Module:Convert: Difference between revisions
From Vigyanwiki
m (1 revision imported) |
m (2 revisions imported from alpha:Module:Convert) |
||
| (3 intermediate revisions by 3 users not shown) | |||
| Line 568: | Line 568: | ||
if key == 'symbol' then | if key == 'symbol' then | ||
value = self.si_prefix .. self._symbol | value = self.si_prefix .. self._symbol | ||
if value == 'l' then value = 'L' end | |||
elseif key == 'sym_us' then | elseif key == 'sym_us' then | ||
value = rawget(self, '_sym_us') | value = rawget(self, '_sym_us') | ||
Latest revision as of 10:35, 28 July 2023
Documentation for this module may be created at Module:Convert/doc
-- Convert a value from one unit of measurement to another.
-- Example: {{convert|123|lb|kg}} --> 123 pounds (56 kg)
-- See [[:en:Template:Convert/Transwiki guide]] if copying to another wiki.
local MINUS = '−' -- Unicode U+2212 MINUS SIGN (UTF-8: e2 88 92)
local abs = math.abs
local floor = math.floor
local format = string.format
local log10 = math.log10
local ustring = mw.ustring
local ulen = ustring.len
local usub = ustring.sub
-- Configuration options to keep magic values in one location.
-- Conversion data and message text are defined in separate modules.
local config, maxsigfig
local numdot -- must be '.' or ',' or a character which works in a regex
local numsep, numsep_remove, numsep_remove2
local data_code, all_units
local text_code
local varname -- can be a code to use variable names that depend on value
local from_en_table -- to translate an output string of en digits to local language
local to_en_table -- to translate an input string of digits in local language to en
-- Use translation_table in convert/text to change the following.
local en_default -- true uses lang=en unless convert has lang=local or local digits
local group_method = 3 -- code for how many digits are in a group
local per_word = 'per' -- for units like "liters per kilometer"
local plural_suffix = 's' -- only other useful value is probably '' to disable plural unit names
local omitsep -- true to omit separator before local symbol/name
-- All units should be defined in the data module. However, to cater for quick changes
-- and experiments, any unknown unit is looked up in an extra data module, if it exists.
-- That module would be transcluded in only a small number of pages, so there should be
-- little server overhead from making changes, and changes should propagate quickly.
local extra_module -- name of module with extra units
local extra_units -- nil or table of extra units from extra_module
-- Some options in the invoking template can set variables used later in the module.
local currency_text -- for a user-defined currency symbol: {{convert|12|$/ha|$=€}} (euro replaces dollar)
local function from_en(text)
-- Input is a string representing a number in en digits with '.' decimal mark,
-- without digit grouping (which is done just after calling this).
-- Return the translation of the string with numdot and digits in local language.
if numdot ~= '.' then
text = text:gsub('%.', numdot)
end
if from_en_table then
text = text:gsub('%d', from_en_table)
end
return text
end
local function to_en(text)
-- Input is a string representing a number in the local language with
-- an optional numdot decimal mark and numsep digit grouping.
-- Return the translation of the string with '.' mark and en digits,
-- and no separators (they have to be removed here to handle cases like
-- numsep = '.' and numdot = ',' with input "1.234.567,8").
if to_en_table then
text = ustring.gsub(text, '%d', to_en_table)
end
if numsep_remove then
text = text:gsub(numsep_remove, '')
end
if numsep_remove2 then
text = text:gsub(numsep_remove2, '')
end
if numdot ~= '.' then
text = text:gsub(numdot, '.')
end
return text
end
local function decimal_mark(text)
-- Return ',' if text probably is using comma for decimal mark, or has no decimal mark.
-- Return '.' if text probably is using dot for decimal mark.
-- Otherwise return nothing (decimal mark not known).
if not text:find('[.,]') then return ',' end
text = text:gsub('^%-', ''):gsub('%+%d+/%d+$', ''):gsub('[Ee]%-?%d+$', '')
local decimal =
text:match('^0?([.,])%d+$') or
text:match('%d([.,])%d?%d?$') or
text:match('%d([.,])%d%d%d%d+$')
if decimal then return decimal end
if text:match('%.%d+%.') then return ',' end
if text:match('%,%d+,') then return '.' end
end
local add_warning, with_separator -- forward declarations
local function to_en_with_check(text, parms)
-- Version of to_en() for a wiki using numdot = ',' and numsep = '.' to check
-- text (an input number as a string) which might have been copied from enwiki.
