imported>Firefly: cleanup

2022-01-30T17:31:41Z

cleanup

New page

local p = {}

local function prepareText( inputString )
local retVal = inputString
local cleanupPatterns = {}
-- Note, these are invoked in order, which may be important!
cleanupPatterns[1] = '=+[^=]+=+[ ]-\n' -- no headers
cleanupPatterns[2] = '[^[]%[[^[ ]+[]%a%d]' -- no external link URLs
cleanupPatterns[3] = '{{[^|]-|' -- no template invocations
cleanupPatterns[4] = '<!%-%-.-%-%->' -- no HTML comments
cleanupPatterns[5] = '\n' -- strip newlines

for _, pattern in ipairs(cleanupPatterns) do
retVal = mw.ustring.gsub(retVal, pattern, '')
end

return retVal
end

local function getWordcount( text )
return (table.maxn(mw.text.split(prepareText(text), '[ ]+')) + 1)
end

local function getWordcountsForSections( targetPage, sectionLevel )

local success, targetTitle = pcall( mw.title.new, targetPage )
if not success or ( success and not targetTitle ) then
return nil
end

local targetContent = targetTitle:getContent()

local headerMarkup = string.rep( "=", sectionLevel )
local sections = mw.text.split(targetContent, '\n[ ]-'.. headerMarkup ..'[^=]+'.. headerMarkup ..'[ ]-\n')

local wordCounts = {}
for i, section in ipairs(sections) do
if (i>1) then -- Skip preamble section
wordCounts[#wordCounts + 1] = getWordcount(section)
end
end

local resultTable = {}
local headersIterator = mw.ustring.gmatch(targetContent, '\n[ ]-'.. headerMarkup ..'([^=]+)'.. headerMarkup ..'[ ]-\n')
local i = 1
for header in headersIterator do
if mw.ustring.match(header, '{your user name}') == nil then -- we don't care about example sections
resultTable[i .. " " .. header] = wordCounts[i] -- first capture group
end
i = i+1
end
return resultTable
end

local function getRowsFromWordcountTable( wordcountTable )
local ret = ""
for section, count in pairs(wordcountTable) do
ret = ret .. string.format("\n|-\n| %s || %s", section, count)
end
return ret
end

local function makeWordcountTable( args )
return
end

function p.main( frame )
local frame = mw.getCurrentFrame()
local targetPage = frame.args["target"]

local sectionLevel = frame.args["sectionlevel"]
sectionLevel = tonumber(sectionLevel, 10)
if sectionLevel == nil then
sectionLevel = 2
end

local tableCss = frame.args["tablecss"]
if tableCss == nil then
tableCss = ''
end

local wordcountTable = getWordcountsForSections( targetPage, sectionLevel )

return mw.ustring.format( '\n{| class="wikitable sortable" %s \n|-\n! Section !! Word Count%s\n|-\n|}', tableCss, getRowsFromWordcountTable(wordcountTable) )
end

return p

Module:Sandbox/Firefly - Revision history

imported>Firefly: cleanup