% \iffalse meta-comment
%
%% File: l3text-map.dtx
%
% Copyright (C) 2022-2025 The LaTeX Project
%
% It may be distributed and/or modified under the conditions of the
% LaTeX Project Public License (LPPL), either version 1.3c of this
% license or (at your option) any later version. The latest version
% of this license is in the file
%
% https://www.latex-project.org/lppl.txt
%
% This file is part of the "l3kernel bundle" (The Work in LPPL)
% and all files in that bundle must be distributed together.
%
% -----------------------------------------------------------------------
%
% The development version of the bundle can be found at
%
% https://github.com/latex3/latex3
%
% for those people who are interested.
%
%<*driver>
\documentclass[full,kernel]{l3doc}
\begin{document}
\DocInput{\jobname.dtx}
\end{document}
%</driver>
% \fi
%
% \title{^^A
% The \pkg{l3text-map} module\\ Text processing (mapping)^^A
% }
%
% \author{^^A
% The \LaTeX{} Project\thanks
% {^^A
% E-mail:
% \href{mailto:latex-team@latex-project.org}
% {latex-team@latex-project.org}^^A
% }^^A
% }
%
% \date{Released 2025-04-14}
%
% \maketitle
%
% \begin{documentation}
%
% \end{documentation}
%
% \begin{implementation}
%
% \section{\pkg{l3text-map} implementation}
%
% \begin{macrocode}
%<*code>
% \end{macrocode}
%
% \begin{macrocode}
%<@@=text>
% \end{macrocode}
%
% \subsection{Mapping to text}
%
% \subsubsection{Common code}
%
% \begin{macro}[EXP]{\@@_map_function:nnN, \@@_map_function:enN}
% \begin{macro}[EXP]{\@@_map_loop:Nnnw}
% \begin{macro}[EXP]{\@@_map_group:Nnnn}
% \begin{macro}[EXP]{\@@_map_space:Nnnw}
% \begin{macro}[EXP]{\@@_map_N_type:NnnN}
% \begin{macro}[EXP]{\@@_map_codepoint:Nnnn}
% \begin{macro}[EXP]{\@@_map_CR:Nnnw}
% \begin{macro}[EXP]{\@@_map_CR:NnnN}
% \begin{macro}[EXP]{\@@_map_class:Nnnn}
% \begin{macro}[EXP]{\@@_map_class:Nnnnn}
% \begin{macro}[EXP]{\@@_map_lookahead:Nnnnnw}
% \begin{macro}[EXP]{\@@_map_lookahead:NnnnnN}
% \begin{macro}[TF,EXP]{\@@_map_if_ignorable:n}
% \begin{macro}[EXP]{\@@_map_output:Nn}
% \begin{macro}[EXP]{\text_map_break:}
% \begin{macro}[EXP]{\text_map_break:n}
% Mapping to text all works the same way: using standard \enquote{action}
% loop on expanded text. There are different ways to determine the boundary
% conditions for breaking: to avoid duplication, the common ideas are covered
% here with the specifics split out. In all cases, anything which is not a
% character token is treated as a boundary.
