lua: 简化和修复 cold_word_drop 逻辑 (#923)

* update cold_word_drop
* Delete lua/cold_word_drop/debugtool.lua
* Delete lua/cold_word_drop/turndown_freq_words.lua
This commit is contained in:
Shingo
2024-06-23 19:37:16 +08:00
committed by GitHub
parent 0b59306d61
commit b2cb11c0d6
10 changed files with 349 additions and 413 deletions

View File

@@ -1,92 +0,0 @@
#! /usr/bin/env lua
--
-- debugtool.lua
-- Copyright (C) 2021 Shewer Lu <shewer@gmail.com>
--
-- Distributed under terms of the MIT license.
--
-- puts(tag,...)
-- DEBUG --> log.error
-- WARN --> log.warning
-- INFO --> log.info
-- CONSOLE --> print
--
-- ex:
-- test.lua
--
-- local puts = require 'tools/debugtool'
-- --set tag D103 C102
-- local D103= DEBUG .. "103"
-- local C102= CONSOLE .. "102"
-- local C103= nil
--
--
-- puts(ERROR,__FILE__(),__LINE__(),__FUNC__(), 1, 2 , 3 )
-- --> log.error( "error" .. tran_msg(...))
--
-- puts(DEBUG,__FILE__(),__LINE__(),__FUNC__(), 1, 2 , 3 )
-- --> log.error( DEBUG .. tran_msg(...))
--
-- puts(D103,__FILE__(),__LINE__(),__FUNC__(), 1 2 3)
-- --> log.error("trace103" .. tran_msg(...)
--
-- puts(C102,__FILE__(),__LINE__(),__FUNC__(), 1 2 3)
-- --> print("console103" .. tran_msg(...)
--
-- puts(C103,__FILE__(),__LINE__(),__FUNC__(), 1 2 3)
-- --> pass
--
--
--
-- puts(DEBUG,__FILE__(),__LINE__(),__FUNC__() , ...)
-- puts(INFO,__FILE__(),__LINE__(),__FUNC__() , ...)
--
-- global variable
function __FILE__(n)
n = n or 2
return debug.getinfo(n, 'S').source
end
function __LINE__(n)
n = n or 2
return debug.getinfo(n, 'l').currentline
end
function __FUNC__(n)
n = n or 2
return debug.getinfo(n, 'n').name
end
INFO = "log"
WARN = "warn"
ERROR = "error"
DEBUG = "trace"
CONSOLE = "console"
local function tran_msg(...)
local msg = "\t"
for i, k in next, { ... } do msg = msg .. ": " .. tostring(k) end
return msg
end
local function puts(tag, ...)
if type(tag) ~= "string" then return end
if INFO and tag:match("^" .. INFO) then
(log and log.info or print)(tag .. tran_msg(...))
elseif WARN and tag:match("^" .. WARN) then
(log and log.warning or print)(tag .. tran_msg(...))
elseif ERROR and tag:match("^" .. ERROR) then
(log and log.error or print)(tag .. tran_msg(...))
elseif DEBUG and tag:match("^" .. DEBUG) then
(log and log.error or print)(tag .. tran_msg(...))
elseif CONSOLE and tag:match("^" .. CONSOLE) then
(print)(tag .. tran_msg(...))
else
return
end
end
return puts

View File

@@ -1,4 +1,3 @@
local drop_words = local drop_words =
{ "示~例~", { "示~例~", "肏女人", }
} return drop_words
return drop_words