-- For example, in '1.234' the '.' could be a decimal mark or a group separator.
-- From viwiki.
if to_en_table then
text = ustring.gsub(text, '%d', to_en_table)
end
if decimal_mark(text) == '.' then
local original = text
text = text:gsub(',', '') -- for example, interpret "1,234.5" as an enwiki value
if parms then
add_warning(parms, 0, 'cvt_enwiki_num', original, with_separator({}, text))
end
else
if numsep_remove then
text = text:gsub(numsep_remove, '')
end
if numsep_remove2 then
text = text:gsub(numsep_remove2, '')
end
if numdot ~= '.' then
text = text:gsub(numdot, '.')
end
end
return text
end
local function omit_separator(id)
-- Return true if there should be no separator before id (a unit symbol or name).
-- For zhwiki, there should be no separator if id uses local characters.
-- The following kludge should be a sufficient test.
if omitsep then
if id:sub(1, 2) == '-{' then -- for "-{...}-" content language variant
return true
end
if id:byte() > 127 then
local first = usub(id, 1, 1)
if first ~= 'Å' and first ~= '°' and first ~= 'µ' then
return true
end
end
end
return id:sub(1, 1) == '/' -- no separator before units like "/ha"
end
local spell_module -- name of module that can spell numbers
local speller -- function from that module to handle spelling (set if needed)
local wikidata_module, wikidata_data_module -- names of Wikidata modules
local wikidata_code, wikidata_data -- exported tables from those modules (set if needed)
local function set_config(args)
-- Set configuration options from template #invoke or defaults.
config = args
maxsigfig = config.maxsigfig or 14 -- maximum number of significant figures
local data_module, text_module
local sandbox = config.sandbox and ('/' .. config.sandbox) or ''
data_module = "Module:Convert/data" .. sandbox
text_module = "Module:Convert/text" .. sandbox
extra_module = "Module:Convert/extra" .. sandbox
wikidata_module = "Module:Convert/wikidata" .. sandbox
wikidata_data_module = "Module:Convert/wikidata/data" .. sandbox
spell_module = "Module:ConvertNumeric"
data_code = mw.loadData(data_module)
text_code = mw.loadData(text_module)
all_units = data_code.all_units
local translation = text_code.translation_table
if translation then
numdot = translation.numdot
numsep = translation.numsep
if numdot == ',' and numsep == '.' then
if text_code.all_messages.cvt_enwiki_num then
to_en = to_en_with_check
end
end
if translation.group then
group_method = translation.group
end
if translation.per_word then
per_word = translation.per_word
end
if translation.plural_suffix then
plural_suffix = translation.plural_suffix
end
varname = translation.varname
from_en_table = translation.from_en
local use_workaround = true
if use_workaround then
-- 2013-07-05 workaround bug by making a copy of the required table.
-- mw.ustring.gsub fails with a table (to_en_table) as the replacement,
-- if the table is accessed via mw.loadData.
local source = translation.to_en
if source then
to_en_table = {}
for k, v in pairs(source) do
to_en_table[k] = v
end
end
else
to_en_table = translation.to_en
end
if translation.lang == 'en default' then
en_default = true -- for hiwiki
end
omitsep = translation.omitsep -- for zhwiki
end
numdot = config.numdot or numdot or '.' -- decimal mark before fractional digits
numsep = config.numsep or numsep or ',' -- group separator for numbers
-- numsep should be ',' or '.' or '' or ' ' or a Unicode character.
