Track the schema files

.gitignore

@@ -15,7 +15,6 @@ _site/
 
 # Ignore locally-generated dictionary files
 /_archived
-/beta/hotfix
 /beta/mabiao
 /beta/schema/yuhao
 
@@ -27,6 +26,5 @@ _site/
 /dist/yuhao.full.dict.yaml
 
 # Ignore beta folder
-/beta/
 /beta/mabiao/*.png
 /beta/schema/*.png

beta/hotfix/lua/yuhao/yuhao_autocompletion_filter.lua

@@ -0,0 +1,24 @@
+-- Name: yuhao_autocompletion_filter.lua
+-- 名稱: 輸入預測開關
+-- Version: 20230901
+-- Author: forFudan 朱宇浩 <[email protected]>
+-- Github: https://github.com/forFudan/
+-- Purpose: 通過開關打開或關閉輸入預測，從而不需要修改 schema.yaml
+-- 版權聲明：
+-- 專爲宇浩輸入法製作 <https://zhuyuhao.com/yuhao/>
+-- 轉載請保留作者名和出處
+-- Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International
+---------------------------------------
+
+local function filter(input, env)
+    local fil = env.engine.context:get_option("yuhao_autocompletion_filter")
+    for cand in input:iter() do
+        if fil and (cand.type == "completion") then
+            return
+        else
+            yield(cand)
+        end
+    end
+end
+
+return { func = filter }

