如何编写Lua过滤器合并Pandoc的相邻多个引用？

To merge adjacent citation elements like [@anand2011]; [@bachman1996] into a single grouped citation [@anand2011; @bachman1996], you can use a Lua filter that traverses Pandoc's Abstract Syntax Tree (AST) to identify and combine consecutive citation nodes.

Filter Code #

Save the following code as merge-citations.lua:

local function is_separator(inline)
  -- Check if inline is semicolon plus optional whitespace
  if inline.t == 'Str' and inline.text == ';' then
    return true
  elseif inline.t == 'Space' or inline.t == 'SoftBreak' or inline.t == 'LineBreak' then
    return true
  end
  return false
end

function Para(para)
  local new_inlines = {}
  local current_cites = {}
  local skip_separators = false

  for i, inline in ipairs(para.content) do
    if inline.t == 'Cite' then
      -- Add citations from current Cite element to the group
      for _, cit in ipairs(inline.citations) do
        table.insert(current_cites, cit)
      end
      skip_separators = true -- Next separators are part of the citation group
    elseif skip_separators and is_separator(inline) then
      -- Ignore separators between citations
      goto continue
    else
      -- If we have a group of citations, create a new Cite element
      if #current_cites > 0 then
        -- Create a single Cite element with merged citations
        local merged_cite = pandoc.Cite(current_cites, {pandoc.Str('')})
        table.insert(new_inlines, merged_cite)
        current_cites = {}
        skip_separators = false
      end
      -- Add non-citation/non-separator element
      table.insert(new_inlines, inline)
    end
    ::continue::
  end

  -- Add any remaining citations at the end of the paragraph
  if #current_cites > 0 then
    local merged_cite = pandoc.Cite(current_cites, {pandoc.Str('')})
    table.insert(new_inlines, merged_cite)
  end

  return pandoc.Para(new_inlines)
end

return {
  {Para = Para}
}

Key Functionality #

• Separator Detection: The is_separator function identifies semicolons and whitespace (spaces, line breaks) that separate citations.
• Citation Grouping: As we iterate through paragraph elements, we collect consecutive Cite nodes, ignoring any separating semicolons or whitespace.
• Merge and Replace: Once we encounter a non-citation/non-separator element, we merge the collected citations into a single Cite element and add it to the output. Any remaining citations at the end of the paragraph are also merged.

Usage #

Run Pandoc with the filter applied:

pandoc your-document.md --lua-filter=merge-citations.lua -o formatted-document.md

Example #

Input #

For example, see [@anand2011];
[@bachman1996]; [@clark2000]

Output #

For example, see [@anand2011; @bachman1996; @clark2000]

内容的提问来源于stack exchange，提问作者TomBen