Module:CR

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

local CR = {}

-- load rules text
CR.text =  mw.loadData( 'Module:CR/rules' ).text

-- locals for performance
local format, find, match, sub, gsub, gmatch, rep, lower, upper = string.format, string.find, string.match, string.sub, string.gsub, string.gmatch, string.rep, string.lower, string.upper
local tonumber, tostring, type, assert, ipairs = tonumber, tostring, type, assert, ipairs
local tinsert = table.insert

-- *_PATTERN: used to match CR text
-- *_FORMAT: used for formatting output

-- We need to split the text via LINE_PATTERN before we can use the beginning-of-string caret
-- Otherwise, we would need a more complicated pattern to match all possible indices
-- (Off the top of my head, I'm not sure Lua's pattern matching *could* do it in one pattern. It's not a full regex suite.)
local LINE_PATTERN = "(.-)\n"
local RULE_LINE_PATTERN = "^%d"
local TOC_END_LINE_PATTERN = "^Glossary"
local BODY_END_LINE_PATTERN  = "^Glossary"
local GLOSSARY_END_LINE_PATTERN = "^Credits"
local GENERAL_RULE_PATTERN = "General"

-- local WIKILINK_FORMAT = "[[%s]]" -- Condider: no-longer used, remove it entirely
local WIKILINK_ALT_FORMAT = "[[%s|%s]]"
local GENERAL_RULE_EXPANDED_TEXT_FORMAT = "General_(%s)"
local CASUAL_FORMAT_EXPANDED_TEXT_FORMAT = "%s_(format)"

-- Note that, per the Introduction to the Comprehensive Rules, the subrules skip the letters L and O to avoid confusion with numerals.
local INDEX_SUBRULE_CHARACTER_SET = "ABCDEFGHIJKMNPQRSTUVWXYZabcdefghijkmnpqrstuvwxyz"

-- This is the format of the line in the Comp Rules where the date is found. As of last edit, it's the second line of text.
-- The date itself is wrapped in parentheses to extract that text alone.
local COMP_RULES_DATE_PATTERN = "These rules are effective as of ([^%.]+)."

-- "From the Comprehensive Rules (March 27, 2015—Dragons of Tarkir)"
local LAST_UPDATED_FORMAT = "<p class=\"crTitle\">From the %s (%s)</p>"

-- String to add one obsolete rules int the RulesDiv.
local OBSOLETE_PATTERN = "&nbsp;'''([[Obsolete]])'''"

--Settings

--Sources
local CR_SOURCE = "''[[Comprehensive Rules]]''"
local GLOSSARY_SOURCE = "glossary of the "..CR_SOURCE

--Categories
local GLOSSARY_CAT = "[[Category:Glossary]]"
local OBSOLETE_CAT = "[[Category:Obsolete]]"

--Colors
local RULES_COLOR = "var(--theme-page-background-color)"
local GLOSSARY_COLOR = "var(--theme-page-accent-mix-color)"
local OBSOLETE_COLOR = "var(--theme-body-background-color)"

-- Escape magic characters in a string.
local sanitize
do
    local matches =
    {
        ["^"] = "%^";
        ["$"] = "%$";
        ["("] = "%(";
        [")"] = "%)";
        ["%"] = "%%";
        ["."] = "%.";
        ["["] = "%[";
        ["]"] = "%]";
        ["*"] = "%*";
        ["+"] = "%+";
        ["-"] = "%-";
        ["?"] = "%?";
    }

    sanitize = function(s)
        return (gsub(s, ".", matches))
    end
end

local MOST_RECENT_SET_NAME = mw.getCurrentFrame():expandTemplate({title = "CURRENTSET" })

local function GetLastUpdate()
    local lastEffectiveDate = match(CR.text, COMP_RULES_DATE_PATTERN)
    -- Italicize and wikilink the set name
    local setName = format("''[[%s]]''", MOST_RECENT_SET_NAME)
    local lastUpdate = lastEffectiveDate .. "—" .. setName
    return lastUpdate
end

local function SplitLine(ruleLine)
    -- Finds the index and the rest. If the index has an extra period, it is considered a formatting issue in the CR and is therefore ignored.
    local i, j, index, rest = find(ruleLine, "^(%d+%.%d*[%.a-kmnp-z]?)%.?%s(.+)")
    return i, j, index, rest
end

