Module:SMWUtils

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

-- Module:SMWUtils
-- Utility functions for Semantic MediaWiki queries
-- Uses frame:preprocess() for #ask queries (supports property chains)

local p = {}

local function trim(s)
    if s == nil then return nil end
    s = tostring(s)
    s = s:gsub('^%s+', ''):gsub('%s+$', '')
    if s == '' then return nil end
    return s
end

-- Parse an HTML table from #ask broadtable format into rows of cells
-- Filters out the "further results" row
local function parseHtmlTable(html)
    local rows = {}
    if not html then return rows end
    
    for rowContent in html:gmatch('<tr[^>]*>(.-)</tr>') do
        -- Skip "further results" rows
        if not rowContent:match('smw%-broadtable%-furtherresults') then
            local cells = {}
            for cellContent in rowContent:gmatch('<td[^>]*>(.-)</td>') do
                -- Keep HTML intact for links
                table.insert(cells, trim(cellContent) or '')
            end
            if #cells > 0 then
                table.insert(rows, cells)
            end
        end
    end
    
    return rows
end

-- Extract plain text from HTML (for numeric comparison)
local function stripHtml(s)
    if not s then return nil end
    return trim(s:gsub('<[^>]+>', ''):gsub('&[^;]+;', ''))
end

-- Deduplicate rows by printout columns, keeping the first occurrence
-- Uses stripped text of printout columns (2..n+1) as the dedup key,
-- since subobject queries produce unique subjects (e.g. Page#sub1, Page#sub2)
-- but identical printout values for the same parent page
local function deduplicateRows(rows, numPrintouts)
    local seen = {}
    local deduped = {}
    for _, row in ipairs(rows) do
        local keyParts = {}
        for j = 2, numPrintouts + 1 do
            table.insert(keyParts, stripHtml(row[j]) or '')
        end
        local key = table.concat(keyParts, '|')
        if not seen[key] then
            seen[key] = true
            table.insert(deduped, row)
        end
    end
    return deduped
end

--- Debug function to see raw query output and parsing
function p.debug(frame)
    local args = frame.args
    local conditions = trim(args.conditions) or ''
    local minProperty = trim(args.minProperty) or ''
    local printouts = trim(args.printouts) or ''
    
    -- Build printout list
    local printoutProps = {}
    local printoutQuery = {}
    for prop in printouts:gmatch('[^,]+') do
        local propTrimmed = trim(prop)
        table.insert(printoutProps, propTrimmed)
        table.insert(printoutQuery, '|?' .. propTrimmed)
    end
    -- Add minProperty to query
    if minProperty ~= '' then
        table.insert(printoutQuery, '|?' .. minProperty .. '=#')
    end
    
    local queryWikitext = '{{#ask:' .. conditions .. '\n' ..
        table.concat(printoutQuery, '\n') .. '\n' ..
        '|sort=' .. minProperty .. '\n' ..
        '|order=asc\n' ..
        '|limit=5\n' ..
        '|format=broadtable\n' ..
        '|headers=hide\n' ..
        '|link=all\n' ..
        '}}'
    
    local result = frame:preprocess(queryWikitext)
    local rows = parseHtmlTable(result)
    
    local output = {}
    table.insert(output, "'''Query wikitext:'''\n<pre>" .. mw.text.nowiki(queryWikitext) .. "</pre>\n\n")
    table.insert(output, "'''Raw HTML result:'''\n<pre>" .. mw.text.nowiki(result):sub(1, 2000) .. "</pre>\n\n")
    table.insert(output, "'''Parsed rows (" .. #rows .. "):'''\n\n")
    
    for i, row in ipairs(rows) do
        table.insert(output, "'''Row " .. i .. " (" .. #row .. " cells):'''\n")
        for j, cell in ipairs(row) do
            local stripped = stripHtml(cell) or "(nil)"
            table.insert(output, "* Cell " .. j .. ": <code>" .. mw.text.nowiki(cell):sub(1, 100) .. "</code> → stripped: <code>" .. stripped .. "</code>\n")
        end
        table.insert(output, "\n")
    end
    
