path: root/lua/clever_tee/migemo_catalog.lua

local domain = require("clever_tee.domain")
local sequence_state = require("clever_tee.sequence_state")
local state_transitions = require("clever_tee.state_transitions")
local text_topology = require("clever_tee.text_topology")

local M = {}
local MigemoCatalog = {}
local MigemoDictionary = {}
M.MigemoCatalog = MigemoCatalog
M.MigemoDictionary = MigemoDictionary

local catalog_records = setmetatable({}, { __mode = "k" })
local dictionary_records = setmetatable({}, { __mode = "k" })

local function fail(message, level)
  error(message, (level or 1) + 1)
end

local function copy_list(values)
  local result = {}
  for index = 1, #values do
    result[index] = values[index]
  end
  return result
end

local EXPECTED_KEYS = {}
for code = string.byte("a"), string.byte("z") do
  EXPECTED_KEYS[#EXPECTED_KEYS + 1] = string.char(code)
end
for code = string.byte("A"), string.byte("Z") do
  EXPECTED_KEYS[#EXPECTED_KEYS + 1] = string.char(code)
end

local EXPECTED_KEY_SET = {}
for _, key in ipairs(EXPECTED_KEYS) do
  EXPECTED_KEY_SET[key] = true
end

local ASSETS = {
  ["utf-8"] = {
    file = "utf8.vim",
    function_name = "clever_tee#migemo#utf8#load_dict",
  },
  cp932 = {
    file = "cp932.vim",
    function_name = "clever_tee#migemo#cp932#load_dict",
  },
  ["euc-jp"] = {
    file = "eucjp.vim",
    function_name = "clever_tee#migemo#eucjp#load_dict",
  },
}

local module_source = debug.getinfo(1, "S").source
local module_file = module_source:sub(1, 1) == "@" and module_source:sub(2) or module_source
local bundled_root = module_file:match("^(.*)/lua/clever_tee/migemo_catalog%.lua$")

local function require_nonempty_string(value, name)
  if type(value) ~= "string" or value == "" then
    fail((name or "value") .. " must be a nonempty string", 2)
  end
  return value
end

local function asset_path(asset)
  if bundled_root == nil then
    fail("Migemo catalog could not locate its bundled assets", 2)
  end
  return bundled_root .. "/autoload/clever_tee/migemo/" .. asset.file
end

local function read_asset_key_order(path)
  local handle, open_error = io.open(path, "rb")
  if handle == nil then
    fail("Migemo asset could not be opened: " .. tostring(open_error), 2)
  end

  local keys = {}
  for line in handle:lines() do
    local key = line:match("^%s*\\%s*'([A-Za-z])'%s*:")
    if key ~= nil then
      keys[#keys + 1] = key
    end
  end
  handle:close()
  return keys
end

local function assert_key_order(keys, encoding)
  if type(keys) ~= "table" or #keys ~= #EXPECTED_KEYS then
    fail(
      "Migemo " .. encoding .. " asset must contain exactly 52 ordered keys",
      2
    )
  end
  for index, expected in ipairs(EXPECTED_KEYS) do
    if keys[index] ~= expected then
      fail(
        "Migemo " .. encoding
          .. " asset keys must be ordered a through z, then A through Z",
        2
      )
    end
  end
end

local function nvim_runtime()
  local runtime = rawget(_G, "vim")
  if type(runtime) ~= "table"
    or runtime.cmd == nil
    or type(runtime.fn) ~= "table"
    or type(runtime.fn.fnameescape) ~= "function"
  then
    fail("Migemo dictionary loading requires Nvim", 2)
  end
  return runtime
end

local function default_asset_loader(encoding, asset)
  local runtime = nvim_runtime()
  local path = asset_path(asset)
  local keys = read_asset_key_order(path)
  assert_key_order(keys, encoding)