% \begin{macrocode}
\cs_new:Npn \@@_map_function:nnN #1#2#3
{
\@@_map_loop:Nnnw #3 {#2} { } #1
\q_@@_recursion_tail \q_@@_recursion_stop
\prg_break_point:Nn \text_map_break: { }
}
\cs_generate_variant:Nn \@@_map_function:nnN { e }
\cs_new:Npn \@@_map_loop:Nnnw #1#2#3#4 \q_@@_recursion_stop
{
\tl_if_head_is_N_type:nTF {#4}
{ \@@_map_N_type:NnnN }
{
\tl_if_head_is_group:nTF {#4}
{ \@@_map_group:Nnnn }
{ \@@_map_space:Nnnw }
}
#1 {#2} {#3} #4 \q_@@_recursion_stop
}
\cs_new:Npn \@@_map_group:Nnnn #1#2#3#4
{
\@@_map_output:Nn #1 {#3}
{
\@@_map_loop:Nnnw #1 {#2} { } #4
\q_@@_recursion_tail \q_@@_recursion_stop
\prg_break_point:Nn \text_map_break: { }
}
\@@_map_loop:Nnnw #1 {#2} { }
}
\use:e
{ \cs_new:Npn \exp_not:N \@@_map_space:Nnnw #1#2#3 \c_space_tl }
{
\@@_map_output:Nn #1 {#3}
#1 { ~ }
\@@_map_loop:Nnnw #1 {#2} { }
}
\cs_new:Npn \@@_map_N_type:NnnN #1#2#3#4
{
\@@_if_q_recursion_tail_stop_do:Nn #4
{
\@@_map_output:Nn #1 {#3}
\text_map_break:
}
\token_if_cs:NTF #4
{
\@@_map_output:Nn #1 {#3}
#1 {#4}
\@@_map_loop:Nnnw #1 {#2} { }
}
{
\@@_codepoint_process:nN
{ \@@_map_codepoint:Nnnn #1 {#2} {#3} } #4
}
}
% \end{macrocode}
% We pull out a few special cases here. Carriage returns case needs a bit of
% context handling so has an auxiliary. Codepoint U+200D is the zero-width
% joiner, which has no context to concern us: just don't break. (These special
% cases apply to all forms of text mapping.)
% \begin{macrocode}
\cs_new:Npn \@@_map_codepoint:Nnnn #1#2#3#4
{
\@@_codepoint_compare:nNnTF {#4} = { "000D }
{
\@@_map_output:Nn #1 {#3}
\@@_map_CR:Nnnw #1 {#2} {#4}
}
{
\@@_codepoint_compare:nNnTF {#4} = { "200D }
{ \@@_map_loop:Nnnw #1 {#2} {#3#4} }
{ \@@_map_class:Nnnn #1 {#2} {#3} {#4} }
}
}
% \end{macrocode}
% A carriage return is a boundary unless it is immediately followed by
% a line feed, in which case that pair is a boundary.
% \begin{macrocode}
\cs_new:Npn \@@_map_CR:Nnnw #1#2#3#4 \q_@@_recursion_stop
{
\tl_if_head_is_N_type:nTF {#4}
{ \@@_map_CR:NnnN #1 {#2} {#3} }
{
#1 {#3}
\@@_map_loop:Nnnw #1 {#2} { }
}
#4 \q_@@_recursion_stop
}
\cs_new:Npn \@@_map_CR:NnnN #1#2#3#4
{
\@@_if_q_recursion_tail_stop_do:Nn #4
{
#1 {#3}
\text_map_break:
}
\bool_lazy_and:nnTF
{ ! \token_if_cs_p:N #4 }
{ \int_compare_p:nNn { `#4 } = { "000A } }
{
\@@_map_output:Nn #1 {#3#4}
\@@_map_loop:Nnnw #1 {#2} { }
}
{ \@@_map_loop:Nnnw #1 {#2} { } #3 }
}
% \end{macrocode}
% There are various classes of character, and we deal with them all in
% the same general way. We need to example the relevant list of codepoints:
% if we get a hit, then we do whatever the relevant action is. To keep names
% short and to allow code sharing, we have two ways of naming the functions:
% most class names are unique, so it's only where we see the same name used
% in both break classes that we need more control.
% \begin{macrocode}
\cs_new:Npn \@@_map_class:Nnnn #1#2#3#4
{
\exp_args:Nnnne \@@_map_class:Nnnnn #1 {#2} {#3} {#4}
{
\use:c { __kernel_codepoint_to_ #2 _class:n }
{ \@@_codepoint_from_chars:Nw #4 }
}
}
\cs_new:Npn \@@_map_class:Nnnnn #1#2#3#4#5
{
\cs_if_exist_use:cF { @@_map_ #5 :Nnnn }
{ \@@_map_Other:Nnnn }
#1 {#2} {#3} {#4}
}
% \end{macrocode}
% A generic loop-ahead setup: we need to handle both the previously collected
% tokens and any \enquote{conditional} ones. The latter occur when looking
% ahead for word-breaking: these \emph{may} be combined with the collected
% tokens, but if we hit the end-of-loop, need to be output separately.
