Module:Dump
This module can dump a table by displaying its contents as text. It can also display formatted html text. That may help develop other modules.
An alternative is to use mw.dumpObject() but the result from this module is clearer and is close to valid Lua source.
Dump of a Wikidata entity
[kulemba source]When working with Wikidata, it can be useful to examine a table representing an entity.
For example, Southern African Large Telescope is d:Q833639. That entity can be viewed as a Lua table by previewing:
{{#invoke:dump|wikidata|Q833639}}
To do that, edit your sandbox and replace its contents with the above line, then click Show preview. Module talk:Dump shows this example.
If wanted, the width of each indented column can be set, for example to two spaces or to use tab characters:
{{#invoke:dump|wikidata|Q833639|indent=2}}{{#invoke:dump|wikidata|Q833639|indent=tab}}
A property such as (P2386) can also be displayed:
{{#invoke:dump|wikidata|P2386}}
Dump of a table for another module
[kulemba source]If there is a problem debugging a module, it can be helpful to use a sandbox copy of the module to display the contents of a table to confirm that it contains expected data. The following shows how a module can dump a table.
local p = {}
function p.main(frame)
local example = {
link = true,
fruit = { yellow = 'bannana', red = 'cherry' },
11, 22, 33,
}
local dump = require('Module:Dump')._dump
return dump(example, 'example')
end
return p
With the above code in Module:Example, the result could be displayed by previewing:
{{#invoke:example|main}}
The module contains a complex table for testing. The table can be displayed by previewing:
{{#invoke:dump|testcase}}
Dump of a formatted html string
[kulemba source]A module can use mw.html to generate html. For testing, it may be useful to display the formatted result. The following shows how a module can create and dump html text.
local function main()
local tbl = mw.html.create('table')
tbl
:addClass('wikitable')
:tag('caption'):wikitext('Table demonstration'):done()
:tag('tr')
:tag('th'):wikitext('Month'):done()
:tag('th'):wikitext('Amount'):done()
:done()
:tag('tr')
:tag('td'):wikitext('January'):done()
:tag('td'):wikitext('$100<br>loss'):done()
:done()
:tag('tr')
:tag('td'):wikitext('February'):done()
:tag('td'):wikitext('$200')
local html = tostring(tbl)
local dumphtml = require('Module:Dump')._dumphtml
return dumphtml(html)
end
return { main = main }
With the above code in Module:Example, the result could be displayed by previewing:
{{#invoke:example|main}}
The result is:
<table class="wikitable">
<caption>Table demonstration</caption>
<tr>
<th>Month</th>
<th>Amount</th>
</tr>
<tr>
<td>January</td>
<td>$100<br>loss</td>
</tr>
<tr>
<td>February</td>
<td>$200</td>
</tr>
</table>
The main() function in the code above could be modified to return the html table by replacing the last two lines with:
return html
In that case, the result could be displayed by previewing the following (the 1= is needed if the text contains "="):
{{#invoke:dump|dumphtml|1={{#invoke:example|main}}}}
Dumping a navbox
[kulemba source]Previewing the following examples in a sandbox may be useful to examine the results of a template, such as {{navbox}}, that generates html.
{{#invoke:dump|dumphtml|1=
{{navbox/sandbox
|group1 = Group1
|list1 = List1
|group2 = Group2
|list2 = List2
|group3 = Group3
|list3 = List3
}}
}}
The dumphtml procedure only works reliably with valid html. In the following example, extra text (<div>) is inserted at the start because the output from a subgroup (child) navbox starts with </div>.
{{#invoke:dump|dumphtml|1=<div>
{{navbox/sandbox|subgroup
|group1 = Group1
|list1 = List1
|group2 = Group2
|list2 = List2
|group3 = Group3
|list3 = List3
}}
}}
Dump of arguments
[kulemba source]Special:ExpandTemplates is useful if there is a need to view the wikitext returned by a template or module. However, ExpandTemplates does not always show exactly what a module would receive. For example, the following template gives the output shown in ExpandTemplates, but the wikitext passed to a module would actually contains strip markers.
