string扩展模块补充utf8各种支持（只是给自带utf8库弄了一套和其他函数风格的名称和注释）

整理代码

Author: MrZ626

stringExtend.lua

@@ -107,7 +107,7 @@ end
 ---Paste new string into original string, won't exceed the length of original string
 ---@param str string
 ---@param str2 string
----@param pos number
+---@param pos integer
 ---@return string
 ---@nodiscard
 function STRING.paste(str,str2,pos)
@@ -156,6 +156,80 @@ function STRING.unshift(str)
     return unshiftMap[str] or lower(str)
 end
 
+local utf8=require'utf8'
+
+---Simple utf8 coding
+---@param num integer
+---@return string
+---@nodiscard
+function STRING.UTF8(num)
+    assertf(type(num)=='number',"Wrong type (%s)",type(num))
+    assertf(num>=0 and num<2^31,"Out of range (%d)",num)
+    -- if     num<2^7  then return char(num)
+    -- elseif num<2^11 then return char(192+floor(num/2^06),128+num%2^6)
+    -- elseif num<2^16 then return char(224+floor(num/2^12),128+floor(num/2^06)%2^6,128+num%2^6)
+    if num<0x10ffff then return utf8.char(num)
+    elseif num<2^21 then return char(240+floor(num/2^18),128+floor(num/2^12)%2^6,128+floor(num/2^06)%2^6,128+num%2^6)
+    elseif num<2^26 then return char(248+floor(num/2^24),128+floor(num/2^18)%2^6,128+floor(num/2^12)%2^6,128+floor(num/2^06)%2^6,128+num%2^6)
+    else                 return char(252+floor(num/2^30),128+floor(num/2^24)%2^6,128+floor(num/2^18)%2^6,128+floor(num/2^12)%2^6,128+floor(num/2^06)%2^6,128+num%2^6)
+    end
+end
+
+---Get the (unicode) char count of (part of) a utf8 string
+---@param str string an utf8 string
+---@param i? integer start byte (cannot start from middle of a char)
+---@param j? integer end byte
+---@param lax? boolean
+---@nodiscard
+---@diagnostic disable-next-line
+function STRING.u8len(str,i,j,lax) end
+STRING.u8len=utf8[('len')]
+
+---Get a utf8 string with codepoint numbers
+---@param ... integer
+---@return string
+---@nodiscard
+---@diagnostic disable-next-line
+function STRING.u8char(...) end
+STRING.u8char=utf8[('char')]
+
+---Get the codepoint(s) of a utf8 string
+---@param str string
+---@param i? integer start byte (cannot start from middle of a char)
+---@param j? integer end byte
+---@param lax? boolean
+---@return integer ...
+---@nodiscard
+---@diagnostic disable-next-line
+function STRING.u8byte(str,i,j,lax) end
+STRING.u8byte=utf8[('codepoint')]
+
+---Get the end byte of `n` characters starting from `i` byte
+---@param str string
+---@param n integer char count
+---@param i? integer start byte
+---@return integer p end byte
+---@nodiscard
+---@diagnostic disable-next-line
+function STRING.u8offset(str,n,i) end
+STRING.u8offset=utf8[('offset')]
+
+---Get a char iterator function of a utf8 string, similar to `str:gmatch('.')`
+---
+--- ```lua
+--- for bytePos,codepoint in STRING.u8codes(str) do
+---    -- code here
+--- end
+---@param str string
+---@param lax? boolean
+---@return fun(s:string, p:integer): integer,integer
+---@diagnostic disable-next-line
+function STRING.u8codes(str,lax) end
+STRING.u8codes=utf8[('codes')]
+
+---Matches exactly one UTF-8 byte sequence, assuming that the subject is a valid UTF-8 string
+STRING.u8pattern="[\0-\x7F\xC2-\xFD][\x80-\xBF]*"
+
 local upperData,lowerData,diaData -- Data is filled later in this file
 
 ---string.upper with utf8 support, warning: low performance
@@ -213,7 +287,7 @@ end
 ---Count the number of occurrences of a regex pattern in a string
 ---@param str string
 ---@param regex string
----@return number
+---@return integer
 ---@nodiscard
 function STRING.count(str,regex)
     local _,count=gsub(str,regex,'')
@@ -231,7 +305,7 @@ end
 ---end
 ---```
 ---@param str string
----@param keep? boolean | number Max number of leading spaces to be trimmed: `nil`- all  `number` - as you want  `true` - same with line#1
+---@param keep? boolean | integer Max number of leading spaces to be trimmed: `nil`- all  `number` - as you want  `true` - same with line#1
 ---@return string
 ---@nodiscard
 function STRING.trimIndent(str,keep)
@@ -313,7 +387,7 @@ end
 ---Calculate the edit distance between two strings
 ---@param s1 string
 ---@param s2 string
----@return number
+---@return integer
 ---@nodiscard
 function STRING.editDist(s1,s2) -- By Copilot
     local len1,len2=#s1,#s2
@@ -389,7 +463,7 @@ end
 