% \begin{macrocode}
\cs_new:Npn \@@_map_lookahead:Nnnnnw #1#2#3#4#5#6 \q_@@_recursion_stop
{
\tl_if_head_is_N_type:nTF {#6}
{ \@@_map_lookahead:NnnnnN #1 {#2} {#3} {#4} {#5} }
{ \@@_map_loop:Nnnw #1 {#2} {#3} #4 }
#6 \q_@@_recursion_stop
}
\cs_new:Npn \@@_map_lookahead:NnnnnN #1#2#3#4#5#6
{
\@@_if_q_recursion_tail_stop_do:Nn #6
{
#1 {#3}
\tl_if_blank:nF {#4} { #1 {#4} }
}
\token_if_cs:NTF #6
{
#1 {#3}
\@@_map_loop:Nnnw #1 {#2} { } #4
}
{ \@@_codepoint_process:nN { #5 #1 {#2} {#3} {#4} } }
#6
}
% \end{macrocode}
% To deal with \enquote{ignored} characters for word break mapping: needed
% for generic |Regional_Indicator| function, so set up here.
% \begin{macrocode}
\prg_new_conditional:Npnn \@@_map_if_ignorable:n #1 { TF }
{
\str_case:nnTF {#1}
{
{ Extend } { }
{ Format } { }
{ ZWJ } { }
}
\prg_return_true:
\prg_return_false:
}
% \end{macrocode}
% For the end of the process.
% \begin{macrocode}
\cs_new:Npn \@@_map_output:Nn #1#2
{ \tl_if_blank:nF {#2} { #1 {#2} } }
\cs_new:Npn \text_map_break:
{ \prg_map_break:Nn \text_map_break: { } }
\cs_new:Npn \text_map_break:n
{ \prg_map_break:Nn \text_map_break: }
% \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}[EXP]
% {
% \@@_map_Control:Nnnn ,
% \@@_map_Newline:Nnnn ,
% \@@_map_Extend:Nnnn ,
% \@@_map_Format:Nnnn ,
% \@@_map_SpacingMark:Nnnn ,
% \@@_map_Other:Nnnn ,
% \@@_map_Regional_Indicator:Nnnn
% }
% \begin{macro}[EXP]{\@@_map_Regional_Indicator_aux:Nnnnn}
% A small number of classes appear in both forms of breaking and have the
% same behavior. For |Control| and |Newline|, we set up here as they are the
% same outcome. We have the same story for |Format|, which is functionally
% the same as |Newline|.
% \begin{macrocode}
\cs_new:Npn \@@_map_Control:Nnnn #1#2#3#4
{
\@@_map_output:Nn #1 {#3}
\@@_map_output:Nn #1 {#4}
\@@_map_loop:Nnnw #1 {#2} { }
}
\cs_new_eq:NN \@@_map_Newline:Nnnn \@@_map_Control:Nnnn
\cs_new:Npn \@@_map_Extend:Nnnn #1#2#3#4
{ \@@_map_loop:Nnnw #1 {#2} {#3#4} }
\cs_new_eq:NN \@@_map_Format:Nnnn \@@_map_Extend:Nnnn
\cs_new_eq:NN \@@_map_SpacingMark:Nnnn \@@_map_Extend:Nnnn
\cs_new:Npn \@@_map_Other:Nnnn #1#2#3#4
{
\@@_map_output:Nn #1 {#3}
\@@_map_loop:Nnnw #1 {#2} {#4}
}
% \end{macrocode}
% The Regional Indicator rule means looking ahead and dealing with the
% case where there are two in a row. So we use a look ahead to pick them
% off. As there is only one range the values are hard-coded. For
% word breaking, we also need to allow for the various extenders.