  runtime.cmd("silent source " .. runtime.fn.fnameescape(path))
  local loader = runtime.fn[asset.function_name]
  if type(loader) ~= "function" then
    fail("Migemo " .. encoding .. " asset did not define its dictionary loader", 2)
  end
  local dictionary = loader()
  return dictionary, keys, path
end

local function explicit_pattern(pattern, case_mode)
  local case_flag = case_mode == domain.CaseMode.INSENSITIVE and "\\c" or "\\C"
  return "\\m" .. case_flag .. "^" .. pattern
end

local function default_pattern_compiler(pattern, key, encoding)
  local runtime = nvim_runtime()
  if type(runtime.regex) ~= "function" or type(runtime.fn.match) ~= "function" then
    fail("Migemo pattern evaluation requires Nvim regular expressions", 2)
  end

  local sensitive = explicit_pattern(pattern, domain.CaseMode.SENSITIVE)
  local insensitive = explicit_pattern(pattern, domain.CaseMode.INSENSITIVE)
  local ok, compile_error = pcall(runtime.regex, sensitive)
  if ok then
    ok, compile_error = pcall(runtime.regex, insensitive)
  end
  if not ok then
    fail(
      "Migemo " .. encoding .. " pattern for '" .. key
        .. "' could not be compiled: " .. tostring(compile_error),
      2
    )
  end

  return function(text, case_mode)
    if type(text) ~= "string" then
      fail("Migemo assertion text must be a string", 2)
    end
    case_mode = domain.CaseMode.from_string(case_mode)
    local selected = case_mode == domain.CaseMode.INSENSITIVE
        and insensitive
      or sensitive
    local matched, start_or_error = pcall(runtime.fn.match, text, selected)
    if not matched then
      fail(
        "Migemo " .. encoding .. " pattern for '" .. key
          .. "' could not be evaluated: " .. tostring(start_or_error),
        2
      )
    end
    return start_or_error == 0
  end
end

local dictionary_metatable = {
  __index = function(dictionary, key)
    local method = MigemoDictionary[key]
    if method ~= nil then
      return method
    end

    local record = dictionary_records[dictionary]
    if key == "encoding" or key == "effective_encoding" then
      return record.encoding
    end
    if key == "entry_count" then
      return #record.keys
    end
    if key == "asset_path" then
      return record.asset_path
    end
    if EXPECTED_KEY_SET[key] then
      return record.predicates[key]
    end
    return nil
  end,
  __newindex = function()
    fail("MigemoDictionary values are immutable", 2)
  end,
  __tostring = function(dictionary)
    return "migemo-dictionary:" .. dictionary_records[dictionary].encoding
  end,
  __metatable = "clever_tee.migemo_catalog.MigemoDictionary",
}

local function validate_dictionary_data(data, ordered_keys, encoding)
  if type(data) ~= "table" then
    fail("Migemo " .. encoding .. " asset must return a dictionary", 3)
  end
  assert_key_order(ordered_keys, encoding)

  local count = 0
  for key, pattern in pairs(data) do
    count = count + 1
    if EXPECTED_KEY_SET[key] ~= true then
      fail("Migemo " .. encoding .. " asset contains an unexpected key", 3)
    end
    if type(pattern) ~= "string" or pattern == "" then
      fail("Migemo " .. encoding .. " patterns must be nonempty strings", 3)
    end
  end
  if count ~= #EXPECTED_KEYS then
    fail("Migemo " .. encoding .. " asset must contain exactly 52 keys", 3)
  end
  for _, key in ipairs(EXPECTED_KEYS) do
    if data[key] == nil then
      fail("Migemo " .. encoding .. " asset is missing key '" .. key .. "'", 3)
    end
  end
end

local function new_dictionary(encoding, data, ordered_keys, path, compiler)
  validate_dictionary_data(data, ordered_keys, encoding)

  local patterns = {}
  local predicates = {}
  for _, key in ipairs(EXPECTED_KEYS) do
    local pattern = data[key]
    patterns[key] = pattern
    local predicate = compiler(pattern, key, encoding)
    if type(predicate) ~= "function" then
      fail("Migemo pattern compiler must return a predicate", 3)
    end
    predicates[key] = predicate
  end

