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)
    local images = {}
    local pattern = "%[%[File:([^|%]]+)"
    for image in string.gmatch(value, pattern) do
        table.insert(images, image)
    end
    return images
end

local function getTab(value, label)
    local pattern = label .. "|"
    local pattern2 = label .. " *= *"
    local matchStart, matchEnd = string.find(value, pattern)
    if not matchStart then
        matchStart, matchEnd = string.find(value, pattern2)
        if not matchStart then
            return nil
        end
    end

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

    while i <= #value do
        local char = value: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
            tab = tab .. char
        end

        i = i + 1
    end

    return tab
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

    local tab = getTab(value, frame.args.tab)
    if tab then
        value = tab
    end

    if not value and frame.args.tab then
        return "Error: Tab not found in infobox."
    end

    if label:lower() == "image" then
        local images = getImages(value)
        local imageIndex = tonumber(frame.args.imageIndex) or 1
        local size = frame.args.size or ""
        local link = frame.args.link or ""

        if imageIndex < 1 or imageIndex > #images then
            return "Error: Invalid image index."
        end

        local selectedImage = images[imageIndex]
        if link ~= "" then
        	return string.format('[[File:%s|%s|link=%s]]', selectedImage, size, link)
        else
        	return string.format('[[File:%s|%s]]', selectedImage, size)
        end
    else
        local processedValue = frame:preprocess(value)
        return processedValue
    end
end

return p