beta/hotfix/lua/yuhao/yuhao_chaifen.lua

@@ -0,0 +1,333 @@
+-- yuhao_chaifen.lua
+-- 作者：ace-who
+-- https://github.com/Ace-Who/rime-xuma/blob/master/schema/lua/ace/xuma_spelling.lua
+
+-- 用于生成词语拆分
+
+function map(table, func)
+  local t = {}
+  for k, v in pairs(table) do
+    t[k] = func(v)
+  end
+  return t
+end
+
+function utf8chars(str)
+  local chars = {}
+  for pos, code in utf8.codes(str) do
+    chars[#chars + 1] = utf8.char(code)
+  end
+  return chars
+end
+
+-- rime.lua
+
+local rime = {}
+package.loaded[...] = rime
+rime.encoder = {}
+
+function rime.encoder.parse_formula(formula)
+  if type(formula) ~= 'string' or formula:gsub('%u%l', '') ~= '' then return end
+  local rule = {}
+  local A, a, U, u, Z, z = ('AaUuZz'):byte(1, -1)
+  for m in formula:gmatch('%u%l') do
+    local upper, lower = m:byte(1, 2)
+    local char_idx = upper < U and upper - A + 1 or upper - Z - 1
+    local code_idx = lower < u and lower - a + 1 or lower - z - 1
+    rule[#rule + 1] = { char_idx, code_idx }
+  end
+  return rule
+end
+
+function rime.encoder.load_settings(setting)
+  -- 注意到公式同则规则同，可通过 f2r 在 rt 中作引用定义，以节省资源。
+  local ft, f2r, rt = {}, {}, {}
+  for _, t in ipairs(setting) do
+    if t.length_equal then
+      ft[t.length_equal] = t.formula
+    elseif t.length_in_range then
+      local min, max = table.unpack(t.length_in_range)
+      for l = min, max do
+        ft[l] = t.formula
+      end
+    end
+  end
+  -- setting 中的 length 不一定连续且一般不包括 1，所以不能用 ipairs()。
+  for k, f in pairs(ft) do
+    local rule = rime.encoder.parse_formula(f)
+    if not rule then return end
+    if not f2r[f] then f2r[f] = rule end
+    rt[k] = f2r[f]
+  end
+  return rt
+end
+
+function rime.switch_option(name, context)
+  context:set_option(name, not context:get_option(name))
+end
+
+-- Cycle options of a switcher. When #options == 1, toggle the only option.
+-- Otherwise unset the first set option and unset the next, or the previous if
+-- 'reverse' is true. When no set option is present, try the key
+-- 'options.save', then 'options.default', then 1.
+function rime.cycle_options(options, env, reverse)
+  local context = env.engine.context
+  if #options == 0 then return 0 end
+  if #options == 1 then
+    rime.switch_option(options[1], context)
+    return 1
+  end
+  local state
+  for k, v in ipairs(options) do
+    if context:get_option(v) then
+      context:set_option(v, false)
+      state = (reverse and (k - 1) or (k + 1)) % #options
+      if state == 0 then state = #options end
+      break
+    end
+  end
+  local k = state or options.save or options.default or 1
+  context:set_option(options[k], true)
+  return k
+end
+
+-- Set an option in 'options' if no one is set yet.
+function rime.init_options(options, context)
+  for k, v in ipairs(options) do
+    if context:get_option(v) then return end
+  end
+  local k = state or options.save or options.default or 1
+  context:set_option(options[k], true)
+end
+
+-- Generate a processor that cycle a group of options with a key.
+-- For now only works when composing.
+function rime.make_option_cycler(
+  options,
+  cycle_key_config_path,
+  switch_key_config_path,
+  reverse
+)
+  local processor, cycle_key, switch_key = {}
+  processor.init = function(env)
+    local config = env.engine.schema.config
+    cycle_key = config:get_string(cycle_key_config_path)
+    switch_key = config:get_string(switch_key_config_path)
+  end
+  processor.func = function(key, env)
+    local context = env.engine.context
+    if context:is_composing() and key:repr() == cycle_key then
+      local state = rime.cycle_options(options, env, reverse)
+      if state > 1 then options.save = state end
+      return 1
+    elseif context:is_composing() and key:repr() == switch_key then
+      -- 选项状态可能在切换方案时被重置，因此需检测更新。但是不能在 filter.init
+      -- 中检测，因为得到的似乎是重置之前的状态，说明组件初始化先于状态重置。为
+      -- 经济计，仅在手动切换开关时检测。
+      -- https://github.com/rime/librime/issues/449
+      -- Todo: 对于较新的 librime-lua，尝试利用 option_update_notifier 更新
+      -- options.save
+      for k, v in ipairs(options) do
+        if context:get_option(v) then
+          if k > 1 then options.save = k end
+        end
+      end
+      local k = options.save or options.default
+      -- Consider the 1st options as OFF state.
+      if context:get_option(options[1]) then
+        context:set_option(options[1], false)
+        context:set_option(options[k], true)
+      else
+        context:set_option(options[k], false)
+        context:set_option(options[1], true)
+      end
+      return 1
+    end
+    return 2 -- kNoop
+  end
+  return processor
+end
+
+-- start of yuhao_chaifen.lua
+
+local config = {}
+config.encode_rules = {
+  { length_equal = 2,          formula = 'AaAbBaBb' },
+  { length_equal = 3,          formula = 'AaBaCaCb' },
+  { length_in_range = { 4, 10 }, formula = 'AaBaCaZa' }
+}
+-- 注意借用编码规则有局限性：取码索引不一定对应取根索引，尤其是从末尾倒数时。
+local spelling_rules = rime.encoder.load_settings(config.encode_rules)
+-- options 要与方案保持一致
+local options = {
+  'yuhao_chaifen.off',
+  'yuhao_chaifen.lv1',
+  'yuhao_chaifen.lv2',
+  'yuhao_chaifen.lv3'
+}
+options.default = 3
+
+local processor = rime.make_option_cycler(options,
+  'yuhao_chaifen/lua/cycle_key',
+  'yuhao_chaifen/lua/switch_key')
+
+local function xform(s)
+  -- input format: "[spelling,code_code...,pinyin_pinyin...]"
+  -- output format: "〔 spelling · code code ... · pinyin pinyin ... 〕"