    -- Show what we'd calculate as min (corrected: +2 for subject column)
    if #rows > 0 then
        local minValueCol = #printoutProps + 2  -- +1 for subject column, +1 for minProperty column
        table.insert(output, "'''Printout count:''' " .. #printoutProps .. "\n")
        table.insert(output, "'''Min value column index:''' " .. minValueCol .. " (printouts + 2 for subject col)\n")
        if rows[1][minValueCol] then
            local minText = stripHtml(rows[1][minValueCol])
            local minNum = tonumber(minText)
            table.insert(output, "'''First row min column value:''' " .. (minText or "nil") .. " → tonumber: " .. tostring(minNum) .. "\n")
        else
            table.insert(output, "'''First row min column:''' (column doesn't exist)\n")
        end
    end
    
    return table.concat(output)
end

--- Find all results where a property equals its minimum value
-- @param frame with args:
--   conditions: SMW conditions (e.g., "[[Has video signal format::Component (YPbPr)]]")
--   minProperty: Property to minimize (e.g., "Has CRT model.Has screen size inches")
--   printouts: Comma-separated properties to display
--   headers: Comma-separated header names
--   mainlabel: Optional header for the subject/page name column
--   format: table, ul, ol (default: table)
--   intro: Custom intro text
function p.findAllWithMin(frame)
    local args = frame.args
    local conditions = trim(args.conditions) or ''
    local minProperty = trim(args.minProperty)
    local printouts = trim(args.printouts) or ''
    local format = trim(args.format) or 'table'
    local headers = trim(args.headers) or ''
    local intro = trim(args.intro)
    local mainlabel = trim(args.mainlabel)
    
    if not minProperty then
        return '<span class="error">Error: minProperty is required</span>'
    end
    
    -- Build printout list
    local printoutProps = {}
    local printoutQuery = {}
    for prop in printouts:gmatch('[^,]+') do
        local propTrimmed = trim(prop)
        table.insert(printoutProps, propTrimmed)
        table.insert(printoutQuery, '|?' .. propTrimmed)
    end
    -- Add minProperty to query
    table.insert(printoutQuery, '|?' .. minProperty .. '=#')
    
    -- Query all results sorted by minProperty ascending
    local queryWikitext = '{{#ask:' .. conditions .. '\n' ..
        table.concat(printoutQuery, '\n') .. '\n' ..
        '|sort=' .. minProperty .. '\n' ..
        '|order=asc\n' ..
        '|limit=500\n' ..
        '|format=broadtable\n' ..
        '|headers=hide\n' ..
        '|link=all\n' ..
        '}}'
    
    local result = frame:preprocess(queryWikitext)
    local rows = parseHtmlTable(result)
    
    if #rows == 0 then
        return "No results found."
    end
    
    -- broadtable format: Column 1 = subject, Columns 2..n+1 = printouts, Column n+2 = minProperty
    -- So minValueCol = #printoutProps + 2
    local minValueCol = #printoutProps + 2
    local minValueText = stripHtml(rows[1][minValueCol])
    local minValue = tonumber(minValueText)
    
    if not minValue then
        return "Could not determine minimum value. Column " .. minValueCol .. ", raw: " .. tostring(rows[1][minValueCol]) .. ", stripped: " .. tostring(minValueText)
    end
    
    -- Filter to only rows matching the minimum
    local filtered = {}
    for _, row in ipairs(rows) do
        local val = tonumber(stripHtml(row[minValueCol]))
        if val == minValue then
            table.insert(filtered, row)
        end
    end
    
    -- Deduplicate (multiple subobjects on same page produce duplicate rows)
    filtered = deduplicateRows(filtered, #printoutProps)
    
    -- Build headers list
    local headerList = {}
    for h in headers:gmatch('[^,]+') do
        table.insert(headerList, trim(h))
    end
    
    local introText = intro or ("Smallest screen size with this input: '''" .. minValue .. "\"'''")
    
    if format == 'ul' then
        return p._formatList(filtered, printoutProps, headerList, 'ul', introText, mainlabel)
    elseif format == 'ol' then
        return p._formatList(filtered, printoutProps, headerList, 'ol', introText, mainlabel)
    else
        return p._formatTable(filtered, printoutProps, headerList, introText, mainlabel)
    end
end

--- Find all results where a property equals its maximum value
function p.findAllWithMax(frame)
    local args = frame.args
    local conditions = trim(args.conditions) or ''
    local maxProperty = trim(args.maxProperty)
    local printouts = trim(args.printouts) or ''
    local format = trim(args.format) or 'table'
    local headers = trim(args.headers) or ''
    local intro = trim(args.intro)
    local mainlabel = trim(args.mainlabel)
    
    if not maxProperty then
        return '<span class="error">Error: maxProperty is required</span>'
    end
    
