How to add the word “Chapter” before chapters in both table of contents using TeX4ht ⁠/ TeX4ebook?

With Michal’s hints, I’ve arrived at the following (overengineered-feeling) solution. It satisfies the following requirements:

• Chapters follow the format Chapter 1: Chapter Title in the table of contents inside the document.
  • Everything is part of the <a> link instead of Chapter 1: being outside, as would normally be the case with TeX4ht.
• Chapters follow the format Chapter 1: Chapter Title in the table of contents shown in the e-reader’s user interface.
• In addition, it lets you change the types of divisions that should be included in the table of contents in one place, and applies that everywhere.

It works by changing these things:

• Use TeX4ht’s \TableOfContents in place of the default \tableofcontents.
• \Configure{tableofcontents} to include e.g. likechapter.
• Use \Configure{NavSection} and \peek_regex_replace_once to capture the division number.
• Add Chapter , the captured division number, and : to the in-document table of contents using \Configure{TocLink}. (This regex solution is highly hacky – let me know if anyone knows of a better way to move the number into the <a> link!)
• Add Chapter and : to the NCX table of contents with a deferred \Configure{NavSection}.

xhtml.cfg

\Preamble{xhtml}

% Edit this to change what the table of contents should record.
% (Doesn’t support KOMA-Script’s `\addchap` – this could probably be
% implemented, but let’s not unless someone needs it.)
\def\tableofcontentsincludes{part,likepart,appendix,chapter,likechapter,section,likesection}


% Replace LaTeX’s `\tableofcontents` (which doesn’t record `\chapter*`s)
% with TeX4Ht’s automatic `\TableOfContents` (which can).
\ExplSyntaxOn
\RenewDocumentCommand{\tableofcontents}{}{%
    % https://tex.stackexchange.com/a/26489/392788
    \exp_args:NNV \TableOfContents[\tableofcontentsincludes]%
}
\ExplSyntaxOff


% What we want is for the table of contents to say “Chapter 1: Chapter Title”
% instead of simply “1 Chapter Title”. The way EPUB 3 works, it has two tables
% of contents for the document: One in the document (X)HTML, and one in the
% NCX file. It also has two different places to *display* the table of contents
% for the document: Inside the actual book, and in the user interface.
% The table of contents inside the actual book is, of course, the one in the
% document (X)HTML. Testing, however, reveals that the user interface ToC comes
% from one of two sources: either from the NCX file, *or* from the document
% (X)HTML *if it has the attribute `epub:type="toc"`. (This might vary from
% e-reader to e-reader.) To that end, if we want “Chapter 1: Chapter Title”,
% we should make sure that it says so both in the document (X)HTML and the NCX
% file.

\catcode`\:=11

% Here we add a custom heading.
\ExplSyntaxOn
\Configure{NavMap}
    {% Before the table of contents.
        \ifvmode\IgnorePar\fi\EndP%
        \boolfalse{tocnoempty}%
        \global\advance\:toccount by1%
        \HCode{<nav\space id="toc\the\:toccount"\space class="toc"\ifnum\:toccount<2 \space epub:type="toc"\space role="doc-toc"\fi>\Hnewline}%
        % Comment the above line and uncomment the line below to use the NCX
        % table of contents in the user interface.
        % \HCode{<nav\space id="toc\the\:toccount"\space class="toc">\Hnewline<ol>}%
        % Custom heading for the table of contents.
        \HCode{<div\space class="heading\space chapter-heading\space without-number"><h2\space class="chapter-title">}\contentsname\HCode{</h2></div>\Hnewline}%
        \HCode{<ol>}%
        \opf:registerfilename{\FileName}%
        \ifnum\:toccount<2 \opf:add:property{nav}\fi%
    }
    {% At the end of the table of contents.
        \exp_args:Ne \usetoclevels{\tableofcontentsincludes}%
        \ifbool{tocnoempty}{}{\HCode{<li><a href="\jobname.\:html">Document</a></li>}}%
        \HCode{</ol></nav>}%
    }
\ExplSyntaxOff