+  return s == '' and s or s:gsub('%[', '〔')
+      :gsub('%]', '〕')
+      :gsub('{', '<')
+      :gsub('}', '>')
+      :gsub('_', ' ')
+      :gsub(',', ' · ')
+      :gsub(' ·  ·  · ', ' · ')
+      :gsub(' ·  · ', ' · ')
+      :gsub('〔〕', '')
+      :gsub('〔 · ', "〔")
+end
+
+
+local function parse_spll(str)
+  -- Handle spellings like "{于下}{四点}丶"(for 求) where some radicals are
+  -- represented by characters in braces.
+  local radicals = {}
+  for seg in str:gsub('%b{}', ' %0 '):gmatch('%S+') do
+    if seg:find('^{.+}$') then
+      table.insert(radicals, seg)
+    else
+      for pos, code in utf8.codes(seg) do
+        table.insert(radicals, utf8.char(code))
+      end
+    end
+  end
+  return radicals
+end
+
+local function parse_raw_tricomment(str)
+  return str:gsub(',.*', ''):gsub('^%[', '')
+end
+
+-- YZ new
+local function parse_raw_code_comment(str)
+  return str:gsub('%[.-,(.-),.*%]', '[%1]'):gsub(',.*', ''):gsub('^%[', '')
+end
+
+local function spell_phrase(s, spll_rvdb)
+  local chars = utf8chars(s)
+  local rule = spelling_rules[#chars]
+  if not rule then return end
+  local radicals = {}
+  for i, coord in ipairs(rule) do
+    local char_idx = coord[1] > 0 and coord[1] or #chars + 1 + coord[1]
+    local raw = spll_rvdb:lookup(chars[char_idx])
+    -- 若任一取码单字没有注解数据，则不对词组作注。
+    if raw == '' then return end
+    local char_radicals = parse_spll(parse_raw_tricomment(raw))
+    local code_idx = coord[2] > 0 and coord[2] or #char_radicals + 1 + coord[2]
+    radicals[i] = char_radicals[code_idx] or '◇'
+  end
+  return table.concat(radicals)
+end
+
+-- YZ new
+local function code_phrase(s, spll_rvdb)
+  local chars = utf8chars(s)
+  local rule = spelling_rules[#chars]
+  if not rule then return end
+  local radicals = {}
+  for i, coord in ipairs(rule) do
+    local char_idx = coord[1] > 0 and coord[1] or #chars + 1 + coord[1]
+    local raw = spll_rvdb:lookup(chars[char_idx])
+    -- 若任一取码单字没有注解数据，则不对词组作注。
+    if raw == '' then return end
+    local char_radicals = parse_spll(parse_raw_code_comment(raw))
+    local code_idx = coord[2] > 0 and coord[2] or #char_radicals + 1 + coord[2]
+    radicals[i] = char_radicals[code_idx] or '◇'
+  end
+  return table.concat(radicals)
+end
+
+-- YZ modified
+local function get_tricomment(cand, env)
+  local text = cand.text
+  if utf8.len(text) == 1 then
+    local raw_spelling = env.spll_rvdb:lookup(text)
+    if raw_spelling == '' then return end
+    return env.engine.context:get_option('yuhao_chaifen.lv1')
+        and xform(raw_spelling:gsub('%[(.-),.*%]', '[%1]'))
+        or env.engine.context:get_option('yuhao_chaifen.lv2')
+        and xform(raw_spelling:gsub('%[(.-,.-),.*%]', '[%1]'))
+        or xform(raw_spelling) -- yuhao_chaifen.lv3 is on
+  elseif utf8.len(text) > 1 then
+    local spelling = spell_phrase(text, env.spll_rvdb)
+    if not spelling then return end
+    spelling = spelling:gsub('{(.-)}', '<%1>')
+    if env.engine.context:get_option('yuhao_chaifen.lv1') then
+      return ('〔%s〕'):format(spelling)
+    end
+    -- local code = env.code_rvdb:lookup(text)
+    local code = code_phrase(text, env.spll_rvdb)
+    if code ~= '' then -- 按长度排列多个编码。
+      local codes = {}
+      for m in code:gmatch('%S+') do codes[#codes + 1] = m end
+      table.sort(codes, function(i, j) return i:len() < j:len() end)
+      return ('〔%s · %s〕'):format(spelling, table.concat(codes, ' '))
+    else -- 以括号类型区分非本词典之固有词
+      return ('〈 %s 〉'):format(spelling)
+      -- Todo: 如果要为此类词组添加编码注释，其中的单字存在一字多码的情况，需先
+      -- 通过比较来确定全码，再提取词组编码。
+    end
+  end
+end
+
+
+local function filter(input, env)
+  if env.engine.context:get_option('yuhao_chaifen.off') then
+    for cand in input:iter() do yield(cand) end
+    return
+  end
+  for cand in input:iter() do
+    --[[
+    用户有时需要通过拼音反查简化字并显示三重注解，但 luna_pinyin 的简化字排序不
+    合理且靠后。用户可开启 simplification 来解决，但是 simplifier 会强制覆盖注
+    释，为了避免三重注解被覆盖，只能生成一个简单类型候选来代替原候选。
+    Todo: 测试在 <simplifier>/tips: none 的条件下，用 cand.text 和
+    cand:get_genuine().text 分别读到什么值。若分别读到转换前后的候选，则可以仅
+    修改 comment 而不用生成简单类型候选来代替原始候选。这样做的问题是关闭
+    yuhao_chaifen 时就不显示 tips 了。
+    --]]
+    if cand.type == 'simplified' then
+      local comment = (get_tricomment(cand, env) or '') .. cand.comment
+      cand = Candidate("simp_rvlk", cand.start, cand._end, cand.text, comment)
+    else
+      local add_comment = cand.type == 'punct'
+          and env.code_rvdb:lookup(cand.text)
+          or cand.type ~= 'sentence'
+          and get_tricomment(cand, env)
+      if add_comment and add_comment ~= '' then
+        -- 混输和反查中的非 completion 类型，原注释为空或主词典的编码。
+        -- 为免重复冗长，直接以新增注释替换之。前提是后者非空。
+        cand.comment = cand.type ~= 'completion'
+            and env.is_mixtyping
+            and add_comment
+            or add_comment .. cand.comment
+      end
+    end
+    yield(cand)
+  end
+end
+
+
+local function init(env)
+  local config = env.engine.schema.config
+  local spll_rvdb = config:get_string('schema_name/spelling')
+  local code_rvdb = config:get_string('schema_name/code')
+  local abc_extags_size = config:get_list_size('abc_segmentor/extra_tags')
+  env.spll_rvdb = ReverseDb('build/' .. spll_rvdb .. '.reverse.bin')
+  env.code_rvdb = ReverseDb('build/' .. code_rvdb .. '.reverse.bin')
+  env.is_mixtyping = abc_extags_size > 0
+  rime.init_options(options, env.engine.context)
+end
+
+
+return { filter = { init = init, func = filter }, processor = processor }