模組:Graph/DataProcess/Source
外观
--Module:Graph/DataProcess/Source
--處理與資料來源相關的工作 (及其依賴函數和提供[[Module:Graph/DataProcess]]工具支援)
local p = {}
local WHITELIST_PAGE = "Module:Graph/APIWhiteList.json"
local cachedApiWhitelist = nil
local function parseBool(val, defaultVal)
if val == nil or val == "" then return defaultVal end
if val == false or val == 0 or val == "0" then return false end
local s = string.lower(mw.text.trim(tostring(val)))
if s == "false" or s == "n" or s == "no" or s == "0" then return false end
return true
end
-- 遮罩保護 MediaWiki 繁簡轉換與標籤之 Strip Marker,避免內部雙引號破壞 JSON 解析或被 encode 損毀
local function maskStripMarkers(s)
if type(s) ~= "string" or not string.find(s, "\127", 1, true) then return s end
s = string.gsub(s, "\127'\"`UNIQ", "__MW_SM_PRE__UNIQ")
s = string.gsub(s, "QINU`\"'\127", "QINU__MW_SM_SUF__")
return s
end
local function numericArray(csv)
if not csv or csv == "" then return nil end
local list = mw.text.split(tostring(csv), "%s*,%s*")
local result = { n = #list }
local isInteger = true
local hasNumber = false
for i = 1, #list do
local item = mw.text.trim(list[i])
if item == "" then
result[i] = false
else
local num = tonumber(item)
if not num then return nil end
hasNumber = true
result[i] = num
if isInteger then
local _, frac = math.modf(num)
isInteger = (frac == 0.0)
end
end
end
if not hasNumber then return nil end
return result, isInteger
end
local function stringArray(text)
if not text or text == "" then return nil end
local escaped = mw.ustring.gsub(tostring(text), "\\,", "<COMMA>")
local list = mw.text.split(escaped, ",", true)
for i = 1, #list do
list[i] = mw.ustring.gsub(mw.text.trim(list[i]), "<COMMA>", ",")
end
list.n = #list
return list
end
-- 解析符號與形狀陣列:先檢查標誌性特徵,一般字串直接走 stringArray;含 <svg>、[[...]] 或引號時以 gsub 遮罩切分,零無窮迴圈風險
local function parseShapeArray(text)
if type(text) == "table" then return text end
if not text or text == "" then return nil end
local raw = mw.text.unstripNoWiki(tostring(text))
if string.find(raw, "<[sS][vV][gG]") then
raw = mw.text.decode(raw)
end
-- 快速通道:若不含 <、[、"、' 等特殊標誌字元,直接走原生 stringArray
if not string.find(raw, "[<%[\"'\\]") then
return stringArray(raw)
end
local tokens = {}
local function saveToken(val)
tokens[#tokens + 1] = val
return "__ GR_TOK_" .. #tokens .. " __"
end
-- 1. 保護完整 <svg>...</svg> 區塊
local masked = string.gsub(raw, "<[sS][vV][gG][^>]*>.-</[sS][vV][gG]>", saveToken)
-- 2. 保護 [[File:...|...]] 雙中括號區塊
masked = string.gsub(masked, "%[%[.-%]%]", saveToken)
-- 3. 保護雙引號 "..." 與單引號 '...' 區塊 (引號保護法,支援檔名含逗號)
masked = string.gsub(masked, '"([^"]*)"', saveToken)
masked = string.gsub(masked, "'([^']*)'", saveToken)
local list = stringArray(masked)
if not list then return nil end
for i = 1, #list do
local item = list[i]
if string.find(item, "__ GR_TOK_", 1, true) then
item = string.gsub(item, "__ GR_TOK_(%d+) __", function(idxStr)
return tokens[tonumber(idxStr)] or ""
end)
list[i] = mw.text.trim(item)
end
end
return list
end
local function isDateArray(arr)
if not arr or #arr == 0 then return false end
local validCount = 0
for i = 1, #arr do
local s = arr[i]
if s and s ~= "" then
if string.match(s, "^%d%d%d%d[%-%/]%d%d?") then
validCount = validCount + 1
else
return false
end
end
end
return validCount > 0
end
local function deserializeXData(serializedX, xType, xMin, xMax)
if not serializedX or mw.text.trim(tostring(serializedX)) == "" then
return { n = 0 }, xType or "number", tonumber(xMin), tonumber(xMax)
end
local x
if not xType or xType == "integer" or xType == "number" then
local isInteger
x, isInteger = numericArray(serializedX)
if x then
xMin = tonumber(xMin)
xMax = tonumber(xMax)
if not xType then
xType = isInteger and "integer" or "number"
end
elseif xType == "integer" or xType == "number" then
x = stringArray(serializedX)
xType = "string"
end
end
if not x then
x = stringArray(serializedX)
if not xType then
xType = isDateArray(x) and "date" or "string"
end
end
return x, xType, xMin, xMax
end
local function deserializeYData(serializedYs, yType, yMin, yMax)