% Here, we configure the table of contents to treat `likechapter` (i.e.
% `\chapter*{…}` like it does `chapter` (i.e. `\chapter{…}`). At the time of
% writing, TeX4ebook does not do this by default.
\ExplSyntaxOn
\seq_new:N \tableofcontentsincludes_seq
\seq_new:N \cur_potential_divisions_to_close_seq
\seq_new:N \cur_divisions_to_close_seq
\seq_set_split:NnV \tableofcontentsincludes_seq {,}{\tableofcontentsincludes}
\Configure{tableofcontents}
    {% Before ToC.
        \a:NavMap%
        %
        \resettoclevels{part,likepart,appendix,chapter,likechapter,section,likesection,subsection,likesubsection,subsubsection,likesubsubsection}%
        %
        % We have to configure each division we want using `\navsection` (see
        % `tex4ebook-epub3.4ht` in the TeX4ebook source).' Since we’ve made the
        % divisions variable, we have to generate these dynamically.
        \tl_new:N \null_tl% Throwaway variable.
        \seq_set_eq:NN \cur_potential_divisions_to_close_seq \tableofcontentsincludes_seq%
        \seq_set_eq:NN \cur_divisions_to_close_seq \tableofcontentsincludes_seq%
        \seq_map_inline:Nn \tableofcontentsincludes_seq {%
            % A little niggle is that, for example, `\navsection{likechapter}` should
            % include `chapter`, i.e. `\navsection{likechapter}{chapter, likechapter, …}`.
            % Thus, we only update the running list of sections to close when
            % we encounter a non-`like` division.
            \tl_if_regex_match:nnF {##1}{^like}{\seq_set_eq:NN \cur_divisions_to_close_seq \cur_potential_divisions_to_close_seq}%
            \exp_args:Nne \navsection
                {##1}
                {\seq_use:Nn \cur_divisions_to_close_seq {,}}%
            \seq_gpop:NN \cur_potential_divisions_to_close_seq \null_tl%
        }%
        %
        \Configure{toTocLink}{}{}%
    }
    {\b:NavMap} % End of ToC.
    {} % After ToC.
    {} % Before non-indented \par.
    {} % Before indented \par.
\ExplSyntaxOff

% We’ll need these for turning “1 Chapter Title” into “Chapter 1: Chapter
% Title”.
\ExplSyntaxOn
\NewDocumentCommand{\curr:sect:prefix}{}{%
    \str_case_e:nnT{\curr:sect:type}{%
        {part}{Part}
        {chapter}{\chaptername}
    }{\space}%
}

\NewDocumentCommand{\curr:sect:colon}{}{%
    \str_case_e:nnT{\curr:sect:type}{%
        {part}{:}
        {chapter}{:}
        {section}{:}
        {subsection}{:}
        {subsubsection}{:}
    }{\space}%
}
\ExplSyntaxOff

% We want the label `Chapter 1:` to be part of the `<a>` link, not outside
% as it is by default.
\NewDocumentCommand{\curr:sect:number}{}{}
\Configure{TocLink}{%
    \Link{#2}{#3}%
    \curr:sect:prefix\curr:sect:number\curr:sect:colon #4%
    \EndLink%
}

% This is a bit of a funky workaround: We’re capturing the division number
% using regex below. In order to do this, we need a single csname to call.
\NewDocumentCommand{\curr:sect:number:set}{m}{%
    \RenewDocumentCommand{\curr:sect:number}{}{#1}%
}

% Here, we change “1 Chapter Title” to “Chapter 1: Chapter Title” in the
% in-document (X)HTML table of contents.
\ExplSyntaxOn
\Configure{NavSection}
    {% Before the division number.
        \booltrue{tocnoempty}%
        \HCode{<li>}%
        \peek_regex_replace_once:nnF{([^\d.]?[\d.]+)}
            {\c{curr:sect:number:set}\{\0\}}
            {\RenewDocumentCommand{\curr:sect:number}{}{}}%
    }
    {\HCode{<ol>\Hnewline}} % Between the division title and its children.
    {} % In between division number and division title.
    {\HCode{</ol></li>}} % After the division’s children.
\ExplSyntaxOff