 ---**Warning:** don't support number format like .26, must have digits before the dot, like 0.26
 ---@param str string
----@return number | nil, string | nil
+---@return integer | nil, string | nil
 ---@nodiscard
 function STRING.cutUnit(str)
     local _s,_e=find(str,'^-?%d+%.?%d*')
@@ -426,22 +500,6 @@ STRING.base64={} for c in gmatch('ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrst
     table.insert(STRING.base64,c)
 end
 
----Simple utf8 coding
----@param num number
----@return string
----@nodiscard
-function STRING.UTF8(num)
-    assertf(type(num)=='number',"Wrong type (%s)",type(num))
-    assertf(num>=0 and num<2^31,"Out of range (%d)",num)
-    if     num<2^7  then return char(num)
-    elseif num<2^11 then return char(192+floor(num/2^06),128+num%2^6)
-    elseif num<2^16 then return char(224+floor(num/2^12),128+floor(num/2^06)%2^6,128+num%2^6)
-    elseif num<2^21 then return char(240+floor(num/2^18),128+floor(num/2^12)%2^6,128+floor(num/2^06)%2^6,128+num%2^6)
-    elseif num<2^26 then return char(248+floor(num/2^24),128+floor(num/2^18)%2^6,128+floor(num/2^12)%2^6,128+floor(num/2^06)%2^6,128+num%2^6)
-    else                 return char(252+floor(num/2^30),128+floor(num/2^24)%2^6,128+floor(num/2^18)%2^6,128+floor(num/2^12)%2^6,128+floor(num/2^06)%2^6,128+num%2^6)
-    end
-end
-
 ---Parse binary number from string
 ---@param str string
 ---@return number
@@ -484,7 +542,7 @@ end
 
 ---Convert a number to binary string
 ---@param num number
----@param len? number
+---@param len? integer
 ---@return string
 ---@nodiscard
 function STRING.toBin(num,len)
@@ -498,7 +556,7 @@ end
 
 ---Convert a number to octal string
 ---@param num number
----@param len? number
+---@param len? integer
 ---@return string
 ---@nodiscard
 function STRING.toOct(num,len)
@@ -515,7 +573,7 @@ b16[0]='0'
 
 ---Convert an integer to hexadecimal string
 ---@param num number
----@param len? number
+---@param len? integer
 ---@return string
 ---@nodiscard
 function STRING.toHex(num,len)
@@ -587,8 +645,8 @@ end
 
 ---Return 16 byte string. Not powerful hash, just simply protect the original text
 ---@param text string
----@param seedRange? number default to 26
----@param seed? number default to 0
+---@param seedRange? integer default to 26
+---@param seed? integer default to 0
 ---@return string
 ---@nodiscard
 function STRING.digezt(text,seedRange,seed)
@@ -628,7 +686,7 @@ end
 
 ---Cut n bytes off a string
 ---@param str string
----@param n number
+---@param n integer
 ---@return string, string #`n` bytes, and the rest of string
 ---@nodiscard
 function STRING.readChars(str,n)
@@ -642,7 +700,7 @@ end
 ---STRING.simplifyPath('Documents/Project/xxx.lua',3) --> 'Doc/Pro/xxx.lua'
 ---```
 ---@param path string
----@param len? number default to 1
+---@param len? integer default to 1
 ---@return string
 ---@nodiscard
 function STRING.simplifyPath(path,len)

tableExtend.lua

@@ -17,7 +17,7 @@ for k,v in next,table do TABLE[k]=v end
 ---https://luajit.org/extensions.html
 ---@generic T
 ---@param val T value to fill
----@param count number how many elements
+---@param count integer how many elements
 ---@return T[]
 ---@nodiscard
 function TABLE.new(val,count)
@@ -28,11 +28,22 @@ function TABLE.new(val,count)
     return L
 end
 
