Module:InfoboxExtractor

Documentation for this module may be created at Module:InfoboxExtractor/doc

local p = {}

local function extractInfobox(wikitext)
    local startInfobox = string.find(wikitext, "{{[^}]*Infobox")
    if not startInfobox then
        return nil
    end

    local braceCount = 0
    local endInfobox = startInfobox

    for i = startInfobox, #wikitext do
        local char = wikitext:sub(i, i)
        if char == '{' then
            braceCount = braceCount + 1
        elseif char == '}' then
            braceCount = braceCount - 1
        end

        if braceCount == 0 then
            endInfobox = i
            break
        end
    end

    return wikitext:sub(startInfobox, endInfobox)
end

local function getLabelValue(infobox, label)
    local pattern = "|" .. label .. " *= *"
    local matchStart, matchEnd = string.find(infobox, pattern)
    if not matchStart then
        return nil
    end

    local value = ""
    local braceCount = 0
    local i = matchEnd + 1
    local isNewLine = false

-- need to do - 2 so it ignores the }} at the end of the infobox
    while i <= #infobox - 2 do
        local char = infobox:sub(i, i)

        if char == '{' then
            braceCount = braceCount + 1
        elseif char == '}' then
            braceCount = braceCount - 1
        end

        if braceCount == 0 and char == '\n' then
            isNewLine = true
        elseif isNewLine and char == '|' then
            break
        else
            isNewLine = false
            value = value .. char
        end

        i = i + 1
    end

    return value
end

local function getImages(value, imageTab)
    local images = {}

    local function processTabber(content)
	    local startPattern = "{{Tabber%s*\n?"
	    local endPattern = "}}"
	    local startPos, endPos, tabberContent = string.find(content, startPattern .. "(.-)" .. endPattern)
	
	    if startPos and endPos then
	        local tabPattern = "|%s*([^|]-)%s*|%s*%[%[File:([^|%]]+)"
	        for tab, image in string.gmatch(tabberContent, tabPattern) do
	            local trimmedTab = mw.text.trim(tab)
	            if not imageTab or mw.ustring.lower(trimmedTab) == mw.ustring.lower(imageTab) then
	                return image
	            end
	        end
	    else
	        local pattern = "%[%[File:([^|%]]+)"
	        local image = string.match(content, pattern)
	        if image then
	            return image
	        end
	    end
	
	    return nil
	end

    processTabber(value)

    return images
end

function p.extract(frame)
	local preprocess = frame.args.template == "true"
    local pageTitle = frame.args[1]
    if preprocess then
        pageTitle = frame:preprocess(pageTitle)
    end
    if not pageTitle then
        return "Error: Please provide a page title."
    end

    local page = mw.title.new(pageTitle)
    if not page or not page.exists then
        return "Error: Page not found."
    end

    local wikitext = page:getContent()
    local infobox = extractInfobox(wikitext)

    if not infobox then
        return "Error: Infobox not found."
    end

    local label = frame.args[2]
    if not label then
        return "Error: Please provide a label."
    end

    local value = getLabelValue(infobox, label)

    if not value then
        return "Error: Label not found in infobox."
    end

    if label:lower() == "image" then
    	local imageTab = frame.args.imageTab or ""
	    local images = getImages(value, imageTab)
	    local selectedImage
	
	    for _, image in ipairs(images) do
	        if image.tab == imageTab then
	            selectedImage = image.image
	            break
	        end
	    end
	
	    if not selectedImage then
	        return "Error: Invalid image tab selection."
	    end
	
	    local size = frame.args.size or ""
	    local link = frame.args.link or ""
	
	    if link ~= "" then
	        return string.format('[[File:%s|%s|link=%s]]', selectedImage, size, link)
	    else
	        return string.format('[[File:%s|%s]]', selectedImage, size)
	    end
    end
end

return p