通用输入算法

wiki

Lua 中的通用函数和算法

有两个特殊的输入迭代器，numbers() 和 words()，它们的工作原理类似于非常有用的 io.lines() 迭代器。要打印标准输入中找到的所有单词

-- words.lua
require 'func'
for w in words() do
  print(w)
end

$ lua words.lua < test.txt

printall(words(),'\n')

require 'func'
local s 
for x in numbers() do
  s = s + x
end
print(s)

local s,n = sum(numbers())
print('average value =',s/n)

f = io.open(arg[1])
print(sum(numbers(f)))
f:close()

function copy(iter)
  local res = {}
  local k = 1
  for v in iter do
     res[k] = v
     k = k + 1
  end
  return res
end

t = copy(numbers '10 20 30') -- will be {10,20,30}
s = sum(list(t))             -- will be 60

t = copy(io.lines())
table.sort(t)
printall(t,'\n')   -- try table.foreach(t,print) and see!

printall(sort(io.lines()),'\n')

printall(slice(io.lines(),1,10),'\n')

require 'func'
print(count(words()))

require 'func'
print(count(io.lines(),matching '^%s*function'))

function matching(s)
  local strfind = string.find
  return function(v)
    return strfind(v,s)
  end
end

当然，您可以在这些操作中使用任何序列。如果您加载了非常有用的 lfs（Lua 文件系统）库，那么t = copy_if(lfs.dir(path),matching '%.cpp$') 将用path 中所有扩展名为.cpp 的文件填充一个列表。

修改count() 输入的另一种有用方法是使用unique()

-- number of unique words in a file
print(count(unique(words())))

function unique(iter)
  local t = count_map(iter)
  return keys(t)
end

table.foreach(count_map(words()),print)

for x,y in join(numbers(f1),numbers(f2)) do
  print(x-y)
end

Lua 中的 AWK 编程风格

{ print $1/$4, $3/$4 }

for x,y,z in fields{1,3,4} do
   print(x/z,y/z)
end

print(count(fields{7},greater_than(44000)))

{ if ($7 > 44000) k++ } END { print(k) }

for x,y in fields({1,2},',',f) do ...

for x,y in fields(2,',',f) do ...  --equivalent--

性能和表达效率

> tt = copy(random(10000))
> = sum(tt)
5039.542771691  10000

local t = {}
local random = math.random
for i = 1,1e6 do
   t[i] = random()
end

第二个反对意见是，它会导致奇怪且不自然的代码。这在 C++ 中当然可能是这样，C++（让我们面对现实）并不真正适合函数式风格；没有闭包，严格的静态类型不断地妨碍，要求所有东西都是模板。这种风格更适合 Lua - 在 C++ 中这样做不会有一半的可读性，即使使用 Boost Lambda 库也是如此

-- sum of squares of input data using an internal iterator
for_each(numbers(),function(v)
    s = s + v*v
end)

-- sum of squares of input data using an external iterator
for v in numbers() do
    s = s + v*v
end

编写自定义输入对象

Files = {}
 
function Files.create(list)
   local files = {}
   files.list = {}
   local n = table.getn(list)
   for i = 1,n do
      files.list[i] = list[i]
   end
   files.open_next = Files.open_next
   files.read = Files.read
   files:open_next()
   return files
end
 
function Files:open_next()
   if self.f then self.f:close() end
   local nf = table.remove(self.list,1)
   if nf then
      self.f = io.open(nf)
      return true
   else
      self.f = nil
      return false
   end
end
 
function Files:read()
  local ret = self.f:read()
  if not ret then
     if not self:open_next() then return nil end
     return self.f:read()
  else
     return ret
  end
end

files = Files.create(arg)
printall(words(files))

实现和进一步开发的说明

'%s*([^,]+),[^,]+,([^,])'

cur = con:execute 'SELECT * FROM [Event Summaries]'
mag = -9
row = cur:fetch({},'a')
while row do
  if row.Magnitude > mag then 
     mag = row.Magnitude
  end
  row = cur:fetch(row,'n')
end
cur:close()

我可以通过创建一个跟踪 row 的迭代器来简化此操作

function rows(cursor)
  local row = {}
  return function()
    return cursor:fetch(row,'a')
  end
end

for row in rows(cur) do
   if row.Magnitude > mag then 
      mag = row.Magnitude    
   end
end

function column(fieldname,cursor)
  local row = {}
  return function()
    row = cur:fetch(row,'a')
    if not row then return nil 
    else return row[fieldname]
    end
  end
end

local minm,maxm = minmax(column('Magnitude',cur))

print(count(column('Magnitude',cur),greater_than(2)))

-- SteveDonovan