Module:VolDiff

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

local p = {}

-- Helper function to parse chapter number from the chapter link
local function getChapterNumber(chapterLink)
    mw.log("getChapterNumber: " .. chapterLink)
    -- Match the entire chapter link
    local fullMatch = mw.ustring.match(chapterLink, "%[%[(.-)%]%]")
    if fullMatch then
        -- Split at the pipe if there is a display title
        local splitMatch = mw.text.split(fullMatch, "|")
        -- Return the first part before the pipe (if exists) or the entire match
        return splitMatch[1]
    end
    return nil
end

-- Helper function to extract content under a specific section header
local function extractSection(content, header)
    mw.log("extractSection: " .. header)
    local pattern = header .. "(.-)(===)"
    local sectionContent = mw.ustring.match(content, pattern)
    if not sectionContent then
        -- Try to match until the end of the content if the section is the last one
        pattern = header .. "(.-)$"
        sectionContent = mw.ustring.match(content, pattern)
    end
    mw.log("extractSection result: " .. (sectionContent or "nil"))
    return sectionContent
end

-- Helper function to validate and sanitize UTF-8 strings
local function validateUtf8(str)
    if mw.ustring.find(str, "[%z\1-\127\194-\244][\128-\191]*") then
        return str
    else
        mw.log("Warning: Invalid UTF-8 sequence found.")
        return mw.text.encode(str)
    end
end

-- Helper function to capture the entire template, including nested braces
local function captureTemplate(content, templateName)
    mw.log("captureTemplate: start for " .. templateName)
    local templateStart, templateEnd = mw.ustring.find(content, "{{%s*" .. templateName .. "%s*[|\n]")
    if not templateStart then
        mw.log("captureTemplate: templateStart not found for " .. templateName)
        return nil
    end

    local finalTemplate = content:sub(templateStart, templateEnd)
    local i = templateEnd + 1
    local braceCount = 2  -- Account for the initial "{{"

    while i <= #content do
        local char = content:sub(i, i)
        finalTemplate = finalTemplate .. char

        if char == '{' then
            braceCount = braceCount + 1
        elseif char == '}' then
            braceCount = braceCount - 1
            if braceCount == 0 then
                local nextChar = content:sub(i + 1, i + 1)
                if nextChar == '}' or nextChar == '\n' or nextChar == '|' then
                    break
                else
                    braceCount = braceCount + 1
                end
            end
        end

        i = i + 1
    end

    -- Validate the final template as a UTF-8 string
    finalTemplate = validateUtf8(finalTemplate)
    mw.log("captureTemplate result: " .. finalTemplate)
    return finalTemplate
end

-- Helper function to capture the entire Desc parameter including nested braces and wikitext links
local function captureDescParameter(diff)
    mw.log("captureDescParameter: start")
    local descStart, descEnd = mw.ustring.find(diff, "Desc%s*=%s*")
    if not descStart then
        mw.log("captureDescParameter: descStart not found")
        return nil
    end
    local finalDesc = ""
    local i = descEnd + 1
    local braceCount = 0

    while i < #diff - 2 do -- Subtract 2 from length to remove the }} from Diff
        local char = diff:sub(i, i)
        finalDesc = finalDesc .. char

        if char == '{' or char == '[' then
            braceCount = braceCount + 1
        elseif char == '}' or char == ']' then
            if braceCount > 0 then
                braceCount = braceCount - 1
            else
                break
            end
        elseif char == '|' and braceCount == 0 then
            break
        end

        i = i + 1
    end

    -- Validate the final description as a UTF-8 string
    finalDesc = validateUtf8(finalDesc)
    mw.log("captureDescParameter result: " .. finalDesc)
    return mw.text.trim(finalDesc)
end

-- Main function to process the DiffBox template
function p.main(frame)
    mw.log("main: start")
    local args = frame:getParent().args
    local currentTitle = mw.title.getCurrentTitle().text
    local sourcePage = args[1]
    local sectionType = args[2] or "magazine"
    local sectionHeader = sectionType == "reprint" and "===Reprint Differences===" or "===Magazine Differences==="
    local result = ""

    if not sourcePage then
        mw.log("main: No source page specified.")
        return "Error: No source page specified."
    end