% \begin{macrocode}
\cs_new:Npn \@@_map_Regional_Indicator:Nnnn #1#2#3#4
{
\@@_map_output:Nn #1 {#3}
\@@_map_lookahead:Nnnnnw #1 {#2} {#4} { }
\@@_map_Regional_Indicator_aux:Nnnnn
}
\cs_new:Npn \@@_map_Regional_Indicator_aux:Nnnnn #1#2#3#4#5
{
\bool_lazy_or:nnTF
{ \@@_codepoint_compare_p:nNn {#5} < { "1F1E6 } }
{ \@@_codepoint_compare_p:nNn {#5} > { "1F1FF } }
{
\str_if_eq:nnTF {#2} { wordbreak }
{
\exp_args:Ne \@@_map_if_ignorable:nTF
{
\__kernel_codepoint_to_grapheme_class:n
{ \@@_codepoint_from_chars:Nw #5 }
}
{
\@@_map_lookahead:Nnnnnw #1 {#2} {#3#5} { }
\@@_map_Regional_Indicator_aux:Nnnnn
}
{ \@@_map_loop:Nnnw #1 {#2} {#3} #5 }
}
{ \@@_map_loop:Nnnw #1 {#2} {#3} #5 }
}
{ \@@_map_loop:Nnnw #1 {#2} {#3#5} }
}
% \end{macrocode}
% \end{macro}
% \end{macro}
%
% \subsection{Grapheme mapping}
%
% \begin{macro}[EXP]{\text_map_function:nN}
% \begin{macro}[EXP]{\@@_map_Prepend:Nnnn}
% \begin{macro}[EXP]{\@@_map_Prepend_aux:Nnnnn}
% \begin{macro}[EXP]{\@@_map_Prepend:Nnn}
% \begin{macro}[EXP]
% {
% \@@_map_L:Nnnn ,
% \@@_map_LV:Nnnn ,
% \@@_map_V:Nnnn ,
% \@@_map_LVT:Nnnn ,
% \@@_map_T:Nnnn
% }
% \begin{macro}[EXP]{\@@_map_hangul:Nnnw}
% \begin{macro}[EXP]{\@@_map_hangul:NnnN}
% \begin{macro}[EXP]{\@@_map_hangul:Nnnn}
% \begin{macro}[EXP]{\@@_map_hangul_aux:Nnnnw}
% \begin{macro}[EXP]{\@@_map_hangul:Nnnnw}
% \begin{macro}[EXP]{\@@_map_hangul_next:Nnnnn}
% \begin{macro}[EXP]{\@@_map_hangul_end:nw}
% \begin{macro}[EXP]
% {
% \@@_map_hangul_L:Nnn ,
% \@@_map_hangul_LV:Nnn ,
% \@@_map_hangul_V:Nnn ,
% \@@_map_hangul_LVT:Nnn ,
% \@@_map_hangul_T:Nnn
% }
% The standard lead-off for an action loop.
% \begin{macrocode}
\cs_new:Npn \text_map_function:nN #1#2
{
\@@_map_function:enN { \text_expand:n {#1} }
{ grapheme } #2
}
% \end{macrocode}
% Outputting anything earlier, the combine with what follows. The only
% exclusions are control characters.
