Module:Unicode data/scripts/make

 localp={}

 localArray=require'Module:Array'

 localfunctionpattern_escape(str)
 return(str:gsub('%p','%%%1'))
 end

 functionp.make_script_name_to_code(page_name)
 localproperty_value_aliases=assert(assert(mw.title.new(page_name)):getContent())

 localscript_aliases=property_value_aliases:match(
 pattern_escape'# Script (sc)'
 ..'%s+(.-)%s+'
 ..pattern_escape'# Script_Extensions (scx)')

 localscript_name_to_code={}

 forcode,nameinscript_aliases:gmatch'sc%s+;%s+(%a+)%s+;%s+([%a_]+)'do
 script_name_to_code[name]=code
 end

 returnscript_name_to_code
 end

 functionp.make_script_data(scripts_txt,property_value_aliases_txt)
 localscript_data=assert(assert(mw.title.new(scripts_txt)):getContent())

 localscript_name_to_code=p.make_script_name_to_code(property_value_aliases_txt)
 setmetatable(script_name_to_code,{__index=function(self,k)
 error(('No code for "%s"'):format(k))
 end})

 localscript_ranges=Array()

 localprev_codepoint,prev_script_name,prev_script_range
 forcodepoint1,codepoint2,script_nameinscript_data:gmatch'%f[^\n%z](%x+)%.?%.?(%x*)%s+;%s*([%w_]+)'do
 codepoint1,codepoint2=tonumber(codepoint1,16),tonumber(codepoint2,16)
 localscript_range
 ifprev_script_rangeandscript_name==prev_script_nameandcodepoint1-prev_codepoint==1then
 prev_script_range[2]=codepoint2orcodepoint1
 else
 script_range={codepoint1,codepoint2orcodepoint1,script_name_to_code[script_name]}
 script_ranges:insert(script_range)
 end
 prev_codepoint,prev_script_name,prev_script_range=
 codepoint2orcodepoint1,script_name,script_rangeorprev_script_range
 end

 localsingles={}
 locali=1

 whilescript_ranges[i]do
 locallow,high,script_code=unpack(script_ranges[i])
 iflow==highthen
 singles[low]=script_code
 script_ranges:remove(i)
 else
 i=i+1
 end
 end

 script_ranges:sort(
 function(range1,range2)
 returnrange1[1]<range2[1]
 end)

 localtemplate=[[
 local data = {
 	singles = {
 ...
 	},

 	ranges = {
 ...
 	},
 	-- Scripts.txt gives full names; here we consider them aliases to save space.
 	aliases = {
 ...
 	},
 }
 ]]

 localprinted_ranges=Array()
 for_,rangeinipairs(script_ranges)do
 locallow,high,script_code=unpack(range)
 printed_ranges:insert(('\t\t{ 0x%05X, 0x%05X, "%s" },'):format(low,high,script_code))
 end

 localprinted_singles=Array()
 forcodepoint,script_codeinrequire'Module:TableTools'.sortedPairs(singles)do
 printed_singles:insert(('\t\t[0x%05X] = "%s",'):format(codepoint,script_code))
 end

 localprinted_script_name_to_code=Array()
 forname,codeinrequire'Module:TableTools'.sortedPairs(script_name_to_code,
 function(name1,name2)
 returnscript_name_to_code[name1]<script_name_to_code[name2]
 end)do
 printed_script_name_to_code:insert(('\t\t%s = "%s",'):format(code,name:gsub('_',' ')))
 end

 localdata=template
 :gsub('%.%.%.',printed_singles:concat('\n'),1)
 :gsub('%.%.%.',printed_ranges:concat('\n'),1)
 :gsub('%.%.%.',printed_script_name_to_code:concat('\n'),1)

 returndata
 end

 functionp.highlight(lua_code)
 returnmw.getCurrentFrame():extensionTag{
 name="syntaxhighlight",
 content=lua_code,
 args={lang="lua"}
 }
 end

 functionp.main(frame)
 localdir='User:Erutuon/Unicode'
 localscripts_txt=dir..'/Scripts.txt'
 localproperty_value_aliases_txt=dir..'/PropertyValueAliases.txt'

 returnp.highlight(p.make_script_data(scripts_txt,property_value_aliases_txt))
 end

 returnp