-- numsep_remove must work in a regex to identify separators to be removed.
if numsep ~= '' then
numsep_remove = (numsep == '.') and '%.' or numsep
end
if numsep ~= ',' and numdot ~= ',' then
numsep_remove2 = ',' -- so numbers copied from enwiki will work
end
end
local function collection()
-- Return a table to hold items.
return {
n = 0,
add = function (self, item)
self.n = self.n + 1
self[self.n] = item
end,
}
end
local function divide(numerator, denominator)
-- Return integers quotient, remainder resulting from dividing the two
-- given numbers, which should be unsigned integers.
local quotient, remainder = floor(numerator / denominator), numerator % denominator
if not (0 <= remainder and remainder < denominator) then
-- Floating point limits may need this, as in {{convert|160.02|Ym|ydftin}}.
remainder = 0
end
return quotient, remainder
end
local function split(text, delimiter)
-- Return a numbered table with fields from splitting text.
-- The delimiter is used in a regex without escaping (for example, '.' would fail).
-- Each field has any leading/trailing whitespace removed.
local t = {}
text = text .. delimiter -- to get last item
for item in text:gmatch('%s*(.-)%s*' .. delimiter) do
table.insert(t, item)
end
return t
end
local function strip(text)
-- If text is a string, return its content with no leading/trailing
-- whitespace. Otherwise return nil (a nil argument gives a nil result).
if type(text) == 'string' then
return text:match("^%s*(.-)%s*$")
end
end
local function table_len(t)
-- Return length (<100) of a numbered table to replace #t which is
-- documented to not work if t is accessed via mw.loadData().
for i = 1, 100 do
if t[i] == nil then
return i - 1
end
end
end
local function wanted_category(catkey, catsort, want_warning)
-- Return message category if it is wanted in current namespace,
-- otherwise return ''.
local cat
local title = mw.title.getCurrentTitle()
if title then
local nsdefault = '0' -- default namespace: '0' = article; '0,10' = article and template
local namespace = title.namespace
for _, v in ipairs(split(config.nscat or nsdefault, ',')) do
if namespace == tonumber(v) then
cat = text_code.all_categories[want_warning and 'warning' or catkey]
if catsort and catsort ~= '' and cat:sub(-2) == ']]' then
cat = cat:sub(1, -3) .. '|' .. mw.text.nowiki(usub(catsort, 1, 20)) .. ']]'
end
break
end
end
end
return cat or ''
end
local function message(parms, mcode, is_warning)
-- Return wikitext for an error message, including category if specified
-- for the message type.
-- mcode = numbered table specifying the message:
-- mcode[1] = 'cvt_xxx' (string used as a key to get message info)
-- mcode[2] = 'parm1' (string to replace '$1' if any in message)
-- mcode[3] = 'parm2' (string to replace '$2' if any in message)
-- mcode[4] = 'parm3' (string to replace '$3' if any in message)
local msg
if type(mcode) == 'table' then
if mcode[1] == 'cvt_no_output' then
-- Some errors should cause convert to output an empty string,
-- for example, for an optional field in an infobox.
return ''
end
msg = text_code.all_messages[mcode[1]]
end
parms.have_problem = true
local function subparm(fmt, ...)
local rep = {}
for i, v in ipairs({...}) do
rep['$' .. i] = v
end
return (fmt:gsub('$%d+', rep))
end
if msg then
local parts = {}
local regex, replace = msg.regex, msg.replace
for i = 1, 3 do
local limit = 40
local s = mcode[i + 1]
if s then
if regex and replace then
s = s:gsub(regex, replace)
limit = nil -- allow long "should be" messages
end
-- Escape user input so it does not break the message.
-- To avoid tags (like {{convert|1<math>23</math>|m}}) breaking
-- the mouseover title, any strip marker starting with char(127) is
-- replaced with '...' (text not needing i18n).
local append
local pos = s:find(string.char(127), 1, true)
if pos then
append = '...'
s = s:sub(1, pos - 1)
end
if limit and ulen(s) > limit then
s = usub(s, 1, limit)
append = '...'
end
s = mw.text.nowiki(s) .. (append or '')
else
s = '?'
end
parts['$' .. i] = s
end
local function ispreview()
-- Return true if a prominent message should be shown.
if parms.test == 'preview' or parms.test == 'nopreview' then
-- For testing, can preview a real message or simulate a preview
-- when running automated tests.
