This commit is contained in:
Dvel 2023-04-20 19:49:29 +08:00
parent 84106b2e26
commit 87940fd97e

View File

@ -62,7 +62,7 @@ function date_translator(input, seg, env)
cand.quality = 100
yield(cand)
end
-- 输出内存
-- -- 输出内存
-- local cand = Candidate("date", seg.start, seg._end, ("%.f"):format(collectgarbage('count')), "")
-- cand.quality = 100
-- yield(cand)
@ -166,10 +166,10 @@ function long_word_filter(input, env)
local idx = config:get_int(env.name_space .. "/idx") or 4
local code = env.engine.context.input -- 当前编码
if string.find(code, "[aeo]") then -- 要提升的词汇拼音一定是包含 a o e 的
if string.find(code, "[aeo]") then -- 要提升的词汇拼音一定是包含 a o e 的
local l = {}
local firstWordLength = 0 -- 记录第一个候选词的长度,提前的候选词至少要比第一个候选词长
local s = 0 -- 记录筛选了多少个词条(只提升 count 个词的权重)
local done = 0 -- 记录筛选了多少个词条(只提升 count 个词的权重)
local i = 1
for cand in input:iter() do
-- 找到要提升的词
@ -178,28 +178,29 @@ function long_word_filter(input, env)
i = i + 1
firstWordLength = leng
yield(cand)
elseif ((leng > firstWordLength) and (s < count)) and (string.find(cand.text, "[%w%p%s]+") == nil) then
elseif ((leng > firstWordLength) and (done < count)) and (string.find(cand.text, "[%w%p%s]+") == nil) then
yield(cand)
s = s + 1
done = done + 1
else
table.insert(l, cand)
end
-- 找齐了或者 l 太大了,就不找了
if s == count then
break
elseif #l > 50 then
local memory = collectgarbage('count')
if memory > 10000 then
collectgarbage('collect')
elseif memory > 5000 then
collectgarbage("step")
end
if (done == count) or (#l > 50) then
break
end
end
-- yield l
for _, cand in ipairs(l) do
yield(cand)
end
-- l 弄完了立马给丫回收了
l = nil
if collectgarbage('count') < 3000 then
collectgarbage("step")
else
collectgarbage('collect')
end
-- yield 其他
for cand in input:iter() do
yield(cand)
end