  local dictionary = setmetatable({}, dictionary_metatable)
  dictionary_records[dictionary] = {
    encoding = encoding,
    keys = copy_list(ordered_keys),
    patterns = patterns,
    predicates = predicates,
    asset_path = path,
  }
  return dictionary
end

function MigemoDictionary.is(value)
  return type(value) == "table" and dictionary_records[value] ~= nil
end

local function dictionary_record(dictionary)
  if not MigemoDictionary.is(dictionary) then
    fail("value must be a MigemoDictionary", 3)
  end
  return dictionary_records[dictionary]
end

function MigemoDictionary:keys()
  return copy_list(dictionary_record(self).keys)
end

function MigemoDictionary:has(key)
  return type(key) == "string"
    and dictionary_record(self).predicates[key] ~= nil
end

function MigemoDictionary:pattern(key)
  require_nonempty_string(key, "Migemo dictionary key")
  local pattern = dictionary_record(self).patterns[key]
  if pattern == nil then
    fail("Migemo dictionary key must be one ASCII alphabetic character", 2)
  end
  return pattern
end

function MigemoDictionary:predicate(key, case_mode)
  require_nonempty_string(key, "Migemo dictionary key")
  local predicate = dictionary_record(self).predicates[key]
  if predicate == nil then
    fail("Migemo dictionary key must be one ASCII alphabetic character", 2)
  end
  if case_mode == nil then
    return predicate
  end

  case_mode = domain.CaseMode.from_string(case_mode)
  return function(text)
    return predicate(text, case_mode)
  end
end

function MigemoDictionary:matches(key, text, case_mode)
  return self:predicate(key)(text, case_mode)
end

function MigemoDictionary:to_table()
  local record = dictionary_record(self)
  return {
    encoding = record.encoding,
    entry_count = #record.keys,
    keys = copy_list(record.keys),
    asset_path = record.asset_path,
  }
end

local function normalize_catalog_options(options)
  if options == nil then
    return {}
  end
  if type(options) ~= "table" then
    fail("MigemoCatalog options must be a table", 3)
  end
  if type(options.disable_migemo_for_unsupported_encoding) == "function"
    and options.policy == nil
    and options.policy_service == nil
    and options.transitions == nil
    and options.state == nil
    and options.asset_loader == nil
    and options.pattern_compiler == nil
  then
    return { policy = options }
  end
  return options
end

local function require_policy(service)
  if service ~= nil and (type(service) ~= "table"
      or type(service.disable_migemo_for_unsupported_encoding) ~= "function")
  then
    fail(
      "MigemoCatalog policy must provide disable_migemo_for_unsupported_encoding",
      3
    )
  end
  return service
end

local function require_transitions(transitions, state)
  transitions = transitions or state_transitions.new(state)
  if type(transitions) ~= "table"
    or type(transitions.CacheMigemo) ~= "function"
    or type(transitions.state) ~= "function"
    or transitions:state() ~= state
  then
    fail("MigemoCatalog transitions must mutate its SequenceState", 3)
  end
  return transitions
end

local function selected_function(value, fallback, name)
  value = value or fallback
  if type(value) ~= "function" then
    fail("MigemoCatalog " .. name .. " must be a function", 3)
  end
  return value
end

local catalog_metatable = {
  __index = MigemoCatalog,
  __newindex = function()
    fail("MigemoCatalog values are immutable", 2)
  end,
  __tostring = function()
    return "migemo-catalog"
  end,
  __metatable = "clever_tee.migemo_catalog.MigemoCatalog",
}

function MigemoCatalog.new(options)
  if MigemoCatalog.is(options) then
    return options
  end
  options = normalize_catalog_options(options)
  local state = options.state or sequence_state.get()
  if not sequence_state.is(state) then
    fail("MigemoCatalog requires the plugin-global SequenceState", 2)
  end