--[[
    Chops up and validates an index.
    Individually breaking down index parts here for ease of comprehension and future maintenance.
    Yes, a clever soul could streamline this with more complex pattern matching.
    Clever and absolute performance are not the goals. Editability by anonymous maintainers is.

    heading, major, and minor should all be number or nil
    returns false instead if the string provided is not an index
--]]
local function ParseIndex(index)
    local heading, major, minor, subrule

    if match(index, "%s") then
        return false
    end

    local i, _, suffix = find(index, "%.(.+)")
    if suffix then
        -- If the last character in the suffix is alphanumeric (and not L or O), set the subrule
        subrule = match(sub(suffix, -1), "(["..INDEX_SUBRULE_CHARACTER_SET.."])")

        -- If that was successful, cut that character off the suffix.
        if subrule then
            suffix = sub(suffix, 1, -2)
        end

        -- Now we can easily check whether the part between the period and the letter is a number.
        -- If so, that's our minor index
        minor = tonumber(suffix)
        assert(type(minor) == "number", "Invalid index!")

        -- Now cut off the entire suffix and let's parse the rest
        index = sub(index, 1, i-1)
    end

    -- Getting the heading and major index is just some number manipulation
    index = tonumber(index)
    -- assert(type(index) == "number", "Invalid index!")
    if not index then
        return false
    end

    if index >= 100 then
        major = index%100
        heading = math.floor(index/100)
    else
        -- The body of the rules starts at 100, lower values are headings
        heading = index
    end

    return heading, major, minor, subrule
end

local function IsSubsequentRule(line, heading, major, minor, subrule)
    local _, _, index = SplitLine(line)
    if not index then
        return false
    end

    local h, a, i, _ = ParseIndex(index)

    if subrule then
        -- We can cheat like hell if we're dealing with subrules because there's no further nesting
        -- Therefore, the next line is the next rule by definition
        -- That said, if this ever changes, this snippet might be useful
        --[[
            -- Yes, this makes assumptions about character encoding and subrules never numbering more than 24 for a given rule
            local i = find(INDEX_SUBRULE_CHARACTER_SET, subrule) + 1
            nextIndex = sub(INDEX_SUBRULE_CHARACTER_SET, i, i)
        --]]
        return true
    elseif i and minor and i > minor then
        return true
    elseif a and major and a > major then
        return true
    elseif h and heading and h > heading then
        return true
    end
end

local function GetNestingDepth(index)
    local depth

    if match(index, "["..INDEX_SUBRULE_CHARACTER_SET.."]") then
        -- Subrules, e.g. 103.7a
        depth = 4
    elseif match(index, "%d+%.%d+%.") then
        -- Rules, e.g. 112.1.
        depth = 3
    elseif match(index, "^%d%d%d%.") or match(index, "^%d+%.%d+") then
        -- Titles, e.g. 102. or 1.2 for normal numbering conventions
        depth = 2
    else
        -- Headings, e.g. 1.
        depth = 1
    end

    return depth
end

local function Titleize(title)
    local link, t, bare
    if match(title,"%(Obsolete%)$") then
        bare = sub(title,0,-12)
        link = lower(bare)
        link = gsub(link, "^(%S)", upper)
        t = format(WIKILINK_ALT_FORMAT, link, bare)
        t = t..OBSOLETE_PATTERN
    else
        link = lower(title)
        -- convert the first letter back to uppercase
        link = gsub(link, "^(%S)", upper)
        t = format(WIKILINK_ALT_FORMAT, link, title)
    end
    return t
end

local function StylizeRule(ruleLine)
    local _, _, index, rest = SplitLine(ruleLine)
    if not index then
        if find(ruleLine, "Example:") then
            ruleLine = "<p class=\"crExample\">''" .. gsub(ruleLine, "(Example:)", "'''%1'''") .. "''</p>"
        end
        return ruleLine
    end