% Here, we turn “1 Chapter Title” into “Chapter 1: Chapter Title” in the NCX
% table of contents. Truly an ugly hack among ugly hacks, but it works!
% We have to `\Configure{NavSection}` at the correct point in `tex4ebook.4ht`,
% because if we do it now, it’ll just get overwritten.
\makeatletter
\ExplSyntaxOn
\NewCommandCopy{\ncx:head@original}{\ncx:head}
\NewCommandCopy{\TableOfContents@original}{\TableOfContents}
\RenewDocumentCommand{\ncx:head}{}{%
    \ncx:head@original%
    %
    \Configure{NavSection}
        {%
            \booltrue{tocnoempty}%
            \HCode{\Hnewline<navPoint\space id="navPoint-}%
            \stepnavpoint%
            \HCode{"\space playOrder="}\the\navpoint%
            \HCode{">\Hnewline}%
            \HCode{<navLabel>\Hnewline<text><navmark\space type="\curr:sect:type">}%
            %
            \curr:sect:prefix%
        }
        {%
            \HCode{</text>\Hnewline</navLabel>\Hnewline}%
            \HCode{<content\space src="\navmapsrc"\space />}%
        }
        {%
            \HCode{</navmark>}%
            \curr:sect:colon%
        }
        {\HCode{</navPoint>\Hnewline}}%
    %
    \RenewDocumentCommand{\TableOfContents}{o}{%
        \exp_args:NNV \TableOfContents@original[\tableofcontentsincludes]%
        \RenewCommandCopy{\TableOfContents}{\TableOfContents@original}%
    }%
}
\ExplSyntaxOff
\makeatother

\catcode`:=12

\begin{document}
\EndPreamble

test.tex

\documentclass{scrbook}

% Everything here in the preamble is only required for the PDF version.

% \iftexforht{[if true…]}{[if false…]}:
% Detect whether you’re running in `tex4ebook` (or similar) or not.
% Some typographical decisions may need to differ between a PDF and an EPUB
% edition. For example, to insert HTML, you may do:
% \iftexforht{\HCode{<cite>}The Time Machine\HCode{</cite>}}{\textit{The Time Machine}}
\makeatletter
\ifdefined\HCode
   \let\iftexforht\@firstoftwo
\else
   \let\iftexforht\@secondoftwo
\fi
\makeatother

% This lets us have `\chapter*` add to the table of contents in the PDF
% version. (This is handled separately for TeX4ht.)
\makeatletter
\NewCommandCopy{\chapter@original}{\chapter}
\RenewDocumentCommand{\chapter}{sm}{%
    \IfBooleanTF{#1}{%
        \chapter@original*{#2}%
        \markboth{#2}{#2}%
        % We have TeX4ht set up to add `\chapter*`s to the table of contents
        % without using `\addcontentsline`. Using it would create a duplicate
        % entry in the table of contents.
        \iftexforht{}{\addcontentsline{toc}{chapter}{#2}}%
    }{%
        \chapter@original{#2}%
    }%
}

\NewCommandCopy{\tableofcontents@original}{\tableofcontents}
\RenewDocumentCommand{\tableofcontents}{}{%
    \iftexforht{%
        \tableofcontents@original%
    }{
        {%
            % This is required because `\tableofcontents` internally calls
            % `\chapter*`, meaning it’ll add itself to the table of contents
            % if we use our redefined version.
            \let\chapter\chapter@original%
            \tableofcontents@original%
        }%
    }
}
\makeatother


\begin{document}

\tableofcontents

\chapter*{Introduction}
\section{Prologue} % This shouldn’t be prefixed with “Chapter”.
\chapter{The Machine}

\end{document}

It’s a complex solution, for sure – I feel like there perhaps should be a more central way to set this format – but from my testing, it works as it should.