local y = {}
local areAllInteger = true
local maxIndex = 0
for k, _ in pairs(serializedYs) do
if k > maxIndex then maxIndex = k end
end
for yNum = 1, maxIndex do
local value = serializedYs[yNum]
if value ~= nil then
local yValues
if not yType or yType == "integer" or yType == "number" then
local isInteger
yValues, isInteger = numericArray(value)
if yValues then
areAllInteger = areAllInteger and isInteger
else
return deserializeYData(serializedYs, "string", yMin, yMax)
end
end
if not yValues then yValues = stringArray(value) end
y[#y + 1] = yValues
end
end
if not yType then
yType = areAllInteger and "integer" or "number"
end
if yType == "integer" or yType == "number" then
yMin = tonumber(yMin)
yMax = tonumber(yMax)
end
return y, yType, yMin, yMax
end
-- 將 mw.loadJsonData 回傳的唯讀代理表轉為標準 Lua Table (避免 Lua 5.1 中 #t 為 0)
local function toPlainTable(obj)
if type(obj) ~= "table" then return obj end
local res = {}
for k, v in pairs(obj) do
res[k] = (type(v) == "table") and toPlainTable(v) or v
end
return res
end
-- 讀取並快取 Module:Graph/APIWhiteList.json (支援 json 內容模型快取與一般頁面文字解析雙通道)
local function getApiWhitelist()
if cachedApiWhitelist then return cachedApiWhitelist end
local data = nil
if mw.loadJsonData then
local ok, res = pcall(mw.loadJsonData, WHITELIST_PAGE)
if ok and type(res) == "table" then
data = res
end
end
if not data then
local titleObj = mw.title.new(WHITELIST_PAGE)
if titleObj then
local rawContent = titleObj:getContent()
if rawContent and rawContent ~= "" then
local okDec, dec = pcall(mw.text.jsonDecode, rawContent)
if okDec and type(dec) == "table" then
data = dec
end
end
end
end
local exactMap = {}
local suffixes = {}
local defaultHost = "wikimedia.org"
if type(data) == "table" then
if type(data.wikirestDefaultHost) == "string" and data.wikirestDefaultHost ~= "" then
defaultHost = string.lower(mw.text.trim(data.wikirestDefaultHost))
end
if type(data.exactHosts) == "table" then
local i = 1
while data.exactHosts[i] ~= nil do
local h = string.lower(mw.text.trim(tostring(data.exactHosts[i])))
if h ~= "" then exactMap[h] = true end
i = i + 1
end
end
if type(data.hostSuffixes) == "table" then
local i = 1
while data.hostSuffixes[i] ~= nil do
local s = string.lower(mw.text.trim(tostring(data.hostSuffixes[i])))
if s ~= "" then
if string.sub(s, 1, 1) ~= "." then s = "." .. s end
suffixes[#suffixes + 1] = s
end
i = i + 1
end
end
else
-- 預設安全保底白名單 (避免 JSON 頁面尚未建立前全站報錯)
exactMap = { ["wikimedia.org"] = true, ["wikidata.org"] = true, ["mediawiki.org"] = true, ["toolforge.org"] = true }
suffixes = { ".wikimedia.org", ".wikipedia.org", ".wikidata.org", ".mediawiki.org", ".toolforge.org", ".wmcloud.org" }
end
cachedApiWhitelist = {
exactMap = exactMap,
suffixes = suffixes,
defaultHost = defaultHost
}
return cachedApiWhitelist
end
-- 取得當前維基站點的主機名稱 (供 wikiraw:/// 與 wikiapi:/// 三斜線本地協議使用)
local function getLocalWikiHost()
if mw and mw.site and type(mw.site.server) == "string" then
local h = string.match(mw.site.server, "//([%w%-%.]+)")
if h and h ~= "" then return string.lower(h) end
end
return "zh.wikipedia.org"
end
-- 清理表頭文字以產生純文字鍵名 (移除 <ref>、HTML 標籤、提取 [[連結|顯示名]] 之顯示名)
local function cleanHeaderName(rawHeader)
if type(rawHeader) ~= "string" then return "" end
local s = rawHeader
s = string.gsub(s, "__MW_SM_PRE__UNIQ.-QINU__MW_SM_SUF__", "")
s = string.gsub(s, "\127'\"`UNIQ.-QINU`\"'\127", "")
s = string.gsub(s, "<!%-%-.-%-%->", "")
s = string.gsub(s, "<[rR][eE][fF][^>]*>.-</[rR][eE][fF]>", "")
s = string.gsub(s, "<[rR][eE][fF][^/>]*/%s*>", "")
s = string.gsub(s, "%[%[([^%[%]|]+)|([^%[%]]+)%]%]", "%2")
s = string.gsub(s, "%[%[([^%[%]]+)%]%]", "%1")
s = string.gsub(s, "<[^>]+>", "")
s = string.gsub(s, " ", " ")
s = string.gsub(s, " ", " ")
return mw.text.trim(s)
end
-- 驗證並正規化動態 API 網址 (嚴格依據 Module:Graph/APIWhiteList.json 進行 HTTPS 與網域白名單檢查)
local function normalizeAndValidateApiUrl(rawUrl)
if type(rawUrl) ~= "string" then return nil end
local u = mw.text.trim(rawUrl)
-- 阻擋控制字元、HTML 標籤字元、引號與路徑穿越 (..)