View File

@@ -1,54 +1,60 @@
local drop_list = require("cold_word_drop.drop_words") local filter = {}
local hide_list = require("cold_word_drop.hide_words")
local turndown_freq_list = require("cold_word_drop.turndown_freq_words")
local function filter(input, env) function filter.init(env)
local idx = 3 -- 降频的词条放到第三个后面, 即第四位, 可在 yaml 里配置 local engine = env.engine
local i = 1 local config = engine.schema.config
local cands = {} env.word_reduce_idx = config:get_int("cold_word_reduce/idx") or 4
local context = env.engine.context env.drop_words = require("cold_word_drop.drop_words") or {}
local preedit_code = context.input env.hide_words = require("cold_word_drop.hide_words") or {}
env.reduce_freq_words = require("cold_word_drop.reduce_freq_words") or {}
end
for cand in input:iter() do function filter.func(input, env)
local cpreedit_code = string.gsub(cand.preedit, ' ', '') local cands = {}
if (i <= idx) then local context = env.engine.context
local tfl = turndown_freq_list[cand.text] or nil local preedit_str = context.input:gsub(" ", "")
-- 前三个 候选项排除 要调整词频的词条, 要删的(实际假性删词, 彻底隐藏罢了) 和要隐藏的词条 local drop_words = env.drop_words
if not local hide_words = env.hide_words
((tfl and table.find_index(tfl, cpreedit_code)) or local word_reduce_idx = env.word_reduce_idx
table.find_index(drop_list, cand.text) or local reduce_freq_words = env.reduce_freq_words
(hide_list[cand.text] and table.find_index(hide_list[cand.text], cpreedit_code)) for cand in input:iter() do
) local cand_text = cand.text:gsub(" ", "")
then local preedit_code = cand.preedit:gsub(" ", "") or preedit_str
i = i + 1
---@diagnostic disable-next-line: undefined-global local reduce_freq_list = reduce_freq_words[cand_text] or {}
yield(cand) if word_reduce_idx > 1 then
else -- 前三个 候选项排除 要调整词频的词条, 要删的(实际假性删词, 彻底隐藏罢了) 和要隐藏的词条
table.insert(cands, cand) if reduce_freq_list and table.find_index(reduce_freq_list, preedit_code) then
end table.insert(cands, cand)
else elseif
table.insert(cands, cand) not (
end table.find_index(drop_words, cand_text)
if (#cands > 50) then or (hide_words[cand_text] and table.find_index(hide_words[cand_text], preedit_code))
break
end )
end then
for _, cand in ipairs(cands) do yield(cand)
local cpreedit_code = string.gsub(cand.preedit, ' ', '') word_reduce_idx = word_reduce_idx - 1
if not end
-- 要删的 和要隐藏的词条不显示 else
( if
table.find_index(drop_list, cand.text) or not (
(hide_list[cand.text] and table.find_index(hide_list[cand.text], cpreedit_code)) table.find_index(drop_words, cand_text)
) or (hide_words[cand_text] and table.find_index(hide_words[cand_text], preedit_code))
then )
---@diagnostic disable-next-line: undefined-global then
yield(cand) table.insert(cands, cand)
end end
end end
for cand in input:iter() do
yield(cand) if #cands >= 80 then
end break
end
end
for _, cand in ipairs(cands) do
yield(cand)
end
end end
return filter return filter

View File

@@ -1,4 +1,5 @@
local hide_words = local hide_words =
{ ["示~例~"] = { "shil", "shili", }, { ["示~例~"] = { "shil", "shili", },
["么特瑞"] = { "meter", },
} }
return hide_words return hide_words

View File

@@ -0,0 +1,48 @@
-- runLog.lua
-- Copyright (C) 2023 yaoyuan.dou <douyaoyuan@126.com>
local M = {}
local dbgFlg = true
--设置 dbg 开关
M.setDbg = function(flg)
dbgFlg = flg
print('runLog dbgFlg is ' .. tostring(dbgFlg))
end
local current_path = string.sub(debug.getinfo(1).source, 2, string.len("/runLog.lua") * -1)
M.logDoc = current_path .. 'runLog.txt'
M.writeLog = function(logStr, newLineFlg)
logStr = logStr or "nothing"
if not newLineFlg then newLineFlg = true end
local f = io.open(M.logDoc, 'a')
if f then
local timeStamp = os.date("%Y/%m/%d %H:%M:%S")
f:write(timeStamp .. '[' .. _VERSION .. ']' .. '\t' .. logStr .. '\n')
f:close()
end
end
--===========================test========================
M.test = function(printPrefix)
if nil == printPrefix then
printPrefix = ' '
end
if dbgFlg then
M.writeLog('this is a test string on new line', true)
M.writeLog('this is a test string appending the last line', false)
M.writeLog('runLogDoc is: ' .. M.logDoc, true)
end
end
function M.init(...)
--如果有需要初始化的动作,可以在这里运行
end
M.init()
return M

