Range processing bug.

[SainoNamkho]

Description

A sentence or two describing the issue.

Add info or delete as appropriate:

https://tex.stackexchange.com/questions/597185/range-option-not-working-properly-in-setmathfont https://tex.stackexchange.com/questions/348481/when-specifying-a-character-in-unicode-maths-range-option-why-are-alternativ

Minimal example demonstrating the issue

\documentclass{article}
\usepackage{unicode-math}
\setmathfont{Latin Modern Math}
\setmathfont{XITS Math}[range={`0-`9}]
%\setmathfont[range={}]{Latin Modern Math}

\begin{document}
    1234567890
    $ x^{2}+y^{2}=z^{2}, \xi\alpha\beta, 1234567890 $
\end{document}

The font for numbers is not set. If add something to the range

\documentclass{article}
\usepackage{unicode-math}
\setmathfont{Latin Modern Math}
\setmathfont{XITS Math}[range={`0-`9,"0391}]

\begin{document}
1234567890
$ x^{2}+y^{2}=z^{2}, \xi\alpha\beta, 1234567890 $
\end{document}

The font is changed is there's at least one code point in the final range which is located in "0391-"03A1,"03A3-"03AF,"03B1-"03BF,"03F4,"2202,"2207,"1D434-"1D454,"1D456-"1D467,"1D6E2-"1D6FA,"1D6FC-"1D714

Further details

Range	Meaning
U+0391- U+03A9	Greek Capital Ketters (U+03A2 is Undefined Character)
U+03AA, U+03AB	ΪΫ Greek Capital Letter with Dialytika
U+03AC - U+03AF	Greek Small Letters with Tonos
U+03B0 (not working)	ΰ Greek Small Letter Upsilon with Dialytika and Tonos
U+03B1 - U+03BF	Greek Small Letters from alpha to omicron
U+03C0 - U+03C9 (not working)	Greek Small Letters from pi to omega
U+03CA - U+03E1 (not working)	Some Viariants of Greak Letters
U+03F4	ϴ Another Capital Theta
U+2202，U+2207	partial and nabla
U+1D434 - U+1D467	Math letters of A-Z and a-z U+1D455 is not defined, mathmatical h is at U+210E (not working)
U+1D6E2 - U+1D6FA	Math Greek Capital Letters
U+1D6E2 - U+1D6FA	Math Greek Small Letters

[EdwinChan]

I have a related but opposite issue when unicode-math 0.8q (2020/01/31) is used with XeTeX 3.141592653-2.6-0.999995 or LuaHBTeX 1.17.0 shipped with TeX Live 2023. If I specify range={up, X}, where X is from the following table, up is ignored:

codepoints	characters
U+00391–U+003A1, U+003A3–U+003A9	Α–Ω
U+003B1–U+003C9	α–ω (incl. ο, ς)
U+003F4	ϴ
U+02202	∂
U+02207	∇
U+1D434–U+1D44D	𝐴–𝑍
U+1D44E–U+1D454, U+1D456–U+1D467	𝑎–𝑔, 𝑖–𝑧
U+1D6E2–U+1D6FA	𝛢–𝛺 (incl. 𝛩, 𝛳)
U+1D6FC–U+1D714	𝛼–𝜔 (incl. 𝜍, 𝜎)

Here is the script that prints out each offending codepoint. Be warned that it may take days to loop through all the codepoints:

import os
import subprocess

latex = 'xelatex' # or 'lualatex'

doc1 = r'''
\documentclass{article}
\usepackage{unicode-math}
\setmathfont{XITS Math}
\setmathfont{TeX Gyre Termes Math}[range={up, "
'''.strip('\n')

doc2 = r'''
}]
\begin{document}
$\symup{ABC}$
\end{document}
'''.strip('\n')

for code in range(0x20000):
  char = chr(code)
  hex  = f'{code:05X}'
  uni  = f'U+{hex}'

  try:
    subprocess.run(
      [latex, f'--jobname={uni}'],
      check=True, input=doc1+hex+doc2, stdout=subprocess.DEVNULL,
      encoding='utf8')
    result = subprocess.run(
      ['pdffonts', f'{uni}.pdf'],
      check=True, capture_output=True, encoding='utf8')
    if 'TeXGyreTermesMath' not in result.stdout:
      print(uni, char, flush=True)
  except:
    print(uni, char, 'error', flush=True)
  finally:
    os.remove(f'{uni}.log')
    os.remove(f'{uni}.aux')
    os.remove(f'{uni}.pdf')