-- centuries.lua
-- A Pandoc Lua filter that automatically formats references to centuries.
-- Copyright 2022 Bastien Dumont (bastien.dumont [at] posteo.net)
-- This file is under the MIT License: see LICENSE for more details

local ERROR_CONTENT_SPAN =
  'The content of a span of class ".cty" must be a positive or negative ' ..
  'number other than 0 possibly preceded by "+" or "-".'

local eras = {}
function eras:new(name, before, after, unspecified)
  self[name] = {
    before = before,
    after = after,
    unspecified = unspecified
  }
end
eras:new('christian', ' BC', ' AD', '')

local default_config = {
  -- default_unit_form is one of 'sg', 'pl' or 'none'
  default_unit_form = 'sg',
  -- default_era, unit_sg and unit_pl are plain strings
  default_era = 'christian',
  unit_sg = ' century',
  unit_pl = ' centuries',
  -- numeral_type is one of 'roman', 'arabic', 'spelled_out'
  numeral_type = 'arabic',
  -- numeral_style is one of 'lowercase', 'uppercase', 'smallcaps'
  numeral_style = 'lowercase',
  -- ordinal_suffix_pos is one of 'normal', 'exponent'
  ordinal_suffix_pos = 'normal',
  --[[
    ordinal_suffixes is a table of ordinal suffixes
    corresponding to their index number.
    If the requested number is superior to the length of the table or 0,
    the last entry will be used.
  --]]
  ordinal_suffixes = { 'rst', 'nd', 'rd', 'th' },
  -- spelled_out_ordinals is an array of strings
  spelled_out_ordinals = { 'first', 'second', 'third', 'fourth',
  'fifth', 'sixth', 'seventh', 'eighth', 'ninth', 'tenth', 'eleventh',
  'twelfth', 'thirteenth', 'fourteenth', 'fifteenth', 'sixteenth',
  'seventeenth', 'eighteenth', 'nineteenth', 'twentieth',
  'twenty-first' },
  -- number_part_for_ordinal_suffix is one of 'whole' or 'last_digit'
  number_part_for_ordinal_suffix = 'last_digit',
  -- parts_order is an array containing
  -- "number", "suffix", "unit" and "era" in any order
  parts_order = { "number", "suffix", "unit", "era" }
}

local config = {}

local function restore_default_config(to_restore)
  if to_restore == nil then
    for key, value in pairs(default_config) do
      config[key] = value
    end
  else
    config[to_restore] = default_config[to_restore]
  end
end

local function metadata_to_config(key)
  return string.gsub(string.gsub(key, '^cty%-', ''), '%-', '_')
end

local function inlines_to_string(inlines)
  result = ''
  for i = 1, #inlines do
    if inlines[i].t == 'Str' then
      result = result .. inlines[i].text
    elseif inlines[i].t == 'Space' then
      result = result .. ' '
    end
  end
  return result
end

local function metalist_to_configlist(metalist)
  -- metalist is a MetaList of MetaInlines containing Str and Space
  -- objects.
  -- Returns an array containing the values of the corresponding
  -- strings.
  configlist = {}
  for i = 1, #metalist do
    configlist[i] = inlines_to_string(metalist[i])
  end
  return configlist
end

local function get_metamap_entry_value(entry)
  if entry[1] then
    return entry[1].text
  else
    return ''
  end
end

local function process_metadata_to_config(key, value)
  local config_key = metadata_to_config(key)
  if key == 'cty-ordinal-suffixes' then
    config.ordinal_suffixes = {}
    for i = 1, #value do
      table.insert(config.ordinal_suffixes, value[i][1].text)
    end
  elseif key == 'cty-other-eras' then
    for i = 1, #value do
      local this_era = value[i]
      eras:new(
        get_metamap_entry_value(this_era['name']),
        get_metamap_entry_value(this_era['before-zero']),
        get_metamap_entry_value(this_era['after-zero']),
        get_metamap_entry_value(this_era['unspecified'])
      )
    end
  elseif key == 'cty-number-part-for-ordinal-suffix'
    or key == 'cty-numeral-type' then
    config[config_key] = metadata_to_config(value[1].text)
  elseif key == 'cty-parts-order' then
    config[config_key] = metalist_to_configlist(value)
  else
    config[config_key] = value[1].text
  end
end

local function set_config(meta)
  restore_default_config()
  for key, value in pairs(meta) do
    if string.match(key, '^cty%-') then
      process_metadata_to_config(key, value)
    end
  end
end

local function get_abs_number(number)
  return math.ceil(math.abs(tonumber(number)))
end

local function number_to_numeral(unformatted_century)
  if not tonumber(unformatted_century) or unformatted_century == '0' then
    error(ERROR_CONTENT_SPAN, 2)
  end
  local abs_century_number = get_abs_number(unformatted_century)
  local century_numeral
  if config.numeral_type == 'roman' then
    century_numeral = pandoc.utils.to_roman_numeral(abs_century_number)
  elseif config.numeral_type == 'arabic' then
    century_numeral = abs_century_number
  elseif config.numeral_type == 'spelled_out' then
    century_numeral = config.spelled_out_ordinals[abs_century_number]
  else
    error('"numeral_type" must be set either to "arabic", "roman" ' ..
          'or "spelled_out". Here set to "' .. config.numeral_type .. '".')
  end
  return century_numeral
end