return parms.test == 'preview'
end
local success, revid = pcall(function ()
return (parms.frame):preprocess('{{REVISIONID}}') end)
return success and (revid == '')
end
local want_warning = is_warning and
not config.warnings and -- show unobtrusive warnings if config.warnings not configured
not msg.nowarn -- but use msg settings, not standard warning, if specified
local title = string.gsub(msg[1] or 'Missing message', '$%d+', parts)
local text = want_warning and '*' or msg[2] or 'Missing message'
local cat = wanted_category(msg[3], mcode[2], want_warning)
local anchor = msg[4] or ''
local fmtkey = ispreview() and 'cvt_format_preview' or
(want_warning and 'cvt_format2' or msg.format or 'cvt_format')
local fmt = text_code.all_messages[fmtkey] or 'convert: bug'
return subparm(fmt, title:gsub('"', '"'), text, cat, anchor)
end
return 'Convert internal error: unknown message'
end
function add_warning(parms, level, key, text1, text2) -- for forward declaration above
-- If enabled, add a warning that will be displayed after the convert result.
-- A higher level is more verbose: more kinds of warnings are displayed.
-- To reduce output noise, only the first warning is displayed.
if level <= (tonumber(config.warnings) or 1) then
if parms.warnings == nil then
parms.warnings = message(parms, { key, text1, text2 }, true)
end
end
end
local function spell_number(parms, inout, number, numerator, denominator)
-- Return result of spelling (number, numerator, denominator), or
-- return nil if spelling is not available or not supported for given text.
-- Examples (each value must be a string or nil):
-- number numerator denominator output
-- ------ --------- ----------- -------------------
-- "1.23" nil nil one point two three
-- "1" "2" "3" one and two thirds
-- nil "2" "3" two thirds
if not speller then
local function get_speller(module)
return require(module).spell_number
end
local success
success, speller = pcall(get_speller, spell_module)
if not success or type(speller) ~= 'function' then
add_warning(parms, 1, 'cvt_no_spell', 'spell')
return nil
end
end
local case
if parms.spell_upper == inout then
case = true
parms.spell_upper = nil -- only uppercase first word in a multiple unit
end
local sp = not parms.opt_sp_us
local adj = parms.opt_adjectival
return speller(number, numerator, denominator, case, sp, adj)
end
------------------------------------------------------------------------
-- BEGIN: Code required only for built-in units.
-- LATER: If need much more code, move to another module to simplify this module.
local function speed_of_sound(altitude)
-- This is for the Mach built-in unit of speed.
-- Return speed of sound in metres per second at given altitude in feet.
-- If no altitude given, use default (zero altitude = sea level).
-- Table gives speed of sound in miles per hour at various altitudes:
-- altitude = -17,499 to 402,499 feet
-- mach_table[a + 4] = s where
-- a = (altitude / 5000) rounded to nearest integer (-3 to 80)
-- s = speed of sound (mph) at that altitude
-- LATER: Should calculate result from an interpolation between the next
-- lower and higher altitudes in table, rather than rounding to nearest.
-- From: http://www.aerospaceweb.org/question/atmosphere/q0112.shtml
local mach_table = { -- a =
799.5, 787.0, 774.2, 761.207051, -- -3 to 0
748.0, 734.6, 721.0, 707.0, 692.8, 678.3, 663.5, 660.1, 660.1, 660.1, -- 1 to 10
660.1, 660.1, 660.1, 662.0, 664.3, 666.5, 668.9, 671.1, 673.4, 675.6, -- 11 to 20
677.9, 683.7, 689.9, 696.0, 702.1, 708.1, 714.0, 719.9, 725.8, 731.6, -- 21 to 30
737.3, 737.7, 737.7, 736.2, 730.5, 724.6, 718.8, 712.9, 707.0, 701.0, -- 31 to 40
695.0, 688.9, 682.8, 676.6, 670.4, 664.1, 657.8, 652.9, 648.3, 643.7, -- 41 to 50
639.1, 634.4, 629.6, 624.8, 620.0, 615.2, 613.2, 613.2, 613.2, 613.5, -- 51 to 60
614.4, 615.3, 616.7, 619.8, 623.4, 629.7, 635.0, 641.1, 650.6, 660.0, -- 61 to 70
672.5, 674.3, 676.1, 677.9, 679.7, 681.5, 683.3, 685.1, 686.8, 688.6, -- 71 to 80
}
altitude = altitude or 0
local a = (altitude < 0) and -altitude or altitude
a = floor(a / 5000 + 0.5)
if altitude < 0 then
a = -a
end
if a < -3 then
a = -3
elseif a > 80 then
a = 80
end
return mach_table[a + 4] * 0.44704 -- mph converted to m/s
end
-- END: Code required only for built-in units.