    -- Fetch the content from the specified source page
    local content = mw.title.new(sourcePage):getContent()
    if not content then
        mw.log("main: Could not fetch content from the specified source page.")
        return "Error: Could not fetch content from the specified source page."
    end
    content = validateUtf8(content)
    mw.log("main: content fetched")

    -- Extract the section content
    local sectionContent = extractSection(content, sectionHeader)
    if not sectionContent then
        mw.log("main: No section found with header " .. sectionHeader)
        return "Error: No section found with header " .. sectionHeader
    end
    sectionContent = validateUtf8(sectionContent)
    mw.log("main: sectionContent extracted")

    -- Extract the entire DiffBox content using brace counting
    local diffBoxContent = captureTemplate(sectionContent, "DiffBox")
    if not diffBoxContent then
        mw.log("main: No DiffBox found in the specified section.")
        return "Error: No DiffBox found in the specified section."
    end

    -- Extract DiffBox parameters
    local part = mw.ustring.match(diffBoxContent, "Part%s*=%s*([^|]+)") or ""
    local title = mw.ustring.match(diffBoxContent, "Title%s*=%s*([^|]+)") or ""
    local label1 = mw.ustring.match(diffBoxContent, "Label1%s*=%s*([^|]*)") or ""
    local label2 = mw.ustring.match(diffBoxContent, "Label2%s*=%s*([^|]*)") or ""

    -- Start the DiffBox template
    result = result .. string.format('{{DiffBox|Part=%s|Title=%s', part, title)
    if label1 ~= "" then result = result .. string.format('|Label1=%s', label1) end
    if label2 ~= "" then result = result .. string.format('|Label2=%s', label2) end
    result = result .. '|\n'

    -- Process the section content to extract Diff entries
    local hasDiffs = false
    local startPos = 1
    local diffCount = -1 -- Subtract DiffBox

    -- Count the number of Diff templates
    for _ in mw.ustring.gmatch(diffBoxContent, "{{%s*Diff%s*[|\n]") do
        diffCount = diffCount + 1
    end
    mw.log("main: diffCount = " .. diffCount)

    local currentDiff = 0

    while true do
        local diffContent = captureTemplate(diffBoxContent:sub(startPos), "Diff")
        if not diffContent then
            mw.log("main: no more diffContent found")
            break
        end

        startPos = startPos + #diffContent
        currentDiff = currentDiff + 1
        mw.log("main: processing diffContent #" .. currentDiff)

        local diffPart = mw.ustring.match(diffContent, "Part%s*=%s*([^|]+)")
        local diffChapter = mw.ustring.match(diffContent, "Chapter%s*=%s*(%[%[.-%]%])")
        local diffPage = mw.ustring.match(diffContent, "Page%s*=%s*([^|]+)")
        local diffImage1 = mw.ustring.match(diffContent, "Image1%s*=%s*([^|]+)")
        local diffImage2 = mw.ustring.match(diffContent, "Image2%s*=%s*([^|]+)")
        local diffDesc = captureDescParameter(diffContent)
        local diffLabel1 = mw.ustring.match(diffContent, "Label1%s*=%s*([^|]*)") or ""
        local diffLabel2 = mw.ustring.match(diffContent, "Label2%s*=%s*([^|]*)") or ""

        if diffPart and diffChapter and diffPage and diffImage1 and diffImage2 and diffDesc then
            local chapterNumber = getChapterNumber(diffChapter)

            -- Check if the current page title matches the chapter number
            if chapterNumber and mw.ustring.match(currentTitle, chapterNumber) then
                hasDiffs = true
                result = result .. string.format('{{Diff|Part=%s|Chapter=%s|Page=%s|Image1=%s|Image2=%s|Desc=%s',
                    diffPart, diffChapter, diffPage, diffImage1, diffImage2, mw.text.trim(diffDesc))

                -- Add optional labels if they are present
                if diffLabel1 ~= "" then result = result .. string.format('|Label1=%s', diffLabel1) end
                if diffLabel2 ~= "" then result = result .. string.format('|Label2=%s', diffLabel2) end

                result = result .. '}}\n'
            end
        end
    end

    -- End the DiffBox template or provide a message if no diffs are found
    if hasDiffs then
        result = result .. '}}'
    else
        result = "No differences found for this chapter."
    end

    mw.log("main: end")
    return frame:preprocess(result)
end

return p