% \begin{macrocode}
\cs_new:Npn \@@_map_Prepend:Nnnn #1#2#3#4
{
\@@_map_output:Nn #1 {#3}
\@@_map_lookahead:Nnnnnw #1 { grapheme } {#4} { }
\@@_map_Prepend_aux:Nnnnn
}
\cs_new:Npn \@@_map_Prepend_aux:Nnnnn #1#2#3#4#5
{
\bool_lazy_or:nnTF
{ \@@_codepoint_compare_p:nNn {#5} = { "000A } }
{ \@@_codepoint_compare_p:nNn {#5} = { "000D } }
{
#1 {#3}
\@@_map_loop:Nnnw #1 { grapheme } {#5}
}
{ \@@_map_Prepend:Nnn #1 {#3} {#5} }
}
\cs_new:Npn \@@_map_Prepend:Nnn #1#2#3
{
\str_if_eq:eeTF
{ Control }
{
\__kernel_codepoint_to_grapheme_class:n
{ \@@_codepoint_from_chars:Nw #3 }
}
{ \@@_map_loop:Nnnw #1 { grapheme } {#2} #3 }
{ \@@_map_loop:Nnnw #1 { grapheme } {#2#3} }
}
% \end{macrocode}
% Hangul needs additional treatment. First we have to deal with
% the start-of-Hangul position: output what we had up to now, then
% move the specialist handler. The idea here is to pick off the
% different codepoint types one at a time, tracking what else can be
% considered at each stage until we hit the end of the viable types.
% Other than that, we just keep building up the Hangul codepoints
% using a dedicated version of the loop from above.
% \begin{macrocode}
\cs_new:Npn \@@_map_L:Nnnn #1#2#3#4
{
\@@_map_output:Nn #1 {#3}
\@@_map_hangul:Nnnw
#1 {#4} { L ; V ; LV ; LVT }
}
\cs_new:Npn \@@_map_LV:Nnnn #1#2#3#4
{
\@@_map_output:Nn #1 {#3}
\@@_map_hangul:Nnnw
#1 {#4} { V ; T }
}
\cs_new_eq:NN \@@_map_V:Nnnn \@@_map_LV:Nnnn
\cs_new:Npn \@@_map_LVT:Nnnn #1#2#3#4
{
\@@_map_output:Nn #1 {#3}
\@@_map_hangul:Nnnw
#1 {#4} { T }
}
\cs_new_eq:NN \@@_map_T:Nnnn \@@_map_LVT:Nnnn
\cs_new:Npn \@@_map_hangul:Nnnw #1#2#3#4 \q_@@_recursion_stop
{
\tl_if_head_is_N_type:nTF {#4}
{ \@@_map_hangul:NnnN #1 {#2} {#3} }
{
#1 {#2}
\@@_map_loop:Nnnw #1 { grapheme } { }
}
#4 \q_@@_recursion_stop
}
\cs_new:Npn \@@_map_hangul:NnnN #1#2#3#4
{
\@@_if_q_recursion_tail_stop_do:Nn #4
{
#1 {#2}
\text_map_break:
}
\token_if_cs:NTF #4
{
#1 {#2}
\@@_map_loop:Nnnw #1 { grapheme } { }
}
{
\@@_codepoint_process:nN
{ \@@_map_hangul:Nnnn #1 {#2} {#3} } #4
}
}
\exp_args_generate:n { Nnne }
\cs_new:Npn \@@_map_hangul:Nnnn #1#2#3#4
{
\exp_args:NNnne \@@_map_hangul_aux:Nnnnw #1 {#2} {#4}
{
\__kernel_codepoint_to_grapheme_class:n
{ \@@_codepoint_from_chars:Nw #4 }
}
#3 ; \q_recursion_tail ; \q_recursion_stop
}
\cs_new:Npn \@@_map_hangul_aux:Nnnnw #1#2#3#4#5 ;
{
\quark_if_recursion_tail_stop_do:nn {#5}
{ \@@_map_loop:Nnnw #1 { grapheme } {#2} #3 }
\@@_map_hangul:Nnnnnw #1 {#2} {#3} {#4} {#5}
}
\cs_generate_variant:Nn \@@_map_hangul_aux:Nnnnw { Nnne }
\cs_new:Npn \@@_map_hangul:Nnnnnw #1#2#3#4#5#6 \q_recursion_stop
{
\str_if_eq:nnTF {#4} {#5}
{ \use:c { @@_map_hangul_ #5 :Nnn } #1 {#2} {#3} }
{ \@@_map_hangul_next:Nnnnn #1 {#2} {#3} {#4} {#6} }
}
\cs_new:Npn \@@_map_hangul_next:Nnnnn #1#2#3#4#5
{ \@@_map_hangul_aux:Nnnnw #1 {#2} {#3} {#4} #5 \q_recursion_stop }
\cs_new:Npn \@@_map_hangul_end:nw #1#2 \q_@@_recursion_stop {#1}