------------------------------------------------------------------------
local function add_style(parms, class)
-- Add selected template style to parms if not already present.
parms.templatestyles = parms.templatestyles or {}
if not parms.templatestyles[class] then
parms.templatestyles[class] = parms.frame:extensionTag({
name = 'templatestyles', args = { src = text_code.titles[class] }
})
end
end
local function get_styles(parms)
-- Return string of required template styles, empty if none.
if parms.templatestyles then
local t = {}
for _, v in pairs(parms.templatestyles) do
table.insert(t, v)
end
return table.concat(t)
end
return ''
end
local function get_range(word)
-- Return a range (string or table) corresponding to word (like "to"),
-- or return nil if not a range word.
local ranges = text_code.ranges
return ranges.types[word] or ranges.types[ranges.aliases[word]]
end
local function check_mismatch(unit1, unit2)
-- If unit1 cannot be converted to unit2, return an error message table.
-- This allows conversion between units of the same type, and between
-- Nm (normally torque) and ftlb (energy), as in gun-related articles.
-- This works because Nm is the base unit (scale = 1) for both the
-- primary type (torque), and the alternate type (energy, where Nm = J).
-- A match occurs if the primary types are the same, or if unit1 matches
-- the alternate type of unit2, and vice versa. That provides a whitelist
-- of which conversions are permitted between normally incompatible types.
if unit1.utype == unit2.utype or
(unit1.utype == unit2.alttype and unit1.alttype == unit2.utype) then
return nil
end
return { 'cvt_mismatch', unit1.utype, unit2.utype }
end
local function override_from(out_table, in_table, fields)
-- Copy the specified fields from in_table to out_table, but do not
-- copy nil fields (keep any corresponding field in out_table).
for _, field in ipairs(fields) do
if in_table[field] then
out_table[field] = in_table[field]
end
end
end
local function shallow_copy(t)
-- Return a shallow copy of table t.
-- Do not need the features and overhead of the Scribunto mw.clone().
local result = {}
for k, v in pairs(t) do
result[k] = v
end
return result
end
local unit_mt = {
-- Metatable to get missing values for a unit that does not accept SI prefixes.
-- Warning: The boolean value 'false' is returned for any missing field
-- so __index is not called twice for the same field in a given unit.
__index = function (self, key)
local value
if key == 'name1' or key == 'sym_us' then
value = self.symbol
elseif key == 'name2' then
value = self.name1 .. plural_suffix
elseif key == 'name1_us' then
value = self.name1
if not rawget(self, 'name2_us') then
-- If name1_us is 'foot', do not make name2_us by appending plural_suffix.
self.name2_us = self.name2
end
elseif key == 'name2_us' then
local raw1_us = rawget(self, 'name1_us')
if raw1_us then
value = raw1_us .. plural_suffix
else
value = self.name2
end
elseif key == 'link' then
value = self.name1
else
value = false
end
rawset(self, key, value)
return value
end
}
local function prefixed_name(unit, name, index)
-- Return unit name with SI prefix inserted at correct position.
-- index = 1 (name1), 2 (name2), 3 (name1_us), 4 (name2_us).
-- The position is a byte (not character) index, so use Lua's sub().
local pos = rawget(unit, 'prefix_position')
if type(pos) == 'string' then
pos = tonumber(split(pos, ',')[index])
end
if pos then
return name:sub(1, pos - 1) .. unit.si_name .. name:sub(pos)
end
return unit.si_name .. name
end
local unit_prefixed_mt = {
-- Metatable to get missing values for a unit that accepts SI prefixes.
-- Before use, fields si_name, si_prefix must be defined.
