Content deleted Content added
comma not period, elseif was okay |
add digits with ordinal |
||
Line 181:
-- These are the element type "constants" for readability mostly
local ELEMENT_INVALID = 1
local ELEMENT_ONETWODIGITS = 2 -- '1' '12' '01'
local ELEMENT_FOURDIGITS = 3 -- '1234'
local ELEMENT_WHITESPACE = 4 -- ' ' ' '
local ELEMENT_MONTHWORD = 5 -- 'May' 'February' 'Aug'
local ELEMENT_COMMA = 6 -- ',' ', '
local ELEMENT_DASH = 7 -- '-' ' - ' ' — ' '- ' ' -'
local ELEMENT_DATESEPARATOR = 8 -- '-'
local ELEMENT_TIMESEPARATOR = 9 -- ':'
local ELEMENT_TIMEPERIOD = 10 -- 'am' 'p.m.'
local ELEMENT_PERIODWHITESPACE = 11 -- '.' or '. '
local ELEMENT_ONETWODIGITSWITHORDINAL = 12 -- '12th' '3rd'
function seekNextElement()
Line 196 ⟶ 197:
-- Profiler says mw.ustring.find is the bottleneck, probably because it's unicode; not sure how to improve though besides writing my own pattern matcher
-- Digits with letters
local foundPositionStart, foundPositionEnd, foundMatch, foundMatch2 = mw.ustring.find(seekString, '^([0-9])+([a-z]+)%.?', currentPosition)
if (foundPositionStart) then
currentPosition = foundPositionEnd + 1 -- this is our new start ___location
-- Additionally check how many digits we actually have, as arbitrary number isn't valid
if (#foundMatch <= 2) then -- most likely a day number
if (foundMatch2 == 'st' or foundMatch2 == 'nd' or foundMatch2 == 'rd' or foundMatch2 == 'th') then -- won't bother checking against a number, no false positives that I saw in 120k cases
return ELEMENT_ONETWODIGITSWITHORDINAL, tonumber(foundMatch), (currentPosition > mw.ustring.len(seekString))
else
return ELEMENT_INVALID -- not a valid ordinal indicator
end
else
return ELEMENT_INVALID -- just the invalid, the number of digits (3+) won't match any patterns
end
end
-- Digits
local foundPositionStart, foundPositionEnd, foundMatch = mw.ustring.find(seekString, '^([0-9]+)', currentPosition)
Line 360 ⟶ 378:
-- Only immediate big improvement is to only seekNextElement() when actually checking that deep, though this will make a (even bigger) mess
if (elements[1] == ELEMENT_ONETWODIGITS or elements[1] == ELEMENT_ONETWODIGITSWITHORDINAL) then -- '3' or '10' or '12th'
if (elements[2] == ELEMENT_WHITESPACE or elements[2] == ELEMENT_PERIODWHITESPACE) then -- '3 ' or '3. '
if (elements[3] == ELEMENT_MONTHWORD) then -- '3 May'
Line 388 ⟶ 406:
end
elseif (elements[6] == ELEMENT_DASH or elements[6] == ELEMENT_DATESEPARATOR) then -- '3 May 2013 - '
if (elements[7] == ELEMENT_ONETWODIGITS or elements[3] == ELEMENT_ONETWODIGITSWITHORDINAL) then -- '3 May 2013 - 12' or '3rd May 2013 - 12th'
if (elements[8] == ELEMENT_WHITESPACE) then -- '3 May 2013 - 12 '
if (elements[9] == ELEMENT_MONTHWORD) then -- '3 May 2013 - 12 February'
Line 402 ⟶ 420:
end
elseif (elements[4] == ELEMENT_DASH or elements[4] == ELEMENT_DATESEPARATOR) then -- '3 May - '