if u == "" or string.find(u, "[%c<>\"'\\]") or string.find(u, "%.%.", 1, true) then
return nil
end
local wl = getApiWhitelist()
local fullUrl = u
-- 1. 支援 wikiraw://、wikiapi:// 與 wikirest:// 協議
if string.match(fullUrl, "^wikiraw:///") then
local pageTitle = string.gsub(string.gsub(fullUrl, "^wikiraw:///", ""), " ", "_")
fullUrl = "https://" .. getLocalWikiHost() .. "/w/index.php?title=" .. pageTitle .. "&action=raw"
elseif string.match(fullUrl, "^wikiraw://") then
local hostPart, pageTitle = string.match(fullUrl, "^wikiraw://([%w%-%.]+)/(.+)$")
if not hostPart or not pageTitle then return nil end
pageTitle = string.gsub(pageTitle, " ", "_")
fullUrl = "https://" .. hostPart .. "/w/index.php?title=" .. pageTitle .. "&action=raw"
elseif string.match(fullUrl, "^wikiapi:///") then
local queryPart = string.gsub(string.gsub(fullUrl, "^wikiapi:///?", ""), " ", "%%20")
fullUrl = "https://" .. getLocalWikiHost() .. "/w/api.php?" .. queryPart
elseif string.match(fullUrl, "^wikirest:///") then
fullUrl = "https://" .. wl.defaultHost .. "/" .. string.gsub(string.gsub(fullUrl, "^wikirest:///", ""), " ", "%%20")
elseif string.match(fullUrl, "^wikirest://") then
fullUrl = "https://" .. string.gsub(string.gsub(fullUrl, "^wikirest://", ""), " ", "%%20")
elseif string.sub(fullUrl, 1, 2) == "//" then
fullUrl = "https:" .. string.gsub(fullUrl, " ", "%%20")
else
fullUrl = string.gsub(fullUrl, " ", "%%20")
end
-- 2. 嚴格禁止明文 http://,僅允許 https:// 且網域須位於白名單內
local host, path = string.match(fullUrl, "^https://([%w%-%.]+)(/.*)$")
if not host or not path then return nil end
host = string.lower(host)
local isAllowedHost = (wl.exactMap[host] == true)
if not isAllowedHost then
for i = 1, #wl.suffixes do
local suf = wl.suffixes[i]
local sLen = #suf
if #host > sLen and string.sub(host, -sLen) == suf then
isAllowedHost = true
break
end
end
end
if not isAllowedHost then return nil end
return fullUrl
end
-- 針對非維基白名單之第三方外部 API (如 NASA、NOAA):驗證為語法安全之外部 HTTPS 網址,供 gr-async-chart-unsupported 容器封裝
local function normalizeUnsupportedExternalUrl(rawUrl)
if type(rawUrl) ~= "string" then return nil end
local u = mw.text.trim(rawUrl)
if u == "" or string.find(u, "[%c<>\"'\\]") or string.find(u, "%.%.", 1, true) then
return nil
end
u = string.gsub(u, " ", "%%20")
if string.sub(u, 1, 2) == "//" then
u = "https:" .. u
elseif string.match(u, "^http://") then
u = "https://" .. string.sub(u, 8)
end
local host, path = string.match(u, "^https://([%w%-%.]+)(/.*)$")
if not host or not path then return nil end
host = string.lower(host)
-- 阻擋 localhost 與純 IPv4 內網位址,僅允許合法外部網域名稱
if host == "localhost" or string.match(host, "^%d+%.%d+%.%d+%.%d+$") or not string.find(host, "%.", 1, true) then
return nil
end
return u
end
-- 將模板參數跳脫模式之表格語法 ({{(!}}, {{!}}, {{!!}}, {{!-}}, {{!)}} 與 {{=}}) 正規化為標準 Wikitext 表格語法
local function normalizeTemplateTableSyntax(s)
if type(s) ~= "string" or s == "" then return "" end
s = string.gsub(s, "{{%s*%(%!%s*}}", "{|")
s = string.gsub(s, "{{%s*%!%)%s*}}", "|}")
s = string.gsub(s, "{{%s*%!%-%s*}}", "|-")
s = string.gsub(s, "{{%s*%!%!%s*}}", "||")
s = string.gsub(s, "{{%s*%!%s*}}", "|")
s = string.gsub(s, "{{%s*=%s*}}", "=")
return s
end
-- 清理資料儲存格內容,若為純數值 (含千分位逗號) 則自動轉為 number,否則保留原始字串 (如 [[User:xxx]])
local function parseWikitableCellValue(rawCell)
if type(rawCell) ~= "string" then return "" end
local s = rawCell
s = string.gsub(s, "<!%-%-.-%-%->", "")
s = string.gsub(s, "<[rR][eE][fF][^>]*>.-</[rR][eE][fF]>", "")
s = string.gsub(s, "<[rR][eE][fF][^/>]*/%s*>", "")
s = string.gsub(s, " ", " ")
s = string.gsub(s, " ", " ")
s = mw.text.trim(s)
if s == "" then return "" end
-- 嘗試解析為數值 (先移除 Strip Marker 與千分位逗號進行檢測)
local numCand = string.gsub(s, "__MW_SM_PRE__UNIQ.-QINU__MW_SM_SUF__", "")
numCand = string.gsub(numCand, "\127'\"`UNIQ.-QINU`\"'\127", "")
numCand = mw.text.trim(numCand)
local noComma = string.gsub(numCand, ",", "")
local numV = tonumber(noComma)
if numV ~= nil then
return numV
end
return s
end
-- 深度感知分離單一儲存格之「HTML 屬性」與「內容」 (例如 style="color:red" | [[連結|文字]])
local function splitCellAttrsAndContent(rawCell)
local i, len = 1, #rawCell
local linkDepth, tmplDepth, tagDepth = 0, 0, 0
local quoteChar = nil
while i <= len do
local two = string.sub(rawCell, i, i + 1)
local ch = string.sub(rawCell, i, i)
if quoteChar then
if ch == quoteChar then quoteChar = nil end
i = i + 1
else
-- 儲存格屬性不可跨行,若在遇到單一 | 前先遇到換行,代表此格無屬性前綴
if ch == "\n" then
return "", rawCell
elseif ch == '"' or ch == "'" then
quoteChar = ch
i = i + 1
elseif two == "[[" then
linkDepth = linkDepth + 1