View File

@@ -2,162 +2,167 @@
orgtype = type orgtype = type
function type(obj) function type(obj)
local _type = orgtype(obj) local _type = orgtype(obj)
if "table" == _type and obj._cname then if "table" == _type and obj._cname then
return obj._cname return obj._cname
end end
return _type return _type
end end
function metatable(...) function metatable(...)
if ... and type(...) == "table" then if ... and type(...) == "table" then
return setmetatable(..., { __index = table }) return setmetatable(..., { __index = table })
else else
return setmetatable({ ... }, { __index = table }) return setmetatable({ ... }, { __index = table })
end end
end end
-- chech metatble -- chech metatble
function metatable_chk(tab) function metatable_chk(tab)
if "table" == type(tab) if "table" == type(tab) then
then return (tab.each and tab) or metatable(tab)
return (tab.each and tab) or metatable(tab) else
else return tab
return tab end
end
end end
table.eachi = function(tab, func) table.eachi = function(tab, func)
for i = 1, #tab do for i = 1, #tab do
func(tab[i], i) func(tab[i], i)
end end
return tab return tab
end end
table.eacha = function(tab, func) table.eacha = function(tab, func)
for i, v in ipairs(tab) do for i, v in ipairs(tab) do
func(v, i) func(v, i)
end end
return tab return tab
end end
table.each = function(tab, func) table.each = function(tab, func)
for k, v in pairs(tab) do for k, v in pairs(tab) do
func(v, k) func(v, k)
end end
return tab return tab
end end
table.find_index = function(tab, elm, ...) table.find_index = function(tab, elm, ...)
local _, i = table.find(tab, elm, ...) local _, i = table.find(tab, elm, ...)
return i return i
end end
table.find = function(tab, elm, func) table.find = function(tab, elm, func)
for i, v in ipairs(tab) do for i, v in ipairs(tab) do
if elm == v then if elm == v then
return v, i return v, i
end end
end end
end end
table.find_with_func = function(tab, elm, ...) table.find_with_func = function(tab, elm, ...)
local i, v = table.find(tab, elm) local i, v = table.find(tab, elm)
end end
table.delete = function(tab, elm, ...) table.delete = function(tab, elm, ...)
local index = table.find_index(tab, elm) local index = table.find_index(tab, elm)
return index and table.remove(tab, index) return index and table.remove(tab, index)
end end
table.find_all = function(tab, elm, ...) table.find_all = function(tab, elm, ...)
local tmptab = setmetatable({}, { __index = table }) local tmptab = setmetatable({}, { __index = table })
local _func = (type(elm) == "function" and elm) or function(v, k, ...) return v == elm end local _func = (type(elm) == "function" and elm) or function(v, k, ...)
for k, v in pairs(tab) do return v == elm
if _func(v, k, ...) then end
tmptab:insert(v) for k, v in pairs(tab) do
end if _func(v, k, ...) then
end tmptab:insert(v)
return tmptab end
end
return tmptab
end end
table.select = table.find_all table.select = table.find_all
table.reduce = function(tab, func, arg) table.reduce = function(tab, func, arg)
local new, old = arg, arg local new, old = arg, arg
for i, v in ipairs(tab) do for i, v in ipairs(tab) do
new, old = func(v, new) new, old = func(v, new)
end end
return new, arg return new, arg
end end
table.map = function(tab, func) table.map = function(tab, func)
local newtab = setmetatable({}, { __index = table }) local newtab = setmetatable({}, { __index = table })
func = func or function(v, i) return v, i end func = func or function(v, i)
for i, v in ipairs(tab) do return v, i
newtab[i] = func(v, i) end
