# Within-word joiner code points (hex, one per line).
#
# The separator set for `seg_word`, which detects single-letter segmentation
# (`v.i.a.g.r.a`). It recognised three characters — . _ - — while Unicode has a
# whole category for joining parts of one word. #750 measured 16 of them silent
# on every path, and U+2010 HYPHEN disagreeing with U+002D HYPHEN-MINUS while
# rendering identically.
#
# This snapshot: 38 code points (10 Pc, 27 Pd, 1 Po).
#
# Derived from the general category, not curated, so a Unicode release that adds
# a dash cannot leave a hole. U+002E FULL STOP is added explicitly: it is Po, and
# widening the test to Po would pull in ? ! @ and every other punctuation mark.
#
# Unicode 17.0.0, generated by CPython 3.15.0a1.
# Regenerate: uv run --python 3.15 python scripts/gen_word_joiners.py
002D
002E
005F
058A
05BE
1400
1806
2010
2011
2012
2013
2014
2015
203F
2040
2054
2E17
2E1A
2E3A
2E3B
2E40
2E5D
301C
3030
30A0
FE31
FE32
FE33
FE34
FE4D
FE4E
FE4F
FE58
FE63
FF0D
FF3F
10D6E
10EAD