imported>MSGJ: more robust tidying of text

2022-11-04T14:46:12Z

more robust tidying of text

New page

require("strict")
local p = {}

local function resolveqid(label)
local resolveEntity = require("Module:ResolveEntityId")._id
local qid
local rawlabel
if label then
rawlabel = string.match(label,'%[%[([^%|%]]+)%|') or string.match(label,'%[%[([^%|%]]+)%]%]')
if rawlabel then
qid = resolveEntity(rawlabel)
else
rawlabel = label
end
end
return qid,rawlabel
end

local function tidystring(string)
local tidy = mw.ustring.gsub(string,"%<ref.+%<%/ref%>","") -- remove any references with <ref .. </ref>
tidy = mw.ustring.gsub(tidy,"%<ref.+%/%>","") -- remove any references with <ref .. />
tidy = mw.text.trim(tidy) -- trim spaces
return tidy
end

function p.import(frame)
local tab = "|"
local wikiqid = "Q328" -- QID for English Wikipedia
local eol = "<br>" -- end of line string
local coord2text = require("Module:Coordinates/sandbox")._coord2text
local args = frame.args or frame:getParent().args
if not args.page then
return "No page specified."
end
if not args.config then
return "No configuation."
end
local config = {}
for c1,col in ipairs(mw.text.split(args.config,",")) do
config[c1] = mw.text.split(col,"-")
end
local common = {}
if args.common then
for c1,prop in ipairs(mw.text.split(args.common,",")) do
common[c1] = mw.text.split(prop,"-")
common[common[c1][1]] = common[c1][2] -- create index, e.g. common["P31"]=Q39715
end
end
local content = mw.title.new(args.page):getContent() -- read page specified
content = string.match(content,"%{%|(.+)%|%}") -- keep table only
content = string.gsub(content,"||","\n|") -- use \n| for column breaks
local rows = mw.text.split(content,"|-",true) -- split table into rows
table.remove(rows,1) -- remove table definition
table.remove(rows,1) -- remove heading row
local output = ""
local v2 = ""
for rn,row in ipairs(rows) do
local columns = mw.text.split(row,"\n|") -- split table rom into columns
table.remove(columns,1) -- remove content before the first \n| character
local label = columns[1]
if label then
label = tidystring(label)
local qid
qid,label = resolveqid(label) -- resolve qid if first column is link
if not qid then
qid = string.match(label,"Q%d+") -- check if QID is specified in first column
end
local entity
if qid then
entity = mw.wikibase.getEntity(qid)
else
v2 = v2 .. "CREATE" .. eol
qid = "LAST" -- creating new item, so use LAST
v2 = v2 .. qid .. tab .. 'Len' .. tab .. '"' .. label .. '"' .. eol
if common["P31"] then -- create auto-description
v2 = v2 .. qid .. tab .. 'Den' .. tab .. '"' .. mw.wikibase.getLabel(common["P31"])
if common["P17"] then -- add country
v2 = v2 .. ' in ' .. mw.wikibase.getLabel(common["P17"])
end
v2 = v2 .. '"' .. eol
end
for c = 1,#common do
v2 = v2 .. qid .. tab .. common[c][1] .. tab .. common[c][2] .. eol
end
end
local function addtov2(prop,val)
v2 = v2 .. qid .. tab .. prop .. tab .. val .. tab .. "S143" .. tab .. wikiqid .. eol
return v2
end
for cn,col in ipairs(columns) do
if config[cn] then
local import = true
if entity then
if config[cn][2] then
if entity:getAllStatements(config[cn][2])[1] then -- statement already present for this property
import = false
end
else -- no property specified to import to
import = false
end
end
if import then
col = tidystring(col)
if config[cn][1] == "wikilink" then
local val = resolveqid(col)
if val then
v2 = addtov2(config[cn][2],val)
end
elseif config[cn][1] == "year" then
local val = string.match(col,"%d%d%d%d")
if val then
v2 = addtov2(config[cn][2],"+" ..val .. "-00-00T00:00:00Z/9")
end
elseif config[cn][1] == "quantity" then
local val = string.gsub(col,",","") -- remove any commas
val = string.match(val,"%d+%.?%d*") -- extract value, possibly from inside convert template
if val then
if string.upper(string.sub(config[cn][3],1,1)) == "Q" then
config[cn][3] = string.sub(config[cn][3],2)
end
v2 = addtov2(config[cn][2],val .. "U" .. config[cn][3])
end
elseif config[cn][1] == "text" then
if col ~= "" then
v2 = addtov2(config[cn][2],'"' .. col .. '"')
end
elseif config[cn][1] == "coord" then
if col ~= "" then
local val = mw.getCurrentFrame():preprocess(col)
v2 = addtov2(config[cn][2],"@" .. coord2text(val,"lat") .. "/" .. coord2text(val,"long"))
end
end
end
end
end
end
end
return v2
end

return p

Module:Import table - Revision history

imported>MSGJ: more robust tidying of text