    -- Build printout list
    local printoutProps = {}
    local printoutQuery = {}
    for prop in printouts:gmatch('[^,]+') do
        local propTrimmed = trim(prop)
        table.insert(printoutProps, propTrimmed)
        table.insert(printoutQuery, '|?' .. propTrimmed)
    end
    table.insert(printoutQuery, '|?' .. maxProperty .. '=#')
    
    -- Query all results sorted by maxProperty descending
    local queryWikitext = '{{#ask:' .. conditions .. '\n' ..
        table.concat(printoutQuery, '\n') .. '\n' ..
        '|sort=' .. maxProperty .. '\n' ..
        '|order=desc\n' ..
        '|limit=500\n' ..
        '|format=broadtable\n' ..
        '|headers=hide\n' ..
        '|link=all\n' ..
        '}}'
    
    local result = frame:preprocess(queryWikitext)
    local rows = parseHtmlTable(result)
    
    if #rows == 0 then
        return "No results found."
    end
    
    -- broadtable format: Column 1 = subject, Columns 2..n+1 = printouts, Column n+2 = maxProperty
    local maxValueCol = #printoutProps + 2
    local maxValueText = stripHtml(rows[1][maxValueCol])
    local maxValue = tonumber(maxValueText)
    
    if not maxValue then
        return "Could not determine maximum value. Column " .. maxValueCol .. ", raw: " .. tostring(rows[1][maxValueCol]) .. ", stripped: " .. tostring(maxValueText)
    end
    
    -- Filter to only rows matching the maximum
    local filtered = {}
    for _, row in ipairs(rows) do
        local val = tonumber(stripHtml(row[maxValueCol]))
        if val == maxValue then
            table.insert(filtered, row)
        end
    end
    
    -- Deduplicate (multiple subobjects on same page produce duplicate rows)
    filtered = deduplicateRows(filtered, #printoutProps)
    
    -- Build headers list
    local headerList = {}
    for h in headers:gmatch('[^,]+') do
        table.insert(headerList, trim(h))
    end
    
    local introText = intro or ("Largest screen size with this input: '''" .. maxValue .. "\"'''")
    
    if format == 'ul' then
        return p._formatList(filtered, printoutProps, headerList, 'ul', introText, mainlabel)
    elseif format == 'ol' then
        return p._formatList(filtered, printoutProps, headerList, 'ol', introText, mainlabel)
    else
        return p._formatTable(filtered, printoutProps, headerList, introText, mainlabel)
    end
end

-- Format results as a wikitable
-- Note: row[1] is subject, row[2..n+1] are printouts, row[n+2] is min/max value
function p._formatTable(rows, printouts, headers, introText, mainlabel)
    local output = {}
    
    table.insert(output, introText .. "\n\n")
    table.insert(output, '{| class="wikitable sortable"\n')
    
    -- Header row
    table.insert(output, '|-\n! #')
    if mainlabel then
        table.insert(output, ' !! ' .. mainlabel)
    end
    for i, prop in ipairs(printouts) do
        local headerText = headers[i] or prop:gsub('.*%.', ''):gsub('Has ', '')
        table.insert(output, ' !! ' .. headerText)
    end
    table.insert(output, '\n')
    
    -- Data rows: printout columns start at index 2 (after subject column)
    for i, row in ipairs(rows) do
        local tr = '|-\n| ' .. tostring(i)
        if mainlabel then
            tr = tr .. ' || ' .. (row[1] or '')
        end
        for j = 1, #printouts do
            -- Printout j is in column j+1 (offset by subject column)
            tr = tr .. ' || ' .. (row[j + 1] or '')
        end
        table.insert(output, tr .. '\n')
    end
    
    table.insert(output, '|}')
    
    return table.concat(output)
end

-- Format results as a list
function p._formatList(rows, printouts, headers, listType, introText, mainlabel)
    local output = {}
    local bullet = listType == 'ol' and '#' or '*'
    
    table.insert(output, introText .. "\n\n")
    
    for _, row in ipairs(rows) do
        local parts = {}
        if mainlabel then
            local subject = row[1]
            if subject and subject ~= '' then
                table.insert(parts, subject)
            end
        end
        for j = 1, #printouts do
            -- Printout j is in column j+1 (offset by subject column)
            local cell = row[j + 1]
            if cell and cell ~= '' then
                table.insert(parts, cell)
            end
        end
        table.insert(output, bullet .. ' ' .. table.concat(parts, ' – ') .. '\n')
    end
    
    return table.concat(output)
end

return p