end for i, v in ipairs(tab) do
return newtab newtab[i] = func(v, i)
end
return newtab
end end
table.map_hash = function(tab, func) -- table to list of array { key, v} table.map_hash = function(tab, func) -- table to list of array { key, v}
local newtab = setmetatable({}, { __index = table }) local newtab = setmetatable({}, { __index = table })
func = func or function(k, v) return { k, v } end func = func or function(k, v)
for k, v in pairs(tab) do return { k, v }
newtab:insert(func(k, v)) end
end for k, v in pairs(tab) do
return newtab newtab:insert(func(k, v))
end
return newtab
end end
function table:push(elm) function table:push(elm)
self:insert(elm) self:insert(elm)
end end
table.append = table.push table.append = table.push
function table:pop() function table:pop()
return self:remove(#self) return self:remove(#self)
end end
function table:shift() function table:shift()
self:remove(1) self:remove(1)
end end
function table:unshift(elm) function table:unshift(elm)
self:insert(1, elm) self:insert(1, elm)
end end
function table.len(t) function table.len(t)
local leng = 0 local leng = 0
for k, v in pairs(t) do for k, v in pairs(t) do
leng = leng + 1 leng = leng + 1
end end
return leng; return leng
end end
-- table to string 序列化 -- table to string 序列化
function table.serialize(obj) function table.serialize(obj)
local serialize_str = "" local serialize_str = ""
local t = type(obj) local t = type(obj)
if t == "number" then if t == "number" then
serialize_str = serialize_str .. obj serialize_str = serialize_str .. obj
elseif t == "boolean" then elseif t == "boolean" then
serialize_str = serialize_str .. tostring(obj) serialize_str = serialize_str .. tostring(obj)
elseif t == "string" then elseif t == "string" then
serialize_str = serialize_str .. string.format("%q", obj) serialize_str = serialize_str .. string.format("%q", obj)
elseif t == "table" then elseif t == "table" then
serialize_str = serialize_str .. "{ " serialize_str = serialize_str .. "{ "
local record_sep = #obj < 4 and ", " or ",\n" local record_sep = #obj < 4 and ", " or ",\n"
local record_prefix = #obj < 4 and "" or "\t" local record_prefix = #obj < 4 and "" or "\t"
for k, v in pairs(obj) do for k, v in pairs(obj) do
if type(k) == "number" then if type(k) == "number" then
serialize_str = serialize_str .. record_prefix .. '"' .. v .. '"' .. record_sep serialize_str = serialize_str .. record_prefix .. '"' .. v .. '"' .. record_sep
else else
serialize_str = serialize_str .. "\t[" .. table.serialize(k) .. "] = " .. table.serialize(v) .. ",\n" serialize_str = serialize_str .. "\t[" .. table.serialize(k) .. "] = " .. table.serialize(v) .. ",\n"
end end
end end
-- local metatable = getmetatable(obj) -- local metatable = getmetatable(obj)
-- if metatable ~= nil and type(metatable.__index) == "table" then -- if metatable ~= nil and type(metatable.__index) == "table" then
-- for k, v in pairs(metatable.__index) do -- for k, v in pairs(metatable.__index) do
-- serialize_str = serialize_str .. "[" .. table.serialize(k) .. "]=" .. table.serialize(v) .. ",\n" -- serialize_str = serialize_str .. "[" .. table.serialize(k) .. "]=" .. table.serialize(v) .. ",\n"
-- end -- end
-- end -- end
serialize_str = serialize_str .. "}" serialize_str = serialize_str .. "}"
elseif t == "nil" then elseif t == "nil" then
return nil return nil
else else
error("can not serialize a " .. t .. " type.") error("can not serialize a " .. t .. " type.")
end end
return serialize_str return serialize_str
end end