{{convert|1+2/3<ref>Example</ref>|ft|in}}- ExpandTemplates output, rearranged on multiple lines for clarity:
<templatestyles src="Fraction/styles.css"></templatestyles><span class="frac" role="math">1<span class="sr-only">+</span><span class="num">2</span>⁄<span class="den">3</span></span><ref>Example</ref>feet (20 in)
The args function shows what a module receives in its arguments.
{{#invoke:dump|args|1={{convert|1+2/3<ref>Example</ref>|ft|in}}}}
The output follows. For clarity, it has been rearranged on multiple lines and each delete character has been replaced with ♢.
♢'"`UNIQ--templatestyles-00000002-QINU`"'♢<span class="frac" role="math">1<span class="sr-only">+</span><span class="num">2</span>⁄<span class="den">3</span></span>♢'"`UNIQ--ref-00000001-QINU`"'♢feet (20 in)
Dump of parameters
[kulemba source]A template might invoke the main function in the example above. Any parameters passed to the template or the module can be displayed for debugging. That would be to investigate an unexpected result in a page, for example, Albedo. To see what parameters are received by a module used in that article, edit the module and insert the following line at the start of the main function:
if true then return require('Module:Dump').parameters(frame) end
Do not save the changes. Instead, enter the name of the article (for example, Albedo) in the box under "Preview page with this module", then click Show preview. Any parameters passed to the module in its frame and parent frame are displayed where the result from the module would normally appear.
Dump of structured data
[kulemba source]A Lua program could execute
local data = mw.ext.data.get('Wikipedia statistics/data.tab')
to read a table of data from c:Data:Wikipedia statistics/data.tab.
An edit in a sandbox can be previewed to see what data the program would receive. To do that, preview the following wikitext:
{{#invoke:dump|Wikipedia statistics/data.tab}}
The dump module accepts any text as the parameter and will apply special processing if the text is recognized. Structured data is identified as text ending with .tab.
Global table _G
[kulemba source]In Lua, _G is a global variable which is a table holding information about all global variables. The _G table can be displayed by previewing (both G and _G work):
{{#invoke:dump|testcase|G}}
If wanted, the width of each indented column can be set, for example to 2 spaces:
{{#invoke:dump|testcase|G|indent=2}}
-- Dump a table to help develop other modules.
-- It is also possible to use mw.dumpObject() but the result from this
-- module is clearer and is close to valid Lua source.
-- The main purpose is to allow easy inspection of Wikidata items.
-- Preview the following in a sandbox to see entity Q833639 as a Lua table:
-- {{#invoke:dump|wikidata|Q833639}}
-- Preview the following to dump a built-in table:
-- {{#invoke:dump|testcase}}
local Collection -- a table to hold items
Collection = {
add = function (self, item)
if item ~= nil then
self.n = self.n + 1
self[self.n] = item
end
end,
join = function (self, sep)
return table.concat(self, sep)
end,
remove = function (self, pos)
if self.n > 0 and (pos == nil or (0 < pos and pos <= self.n)) then
self.n = self.n - 1
return table.remove(self, pos)
end
end,
sort = function (self, comp)
table.sort(self, comp)
end,
new = function ()
return setmetatable({n = 0}, Collection)
end
}
Collection.__index = Collection
local function pre_block(text)
-- Pre tags returned by a module do not act like wikitext <pre>...</pre>.
return '<pre>\n' ..
mw.text.nowiki(text) ..
(text:sub(-1) == '\n' and '' or '\n') ..
'</pre>\n'
end
local function make_tabstr(indent)
-- Return a string to generate one level of indent.
if indent == 'tab' then
-- Tabs do not work well in a browser edit window, but can force them.
return '\t'
end
indent = tonumber(indent)
if not (type(indent) == 'number' and 1 <= indent and indent <= 32) then
indent = 4
end
return string.rep(' ', indent)
end
local function _dumphtml(html, tabwidth)
-- Return a pretty-text formatted dump of an html string.