    local h, a, i, _ = ParseIndex(index)
    -- Major indices and any rule shorter than five words should be a title, so try linking it!
    -- (this is probably a stupid assumption let's see how long before we get burned)
    if (h and a and not i) then
        if find(rest, GENERAL_RULE_PATTERN) then
            -- Each heading in the CR has a "General" section, expand that to "General_(<name of header>)"
            local headingName
            for line in gmatch(CR.text, LINE_PATTERN) do
                headingName = match(line, "^"..h.."%. (.+)")
                if headingName then
                    break
                end
            end
            assert(headingName)
            headingName = lower(headingName)
            headingName = gsub(headingName, "^(%S)", upper)
            local expandedLink = format(GENERAL_RULE_EXPANDED_TEXT_FORMAT, headingName)
            if match(rest,"%(Obsolete%)$") then
                local bare = sub(rest,0,-12)
                rest = format(WIKILINK_ALT_FORMAT, expandedLink, bare)..OBSOLETE_PATTERN
            else
                rest = format(WIKILINK_ALT_FORMAT, expandedLink, rest)
            end
        elseif match(index, "90[1-4]") then
            -- The casual format articles usually have a "<Name of format>_(format)" nameing pattern

            local expandedLink = format(CASUAL_FORMAT_EXPANDED_TEXT_FORMAT, rest)
            rest = format(WIKILINK_ALT_FORMAT, expandedLink, rest)
        else
            rest = Titleize(rest)
        end
    else
        local _, numWords = gsub(rest, "%S+", "")
        if numWords < 5 then
            rest = Titleize(rest)
        end
    end

    return format("'''%s''' %s", index, rest)
end

-- Creates a mw.html object matching the styling of the old CR template
local function CreateRulesDiv(output,source,update,color)
    local div = mw.html.create("div")
    div:addClass("crDiv")
    div:addClass("crBox")
    div:css("background-color", color)
    local ns = mw.title.getCurrentTitle().namespace
    if color == OBSOLETE_COLOR and ns == 0 then
        div:wikitext(OBSOLETE_CAT)
    end
    div:wikitext(format(LAST_UPDATED_FORMAT, source, update))
    div:newline()
    if type(output) == "string" then
        local line = StylizeRule(output)
        div:wikitext("* ", line)
    else
        local indentLevel
        local prevMax = 0
        local outputLine, maxIndent, index, _
        for _, line in ipairs(output) do
            outputLine = StylizeRule(line)
            _, _, index = SplitLine(line)
            if index then
                div:newline()
                maxIndent = GetNestingDepth(index)
                if not indentLevel then
                    indentLevel = 1
                else
                    indentLevel = indentLevel + (maxIndent - prevMax)
                end
                prevMax = maxIndent
                outputLine = rep("*",  indentLevel) .. outputLine
            else
                if not find(outputLine, "Example:") then
                    outputLine = "<br/>" .. outputLine
                else
                    outputLine = outputLine
                end
            end

            div:wikitext(outputLine)
        end
    end

    return div
end

-- Creates a mw.html object with a slightly different background color
local function CreateGlossaryDiv(output,source,update)
    local div = mw.html.create("div")
    div:addClass("crBox")

    local ns = mw.title.getCurrentTitle().namespace
    div:wikitext(format(LAST_UPDATED_FORMAT, source, update))
    div:newline()

    for i, line in ipairs(output) do
        -- Bold the name of the entry for clarity
        if i == 1 then
            if match(line,"%(Obsolete%)$") then
                -- Link and categorize obsolete terms
                div:css("background-color", OBSOLETE_COLOR)
                div:wikitext(";" .. gsub(line,"%(Obsolete%)","%([[Obsolete]]%)") .. "\n")
                if ns == 0 then
                    div:wikitext(OBSOLETE_CAT)
                end
            else
                div:css("background-color", GLOSSARY_COLOR)
                div:wikitext(";" .. line .. "\n")
            end
        else
            div:wikitext(":" .. line .. "\n")
        end
    end
    -- Add glossary category in mainspace articles
    if ns == 0 then
        div:wikitext(GLOSSARY_CAT)
    end

