Commit 4c89010f authored by Marco Cavalli's avatar Marco Cavalli
Browse files

fix: prevent adding punctuation in links

parent 5008002f

md_to_docx_converter/md_to_html_3.lua

+24 −16
Original line number Diff line number Diff line
@@ -44,23 +44,27 @@ function CustomTagsToLinks(el, prefix) 
      end



      local remaining_text = child.text



      -- check if there is some text before the link (punctuation or words) and keep it

      local start_index = remaining_text:find(prefix .. "+++")

      -- logfile:write("DEBUG: Par with content -> " .. pandoc.utils.stringify(child) .. "\n")

      -- logfile:write("DEBUG: Par with content -> " .. prefix .. "+++" .. "\n")

      -- logfile:flush()



      if start_index > 1 then

        new_content:insert(pandoc.Str(child.text:sub(1, start_index - 1)))

        remaining_text = child.text:sub(start_index)

      end

      local last_char = remaining_text:sub(-1)

      if last_char:find("[ ,.;:!?)]") then

        extract_text = remaining_text:sub(start_index, #remaining_text - 1)

      else

        -- If no delimiter found, assume the reference goes to the end of the text

        end_index = #remaining_text + 1

        extract_text = remaining_text:sub(start_index)



      -- check if there is some text after the link (punctuation or words), and extract only the link part

      local end_index = #remaining_text

      for i = #remaining_text, 1, -1 do

        if remaining_text:sub(i, i):find("[%w]") then

          end_index = i

          break

        end

      end

      

      extract_text = remaining_text:sub(start_index, end_index)



      -- extract the individual parts of the link (up to 2 parts: optional filename, id)

      local parts = split(extract_text, "+++")



      local filename = nil
@@ -71,24 +75,24 @@ function CustomTagsToLinks(el, prefix) 
      elseif #parts == 2 then

        id = parts[2]

      else

        -- skip, it is malformed

        new_content:insert(child)

        goto continue

        error(string.format("%s\nInvalid reference: '%s' does not have the correct format. It should be %s+++<id> or %s+++<filename>+++<id>", pandoc.utils.stringify(child), child.text, prefix, prefix))

      end



      local id_prefix = ""

      local html_filename = ""



      if prefix == "Figure" and #parts == 2  and (id:find("^[%w_-]+%.png$") or id:find("^[%w_-]+%.jpg$") or id:find("^[%w_-]+%.jpeg$") or id:find("^[%w_-]+%.svg$")) then

        id_prefix = "Figure+++"

      elseif prefix ~= "Clause" then

        id_prefix = prefix .. "_"

      end



      if filename then

        if filename_numbers_mapping[filename] == nil then

          error(string.format("Filename '%s' not found. Can't create a link for element %s", filename, pandoc.utils.stringify(child)))

        end

        local clause_number = filename_numbers_mapping[filename]

        -- check that number is not a letter

        -- check that number is not a letter to differentiate between clauses and annexes

        if type(clause_number) ~= "number" then

          html_filename = "annex-" .. clause_number .. ".html"

          if id:find("root.") then
@@ -101,11 +105,15 @@ function CustomTagsToLinks(el, prefix) 
          end

        end

      end



      -- create the HTML link

      local html_link = string.format('<a href="%s#%s%s">%s</a>', html_filename, id_prefix, id, prefix:lower() .. " " .. id)

      new_content:insert(pandoc.RawInline('html', html_link))



      if last_char:find("[ ,.;:!?)]") then

        new_content:insert(pandoc.Str(last_char))

      -- if we extracted the link (end_index < #remaining_text), keep the remaining text after the link

      if end_index < #remaining_text then

        local after_text = remaining_text:sub(end_index + 1)

        new_content:insert(pandoc.Str(after_text))

      end

    else

      new_content:insert(child)