\cs_new:Npn \@@_map_hangul_L:Nnn #1#2#3
{
\@@_map_hangul:Nnnw
#1 {#2#3} { L V { LV } { LVT } }
}
\cs_new:Npn \@@_map_hangul_LV:Nnn #1#2#3
{
\@@_map_hangul:Nnnw
#1 {#2#3} { VT }
}
\cs_new_eq:NN \@@_map_hangul_V:Nnn \@@_map_hangul_LV:Nnn
\cs_new:Npn \@@_map_hangul_LVT:Nnn #1#2#3
{
\@@_map_hangul:Nnnw
#1 {#2#3} { T }
}
\cs_new_eq:NN \@@_map_hangul_T:Nnn \@@_map_hangul_LVT:Nnn
% \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \subsection{Word break mapping}
%
% \begin{macro}[EXP]{\text_map_function:nN}
% \begin{macro}[EXP]{\@@_map_collect:Nnnnn}
% \begin{macro}[EXP]{\@@_map_collect_auxi:nnnNnnnn}
% \begin{macro}[EXP]{\@@_map_collect_auxii:nNnnnnn}
% \begin{macro}[EXP]{\@@_map_collect_auxiii:n}
% \begin{macro}[EXP]{\@@_map_collect_auxiv:nnNnnnn}
% \begin{macro}[EXP]{\@@_map_collect_auxv:nNnnnnn}
% \begin{macro}[EXP]
% {
% \@@_map_ALetter:Nnnn ,
% \@@_map_Hebrew_Letter:Nnnn ,
% \@@_map_Katakana:Nnnn ,
% \@@_map_Numeric:Nnnn ,
% \@@_map_WSegSpace:Nnnn ,
% \@@_map_ExtendNumLet:Nnnn
% }
% \begin{macro}[EXP]{\@@_map_ExtendNumLet_auxi::Nnnnn}
% \begin{macro}[EXP]{\@@_map_ExtendNumLet_auxii:nNnn}
% The standard lead-off for an action loop.
% \begin{macrocode}
\cs_new:Npn \text_words_map_function:nN #1#2
{
\@@_map_function:enN { \text_expand:n {#1} }
{ wordbreak } #2
}
% \end{macrocode}
% The main rule for word breaking is that characters bind to following
% ones, potentially either allowing for \emph{or} totally ignoring
% intervening ones. For each class, we are passed a list of classes that
% bind and ones that we should allow in between. In all cases, the classes
% |Extend|, |Format| and |ZWJ| need to be entirely ignored: they are hard
% coded and handled separately from the in-between ones. Notice that we use
% \cs{str_case:nnTF} to make our boolean here: that way, all that needs to be
% passed internally are lists of classes.
% \begin{macrocode}
\cs_new:Npn \@@_map_collect:Nnnnn #1#2#3#4#5
{
\@@_map_lookahead:Nnnnnw #1 { wordbreak } {#2} { }
{ \@@_map_collect_auxi:nnnNnnnn {#3} {#4} {#5} }
}
\cs_new:Npn \@@_map_collect_auxi:nnnNnnnn #1#2#3#4#5#6#7#8
{
\exp_args:Ne \@@_map_collect_auxii:nNnnnnn
{
\__kernel_codepoint_to_wordbreak_class:n
{ \@@_codepoint_from_chars:Nw #8 }
}
#4 {#6} {#1} {#2} {#3} {#8}
}
% \end{macrocode}
% We now need to deal with the three possible positive outcomes of examining
% the next character. The first is that we have found one of the binding
% characters that ends the current cycle: we then pass on to the appropriate
% function. Second, we have the ignored characters: if we find these, we
% loop back around. Finally, we look at the \enquote{in-between} characters:
% if one is found, we need a further look ahead to reach a decision. Rather
% than have extra complexity in the setup, we have a hard-coded skipping of
% |ExtendNumLet| for |WSegSpace| (as |ExtendNumLet| only applies to
% |ALetter|, |Hebrew_Letter|, |Numeric| and |Katakana|).