i = i + 2
elseif two == "]]" and linkDepth > 0 then
linkDepth = linkDepth - 1
i = i + 2
elseif two == "{{" then
tmplDepth = tmplDepth + 1
i = i + 2
elseif two == "}}" and tmplDepth > 0 then
tmplDepth = tmplDepth - 1
i = i + 2
elseif ch == "<" then
tagDepth = tagDepth + 1
i = i + 1
elseif ch == ">" and tagDepth > 0 then
tagDepth = tagDepth - 1
i = i + 1
elseif ch == "|" and two ~= "||" and linkDepth == 0 and tmplDepth == 0 and tagDepth == 0 then
return string.sub(rawCell, 1, i - 1), string.sub(rawCell, i + 1)
else
i = i + 1
end
end
end
return "", rawCell
end
-- 深度感知切分同行儲存格:在 [[...]]、{{...}}、<...> 深度為 0 時,以 || (或表頭列的 !!) 切分
local function splitInlineWikitableCells(lineBody, isHeaderLine)
local cells = {}
local cur = {}
local i, len = 1, #lineBody
local linkDepth, tmplDepth, tagDepth = 0, 0, 0
while i <= len do
local two = string.sub(lineBody, i, i + 1)
local ch = string.sub(lineBody, i, i)
if two == "[[" then
linkDepth = linkDepth + 1
cur[#cur + 1] = "[["
i = i + 2
elseif two == "]]" and linkDepth > 0 then
linkDepth = linkDepth - 1
cur[#cur + 1] = "]]"
i = i + 2
elseif two == "{{" then
tmplDepth = tmplDepth + 1
cur[#cur + 1] = "{{"
i = i + 2
elseif two == "}}" and tmplDepth > 0 then
tmplDepth = tmplDepth - 1
cur[#cur + 1] = "}}"
i = i + 2
elseif ch == "<" then
tagDepth = tagDepth + 1
cur[#cur + 1] = ch
i = i + 1
elseif ch == ">" and tagDepth > 0 then
tagDepth = tagDepth - 1
cur[#cur + 1] = ch
i = i + 1
elseif linkDepth == 0 and tmplDepth == 0 and tagDepth == 0 and (two == "||" or (isHeaderLine and two == "!!")) then
cells[#cells + 1] = table.concat(cur)
cur = {}
i = i + 2
else
cur[#cur + 1] = ch
i = i + 1
end
end
cells[#cells + 1] = table.concat(cur)
return cells
end
-- 解析單一標準化後的 {| ... |} 表格區塊為二維展開網格與紀錄陣列
local function parseSingleWikitable(tableText)
local normText = normalizeTemplateTableSyntax(tableText)
normText = string.gsub(normText, "\r\n", "\n")
normText = string.gsub(normText, "\r", "\n")
local lines = mw.text.split(normText, "\n", true)
local rawRows = {}
local curRow = nil
local caption = nil
local inTable = false
local function ensureRow()
if not curRow then
curRow = { cells = {}, hasHeaderCell = false, hasDataCell = false }
rawRows[#rawRows + 1] = curRow
end
end
for lIdx = 1, #lines do
local line = mw.text.trim(lines[lIdx])
if not inTable then
if string.sub(line, 1, 2) == "{|" then
inTable = true
end
else
if string.sub(line, 1, 2) == "|}" then
break
elseif string.sub(line, 1, 2) == "|+" then
local capBody = string.sub(line, 3)
local _, capContent = splitCellAttrsAndContent(capBody)
caption = cleanHeaderName(capContent)
elseif string.sub(line, 1, 2) == "|-" then
if curRow and #curRow.cells > 0 then
curRow = nil
end
elseif string.sub(line, 1, 1) == "!" or string.sub(line, 1, 1) == "|" then
ensureRow()
local isHeaderLine = (string.sub(line, 1, 1) == "!")
local lineBody
-- 相容行首連續寫 !! 或 || 的情況
if string.sub(line, 1, 2) == "!!" or string.sub(line, 1, 2) == "||" then
lineBody = string.sub(line, 3)
else
lineBody = string.sub(line, 2)
end
if isHeaderLine then
curRow.hasHeaderCell = true
else
curRow.hasDataCell = true
end
local inlineList = splitInlineWikitableCells(lineBody, isHeaderLine)
for cIdx = 1, #inlineList do
local attrs, content = splitCellAttrsAndContent(inlineList[cIdx])
local rowspan = tonumber(string.match(attrs, '[rR][oO][wW][sS][pP][aA][nN]%s*=%s*["\']?(%d+)')) or 1
local colspan = tonumber(string.match(attrs, '[cC][oO][lL][sS][pP][aA][nN]%s*=%s*["\']?(%d+)')) or 1
local sortVal = string.match(attrs, 'data%-sort%-value%s*=%s*"([^"]*)"')
or string.match(attrs, "data%-sort%-value%s*=%s*'([^']*)'")
curRow.cells[#curRow.cells + 1] = {
raw = content,
isHeader = isHeaderLine,
rowspan = math.max(1, math.min(50, rowspan)),
colspan = math.max(1, math.min(50, colspan)),
sortVal = sortVal
}
end
elseif curRow and #curRow.cells > 0 and line ~= "" then
-- 跨行儲存格內容續接
local lastCell = curRow.cells[#curRow.cells]
lastCell.raw = lastCell.raw .. "\n" .. line
end
end
end
if #rawRows == 0 then return nil end
-- 將含有 rowspan / colspan 的列展開為對齊的二維矩陣 grid[r][c]
local grid = {}
local rowMeta = {}
local maxCols = 0
for r = 1, #rawRows do
grid[r] = grid[r] or {}
rowMeta[r] = { hasHeader = rawRows[r].hasHeaderCell, hasData = rawRows[r].hasDataCell }
local colPtr = 1
for _, cell in ipairs(rawRows[r].cells) do
while grid[r][colPtr] ~= nil do
colPtr = colPtr + 1
end
for dr = 0, cell.rowspan - 1 do
local tr = r + dr
grid[tr] = grid[tr] or {}
for dc = 0, cell.colspan - 1 do
local tc = colPtr + dc
grid[tr][tc] = cell
if tc > maxCols then maxCols = tc end