  local catalog = setmetatable({}, catalog_metatable)
  catalog_records[catalog] = {
    state = state,
    transitions = require_transitions(options.transitions, state),
    policy = require_policy(options.policy or options.policy_service),
    disable_migemo = options.disable_migemo,
    asset_loader = selected_function(
      options.asset_loader,
      default_asset_loader,
      "asset_loader"
    ),
    pattern_compiler = selected_function(
      options.pattern_compiler,
      default_pattern_compiler,
      "pattern_compiler"
    ),
    load_counts = {},
  }
  if catalog_records[catalog].disable_migemo ~= nil
    and type(catalog_records[catalog].disable_migemo) ~= "function"
  then
    fail("MigemoCatalog disable_migemo must be a function", 2)
  end
  return catalog
end

function MigemoCatalog.is(value)
  return type(value) == "table" and catalog_records[value] ~= nil
end

function M.new(options)
  return MigemoCatalog.new(options)
end

setmetatable(M, {
  __call = function(_, options)
    return MigemoCatalog.new(options)
  end,
})

local function catalog_record(catalog)
  if not MigemoCatalog.is(catalog) then
    fail("value must be a MigemoCatalog", 3)
  end
  return catalog_records[catalog]
end

local function unsupported(catalog, requested_encoding, policy_override)
  local record = catalog_record(catalog)
  local active_policy = policy_override or record.policy
  if active_policy ~= nil then
    require_policy(active_policy):disable_migemo_for_unsupported_encoding()
  elseif record.disable_migemo ~= nil then
    record.disable_migemo()
  end
  error(
    "clever-tee: Encoding '" .. requested_encoding
      .. "' is not supported. Migemo is disabled",
    0
  )
end

function MigemoCatalog:get(effective_encoding, policy_override)
  local requested = require_nonempty_string(effective_encoding, "effective encoding")
  local encoding = text_topology.normalize_encoding(requested)
  local asset = ASSETS[encoding]
  if asset == nil then
    return unsupported(self, requested, policy_override)
  end

  local record = catalog_record(self)
  local cached = record.state:get_migemo(encoding)
  if cached ~= nil then
    if not MigemoDictionary.is(cached) then
      fail("Migemo cache contains an invalid dictionary", 2)
    end
    return cached
  end

  local data, ordered_keys, path = record.asset_loader(encoding, asset)
  local dictionary = new_dictionary(
    encoding,
    data,
    ordered_keys,
    path,
    record.pattern_compiler
  )
  record.transitions:CacheMigemo(encoding, dictionary)
  record.load_counts[encoding] = (record.load_counts[encoding] or 0) + 1
  return dictionary
end

function MigemoCatalog:load_count(effective_encoding)
  local encoding = text_topology.normalize_encoding(effective_encoding)
  return catalog_record(self).load_counts[encoding] or 0
end

function MigemoCatalog:cached(effective_encoding)
  local encoding = text_topology.normalize_encoding(effective_encoding)
  local value = catalog_record(self).state:get_migemo(encoding)
  if value ~= nil and not MigemoDictionary.is(value) then
    fail("Migemo cache contains an invalid dictionary", 2)
  end
  return value
end

MigemoCatalog.load = MigemoCatalog.get
MigemoCatalog.select = MigemoCatalog.get
MigemoCatalog.dictionary = MigemoCatalog.get

function M.expected_keys()
  return copy_list(EXPECTED_KEYS)
end

function M.supported_encodings()
  return { "utf-8", "cp932", "euc-jp" }
end

function M.bundled_asset_path(effective_encoding)
  local encoding = text_topology.normalize_encoding(effective_encoding)
  local asset = ASSETS[encoding]
  if asset == nil then
    return nil
  end
  return asset_path(asset)
end

M.load = function(effective_encoding, options)
  return MigemoCatalog.new(options):get(effective_encoding)
end
M.EXPECTED_ENTRY_COUNT = #EXPECTED_KEYS

return M