-- This assumes clean html, for example, tag "<table>" not "< table >".
if type(html) ~= 'string' then
return ''
end
local selfClosingTags = { -- from mw.html.lua
area = true,
base = true,
br = true,
col = true,
command = true,
embed = true,
hr = true,
img = true,
input = true,
keygen = true,
link = true,
meta = true,
param = true,
source = true,
track = true,
wbr = true,
}
local tabstr = make_tabstr(tabwidth)
local function indent_pad(depth, isfirst)
-- Return a string with an indent to match depth.
if depth > 0 then
return '\n' .. string.rep(tabstr, depth)
end
return isfirst and '' or '\n'
end
local function extract(result, html, pos, len, depth, currenttag)
-- Dump more of html into table result and return new pos.
local has_child
while pos <= len do
local s, e = html:find('<[^<>]*>', pos)
if s then
if s > pos then
table.insert(result, html:sub(pos, s-1))
end
if html:sub(s+1, s+1) == '/' then
-- A closing tag.
local tag = html:match('^([a-zA-Z0-9]+)>', s+2) or 'NOTAG'
if tag == currenttag then
local indent = has_child and indent_pad(depth - 1) or ''
table.insert(result, indent .. '</' .. tag .. '>')
else
-- Should never happen.
table.insert(result, '\n</' .. tag .. '>')
end
return e + 1
end
local tag = html:match('^[a-zA-Z0-9]+', s+1) or 'NOTAG'
if html:sub(e-1, e-1) == '/' or selfClosingTags[tag] then
-- A self-closing tag.
table.insert(result, html:sub(s, e))
pos = e + 1
else
-- An opening tag.
table.insert(result, indent_pad(depth, pos == 1) .. html:sub(s, e))
pos = extract(result, html, e+1, len, depth+1, tag)
has_child = true
end
else
table.insert(result, html:sub(pos))
break
end
end
return len + 1
end
local result = {}
html = html:gsub('>%s+<', '><'):gsub('\n%s*', ' ')
extract(result, html, 1, #html, 0)
return pre_block(table.concat(result))
end
local function dumphtml(frame)
local args = frame.args
local pargs = frame:getParent().args
local text = args[1] or pargs[1]
local indent = args.indent or pargs.indent
return _dumphtml(text, indent)
end
local function quoted(str)
return (string.format('%q', str):gsub('\\\n', '\\n'))
end
local function iterkeys(var, control)
-- Return an iterator over the keys of var (which should be a table).
-- The keys are sorted with numbered keys first, then other types.
-- The iterator returns key, repr where key is the actual key, and
-- repr is its representation: a number for the ipairs keys, or
-- a string, including for number keys above the table length.
if type(var) ~= 'table' then
return function () return nil end
end
local nums = {}
local results = Collection.new()
for i, _ in ipairs(var) do
nums[i] = true
results:add({ i, i })
end
local keys = Collection.new()
for k, _ in pairs(var) do
if not nums[k] then
keys:add(k)
end
end
local autoname = control.autoname
keys:sort(function (a, b)
local ta, tb = type(a), type(b)
if ta == tb then
if ta == 'number' or ta == 'string' then
return a < b
end
if ta == 'boolean' then
return b and not a
end
return autoname(a) < autoname(b)
end
if ta == 'number' then
return true
elseif tb == 'number' then
return false
else
return ta < tb
end
end)
for _, k in ipairs(keys) do
local repr
local tk = type(k)
if tk == 'number' then
repr = '[' .. k .. ']'
elseif tk == 'string' then
if k:match('^[%a_][%w_]*$') then
repr = k
else
repr = '[' .. quoted(k) .. ']'
end
elseif tk == 'boolean' then
repr = '[' .. tostring(k) .. ']'
else
repr = autoname(k)
control.needed[repr] = true
end
results:add({ k, repr })
end
local last = 0
return function ()
if last < results.n then
last = last + 1
return unpack(results[last])
end
end
end
local function vardump(var, vname, depth, control, self, parents)
-- Update items in control with results from dumping a variable.