end
end
colPtr = colPtr + cell.colspan
end
end
if maxCols == 0 then return nil end
-- 判定表頭列數:前導含有 ! 的列視為表頭;若第 1 列無 ! 但全為非數值文字且第 2 列含數值,亦視為表頭
local headerRowCount = 0
for r = 1, #rawRows do
if rowMeta[r].hasHeader and not rowMeta[r].hasData then
headerRowCount = r
elseif r == 1 and rowMeta[r].hasHeader then
headerRowCount = 1
break
else
break
end
end
if headerRowCount == 0 and #rawRows >= 2 then
local r1HasNum, r2HasNum = false, false
for c = 1, maxCols do
local c1 = grid[1] and grid[1][c]
local c2 = grid[2] and grid[2][c]
if c1 and type(parseWikitableCellValue(c1.raw)) == "number" then r1HasNum = true end
if c2 and type(parseWikitableCellValue(c2.raw)) == "number" then r2HasNum = true end
end
if not r1HasNum and r2HasNum then
headerRowCount = 1
end
end
local headers = {}
local rawHeaders = {}
for c = 1, maxCols do
local hParts = {}
local lastRawH = nil
for r = 1, headerRowCount do
local cell = grid[r] and grid[r][c]
if cell then
local ch = cleanHeaderName(cell.raw)
if ch ~= "" and (hParts[#hParts] ~= ch) then
hParts[#hParts + 1] = ch
lastRawH = mw.text.trim(cell.raw)
end
end
end
local finalH = (#hParts > 0) and hParts[#hParts] or ("col" .. c)
headers[c] = finalH
rawHeaders[c] = lastRawH or finalH
end
local records = {}
local startDataRow = headerRowCount + 1
for r = startDataRow, #rawRows do
-- 略過夾在資料中間的純重複表頭列
if not (rowMeta[r].hasHeader and not rowMeta[r].hasData) then
local rec = {}
local hasNonEmpty = false
for c = 1, maxCols do
local cell = grid[r] and grid[r][c]
local val = cell and parseWikitableCellValue(cell.raw) or ""
if val ~= "" then hasNonEmpty = true end
local hClean = headers[c]
local hRaw = rawHeaders[c]
rec[hClean] = val
if hRaw and hRaw ~= hClean and rec[hRaw] == nil then
rec[hRaw] = val
end
-- 同步建立底線與空白別名,並支援 1-based 索引供 getRecordField ("0", "1") 讀取
local spaceAlias = string.gsub(hClean, "_", " ")
if spaceAlias ~= hClean and rec[spaceAlias] == nil then rec[spaceAlias] = val end
local underAlias = string.gsub(hClean, " ", "_")
if underAlias ~= hClean and rec[underAlias] == nil then rec[underAlias] = val end
end
if hasNonEmpty then
records[#records + 1] = rec
end
end
end
if #records == 0 then return nil end
records._headers = headers
records._caption = caption
return records
end
-- 從頁面或段落文字中掃描所有頂層 {| ... |} 表格並依條件或欄位智慧挑選最佳表格
local function parseWikitableFromContent(content, selector, hintFields, wf)
if type(content) ~= "string" or content == "" then return nil end
local text = normalizeTemplateTableSyntax(content)
if not string.find(text, "{|", 1, true) then return nil end
-- 掃描並提取所有頂層 {| ... |} 區塊及其所屬章節標題
local candidates = {}
local i, len = 1, #text
local depth = 0
local startIdx = nil
local currentSection = ""
while i <= len do
local two = string.sub(text, i, i + 1)
if depth == 0 and (i == 1 or string.byte(text, i - 1) == 10) and string.sub(text, i, i + 1) == "==" then
local lineEnd = string.find(text, "\n", i, true) or (len + 1)
local headingLine = string.sub(text, i, lineEnd - 1)
local hText = string.match(headingLine, "^=+%s*(.-)%s*=+%s*$")
if hText and hText ~= "" then
currentSection = cleanHeaderName(hText)
end
i = lineEnd
elseif two == "{|" then
if depth == 0 then startIdx = i end
depth = depth + 1
i = i + 2
elseif two == "|}" and depth > 0 then
depth = depth - 1
if depth == 0 and startIdx then
local block = string.sub(text, startIdx, i + 1)
local firstLine = string.match(block, "^([^\r\n]+)") or ""
local isWikitableClass = string.find(string.lower(firstLine), "wikitable", 1, true) ~= nil
local tableId = string.match(firstLine, '[iI][dD]%s*=%s*"([^"]+)"')
or string.match(firstLine, "[iI][dD]%s*=%s*'([^']+)'")
candidates[#candidates + 1] = {
idx = #candidates + 1,
raw = block,
isWikitable = isWikitableClass,
id = tableId,
section = currentSection
}
startIdx = nil
end
i = i + 2
else
i = i + 1
end
end
if #candidates == 0 then return nil end
-- 對單一候選表格進行按需局部 preprocess 與解析
local function decodeCandidate(cand)
if cand._decoded ~= nil then return cand._decoded or nil end
local rawBlock = cand.raw
if wf and wf.preprocess and string.find(rawBlock, "{{", 1, true) then
local okPre, preRes = pcall(function() return wf:preprocess(rawBlock) end)
if okPre and type(preRes) == "string" and preRes ~= "" then
rawBlock = preRes
end
end
local recs = parseSingleWikitable(rawBlock)
cand._decoded = recs or false
return recs
end
-- 1. 若有明確指定選擇器 (如 #1, #2 或章節/標題/ID 名稱)
if selector and selector ~= "" then
local selTrim = mw.text.trim(tostring(selector))
selTrim = string.gsub(selTrim, "^#", "")
local selNum = tonumber(selTrim)
