting the final glyph name returns false results for almost all five-digit code points:

import os
import unicodedata
from afdko import agd, fdkutils
print(agd.__file__)
resources_dir = fdkutils.get_resources_dir()
kAGD_TXTPath = os.path.join(resources_dir, "AGD.txt")
print('AGD file used:', kAGD_TXTPath)
fp = open(kAGD_TXTPath, "r")
agdTextPath = fp.read()
fp.close()
gAGDDict = agd.dictionary(agdTextPath)

codepoint_start = int('10000', 16)
codepoint_end = int('FFFFF', 16) + 1
for i in range(codepoint_start, codepoint_end):
    hex_value = '{0:5X}'.format(i)
    u_name = 'u' + hex_value

    if gAGDDict.glyph(u_name):
        print(u_name)
        try:
            ud_name = unicodedata.name(chr(i))
            print(ud_name)
        except ValueError:
            print('no unicode name')
        print(gAGDDict.glyph(u_name).name)
        print(gAGDDict.glyph(u_name).uni)  # gets truncated to 4 digits
        print()

(this code is mostly pulled from comparefamily.py)

Looking through agd.py it seems that five-digit Unicode names are simply not expected:

afdko/python/afdko/agd.py

Line 47 in 20b4817

self.uni = re.compile(r'[0-9A-F]{4}').search(x).group() # Unicode value

afdko/python/afdko/agd.py

Line 63 in 20b4817

u = re.compile(r'uni([0-9A-F]{4})$') # Unicode value regexp

I fixed this in my local agd.py, and will send a PR in a minute.