epub2md/pandoc/book.lua

--[[
book.lua – Pandoc filter for book files from epub2md_merge.py (HTML output, e.g. PDF via weasyprint).

Pandoc's --toc always puts the table of contents (and its own title block) before the whole text,
i.e. before the book's cover, title page and imprint. epub2md_merge.py marks the right places:

    <div id="epub2md-titlepage"></div>   after the cover, only if the book has no title page of its own
    <div id="epub2md-toc"></div>         after the front matter (cover, title page, imprint …)

This filter, for HTML output:
  * builds the table of contents at the placeholder (use it INSTEAD of --toc; --toc-depth is honoured),
  * drops Pandoc's title block when the book has its own title page, or moves it to the placeholder,
  * shows the cover picture (chapter "NNN-cover") without the caption "Cover",
  * turns the chapter anchors <a id="001-…"></a> into <div class="chapter-start"> so that print.css can
    start every chapter – including cover, title page and imprint – on a new page.

For other formats it only removes the placeholders (EPUB, DOCX, ODT: use --toc there).

Usage:
    pandoc buch.md -s --lua-filter /pfad/zu/epub2md/pandoc/book.lua \
      --lua-filter /pfad/zu/epub2md/pandoc/callouts.lua --css /pfad/zu/epub2md/pandoc/print.css \
      --pdf-engine=weasyprint -o buch.pdf
]]

local TOC_ID, TITLE_ID = "epub2md-toc", "epub2md-titlepage"
local TOC_TITLE = { de = "Inhaltsverzeichnis", en = "Contents" }

local function is_placeholder(block, id)
  return block.t == "Div" and block.identifier == id and #block.content == 0
end

local function esc(s)
  return (s:gsub("&", "&amp;"):gsub("<", "&lt;"):gsub(">", "&gt;"))
end

-- <header id="title-block-header"> as Pandoc's HTML template writes it.
local function title_block(meta)
  local html = { '<header id="title-block-header">', '<h1 class="title">' .. esc(pandoc.utils.stringify(meta.title)) .. "</h1>" }
  if meta.subtitle then
    html[#html + 1] = '<p class="subtitle">' .. esc(pandoc.utils.stringify(meta.subtitle)) .. "</p>"
  end
  local authors = meta.author
  if authors and pandoc.utils.type(authors) ~= "List" then
    authors = { authors }
  end
  for _, a in ipairs(authors or {}) do
    html[#html + 1] = '<p class="author">' .. esc(pandoc.utils.stringify(a)) .. "</p>"
  end
  if meta.date then
    html[#html + 1] = '<p class="date">' .. esc(pandoc.utils.stringify(meta.date)) .. "</p>"
  end
  html[#html + 1] = "</header>"
  return pandoc.RawBlock("html", table.concat(html, "\n"))
end

-- Paragraph that only holds chapter start anchors (<a id="001-cover"></a>) → id of the first one.
local function chapter_anchor(block)
  if block.t ~= "Para" and block.t ~= "Plain" then
    return nil
  end
  local id = nil
  for _, inl in ipairs(block.content) do
    if inl.t ~= "RawInline" then
      return nil
    end
    local found = inl.text:match('^<a id="(%d%d%d%-[^"]*)">$') or inl.text:match('^<a id="(%d%d%d%-[^"]*)"></a>$')
    if found then
      id = id or found
    elseif inl.text ~= "</a>" then
      return nil
    end
  end
  return id
end

function Pandoc(doc)
  local html = FORMAT:match("html") and not FORMAT:match("epub")
  local toc_at, title_at = nil, nil
  for i, block in ipairs(doc.blocks) do
    if is_placeholder(block, TOC_ID) then
      toc_at = i
    elseif is_placeholder(block, TITLE_ID) then
      title_at = i
    end
  end
  if not toc_at and not title_at then
    return nil -- not a file from epub2md_merge.py
  end
  if not html then -- placeholders are top-level blocks: just drop them
    local kept = pandoc.Blocks({})
    for _, block in ipairs(doc.blocks) do
      if not is_placeholder(block, TOC_ID) and not is_placeholder(block, TITLE_ID) then
        kept:insert(block)
      end
    end
    doc.blocks = kept
    return doc
  end

  local blocks = pandoc.Blocks({})
  for i, block in ipairs(doc.blocks) do
    local anchor = chapter_anchor(block)
    if i == toc_at then
      if PANDOC_WRITER_OPTIONS.table_of_contents then
        -- --toc given as well: Pandoc's own table of contents stays at the top, no second one
      else
        local lang = doc.meta.lang and pandoc.utils.stringify(doc.meta.lang):sub(1, 2):lower() or "de"
        local toc = pandoc.structure.table_of_contents(doc.blocks, PANDOC_WRITER_OPTIONS)
        blocks:insert(pandoc.Div({
          pandoc.Header(1, TOC_TITLE[lang] or TOC_TITLE.de, pandoc.Attr("toc-title", { "unnumbered", "unlisted" })),
          toc,
        }, pandoc.Attr("TOC", { "toc" })))
      end
    elseif i == title_at then
      if doc.meta.title then
        blocks:insert(title_block(doc.meta))
      end
    elseif anchor then
      blocks:insert(pandoc.Div({}, pandoc.Attr(anchor, { "chapter-start" })))
      local nxt = doc.blocks[i + 1]
      if anchor:match("^%d%d%d%-cover") and nxt and nxt.t == "Figure" then
        -- cover chapter: the picture without the caption "Cover", centred on its own page
        local imgs = {}
        nxt:walk({ Image = function(img) imgs[#imgs + 1] = img end })
        if #imgs == 1 then
          doc.blocks[i + 1] = pandoc.Div({ pandoc.Plain({ imgs[1] }) }, pandoc.Attr("", { "cover" }))
        end
      end
    else
      blocks:insert(block)
    end
  end
  doc.blocks = blocks
  -- The book's own title page (or the moved block) replaces Pandoc's title block at the very top.
  if doc.meta.title then
    doc.meta.pagetitle = doc.meta.pagetitle or pandoc.utils.stringify(doc.meta.title)
    doc.meta.title = nil
  end
  return doc
end