if selNum and selNum >= 1 then
-- 優先取第 N 個 class="wikitable" 的表格;若不足則取第 N 個任意表格
local wtList = {}
for _, c in ipairs(candidates) do
if c.isWikitable then wtList[#wtList + 1] = c end
end
local target = wtList[selNum] or candidates[selNum]
if target then
local recs = decodeCandidate(target)
if recs then return recs end
end
else
for _, c in ipairs(candidates) do
if (c.id and c.id == selTrim) or (c.section ~= "" and string.find(c.section, selTrim, 1, true)) then
local recs = decodeCandidate(c)
if recs then return recs end
end
end
for _, c in ipairs(candidates) do
local recs = decodeCandidate(c)
if recs and recs._caption and string.find(recs._caption, selTrim, 1, true) then
return recs
end
end
end
end
-- 2. 收集期望匹配的欄位名 (供多表格智慧嗅探)
local wantedFields = {}
if type(hintFields) == "table" then
local checkKeys = { "xField", "xfield", "yField", "yfield", "zField", "zfield", "groupField", "groupfield", "seriesField" }
for _, k in ipairs(checkKeys) do
if type(hintFields[k]) == "string" and hintFields[k] ~= "" then
wantedFields[#wantedFields + 1] = cleanHeaderName(hintFields[k])
end
end
local fIdx = 1
while hintFields["y" .. fIdx .. "Field"] or hintFields["y" .. fIdx .. "field"] do
wantedFields[#wantedFields + 1] = cleanHeaderName(hintFields["y" .. fIdx .. "Field"] or hintFields["y" .. fIdx .. "field"])
fIdx = fIdx + 1
end
end
if #wantedFields > 0 then
local bestRecs = nil
local bestScore = 0
for _, c in ipairs(candidates) do
local recs = decodeCandidate(c)
if recs and recs[1] then
local score = 0
for _, wfName in ipairs(wantedFields) do
if recs[1][wfName] ~= nil then
score = score + 10
end
end
if score > 0 and c.isWikitable then
score = score + 1
end
if score > bestScore then
bestScore = score
bestRecs = recs
end
end
end
if bestRecs then return bestRecs end
end
-- 3. 預設退回:優先取第一個帶有 class="wikitable" 且有資料的表格,其次取第一個有效表格
for _, c in ipairs(candidates) do
if c.isWikitable then
local recs = decodeCandidate(c)
if recs then return recs end
end
end
for _, c in ipairs(candidates) do
local recs = decodeCandidate(c)
if recs then return recs end
end
return nil
end
-- 從巢狀或 Metatable 代理表中提取資料陣列 (支援 formatSpec.property == "fields" 與欄位別名映射)
local function extractRecordList(rawObj, formatSpec)
if type(rawObj) ~= "table" then return nil end
local reqProp = type(formatSpec) == "table" and formatSpec.property or nil
-- 若 Vega 明確要求讀取 schema.fields (如 Template:Graph:Lines 的 labels 資料源)
if reqProp == "fields" and type(rawObj.schema) == "table" and type(rawObj.schema.fields) == "table" then
local outFields = {}
local fIdx = 1
while rawObj.schema.fields[fIdx] ~= nil do
local f = rawObj.schema.fields[fIdx]
local tVal = f.title
if type(tVal) == "table" then
tVal = tVal.zh or tVal["zh-hant"] or tVal["zh-hans"] or tVal.en or next(tVal) and select(2, next(tVal))
end
outFields[fIdx] = {
name = f.name or tostring(fIdx),
title = tostring(tVal or f.name or fIdx)
}
fIdx = fIdx + 1
end
return outFields
end
if rawObj[1] ~= nil then return rawObj end
-- 常見的外層鍵名(statistics 為 Template:Unblock-zh backlog statistics/data.json 所用)
local knownKeys = { "statistics", "records", "rows", "items" }
for _, k in ipairs(knownKeys) do
local v = rawObj[k]
if type(v) == "table" and v[1] ~= nil and type(v[1]) == "table" then
return v
end
end
if type(rawObj.data) == "table" then
-- 檢查是否為 Commons Tabular Data 格式 (含 schema.fields 與二維陣列/字典 data)
if type(rawObj.schema) == "table" and type(rawObj.schema.fields) == "table" then
local fields = {}
local titles = {}
local fIdx = 1
while rawObj.schema.fields[fIdx] ~= nil do
local f = rawObj.schema.fields[fIdx]
local fName = f.name or tostring(fIdx)
fields[fIdx] = fName
local tVal = f.title
if type(tVal) == "table" then
tVal = tVal.en or tVal.zh or tVal["zh-hant"]
end
if type(tVal) == "string" and tVal ~= "" and tVal ~= fName then
titles[fIdx] = tVal
end
fIdx = fIdx + 1
end
local records = {}
local rIdx = 1
while rawObj.data[rIdx] ~= nil do
local row = rawObj.data[rIdx]
local rec = {}
for cIdx = 1, #fields do
local fName = fields[cIdx]
local val = row[cIdx]
if val == nil then val = row[fName] end
rec[fName] = val
-- 同步建立「底線轉空白」與「空白轉底線」的雙向別名 (如 New_Zealand <-> New Zealand)
local spaceAlias = string.gsub(fName, "_", " ")
if spaceAlias ~= fName and rec[spaceAlias] == nil then
rec[spaceAlias] = val
end
local underAlias = string.gsub(fName, " ", "_")
if underAlias ~= fName and rec[underAlias] == nil then
rec[underAlias] = val
end
-- 同步建立 schema.fields[].title 別名
local tAlias = titles[cIdx]
if tAlias and rec[tAlias] == nil then
rec[tAlias] = val
end
end
records[rIdx] = rec