local function put(value, options)
options = options or {}
local indent = options.indent or depth
local comma = (options.kind == 'open' or indent == 0) and '' or ','
control.items:add({
key = (type(vname) == 'string' and options.kind ~= 'close') and vname or nil,
value = value .. comma,
depth = indent,
note = options.note
})
end
if var == nil then
put('nil')
elseif type(var) == 'string' then
put(quoted(var))
elseif type(var) == 'table' then
local this = control.autoname(var)
if depth >= control.limitdepth then
put(this)
elseif parents and parents[this] then
control.needed[this] = true
if self == this then
put(this, {note = 'self'})
control.needed['self'] = true
else
put(this, {note = 'repeat'})
control.needed['repeat'] = true
end
else
parents = parents or {}
parents[this] = true
self = this
put('{', {kind = 'open', note = this})
local mt = getmetatable(var)
if mt then
vardump(mt, '__metatable', depth + 1, control, self, parents)
end
local maxsize = control.items.n + control.limititems
for key, keyrep in iterkeys(var, control) do
if control.items.n > maxsize then
put('...more...')
break
end
vardump(var[key], keyrep, depth + 1, control, self, parents)
end
put('}', { kind = 'close' })
end
elseif type(var) == 'boolean' or type(var) == 'number' then
put(tostring(var))
else -- function (or userdata or thread)
put(control.autoname(var))
end
end
local function dumper(var, vname, tabwidth, wantraw, limititems, limitdepth)
-- Return a string representing var in almost-correct Lua syntax.
-- There is no newline at the end of the result.
local onames = {}
local tcounts = {}
local function autoname(var)
-- Return a string that is a unique name for var, given it is not
-- a number or string.
if not onames[var] then
local name = type(var)
tcounts[name] = (tcounts[name] or 0) + 1
onames[var] = name .. '_' .. tcounts[name]
end
return onames[var]
end
local control = {
autoname = autoname,
limititems = limititems or 10000,
limitdepth = limitdepth or 50,
items = Collection.new(),
needed = {},
}
vardump(var, tostring(vname or 'variable'), 0, control)
local tabstr = make_tabstr(tabwidth)
local lines = Collection.new()
for i, v in ipairs(control.items) do
local indent = string.rep(tabstr, v.depth)
local note = v.note
if note and control.needed[note] then
note = ' -- ' .. note
else
note = ''
end
local k = v.key and (v.key .. ' = ') or ''
lines:add(indent .. k .. v.value .. note)
end
local raw = lines:join('\n')
return wantraw and raw or pre_block(raw)
end
local function dump_testcase(frame)
local item
if type(frame) == 'table' then
item = frame.args[1]
else
item = frame
end
if item == 'G' or item == '_G' then
return dumper(_G, '_G', frame.args.indent)
end
local fruit = { 'apple', 'banana', [0] = 'zero', [{'anon'}] = 'anon' }
local testcase = {
[100] = 'one hundred',
[99] = 'ninety nine',
[0.5] = 'one half',
[-1] = 'negative one',
'one',
'two',
[' '] = 'space',
['1 –◆— z'] = 'unicode',
alpha = 'aaa',
beta = 'bbb',
c = 123,
data = {
dumper = dumper,
[dumper] = 'dumper',
'three',
'four',
T = true,
[true] = 'T',
alpha2 = 'aaa2',
beta2 = 'bbb2',
F = false,
[false] = 'F',
c2 = 1234,
data2 = {
'five',
'six',
alpha3 = 'aaa3',
beta3 = 'bbb3',
c3 = 12345,
fruit = fruit,
[fruit] = 'fruit',
},
},
z = 'zoo',
}
testcase.testcase = testcase
testcase.data.me = testcase.data
testcase.data.data2.me = testcase
testcase.data.data2.fruit.back = testcase.data
setmetatable(testcase.data, {
__index = function (self, key) return type(key) == 'string' and #key or nil end,
__tostring = function (self) return tostring(#self) end,
})
if item == 'return table' then
return testcase
end
return dumper(testcase, 'testcase', frame.args.indent)
end
local function execute(frame)
-- Return a dump of the result from executing {{#invoke:dump|execute|EXPRESSION}}.
-- In general that is not possible in Scribunto so this has built-in code
-- to parse some expressions of interest.
-- The primary aim is to test the result of calling a Wikidata function
-- while previewing an edit in an article.
-- Examples of EXPRESSION:
-- mw.wikibase.getEntityIdForCurrentPage()
-- mw.wikibase.getBestStatements('Q868', 'P214')
-- mw.wikibase.getBestStatements(Q868, P214) -- also accepted
-- mw.wikibase.getEntity():getDescription('de')
-- mw.wikibase.getEntity('Q868'):getDescription('de')
-- getEntityObject is an alias for getEntity.