if (elements[5] == ELEMENT_ONETWODIGITS or elements[5] == ELEMENT_ONETWODIGITSWITHORDINAL) then -- '3 May - 12' or '3rd May - 12th'
if (elements[6] == ELEMENT_WHITESPACE) then -- '3 May - 12 '
if (elements[7] == ELEMENT_MONTHWORD) then -- '3 May - 12 October'
Line 417 ⟶ 435:
end
elseif (elements[2] == ELEMENT_DASH or elements[2] == ELEMENT_DATESEPARATOR) then -- '3 - '
if (elements[3] == ELEMENT_ONETWODIGITS or elements[3] == ELEMENT_ONETWODIGITSWITHORDINAL) then -- '3 - 12' or '3rd - 12th'
if (elements[4] == ELEMENT_WHITESPACE) then -- '3 - 12 '
if (elements[5] == ELEMENT_MONTHWORD) then -- '3 - 12 May'
Line 440 ⟶ 458:
local i = 0 -- this is our offset from the closest possible ___location for date seeking
if (elements[2] == ELEMENT_TIMESEPARATOR and elements[1] ~= ELEMENT_ONETWODIGITSWITHORDINAL) then -- '10:' but not
possibleHour = values[1] -- only once we see ':' (or 'am' below) it is likely a time
if (elements[3] == ELEMENT_ONETWODIGITS) then -- '10:28'
Line 571 ⟶ 589:
if (numberOfElements == 1) then return checkAndOutput(nil, values[1], nil, nil, nil, nil) end
if (elements[2] == ELEMENT_WHITESPACE or elements[2] == ELEMENT_PERIODWHITESPACE) then -- 'May ' or 'Feb. '
if (elements[3] == ELEMENT_ONETWODIGITS or elements[3] == ELEMENT_ONETWODIGITSWITHORDINAL) then -- 'May 3' or 'May 3rd'
if (numberOfElements == 3) then return checkAndOutput(nil, values[1], values[3], nil, nil, nil) end
if (elements[4] == ELEMENT_COMMA or elements[4] == ELEMENT_WHITESPACE) then -- 'May 3, '
Line 599 ⟶ 617:
if (elements[7] == ELEMENT_MONTHWORD) then -- 'May 3, 2013 - February'
if (elements[8] == ELEMENT_WHITESPACE) then -- 'May 3, 2013 - February '
if (elements[9] == ELEMENT_ONETWODIGITS or elements[3] == ELEMENT_ONETWODIGITSWITHORDINAL) then -- 'May 3, 2013 - February 12' or 'May 3rd, 2013 - February 12th'
if (elements[10] == ELEMENT_COMMA or elements[10] == ELEMENT_WHITESPACE) then -- 'May 3, 2013 - February 12, '
if (elements[11] == ELEMENT_FOURDIGITS) then -- 'May 3, 2013 - February 12, 2014'
Line 613 ⟶ 631:
if (elements[5] == ELEMENT_MONTHWORD) then -- 'May 3 - June'
if (elements[6] == ELEMENT_WHITESPACE) then -- 'May 3 - June '
if (elements[7] == ELEMENT_ONETWODIGITS or elements[3] == ELEMENT_ONETWODIGITSWITHORDINAL) then -- 'May 3 - June 12' or 'May 3rd - June 12th'
if (numberOfElements == 7) then return checkAndOutput(nil, values[1], values[3], nil, nil, nil, nil, values[5], values[7], nil, nil, nil) end
if (elements[8] == ELEMENT_COMMA or elements[8] == ELEMENT_WHITESPACE) then -- 'May 3 - June 12, '
Line 622 ⟶ 640:
end
end
elseif (elements[5] == ELEMENT_ONETWODIGITS or elements[3] == ELEMENT_ONETWODIGITSWITHORDINAL) then -- 'May 3 - 12' or 'May 3rd - 12th'
if (numberOfElements == 5) then return checkAndOutput(nil, values[1], values[3], nil, nil, nil, nil, values[1], values[5], nil, nil, nil) end
if (elements[6] == ELEMENT_COMMA or elements[6] == ELEMENT_WHITESPACE) then -- 'May 3 - 12, '
| |||