unicodedomino_kernel_better_decode.def

% -*- mode: tex -*-
%-
% See unicodedomino.sty for copyright and licence terms. Furthermore
% this file is dual-licenced under the LPPL version 1.3c or later.
%-
% Fix check for illegal sequences to fail overlong encoded sequences
% as well as codepoints outside of the UCS range [0;10FFFF]; improve
% UTF-8 decoding with fixed-up check code and a safer UTF-8 decoder.
% This is https://github.com/latex3/latex2e/pull/83 (rejected).

% new check for illegal sequences
\gdef\UTFviii@checkseq#1:#2#3\empty{%
 \ifnum`#2<"80 %
  \ifx\empty#3\empty%
  \else%
   1%
  \fi%
 \else%
  \ifnum`#2<"C2 %
   1%
  \else%
   \ifnum`#2<"E0 %
    % one "80‥"BF
    \UTFviii@check@one#3\empty%
   \else%
    \ifnum`#2<"E1 %
     % "A0‥"BF + one "80‥"BF
     \UTFviii@check@two"A0.#3\empty%
    \else%
     \ifnum`#2<"F0 %
      % two "80‥"BF
      \UTFviii@check@two"80.#3\empty%
     \else%
      \ifnum`#2<"F1 %
       % "90‥"BF + two "80‥"BF
       \UTFviii@check@three"90."BF.#3\empty%
      \else%
       \ifnum`#2<"F4 %
        % three "80‥"BF
        \UTFviii@check@three"80."BF.#3\empty%
       \else%
        \ifnum`#2<"F5 %
         % "80‥"8F + two "80‥"BF
         \UTFviii@check@three"80."8F.#3\empty%
        \else%
         1%
        \fi%
       \fi%
      \fi%
     \fi%
    \fi%
   \fi%
  \fi%
 \fi%
}%

% check last trail octet in the range "80‥"BF
\gdef\UTFviii@check@one#1#2\empty{%
 \ifx\empty#2\empty%
  \ifnum`#1<"80 %
   1%
  \else%
   \ifnum`#1>"BF %
    1%
   \fi%
  \fi%
 \else%
  1%
 \fi%
}%

% check second-to-last trail octet in the range #1‥"BF
\gdef\UTFviii@check@two#1.#2#3\empty{%
 \ifx\empty#3\empty%
  1%
 \else%
  \ifnum`#2<#1 %
   1%
  \else%
   \ifnum`#2>"BF %
    1%
   \else%
    \UTFviii@check@one#3\empty%
   \fi%
  \fi%
 \fi%
}%

% check third-to-last trail octet in the range #1‥#2
\gdef\UTFviii@check@three#1.#2.#3#4\empty{%
 \ifx\empty#4\empty%
  1%
 \else%
  \ifnum`#3<#1 %
   1%
  \else%
   \ifnum`#3>#2 %
    1%
   \else%
    \UTFviii@check@two"80.#4\empty%
   \fi%
  \fi%
 \fi%
}%

% changed calling API for \UTFviii@checkseq (empty+relax)
\def\UTFviii@defined#1{%
 \ifx#1\relax%
  \if\relax\expandafter\UTFviii@checkseq\string#1\empty\relax%
   \UTFviii@undefined@err{#1}% from v1.2a 2018/03/24
   % not needed in unicodedomino_compat.def though because the
   % \UTFviii@defined macro is reimplemented by unicodedomino.sty
  \else%
   \PackageError{inputenc}{Invalid UTF-8 byte sequence}%
    \UTFviii@invalid@help
  \fi%
 \else%
  \expandafter#1%
 \fi%
}%

% decode octets to codepoint number, safer
\gdef\decode@UTFviii#1\relax{%
 \the\numexpr(\UTFviii@decode0:#1\relax)%
}%

% safer decode, returns "1FFFFF for illegal sequences
\gdef\UTFviii@decode#1\relax{%
 \if\relax\expandafter\UTFviii@checkseq\string#1\empty\relax%
  \UTFviii@dec@lead#1\relax%
 \else%
  2097151%
 \fi%
}%

% decode lead octet
\gdef\UTFviii@dec@lead#1:#2#3\relax{%
 % we know #2 is in "00‥"7F, "C2‥"F4
 \ifnum`#2<"80 %
  `#2%
 \else%
  \ifnum`#2<"E0 %
   (`#2-"C0%
  \else%
   \ifnum`#2<"F0 %
    ((`#2-"E0%
   \else%
    (((`#2-"F0%
   \fi%
  \fi%
  \UTFviii@dec@trail#3\relax%
 \fi%
}%

% decode trail octets recursively
\gdef\UTFviii@dec@trail#1#2\relax{%
 )*64+(`#1-"80)%
 \ifx\relax#2\else\UTFviii@dec@trail#2\relax\fi%
}%