मॉड्यूल:अक्षर गणना

Module documentation[view] [edit] [history] [purge]

यह मॉड्यूल इनपुट पाठ में कैरेक्टरों, अक्षरों और स्पेसों की गणना करता है। यह साधारण पाठ के साथ काम करता है और बहुत सीमित स्तर तक विकिपाठ समर्थित करता है। इसका प्रयोग साँचा:अक्षर गणना द्वारा किया जाता है।

The above documentation is transcluded from मॉड्यूल:अक्षर गणना/doc. (edit | history)
Editors can experiment in this module's sandbox (create | mirror) and testcases (create) pages.
Subpages of this module.

--[[Module for counting akshars in input text
	Call using {{#invoke:modname|count|param1|param2}}

	param1:
	String
	input text
	Default: '' (Empty string)
	Plain text: fully supported
	Wikitext: Partly supported (see Support below)

	param2:
	Boolean
	Specifies whether to ignore wikisyntax
	or not when counting
	Default:true
	By default all counts ignore wikisyntax
	If set to false, all counts will include wikisyntax

	Support:
	All input text is preprocessed as wikitext
	Anything inside xml tags is then removed
	Non-visible part of wikilinks and external
	links is removed
	[=[अगर]=]-->अगर
	[=[आगरा|अगर]=]-->अगर
	[http://google.com अगर]-->अगर
	Removes bullets and numbering #es
	*अगर-->अगर
	#अगर-->अगर

	outputs the character length,
	akshar length and
	the number of spaces used
]]

local ustring = mw.ustring

--Preprocesses the input
local function preprocess(text, frame)
	local ptext = frame:preprocess(text)
	return ptext
end

--Removes the wikisyntax from the input
local function dewiki(text)
	local dewikitext = text
--	mw.log('dewikitext is ' .. dewikitext)
	--the following table is iterated on first to last
	local wikitext = {'((\127[^\127]*\127))', --xml tags from preprocessing
						'(%[%[[^%]%]|]*|)', --piped wikilinks start
						'(%[%[([^%]%]]*))', --plain wikilinks start
						'(([^%[]?)%[[^%s%[]+%s)', --named external links start
						'(([^%[]?)%[[^%]]+)', --unnamed external links
						'(%])', --ending brackets for all links
						'[*#]' --numbers and bullets
					}
	local replacement = {'', '', '%2', '%2', '%2', '', ''}
	local flag = 1
	while flag <= #wikitext do
		dewikitext = ustring.gsub(dewikitext, wikitext[flag], replacement[flag])
--		mw.log('dewikitext is ' .. dewikitext)
		flag = flag + 1
	end
	return dewikitext
end	

--Does the counting
local function count(frame)
	local deva = '[ँ-॰%s]+'
	local dia = '[ँंः़ािीुूृॅेैॉोौ॒॰]'
	local halant = '[्]'
	local text = frame.args[1]
	if text == nil then
		return 'कैरैक्टर:0, अक्षर:0, स्पेस:0'
	end
	local ignorewikitext = frame.args[2]
	local charnum = ustring.len(text)
	local spacenum = 0
--	mw.log('input is ' .. text)

	if ignorewikitext ~= false then --ignore wikisyntax, default state
		text = preprocess(text, frame)
		text = dewiki(text)
		charnum = ustring.len(text)
--	else --non-default state, count wikisyntax
--		aksharnum = ustring.len(ustring.gsub(text, wikitext, ''))
	end
	
	local aksharnum = ustring.len(text)
--	charnum = ustring.len(text)
--	mw.log('1. charnum is ' .. charnum)
--	mw.log('2. aksharnum is ' .. aksharnum)
--	mw.log('3. spacenum is ' .. aksharnum)

	matches = ustring.gmatch(text, deva)

	for fmatch in matches do
--		mw.log('fmatch in matches')
		diacritics = ustring.gmatch(fmatch, dia)
		halcount = ustring.gmatch(fmatch, halant)
		space = ustring.gmatch(fmatch, '[%s]')

--		mw.log('fmatch is ' .. fmatch)

		for fmatch1 in diacritics do
			aksharnum = aksharnum - 1
		end

--		mw.log('4. aksharnum after subtracting diacritic length is ' .. aksharnum)

		for fmatch2 in halcount do
			aksharnum = aksharnum - 1.5
		end

--		mw.log('5. aksharnum after subtracting halant length is ' .. aksharnum)

		for fmatch3 in space do
			aksharnum = aksharnum - 1
			spacenum = spacenum + 1
		end
		
--		mw.log('5. num after subtracting spaces is ' .. aksharnum)
		
	end
	local output = 'कैरैक्टर: ' .. charnum .. ', अक्षर: ' .. aksharnum .. ', स्पेस: ' .. spacenum
--	output = text .. '\n' .. output
	return output
end

return {count = count, preprocess = preprocess, dewiki = dewiki}