%%%%%
%%
%% extraction.sty creates a set of commands for extracting text from
%% files.  TeX wasn't really designed for this, so other programming
%% languages might be more effective or powerful in this regard.
%% However, this works reliably.
%%
%% Commands provided:
%%
%%   \getextracts{<mark1>}{<mark2>}{<filename>}
%%   \procextracts{<mark1>}{<mark2>}{<filename>}{<out1>}{<out2>}{<in1>}{<in2>}
%%
%%   \getextractenvs{<envname>}{<filename>}
%%   \procextractenvs{<envname>}{<filename>}{<out1>}{<out2>}{<in1>}{<in2>}
%%
%%%%%


%%%%%
%% allows ! to be used in TeX command names, for this file, so
%% internal commands that use it can't be used directly/messed with
%% outside of this file.
\catcode`\!11\relax


%%%%%
%% LaTeX changes the def of TeX's \input command and makes \@@input
%% the old command.  Using \@@input technically breaks LaTeX's
%% abstraction barrier, so it's best to do it only once, here.
\let\!input\@@input


%%%%%
%% \getextracts{<mark1>}{<mark2>}{<filename>}
%%
%% Extracts text that is between <mark1> and <mark2> in <filename>.
%% If there are multiple <mark1>/<mark2> pairs in <filename>, it
%% extracts the text between all of them, in order.
%%
%% TeX groupings are ignored during extraction, so if <filename> is a
%% TeX file, there could be unbalanced braces etc. between <mark1> and
%% <mark2>.  \catcode changes _may_ cause odd effects, though often
%% DTRT; for example, \obeyspaces (if in effect when an extraction
%% command is used) causes problems with commands (in the input file)
%% that take arguments (i.e. they will take " " as the first
%% argument).
%%
%%
%% \procextracts{<mark1>}{<mark2>}{<filename>}{<out1>}{<out2>}{<in1>}{<in2>}
%%
%% Acts like \getextracts, except puts <in1> before each set of
%% extractions, <in2> after each set of extractions, <out1> before the
%% whole extraction, and <out2> after the whole extraction (set of
%% extractions -> 1 per each <mark1>/<mark2> pair).
%%
\newtoks\!toks

\def\!scantoks{%
  \expandafter\global\expandafter\!toks\expandafter{\expandafter}%
  \expandafter\scantokens\expandafter{\the\!toks}%
}

\def\!end{\!end}

\def\!changecats{%
  \advance\count0by1%
  \ifnum\count0<255%
    \let\!do\!changecats%
    \ifnum\catcode\count0=1%
      \catcode\count0=12%
    \fi%
    \ifnum\catcode\count0=2%
      \catcode\count0=12%
    \fi%
    \ifnum\catcode\count0=6%
      \catcode\count0=12%
    \fi%
  \else%
    \let\!do\relax%
  \fi%
  \!do%
}

\long\gdef\getextracts#1#2#3{\procextracts{#1}{#2}{#3}{}{}{}{}}%
\long\gdef\procextracts#1#2#3#4#5#6#7{%
  \begingroup%
    \openin0=#3\relax%
    \ifeof0%
      \typeout{! \string\getextracts: "#3" not found.}%
    \else%
      \long\def\!get##1#1##2#2##3{%
        \def\!tmp{##3}%
        \ifx\!tmp\!end%
          \let\!do\relax%
        \else%
          \def\!tmp{##2}%
          \ifx\!tmp\!end\relax\else%
            \let\!tmp\empty%
            \global\!toks\expandafter{\the\!toks#6##2#7}%
          \fi%
          \def\!do{\!get##3}%
        \fi%
        \!do%
      }%
      \count0=0%
      \!changecats%
      \global\!toks{#4}%
      \everyeof{#1\!end#2\!end}%
      \expandafter\!get\!input#3\relax%
      \global\!toks\expandafter{\the\!toks#5}%
      \aftergroup\!scantoks%
    \fi%
    \closein0\relax%
  \endgroup%
}%


%%%%%
%% \getextractenvs{<envname>}{<filename>}
%%
%% Acts like \getextracts, except uses \begin{<envname>} and
%% \end{<envname>} as <mark1> and <mark2>, respectively.
%%
%%
%% \procextractenvs{<envname>}{<filename>}{<out1>}{<out2>}{<in1>}{<in2>}
%%
%% Acts like \getextractenvs, except uses <in1>, <in2>, <out1>, and
%% <out2> like \procextracts.
%%
\def\getextractenvs#1#2{\procextractenvs{#1}{#2}{}{}{}{}}
\begingroup
  \catcode`[=1\relax%
  \catcode`]=2\relax%
  \catcode`{=12\relax%
  \catcode`}=12\relax%
  \long\gdef\procextractenvs#1[\procextracts[\begin{#1}][\end{#1}]]
\endgroup


%%%%%
%% Returns ! to its default behavior, so it can no longer be used in
%% TeX command names.
\catcode`\!12\relax


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%