+---Create a new table with specific size allocated
+---
+---Fallback to `return {}` if failed to require `table.new`
+---@param nArray? integer the size of "list part" of the table
+---@param nHash? integer the size of "hash part" of the table
+---@return table
+---@nodiscard
+---@diagnostic disable-next-line
+function TABLE.newSize(nArray,nHash) return {} end
+pcall(function() TABLE[('newSize')]=require'table.new' end)
+
 ---Create a new filled matrix
 ---@generic T
 ---@param val T value to fill
----@param height number
----@param width number
+---@param height integer
+---@param width integer
 ---@return Mat<T>
 ---@nodiscard
 function TABLE.newMat(val,height,width)
@@ -51,8 +62,8 @@ end
 ---leave `start&stop` as `nil` will simply copy
 ---@generic T
 ---@param org T original table
----@param start? number start pos (default 1)
----@param stop? number end pos (default #org)
+---@param start? integer start pos (default 1)
+---@param stop? integer end pos (default #org)
 ---@return T
 ---@nodiscard
 function TABLE.sub(org,start,stop)
@@ -67,7 +78,7 @@ end
 ---Create a copy of [1~#] elements
 ---@generic T
 ---@param org T original table
----@param depth? number how many layers will be recreate, default to inf
+---@param depth? integer how many layers will be recreate, default to inf
 ---@return T
 ---@nodiscard
 function TABLE.copy(org,depth)
@@ -86,7 +97,7 @@ end
 ---Create a full copy of org, depth = how many layers will be recreate, default to inf
 ---@generic T
 ---@param org T original table
----@param depth? number how many layers will be recreate, default to inf
+---@param depth? integer how many layers will be recreate, default to inf
 ---@return T
 ---@nodiscard
 function TABLE.copyAll(org,depth)
@@ -303,7 +314,7 @@ end
 ---Update old table with new table (recursive when both table type and below specifiled depth)
 ---@param new table
 ---@param old table
----@param depth? number how many layer will be entered, default to inf
+---@param depth? integer how many layer will be entered, default to inf
 function TABLE.update(old,new,depth)
     if not depth then depth=1e99 end
     for k,v in next,new do
@@ -506,7 +517,7 @@ end
 ---]]
 ---```
 ---@param org table
----@param depth? number how many layer will be entered and flattened, default to inf
+---@param depth? integer how many layer will be entered and flattened, default to inf
 function TABLE.flatten(org,depth)
     if not depth then depth=1e99 end
     while depth>0 do
@@ -534,8 +545,8 @@ end
 ---Find value in [1~#], like string.find
 ---@param t any[]
 ---@param val any
----@param start? number
----@return number? key
+---@param start? integer
+---@return integer? key
 ---@nodiscard
 function TABLE.find(t,val,start)
     for i=start or 1,#t do if t[i]==val then return i end end
@@ -544,7 +555,7 @@ end
 ---TABLE.find for ordered list only, faster (binary search)
 ---@param t any[]
 ---@param val any
----@return number | nil key
+---@return integer | nil key
 ---@nodiscard
 function TABLE.findOrdered(t,val)
     if val<t[1] or val>t[#t] then return nil end
@@ -576,7 +587,7 @@ end
 ---@param t any[]
 ---@param v_old any
 ---@param v_new any
----@param start? number
+---@param start? integer
 function TABLE.replace(t,v_old,v_new,start)
     for i=start or 1,#t do
         if t[i]==v_old then
@@ -673,7 +684,7 @@ do -- function TABLE.dumpDeflate(t,depth)
     end
     ---Dump a simple lua table (no whitespaces)
     ---@param t table
-    ---@param depth? number how many layers will be dumped, default to inf
+    ---@param depth? integer how many layers will be dumped, default to inf
     ---@return string
     ---@nodiscard
     function TABLE.dumpDeflate(t,depth)
@@ -740,7 +751,7 @@ do -- function TABLE.dump(t,depth)
     end
     ---Dump a simple lua table
     ---@param t table
-    ---@param depth? number how many layers will be dumped, default to inf
+    ---@param depth? integer how many layers will be dumped, default to inf
     ---@return string
     ---@nodiscard
     function TABLE.dump(t,depth)
@@ -754,7 +765,7 @@ end
 
 ---Get element count of table
 ---@param t table
----@return number
+---@return integer
 ---@nodiscard
 function TABLE.getSize(t)
     local size=0
@@ -765,7 +776,7 @@ end
 ---Count value repeating time in [1~#]
 ---@param t any[]
 ---@param val any
----@return number
+---@return integer
 ---@nodiscard
 function TABLE.count(t,val)
     local count=0
@@ -780,7 +791,7 @@ end
 ---Count value repeating time in whole table
 ---@param t table
 ---@param val any
----@return number
+---@return integer
 ---@nodiscard
 function TABLE.countAll(t,val)
     local count=0
@@ -810,7 +821,7 @@ end
 ---Execute func(table[i],i) in [1~#]
 ---@generic T
 ---@param t T[]
----@param f fun(v:T, i:number)
+---@param f fun(v:T, i:integer)
 ---@param rev? boolean Reverse the order, allow removing elements from list
 function TABLE.foreach(t,f,rev)
     if rev then
@@ -826,7 +837,7 @@ end
 
 ---Execute func(table[k],k) for all elements in table (only allow removing elements with t[k]=nil)
 ---@param t table
----@param f fun(v:any, k:number)
+---@param f fun(v:any, k:any)
 function TABLE.foreachAll(t,f)
     for k,v in next,t do
         f(v,k)