local function format_numeral(numeral)
  local formatted_numeral
  if config.numeral_style == 'lowercase' then
    formatted_numeral = pandoc.Str(string.lower(numeral))
  elseif config.numeral_style == 'uppercase' then
    formatted_numeral = pandoc.Str(string.upper(numeral))
  elseif config.numeral_style == 'smallcaps' then
    formatted_numeral = pandoc.SmallCaps(pandoc.Str(string.lower(numeral)))
  else
    error('"numeral_style" must be set either to "lowercase", "uppercase" ' ..
    'or "smallcaps". Here set to "' .. config.numeral_style .. '".')
  end
  return formatted_numeral
end

local function format_century_number(unformatted_century)
  century_numeral = number_to_numeral(unformatted_century)
  formatted_century_number = format_numeral(century_numeral)
  return formatted_century_number
end

local function get_last_digit(number)
  return tonumber(string.sub(tostring(number), -1))
end

local function get_ordinal_suffix_for_number(raw_number)
  if config.numeral_type == 'spelled_out' then
    return ''
  else
    local abs_number = get_abs_number(raw_number)
    return config.ordinal_suffixes[abs_number]
      or config.ordinal_suffixes[#config.ordinal_suffixes]
  end
end

local function get_raw_ordinal_suffix(unformatted_century)
  -- unformatted_century is an integer
  local raw_ordinal_suffix
  if config.number_part_for_ordinal_suffix == 'whole' then
    raw_ordinal_suffix = get_ordinal_suffix_for_number(unformatted_century)
  elseif config.number_part_for_ordinal_suffix == 'last_digit' then
    local the_last_digit = get_last_digit(unformatted_century)
    raw_ordinal_suffix = get_ordinal_suffix_for_number(the_last_digit)
  else
    error('"number_part_for_ordinal_suffix" must be set either to "whole" ' ..
          'or to "last_digit". Here set to "' ..
          config.number_part_for_ordinal_suffix .. '".')
  end
  return raw_ordinal_suffix
end

local function format_ordinal_suffix(raw_ordinal_suffix)
  local formatted_ordinal_suffix
  if config.ordinal_suffix_pos == 'normal' then
    formatted_ordinal_suffix = pandoc.Str(raw_ordinal_suffix)
  elseif config.ordinal_suffix_pos == 'exponent' then
    formatted_ordinal_suffix = pandoc.Superscript(pandoc.Str(raw_ordinal_suffix))
  else
    error('"ordinal_suffix_pos" must be set either to "normal" ' ..
          'or to "exponent". Here set to "' .. config.ordinal_suffix_pos .. '".')
  end
  return formatted_ordinal_suffix
end

local function create_ordinal_suffix(unformatted_century)
  -- unformatted_century is an integer
  local raw_ordinal_suffix = get_raw_ordinal_suffix(unformatted_century)
  local formatted_ordinal_suffix = format_ordinal_suffix(raw_ordinal_suffix)
  return formatted_ordinal_suffix
end

local function format_unit(unit_form)
  local unit
  if unit_form == 'sg' then
    unit = config.unit_sg
  elseif unit_form == 'pl' then
    unit = config.unit_pl
  elseif unit_form == 'none' then
    unit = ''
  else
    error('Attribute "ctyunit" and "config.default_unit_form" must be set ' ..
          'to either "sg", "pl" or "none". Here set to "' .. unit_form .. '".')
  end
  return pandoc.Str(unit)
end

local function get_era_indication(unformatted_century, era)
  local first_char = string.sub(unformatted_century, 1, 1)
  local era_indication
  if eras[era] == nil then
    error('Use of undefined era "' .. era .. '".')
  end
  if first_char == '+' then
    era_indication = eras[era].after
  elseif first_char == '-' then
    era_indication = eras[era].before
  else
    era_indication = eras[era].unspecified
  end
  return pandoc.Str(era_indication)
end

local function reorder_parts(parts_array)
  local part_name_to_value = {
    number = parts_array[1],
    suffix = parts_array[2],
    unit = parts_array[3],
    era = parts_array[4]
  }
  local reordered_array = {}
  for i = 1, #config.parts_order do
    reordered_array[i] = part_name_to_value[config.parts_order[i]]
  end
  return reordered_array
end

local function format_century(unformatted_century, unit_form, era)
  -- unformatted_century is an integer possibly preceded by '+' or '-'
  local century_number = format_century_number(unformatted_century)
  local ordinal_suffix = create_ordinal_suffix(unformatted_century)
  local unit = format_unit(unit_form)
  local era_indication = get_era_indication(unformatted_century, era)
  local formatted_century = reorder_parts(
    { century_number, ordinal_suffix, unit, era_indication })
  return formatted_century
end

function Span(span)
  if span.classes:includes('cty', 1) then
    if #span.content == 1 and span.content[1].t == 'Str' then
      local unformatted_century = span.content[1].text
      local unit_form = span.attributes.ctyunit or config.default_unit_form
      local era = span.attributes.ctyera or config.default_era
      span.content = format_century(unformatted_century, unit_form, era)
      return span
    else
      error(ERROR_CONTENT_SPAN)
    end
  end
end

return {
   { Meta = set_config },
   { Span = Span }
 }