Module:A or an: Difference between revisions

Revision as of 22:37, 30 December 2019

Latest revision as of 19:06, 18 May 2020

(3 intermediate revisions by 2 users not shown)

Line 1:

local p = {}

local words = mw.loadData('Module:A or an/words')

local find = mw.ustring.find

local gsub = mw.ustring.gsub

local lower = mw.ustring.lower

local match = mw.ustring.match

local lcVChars = 'aeiouà-æè-ïò-öø-üāăąēĕėęěĩīĭįıijōŏőœũūŭůűų'

Line 12 ⟶ 7:

local function findWord(s, t)

for i, v in ipairs(t) do

if mw.ustring.find(s, '^' .. v .. '$') then

return true

end

Line 26 ⟶ 21:

local origStr = s

s = mw.ustring.gsub(s, '</?[A-Za-z][^>]->', '') -- Remove HTML tags

s = mw.ustring.gsub(s, '%[%[[^%|]+%|(..-)%]%]', '%1') -- Remove wikilinks

s = mw.ustring.gsub(mw.ustring.gsub(s, '%[%[', ''), '%]%]', '')

s = mw.ustring.gsub(s, '^["%$\'%(<%[%{¢-¥₠-₿]+', '') -- Strip some symbols at the beginning

s = mw.ustring.match(s, '^%.?[0-9%u%l]+') or s -- Extract the first word

if mw.ustring.find(s, '^[0-9]') then -- It begins with a number

s = mw.ustring.match(s, '^[0-9]+') -- Extract the number

if findWord(s, words['.vNums']) then -- '18' etc.

pron = 'an'

end

elseif mw.ustring.match(s, '^[0-9%u]+$') then -- It looks like an acronym

if mw.ustring.find(s, '^[' .. ucVvChars .. ']')

and not findWord(s, words['.cvAcronyms']) -- Exclude 'NASA' etc.

or findWord(s, words.vvAcronyms) -- 'UNRWA' etc.

then

pron = 'an'

end

else

s = mw.ustring.lower(s) -- Uncapitalize

if mw.ustring.find(s, '^['.. lcVChars .. ']') then -- It begins with a vowel

if not findWord(s, words['.vcWords']) -- Exclude 'euro' etc.

or findWord(s, words['.vvWords']) -- But not 'Euler' etc.

then

pron = 'an'

end

elseif args.variety and mw.ustring.lower(args.variety) == 'us' -- 'herb' etc.

and findWord(s, words['.cvWordsUS'])

or findWord(s, words['.cvWords']) -- 'hour' etc.

then

pron = 'an'