rIdx = rIdx + 1
end
return records
end
if rawObj.data[1] ~= nil then
if type(rawObj.data[1]) == "table" and rawObj.data[1].values then
return rawObj.data[1].values
end
return rawObj.data
end
end
if type(rawObj.values) == "table" and rawObj.values[1] ~= nil then
return rawObj.values
end
if type(rawObj.series) == "table" and rawObj.series[1] ~= nil then
return rawObj.series
end
-- 通用備案:依鍵名排序,取第一個「陣列且元素為記錄表」的欄位
local keys = {}
for k, v in pairs(rawObj) do
if type(k) == "string" and type(v) == "table" and v[1] ~= nil and type(v[1]) == "table" then
keys[#keys + 1] = k
end
end
table.sort(keys)
if #keys > 0 then
return rawObj[keys[1]]
end
return nil
end
-- 5 階段外部資料路由器 (Data Router)
local function fetchExternalData(rawUrl, frame, formatSpec, hintFields)
local DataProcesser = require("Module:Graph/DataProcess")
if not rawUrl or mw.text.trim(tostring(rawUrl)) == "" then
return nil, false
end
local url = mw.text.trim(tostring(rawUrl))
local wf = (frame and frame.preprocess) and frame or mw.getCurrentFrame()
local explicitSelector = (type(formatSpec) == "table" and (formatSpec.table or formatSpec.selector or formatSpec.caption or formatSpec.section or formatSpec.id))
or (type(hintFields) == "table" and (hintFields.tableIndex or hintFields.tableindex or hintFields.tableSelector or hintFields.tableselector))
or nil
-- 【路由 A-Inline】若直接傳入行內 Wikitable ({|...|}) 或模板跳脫表格 ({{(!}}...{{!)}}),直接解析無須讀取頁面
if string.find(url, "{|", 1, true) or string.find(url, "{{(!", 1, true) then
local wtRecords = parseWikitableFromContent(url, explicitSelector, hintFields, wf)
if wtRecords then
return wtRecords, false
end
return nil, true
end
-- 【路由 A0】Wikidata SPARQL 雙通道:無論寫 wikidatasparql:// 或 https://query.wikidata.org/sparql,
-- 一律優先由後端 Lua (Module:Graph/Wikidata) 執行並輸出靜態 SVG;僅當查詢超出 Lua 能力時才轉交前端 "async"
if string.match(url, "^wikidatasparql:/")
or string.match(url, "^https?://query%.wikidata%.org/")
or string.match(url, "^//query%.wikidata%.org/") then
local okW, Wikidata = pcall(require, "Module:Graph/Wikidata")
if okW and type(Wikidata) == "table" then
if Wikidata.query then
local records, shouldAbort = Wikidata.query(url)
if not shouldAbort and type(records) == "table" and #records > 0 then
return records, false
end
end
if Wikidata.toStandardSparqlEndpointUrl then
local fallbackSparqlUrl = Wikidata.toStandardSparqlEndpointUrl(url)
if fallbackSparqlUrl then
return fallbackSparqlUrl, "async"
end
end
end
return nil, true
end
-- 【路由 A】wikirest:// 或外部動態 HTTP API:
-- 1. 通過維基官方白名單者標記為 "async" (輸出 gr-async-chart)
-- 2. 非維基白名單之第三方外部 API (如 NASA、NOAA) 標記為 "async_unsupported" (輸出 gr-async-chart-unsupported)
if string.match(url, "^wikirest://") or string.match(url, "^https?://") or string.match(url, "^//") then
local validApiUrl = normalizeAndValidateApiUrl(url)
if validApiUrl then
return validApiUrl, "async"
end
local thirdPartyUrl = normalizeUnsupportedExternalUrl(url)
if thirdPartyUrl then
return thirdPartyUrl, "async_unsupported"
end
return nil, true
end
-- 清理 wikitable:///、wikiraw:///、tabular:///、map:/// 前綴並還原 URL 編碼路徑
local pageName = url
local isTabularProtocol = false
local isWikitableProtocol = false
if string.match(pageName, "^wikitable:/") then
pageName = string.gsub(pageName, "^wikitable:///?", "")
pageName = mw.uri.decode(pageName, "PATH")
isWikitableProtocol = true
elseif string.sub(pageName, 1, 9) == "wikiraw:/" then
pageName = string.gsub(pageName, "^wikiraw:///?", "")
pageName = mw.uri.decode(pageName, "PATH")
elseif string.match(pageName, "^tabular:/") or string.match(pageName, "^map:/") then
pageName = string.gsub(pageName, "^tabular:///?", "")
pageName = string.gsub(pageName, "^map:///?", "")
pageName = mw.uri.decode(pageName, "PATH")
isTabularProtocol = true
end
pageName = mw.text.trim(pageName)
-- 分離頁面名稱後的 #選擇器 (例如 "Wikipedia:用户脚本#2" 或 "#1")
local hashPage, hashSel = string.match(pageName, "^(.-)#(.+)$")
if hashSel then
pageName = mw.text.trim(hashPage)
explicitSelector = explicitSelector or mw.text.trim(hashSel)
end
-- 若為 wikitable:/// (或 self) 未指定頁面名,預設讀取當前頁面自身之 Wikitable
if (pageName == "" or string.lower(pageName) == "self" or string.lower(pageName) == "this") and mw.title.getCurrentTitle then
local curTitle = mw.title.getCurrentTitle()
if curTitle and curTitle.prefixedText ~= "" then
pageName = curTitle.prefixedText
isWikitableProtocol = true
end
end
-- 【路由 B】維基共享資源 Data: 命名空間 (mw.ext.data.get,若失敗則繼續往路由 D/E 尋找本地 Data: 頁面)
if not isWikitableProtocol and (isTabularProtocol or string.match(pageName, "^[Cc]ommons:[Dd]ata:") or string.match(pageName, "^[Dd]ata:")) then