View File

@@ -1,151 +1,120 @@
require('cold_word_drop.string') require("cold_word_drop.string")
require("cold_word_drop.metatable") require("cold_word_drop.metatable")
-- local puts = require("tools/debugtool") local processor = {}
local drop_list = require("cold_word_drop.drop_words")
local hide_list = require("cold_word_drop.hide_words")
local turndown_freq_list = require("cold_word_drop.turndown_freq_words")
local tbls = {
['drop_list'] = drop_list,
['hide_list'] = hide_list,
['turndown_freq_list'] = turndown_freq_list
}
-- local cold_word_drop = {}
local function get_record_filername(record_type) local function get_record_filername(record_type)
local user_distribute_name = rime_api:get_distribution_name() local user_distribute_name = rime_api:get_distribution_code_name()
if user_distribute_name == '小狼毫' then if user_distribute_name:lower():match("weasel") then
return string.format("%s\\Rime\\lua\\cold_word_drop\\%s_words.lua", os.getenv("APPDATA"), record_type) return string.format("%s\\lua\\cold_word_drop\\%s_words.lua", rime_api:get_user_data_dir(), record_type)
end elseif user_distribute_name:lower():match("squirrel") then
return string.format("%s/lua/cold_word_drop/%s_words.lua", rime_api:get_user_data_dir(), record_type)
local system = io.popen("uname -s"):read("*l") elseif user_distribute_name:lower():match("fcitx") then
local filename = nil return string.format("%s/lua/cold_word_drop/%s_words.lua", rime_api:get_user_data_dir(), record_type)
-- body elseif user_distribute_name:lower():match("ibus") then
if system == "Darwin" then return string.format(
filename = string.format("%s/Library/Rime/lua/cold_word_drop/%s_words.lua", os.getenv('HOME'), record_type) "%s/rime/lua/cold_word_drop/%s_words.lua",
elseif system == "Linux" then os.getenv("HOME") .. "/.config/ibus",
filename = string.format("%s/%s/rime/lua/cold_word_drop/%s_words.lua", record_type
os.getenv('HOME'), )
(string.find(os.getenv('GTK_IM_MODULE'), 'fcitx') and '.local/share/fcitx5' or '.config/ibus'), end
record_type)
end
return filename
end end
local function write_word_to_file(record_type) local function write_word_to_file(env, record_type)
-- local filename = string.format("%s/Library/Rime/lua/cold_word_drop/%s_words.lua", os.getenv('HOME'), record_type) local filename = get_record_filername(record_type)
local filename = get_record_filername(record_type) local record_header = string.format("local %s_words =\n", record_type)
local record_header = string.format("local %s_words =\n", record_type) local record_tailer = string.format("\nreturn %s_words", record_type)
local record_tailer = string.format("\nreturn %s_words", record_type) if not filename then
local fd = assert(io.open(filename, "w")) --打开 return false
fd:setvbuf("line") end
fd:write(record_header) --写入文件头部 local fd = assert(io.open(filename, "w")) --打开
-- df:flush() --刷新 -- fd:flush() --刷新
local x = string.format("%s_list", record_type) local x = string.format("%s_list", record_type)
local record = table.serialize(tbls[x]) -- lua 的 table 对象 序列化为字符串 local record = table.serialize(env.tbls[x]) -- lua 的 table 对象 序列化为字符串
fd:write(record) --写入 序列化的字符串 fd:setvbuf("line")
fd:write(record_tailer) --写入文件尾部, 结束记录 fd:write(record_header) --写入文件头部
fd:close() --关闭 fd:write(record) --写入 序列化的字符串
fd:write(record_tailer) --写入文件尾部, 结束记录
fd:close() --关闭
end end
local function check_encode_matched(cand_code, word, input_code_tbl, reversedb) local function append_word_to_droplist(env, ctx, action_type)
if #cand_code < 1 and utf8.len(word) > 1 then -- 二字词以上的词条反查, 需要逐个字去反查 local word = ctx.word:gsub(" ", "")
local word_cand_code = string.split(word, "") local input_code = ctx.code:gsub(" ", "")
for i, v in ipairs(word_cand_code) do
-- 如有 `[` 引导的辅助码情况, 去掉引导符及之后的所有形码字符 if action_type == "drop" then
local char_code = string.gsub(reversedb:lookup(v), '%[%l%l', '') table.insert(env.drop_words, word) -- 高亮选中的词条插入到 drop_list
local _char_preedit_code = input_code_tbl[i] or " " return true
-- 如有 `[` 引导的辅助码情况, 同上, 去掉之 end
local char_preedit_code = string.gsub(_char_preedit_code, '%[%l+', '')
if not string.match(char_code, char_preedit_code) then if action_type == "hide" then
-- 输入编码串和词条反查结果不匹配(考虑到多音字, 开启了模糊音, 纠错音), 返回false, 表示隐藏这个词条 if not env.hide_words[word] then
return false env.hide_words[word] = { input_code }
end -- 隐藏的词条如果已经在 hide_list 中, 则将输入串追加到 值表中, 如: ['藏'] = {'chang', 'zhang'}
end elseif not table.find_index(env.hide_words[word], input_code) then