-- Using the following gives an "out of memory" error presumably because
-- the result is a table with a metatable that dump repeatedly expands.
-- mw.title.getCurrentTitle()
local function params(ptext, first)
local p = { first }
for item in (ptext .. ','):gmatch('(%S.-)%s*,') do
-- Remove any quotes around each parameter because it is already a string.
local _, s = item:match([[^%s*(['"])(.*)%1%s*$]])
table.insert(p, s or tonumber(item) or item)
end
return unpack(p)
end
local expression = frame.args[1] or ''
local text = expression:match('^%s*mw(%..-)%s*$')
if not text then
return 'Expression not recognized: "' .. expression .. '"'
end
-- Look for a supported expression of form 'mw.a.b(c):d.e(f)'.
local entity
local object = mw
local item, ptext, rest = text:match('^%.wikibase%.(%w+)%s*%((.*)%):(.*)$')
if item == 'getEntity' or item == 'getEntityObject' then
entity = mw.wikibase.getEntity(params(ptext))
if not entity then
return 'No entity found for (' .. ptext .. ')'
end
object = entity
text = '.' .. rest -- treat ':' as '.'
end
local upto = 1
for i1, item, i2 in text:gmatch('()%.(%w+)()') do
if i1 == upto and type(object) == 'table' then
object = object[item]
else
object = nil
end
if object == nil then
return 'Invalid item "' .. item .. '"'
end
if type(object) == 'function' then
if text:sub(i2, i2 + 1) == '()' then
object = object()
i2 = i2 + 2
end
end
upto = i2
end
local parm = text:sub(upto):match('^%((.*)%)%s*$')
if parm then
object = object(params(parm, entity))
end
return dumper(object, expression)
end
local function dumpargs(frame)
-- Return text dump of frame.args.
-- {{#invoke:dump|args|<ref>Example</ref>}} → display ref strip marker
local control = {
autoname = function (var) return tostring(var) end, -- should not be called since keys should be numbers or strings
}
local lines = Collection.new()
for key, keyrep in iterkeys(frame.args, control) do
lines:add(keyrep .. ' = <code>' .. mw.text.nowiki(frame.args[key]) .. '</code>')
end
return lines:join('<br>\n')
end
local function parameters(frame)
-- Return text dump of args and parent args from frame.
-- This is for debugging a module to show what parameters it received.
local control = {
autoname = function (var) return tostring(var) end, -- should not be called since keys should be numbers or strings
}
local lines = Collection.new()
lines:add('')
for _, f in ipairs({ frame, frame:getParent() }) do
lines:add('[[' .. f:getTitle() .. ']]')
for key, keyrep in iterkeys(f.args, control) do
lines:add(' ' .. mw.text.nowiki(keyrep .. '=' .. f.args[key]))
end
end
lines:add('')
return lines:join('<br>\n')
end
local function wikidata(frame)
local item = frame.args[1]
if item then
local id = item:match('^%s*([PQ]%d+)%s*$')
if id then
local entity = mw.wikibase.getEntity(id)
return dumper(entity, id, frame.args.indent)
end
end
return 'Parameter should be a Wikidata identifier such as P2386 or Q833639'
end
local builtins = {
-- Handle preview of wikitext like {{#invoke|dump|TEXT}}
-- where TEXT is a built-in value that can be dumped.
__index = function (self, key)
local result
local function caller()
return result
end
if type(key) == 'string' then
local title = key:match('^%s*[\'"]?(.*%.tab)[\'"]?%s*$')
if title then
-- Assume structured data from Commons at [[c:Data:<title>]].
if title:match('^[Dd]ata:') then
title = title:sub(6)
end
local data = mw.ext.data.get(title) -- false if page does not exist
result = dumper(data, '[[c:Data:' .. title .. ']]')
end
end
result = result or ('UNKNOWN: ' .. tostring(key))
return caller
end
}
return setmetatable({
args = dumpargs,
_dump = dumper,
_dumphtml = _dumphtml,
dumphtml = dumphtml,
execute = execute,
parameters = parameters,
testcase = dump_testcase,
wikidata = wikidata,
}, builtins)