    return div
end

function CR.TOC(frame)
    local index = frame.args[1]
    local heading, major, minor, subrule
    -- If there's no index, we want the full TOC. Otherwise, pass it in for validation.
    local fullTOC
    if (not index) or type(index)=="string" and index=="" then
        heading = 1
        fullTOC = true
    else
        heading, major, minor, subrule = ParseIndex(index)
        assert(heading and heading>=1 and heading<=10 and not major and not minor and not subrule, "Invalid table of contents index!")
    end

    local output = {}

    local collecting = false
    for line in gmatch(CR.text, LINE_PATTERN) do
        if match(line, RULE_LINE_PATTERN) then
            if match(line, "^"..heading.."%.") then
                collecting = true
            elseif (not fullTOC) and IsSubsequentRule(line, heading, major, minor, subrule) then
                break
            end

            -- NOT elseif. We want to start collecting lines on the same line we match the target heading
            if collecting then
                tinsert(output, line)
            end
        elseif match(line, TOC_END_LINE_PATTERN) then
            break
        end
    end

    assert(#output > 0, "Index not found! ", index)
    return tostring(CreateRulesDiv(output,CR_SOURCE,GetLastUpdate(),RULES_COLOR))
end

-- Basically CR.full but with the full text of an index
function CR.title(frame)
    local title = frame.args[1]
    title = sanitize(title)

    local output = {}
    local passedTOC = false
    local collecting = false
    local heading, major, minor, subrule -- this is a stupid hack to continue using the original index-based stuff
    for line in gmatch(CR.text, LINE_PATTERN) do
        if (not passedTOC) and match(line, TOC_END_LINE_PATTERN) then
            passedTOC = true
        elseif match(line, BODY_END_LINE_PATTERN) then
            break
        elseif passedTOC then
            if match(line, title.."$") then
                collecting = true
                -- Stupid hack see above
                local _, _, i = SplitLine(line)
                heading, major, minor, subrule = ParseIndex(i)
            elseif collecting and IsSubsequentRule(line, heading, major, minor, subrule) then
                break
            end

            -- NOT elseif. We want to start collecting lines on the same line we match the target heading
            -- ignore whitespace
            if collecting and match(line, "%S") then
                tinsert(output, line)
            end
        end
    end

    assert(#output > 0, "Index not found! " .. title)
    if output then
        return tostring(CreateRulesDiv(output,CR_SOURCE,GetLastUpdate(),RULES_COLOR))
    else
        return nil
    end
end

function CR.only(frame)
    local index, additionalLevels = frame.args[1], tonumber(frame.args[2])
    local heading, major, minor, subrule = ParseIndex(index)

    local output = {}
    local passedTOC = false
    local collecting = false
    local ruleDepth, lineDepth = GetNestingDepth(index)
    for line in gmatch(CR.text, LINE_PATTERN) do
        if (not passedTOC) and match(line, TOC_END_LINE_PATTERN) then
            passedTOC = true
        elseif match(line, BODY_END_LINE_PATTERN) then
            break
        elseif passedTOC then
            if match(line, RULE_LINE_PATTERN) then
                if match(line, "^"..index) then
                    collecting = true
                elseif collecting and IsSubsequentRule(line, heading, major, minor, subrule) then
                    break
                end
            end

            -- NOT elseif. We want to start collecting lines on the same line we match the target heading
            -- ignore whitespace
            if collecting and match(line, "%S") then
                if additionalLevels then
                    local _, _, additionalIndex = SplitLine(line)
                    -- This looks a little weird.
                    -- We only update lineDepth in the case that we're looking at a rules index
                    -- But we capture any line for which it or the preceding index is within our targeting scope
                    -- (examples, mostly)
                    if additionalIndex then
                        lineDepth = GetNestingDepth(additionalIndex)
                    end
                    if lineDepth <= ruleDepth + additionalLevels then
                        tinsert(output, line)
                    end
                else
                    tinsert(output, line)
                    break
                end
            end
        end
    end

    assert(#output > 0, "Index not found! " .. index)
    return tostring(CreateRulesDiv(output,CR_SOURCE,GetLastUpdate(),RULES_COLOR))
end