end table.insert(env.hide_words[word], input_code)
-- 输入编码串和词条反查结果匹配, 返回true, 表示对这个词条降频 end
return true return true
end
if action_type == "reduce_freq" then
if env.reduce_freq_words[word] then
table.insert(env.reduce_freq_words[word], input_code)
else
env.reduce_freq_words[word] = { input_code }
end
return true
end
end end
local function append_word_to_droplist(ctx, action_type, reversedb) function processor.init(env)
local word = ctx.word local engine = env.engine
local input_code = ctx.code local config = engine.schema.config
if action_type == 'drop' then env.drop_cand_key = config:get_string("key_binder/drop_cand") or "Control+d"
table.insert(drop_list, word) -- 高亮选中的词条插入到 drop_list env.hide_cand_key = config:get_string("key_binder/hide_cand") or "Control+x"
return true env.reduce_cand_key = config:get_string("key_binder/reduce_freq_cand") or "Control+j"
end env.drop_words = require("cold_word_drop.drop_words") or {}
local input_code_tbl = string.split(input_code, " ") env.hide_words = require("cold_word_drop.hide_words") or {}
local cand_code = reversedb:lookup(word) or "" -- 反查候选项文字编码 env.reduce_freq_words = require("cold_word_drop.reduce_freq_words") or {}
-- 二字词 的匹配检查, 匹配返回true, 不匹配返回false env.tbls = {
local match_result = check_encode_matched(cand_code, word, input_code_tbl, reversedb) ["drop_list"] = env.drop_words,
local ccand_code = string.gsub(cand_code, '%[%l%l', '') ["hide_list"] = env.hide_words,
-- 如有 `[` 引导的辅助码情况, 去掉引导符及之后的所有形码字符 ["reduce_freq_list"] = env.reduce_freq_words,
local input_str = string.gsub(input_code, '%[%l+', '') }
local input_code_str = table.concat(input_code_tbl, '')
-- 单字和二字词 的匹配检查, 如果匹配, 降频
if string.match(ccand_code, input_str) or match_result then
if turndown_freq_list[word] then
table.insert(turndown_freq_list[word], input_code_str)
else
turndown_freq_list[word] = { input_code_str }
end
return 'turndown_freq'
end
-- 单字和二字词 如果不匹配 就隐藏
if not hide_list[word] then
hide_list[word] = { input_code_str }
return true
else
-- 隐藏的词条如果已经在 hide_list 中, 则将输入串追加到 值表中, 如: ['藏'] = {'chang', 'zhang'}
if not table.find_index(hide_list[word], input_code_str) then
table.insert(hide_list[word], input_code_str)
return true
else
return false
end
end
end end
local function processor(key, env) function processor.func(key, env)
local engine = env.engine local engine = env.engine
local config = engine.schema.config local context = engine.context
local context = engine.context local preedit_code = context:get_script_text()
-- local top_cand_text = context:get_commit_text() local action_map = {
-- local preedit_code = context.input [env.drop_cand_key] = "drop",
local preedit_code = context:get_script_text() [env.hide_cand_key] = "hide",
local turndown_cand_key = config:get_string("key_binder/turn_down_cand") or "Control+j" [env.reduce_cand_key] = "reduce_freq",
local drop_cand_key = config:get_string("key_binder/drop_cand") or "Control+d" }
local action_map = {
[turndown_cand_key] = 'hide',
[drop_cand_key] = 'drop'
}
-- local schema_id = config:get_string("schema/schema_id") if context:has_menu() and action_map[key:repr()] then
local schema_id = config:get_string("translator/dictionary") -- 多方案共用字典取主方案名称 local cand = context:get_selected_candidate()
---@diagnostic disable-next-line: undefined-global local action_type = action_map[key:repr()]
local reversedb = ReverseLookup(schema_id) local ctx_map = {
if key:repr() == turndown_cand_key or key:repr() == drop_cand_key then ["word"] = cand.text,
local cand = context:get_selected_candidate() ["code"] = preedit_code,
local action_type = action_map[key:repr()] }
local ctx_map = { local res = append_word_to_droplist(env, ctx_map, action_type)
['word'] = cand.text,
['code'] = preedit_code
}
local res = append_word_to_droplist(ctx_map, action_type, reversedb)
context:refresh_non_confirmed_composition() -- 刷新当前输入法候选菜单, 实现看到实时效果 context:refresh_non_confirmed_composition() -- 刷新当前输入法候选菜单, 实现看到实时效果
if type(res) == "boolean" then if not res then
-- 期望被删的词和隐藏的词条写入文件(drop_words.lua, hide_words.lua) return 2
write_word_to_file(action_type) end
else
-- 期望 要调整词频的词条写入 turndown_freq_words.lua 文件
write_word_to_file(res)
end
return 1 -- kAccept
end
return 2 -- kNoop, 不做任何操作, 交给下个组件处理 if res then
-- 期望被删的词和隐藏的词条写入文件(drop_words.lua, hide_words.lua)
write_word_to_file(env, action_type)
end
return 1 -- kAccept
end
return 2 -- kNoop, 不做任何操作, 交给下个组件处理
end end
return processor return processor