% \begin{macrocode}
\cs_new:Npn \@@_map_collect_auxii:nNnnnnn #1#2#3#4#5#6#7
{
\str_case:neTF {#1}
{
\tl_map_function:eN
{
#4
\str_if_eq:nnF {#4} { { WSegSpace } } { { ExtendNumLet } }
}
\@@_map_collect_auxiii:n
}
{
\cs_if_exist_use:cF { @@_map_ #1 :Nnnn }
{ \@@_map_Other:Nnnn }
#2 { wordbreak } { } {#3#7}
}
{
\@@_map_if_ignorable:nTF {#1}
{ \@@_map_collect:Nnnnn #2 {#3#7} {#4} {#5} {#6} }
{
\str_case:neTF {#1}
{ \tl_map_function:nN {#5} \@@_map_collect_auxiii:n }
{
\@@_map_lookahead:Nnnnnw #2 { wordbreak } {#3} {#7}
{ \@@_map_collect_auxiv:nnNnnnn {#5} {#6} }
}
{
\@@_map_output:Nn #2 {#3}
\@@_map_loop:Nnnw #2 { wordbreak } { } #7
}
}
}
}
\cs_new:Npn \@@_map_collect_auxiii:n #1
{ \exp_not:n { {#1} { } } }
% \end{macrocode}
% We are now have a character which \emph{may} bind to the previous one if
% the next character is of the correct class also. So we carry forward the
% collected material and the conditional character, then look ahead again.
% If successful, combine together and move on using the new class, otherwise
% output and restart where we were.
% \begin{macrocode}
\cs_new:Npn \@@_map_collect_auxiv:nnNnnnn #1#2#3#4#5#6#7
{
\exp_args:Ne \@@_map_collect_auxv:nNnnnnn
{
\__kernel_codepoint_to_wordbreak_class:n
{ \@@_codepoint_from_chars:Nw #7 }
}
#3 {#5} {#6} {#1} {#2} {#7}
}
\cs_new:Npn \@@_map_collect_auxv:nNnnnnn #1#2#3#4#5#6#7
{
\str_case:neTF {#1}
{ \tl_map_function:nN {#6} \@@_map_collect_auxiii:n }
{ \use:c { @@_map_ #1 :Nnnn } #2 { wordbreak } { } {#3#4#7} }
{
\@@_map_if_ignorable:nTF {#1}
{
\@@_map_lookahead:Nnnnnw #2 { wordbreak } {#3} {#4#7}
{ \@@_map_collect_auxiv:nnNnnnn {#5} {#6} }
}
{
\@@_map_output:Nn #2 {#3}
\@@_map_loop:Nnnw #2 { wordbreak } { } #4#7
}
}
}
% \end{macrocode}
% Use the generic collector.