function CR.full(frame)
    local index = frame.args[1]
    local heading, major, minor, subrule = ParseIndex(index)

    local output = {}
    local passedTOC = false
    local collecting = false
    for line in gmatch(CR.text, LINE_PATTERN) do
        if (not passedTOC) and match(line, TOC_END_LINE_PATTERN) then
            passedTOC = true
        elseif match(line, BODY_END_LINE_PATTERN) then
            break
        elseif passedTOC then
            if match(line, RULE_LINE_PATTERN) then
                if match(line, "^"..index) then
                    collecting = true
                elseif collecting and IsSubsequentRule(line, heading, major, minor, subrule) then
                    break
                end
            end

            -- NOT elseif. We want to start collecting lines on the same line we match the target heading
            -- ignore whitespace
            if collecting and match(line, "%S") then
                tinsert(output, line)
            end
        end
    end

    assert(#output > 0, "Index not found! " .. index)
    return tostring(CreateRulesDiv(output,CR_SOURCE,GetLastUpdate(),RULES_COLOR))
end

function CR.glossary(frame)
    local lookup = frame.args[1]
    lookup = sanitize(lookup)

    local output = {}
    local passedTOC, passedBody = false, false
    local collecting = false
    for line in gmatch(CR.text, LINE_PATTERN) do
        if (not passedTOC) and match(line, TOC_END_LINE_PATTERN) then
            passedTOC = true
        elseif (not passedBody) and match(line, BODY_END_LINE_PATTERN) then
            passedBody = true
        elseif passedTOC and passedBody and match(line, GLOSSARY_END_LINE_PATTERN) then
            break
        elseif passedBody then
            if match(line, "^"..lookup.."$") or match(line, "^"..lookup.." %(Obsolete%)$") then
                collecting = true
            elseif collecting and not match(line, "%w") then
                break
            end

            -- NOT elseif. We want to start collecting lines on the same line we match the target heading
            -- ignore whitespace
            if collecting then
                tinsert(output, line)
            end
        end
    end

    assert(#output > 0, "Glossary entry not found! " .. lookup)
    return tostring(CreateGlossaryDiv(output,GLOSSARY_SOURCE,GetLastUpdate()))
end

function CR.CRTemplateCall(frame)
    local toc, exact, lookup, glossary

    for key, value in pairs(frame.args) do
        if ((key == "toc") and value ~= "") or (value == "toc") then
            toc = true
        elseif ((key == "exact") and value ~= "") or (value == "exact") then
            exact = true
        elseif ((key == "glossary") and value ~= "") or (value == "glossary") then
            glossary = true
        elseif value and value ~= "" then
            assert(not lookup, "Unknown error, multiple lookups ")
            lookup = value
        end
    end
    assert(lookup or toc, "No lookup provided")

    if toc then
        if not lookup then
            return CR.TOC({args={}})
        elseif tonumber(lookup) < 10 then
            return CR.TOC({args={lookup}})
        else
            return CR.only({args={lookup, 1}})
        end
    elseif exact then
        return CR.only({args={lookup}})
    elseif glossary then
        return CR.glossary({args={lookup}})
    else
        if ParseIndex(lookup) then
            return CR.full({args={lookup}})
        else
            local output = CR.title({args={lookup}})
            if output then
                return output
            else
                return CR.glossary({args={lookup}})
            end
        end
    end
end

function CR.FormatRaw(frame)
    -- can't just unpack these because Scribunto screws with the metatable
    local source, update, rawRulesText, rulesFormat = frame.args[1], frame.args[2], frame.args[3], frame.args[4]

    -- split the lines so that it'll format correctly
    local lines = {}
    for line in gmatch(rawRulesText, LINE_PATTERN) do
        tinsert(lines, line)
    end
    if rulesFormat == "glossary" then
        return tostring(CreateGlossaryDiv(lines,source,update))
    elseif find(rawRulesText,"(Obsolete)") then
        return tostring(CreateRulesDiv(lines,source,update,OBSOLETE_COLOR))
    else
        return tostring(CreateRulesDiv(lines,source,update,RULES_COLOR))
    end
end

return CR