local cleanDataPage = string.gsub(pageName, "^[Cc]ommons:[Dd]ata:", "")
cleanDataPage = string.gsub(cleanDataPage, "^[Dd]ata:", "")
local spaceDataPage = string.gsub(cleanDataPage, "_", " ")
if mw.ext and mw.ext.data and mw.ext.data.get then
local ok, tabData = pcall(mw.ext.data.get, spaceDataPage)
if not ok or type(tabData) ~= "table" then
ok, tabData = pcall(mw.ext.data.get, cleanDataPage)
end
if ok and type(tabData) == "table" then
local list = extractRecordList(tabData, formatSpec)
if list then return list, false end
end
end
pageName = "Data:" .. cleanDataPage
end
-- 【路由 C】Module: 命名空間 (mw.loadData)
if not isWikitableProtocol and (string.match(pageName, "^[Mm]odule:") or string.match(pageName, "^模組:") or string.match(pageName, "^模块:")) then
local ok, modData = pcall(mw.loadData, pageName)
if not ok or type(modData) ~= "table" then
ok, modData = pcall(require, pageName)
end
if ok and type(modData) == "table" then
local list = extractRecordList(modData, formatSpec)
if list then return list, false end
end
return nil, true
end
-- 取得頁面 Title 物件,檢查【路由 D】與【路由 E】
local titleObj = mw.title.new(pageName)
if not titleObj then return nil, true end
-- 【路由 D】檢查 contentModel === 'json',優先使用 mw.loadJsonData 快取通道
if not isWikitableProtocol and titleObj.contentModel == "json" and mw.loadJsonData then
local ok, jsonData = pcall(mw.loadJsonData, pageName)
if ok and type(jsonData) == "table" then
local list = extractRecordList(jsonData, formatSpec)
if list then return toPlainTable(list), false end
end
end
-- 【路由 E】一般維基頁面:支援 Wikitable、JSON 或 CSV
local content = titleObj:getContent()
if not content or content == "" then return nil, true end
content = string.gsub(content, "<noinclude>.-</noinclude>", "")
content = string.gsub(content, "</?includeonly>", "")
content = string.gsub(content, "</?onlyinclude>", "")
local isWikitableFormat = isWikitableProtocol
or (type(formatSpec) == "table" and string.lower(tostring(formatSpec.type or "")) == "wikitable")
or string.find(content, "{|", 1, true) ~= nil
or string.find(content, "{{(!", 1, true) ~= nil
-- E1. 若頁面含有 Wikitable ({| 或 {{(!}}),優先以局部按需展開方式解析表格 (避免對整篇長條目執行 preprocess)
if isWikitableFormat then
local wtList = parseWikitableFromContent(content, explicitSelector, hintFields, wf)
if wtList then return wtList, false end
if isWikitableProtocol then return nil, true end
end
-- E2. 一般 JSON 或 CSV 資料頁:若含模板則展開後解析
if wf and wf.preprocess and string.find(content, "{{", 1, true) then
local okPre, preRes = pcall(function() return wf:preprocess(content) end)
if okPre and preRes then content = preRes end
end
-- 若展開模板後才浮現 {| 表格,再次嘗試解析為 Wikitable
if string.find(content, "{|", 1, true) or string.find(content, "{{(!", 1, true) then
local wtList = parseWikitableFromContent(content, explicitSelector, hintFields, nil)
if wtList then return wtList, false end
end
local isCsvFormat = (type(formatSpec) == "table" and string.lower(tostring(formatSpec.type or "")) == "csv")
or string.match(pageName, "[%-%.]csv$") ~= nil
if isCsvFormat then
local csvList = DataProcesser.parseCSV(mw.text.trim(content))
if csvList then return csvList, false end
end
local cleanJson = DataProcesser.stripJsonComments(mw.text.trim(content))
local okDec, decoded = pcall(mw.text.jsonDecode, cleanJson, mw.text.JSON_TRY_FIXING)
if okDec and type(decoded) == "table" then
local list = extractRecordList(decoded, formatSpec)
if list then return list, false end
end
-- 若未標記 csv 但 JSON 解析失敗,嘗試當作 CSV 解析
if not isCsvFormat then
local csvList = DataProcesser.parseCSV(mw.text.trim(content))
if csvList then return csvList, false end
end
return nil, true
end
p.WHITELIST_PAGE = WHITELIST_PAGE
p.parseBool = parseBool
p.maskStripMarkers = maskStripMarkers
p.numericArray = numericArray
p.stringArray = stringArray
p.parseShapeArray = parseShapeArray
p.isDateArray = isDateArray
p.deserializeXData = deserializeXData
p.deserializeYData = deserializeYData
p.getApiWhitelist = getApiWhitelist
p.getLocalWikiHost = getLocalWikiHost
p.normalizeAndValidateApiUrl = normalizeAndValidateApiUrl
p.normalizeUnsupportedExternalUrl = normalizeUnsupportedExternalUrl
p.normalizeTemplateTableSyntax = normalizeTemplateTableSyntax
p.parseWikitableCellValue = parseWikitableCellValue
p.splitCellAttrsAndContent = splitCellAttrsAndContent
p.splitInlineWikitableCells = splitInlineWikitableCells
p.parseSingleWikitable = parseSingleWikitable
p.parseWikitableFromContent = parseWikitableFromContent
p.extractRecordList = extractRecordList
p.fetchExternalData = fetchExternalData
return p