View File

@@ -0,0 +1,5 @@
local reduce_freq_words =
{ ["示~例~"] = { "shili", },
["颜射"] = { "yanshe", },
}
return reduce_freq_words

View File

@@ -4,35 +4,34 @@
-- string.utf8_offset= utf8.offset -- string.utf8_offset= utf8.offset
-- string.utf8_sub= utf8.sub -- string.utf8_sub= utf8.sub
function string.split(str, sp, sp1) function string.split(str, sp, sp1)
sp = type(sp) == "string" and sp or " " sp = type(sp) == "string" and sp or " "
if #sp == 0 then if #sp == 0 then
sp = "([%z\1-\127\194-\244][\128-\191]*)" sp = "([%z\1-\127\194-\244][\128-\191]*)"
elseif #sp == 1 then elseif #sp == 1 then
sp = "[^" .. (sp == "%" and "%%" or sp) .. "]*" sp = "[^" .. (sp == "%" and "%%" or sp) .. "]*"
else else
sp1 = sp1 or "^" sp1 = sp1 or "^"
str = str:gsub(sp, sp1) str = str:gsub(sp, sp1)
sp = "[^" .. sp1 .. "]*" sp = "[^" .. sp1 .. "]*"
end end
local tab = {} local tab = {}
for v in str:gmatch(sp) do for v in str:gmatch(sp) do
table.insert(tab, v) table.insert(tab, v)
end end
return tab return tab
end end
function utf8.gsub(str, si, ei) function utf8.gsub(str, si, ei)
local function index(ustr, i) local function index(ustr, i)
return i >= 0 and (ustr:utf8_offset(i) or ustr:len() + 1) return i >= 0 and (ustr:utf8_offset(i) or ustr:len() + 1) or (ustr:utf8_offset(i) or 1)
or (ustr:utf8_offset(i) or 1) end
end
local u_si = index(str, si) local u_si = index(str, si)
ei = ei or str:utf8_len() ei = ei or str:utf8_len()
ei = ei >= 0 and ei + 1 or ei ei = ei >= 0 and ei + 1 or ei
local u_ei = index(str, ei) - 1 local u_ei = index(str, ei) - 1
return str:sub(u_si, u_ei) return str:sub(u_si, u_ei)
end end
string.utf8_len = utf8.len string.utf8_len = utf8.len

View File

@@ -1,4 +0,0 @@
local turndown_freq_words =
{ ["示~例~"] = { "shili", },
}
return turndown_freq_words