% \begin{macrocode}
\cs_new:Npn \@@_map_ALetter:Nnnn #1#2#3#4
{
\@@_map_output:Nn #1 {#3}
\@@_map_collect:Nnnnn #1 {#4}
{ { ALetter } { Hebrew_Letter } { Numeric } }
{ { MidLetter } { MidNumLet } { Single_Quote } }
{ { ALetter } { Hebrew_Letter } }
}
\cs_new:Npn \@@_map_Hebrew_Letter:Nnnn #1#2#3#4
{
\@@_map_output:Nn #1 {#3}
\@@_map_collect:Nnnnn #1 {#4}
{ { ALetter } { Hebrew_Letter } { Numeric } { Single_Quote } }
{ { MidLetter } { MidNumLet } { Double_Quote } }
{ { Hebrew_Letter } }
}
\cs_new:Npn \@@_map_Katakana:Nnnn #1#2#3#4
{
\@@_map_output:Nn #1 {#3}
\@@_map_collect:Nnnnn #1 {#4} { { Katakana } } { } { }
}
\cs_new:Npn \@@_map_Numeric:Nnnn #1#2#3#4
{
\@@_map_output:Nn #1 {#3}
\@@_map_collect:Nnnnn #1 {#4}
{ { ALetter } { Hebrew_Letter } { Numeric } }
{ { MidNum } { MidNumLet } { Single_Quote } }
{ { Numeric } }
}
\cs_new:Npn \@@_map_WSegSpace:Nnnn #1#2#3#4
{
\@@_map_output:Nn #1 {#3}
\@@_map_collect:Nnnnn #1 {#4} { { WSegSpace } } { } { }
}
% \end{macrocode}
% We should only get here in the case we have a \enquote{dangling} extender.
% If so, look ahead for characters to bind to, then for the set of three
% that we need to skip over.
% \begin{macrocode}
\cs_new:Npn \@@_map_ExtendNumLet:Nnnn #1#2#3#4
{
\@@_map_output:Nn #1 {#3}
\@@_map_lookahead:Nnnnnw #1 { wordbreak } {#4} { }
\@@_map_ExtendNumLet_auxi:Nnnnn
}
\cs_new:Npn \@@_map_ExtendNumLet_auxi:Nnnnn #1#2#3#4#5
{
\exp_args:Ne \@@_map_ExtendNumLet_auxii:nNnn
{
\__kernel_codepoint_to_wordbreak_class:n
{ \@@_codepoint_from_chars:Nw #5 }
}
#1 {#3} {#5}
}
\cs_new:Npn \@@_map_ExtendNumLet_auxii:nNnn #1#2#3#4
{
\str_case:nnTF {#1}
{
{ ALetter } { }
{ Hebrew_Letter } { }
{ Numeric } { }
{ Katakana } { }
{ ExtendNumLet } { }
}
{
\cs_if_exist_use:cF { @@_map_ #1 :Nnnn } % TEMP?
{ \@@_map_Other:Nnnn }
#2 { wordbreak } { } {#3#4}
}
{
\@@_map_if_ignorable:nTF {#1}
{
\@@_map_lookahead:Nnnnnw #2 { wordbreak } {#3#4} { }
\@@_map_ExtendNumLet_auxi:Nnnnn
}
{
\@@_map_output:Nn #2 {#3}
\@@_map_loop:Nnnw #2 { wordbreak } { } #4
}
}
}
% \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \subsection{Inline mappings}
%
% \begin{macro}{\text_map_inline:nn, \text_words_map_inline:nn}
% The standard non-expandable inline version.
% \begin{macrocode}
\cs_new_protected:Npn \text_map_inline:nn #1#2
{
\int_gincr:N \g__kernel_prg_map_int
\cs_gset_protected:cpn
{ @@_map_ \int_use:N \g__kernel_prg_map_int :w } ##1 {#2}
\exp_args:Nnc \text_map_function:nN {#1}
{ @@_map_ \int_use:N \g__kernel_prg_map_int :w }
\prg_break_point:Nn \text_map_break:
{ \int_gdecr:N \g__kernel_prg_map_int }
}
\cs_new_protected:Npn \text_words_map_inline:nn #1#2
{
\int_gincr:N \g__kernel_prg_map_int
\cs_gset_protected:cpn
{ @@_map_ \int_use:N \g__kernel_prg_map_int :w } ##1 {#2}
\exp_args:Nnc \text_words_map_function:nN {#1}
{ @@_map_ \int_use:N \g__kernel_prg_map_int :w }
\prg_break_point:Nn \text_map_break:
{ \int_gdecr:N \g__kernel_prg_map_int }
}
% \end{macrocode}
% \end{macro}
%
% \begin{macrocode}
%</code>
% \end{macrocode}
%
% \end{implementation}
%
% \PrintIndex