-- The unit must define _symbol, _name1 and
-- may define _sym_us, _name1_us, _name2_us
-- (_sym_us, _name2_us may be defined for a language using sp=us
-- to refer to a variant unrelated to U.S. units).
__index = function (self, key)
local value
if key == 'symbol' then
value = self.si_prefix .. self._symbol
if value == 'l' then value = 'L' end
elseif key == 'sym_us' then
value = rawget(self, '_sym_us')
if value then
value = self.si_prefix .. value
else
value = self.symbol
end
elseif key == 'name1' then
value = prefixed_name(self, self._name1, 1)
elseif key == 'name2' then
value = rawget(self, '_name2')
if value then
value = prefixed_name(self, value, 2)
else
value = self.name1 .. plural_suffix
end
elseif key == 'name1_us' then
value = rawget(self, '_name1_us')
if value then
value = prefixed_name(self, value, 3)
else
value = self.name1
end
elseif key == 'name2_us' then
value = rawget(self, '_name2_us')
if value then
value = prefixed_name(self, value, 4)
elseif rawget(self, '_name1_us') then
value = self.name1_us .. plural_suffix
else
value = self.name2
end
elseif key == 'link' then
value = self.name1
else
value = false
end
rawset(self, key, value)
return value
end
}
local unit_per_mt = {
-- Metatable to get values for a per unit of form "x/y".
-- This is never called to determine a unit name or link because per units
-- are handled as a special case.
-- Similarly, the default output is handled elsewhere, and for a symbol
-- this is only called from get_default() for default_exceptions.
__index = function (self, key)
local value
if key == 'symbol' then
local per = self.per
local unit1, unit2 = per[1], per[2]
if unit1 then
value = unit1[key] .. '/' .. unit2[key]
else
value = '/' .. unit2[key]
end
elseif key == 'sym_us' then
value = self.symbol
elseif key == 'scale' then
local per = self.per
local unit1, unit2 = per[1], per[2]
value = (unit1 and unit1.scale or 1) * self.scalemultiplier / unit2.scale
else
value = false
end
rawset(self, key, value)
return value
end
}
local function make_per(unitcode, unit_table, ulookup)
-- Return true, t where t is a per unit with unit codes expanded to unit tables,
-- or return false, t where t is an error message table.
local result = {
unitcode = unitcode,
utype = unit_table.utype,
per = {}
}
override_from(result, unit_table, { 'invert', 'iscomplex', 'default', 'link', 'symbol', 'symlink' })
result.symbol_raw = (result.symbol or false) -- to distinguish between a defined exception and a metatable calculation
local prefix
for i, v in ipairs(unit_table.per) do
if i == 1 and v == '' then
-- First unit symbol can be empty; that gives a nil first unit table.
elseif i == 1 and text_code.currency[v] then
prefix = currency_text or v
else
local success, t = ulookup(v)
if not success then return false, t end
result.per[i] = t
end
end
local multiplier = unit_table.multiplier
if not result.utype then
-- Creating an automatic per unit.
local unit1 = result.per[1]
local utype = (unit1 and unit1.utype or prefix or '') .. '/' .. result.per[2].utype
local t = data_code.per_unit_fixups[utype]
if t then
if type(t) == 'table' then
utype = t.utype or utype
result.link = result.link or t.link
multiplier = multiplier or t.multiplier
else
utype = t
end
end
result.utype = utype
end
result.scalemultiplier = multiplier or 1
result.vprefix = prefix or false -- set to non-nil to avoid calling __index
return true, setmetatable(result, unit_per_mt)
end
local function lookup(parms, unitcode, what, utable, fails, depth)
-- Return true, t where t is a copy of the unit's converter table,
-- or return false, t where t is an error message table.
-- Parameter 'what' determines whether combination units are accepted:
-- 'no_combination' : single unit only
-- 'any_combination' : single unit or combination or output multiple
-- 'only_multiple' : single unit or output multiple only
-- Parameter unitcode is a symbol (like 'g'), with an optional SI prefix (like 'kg').
-- If, for example, 'kg' is in this table, that entry is used;
-- otherwise the prefix ('k') is applied to the base unit ('g').
-- If unitcode is a known combination code (and if allowed by what),
-- a table of output multiple unit tables is included in the result.
-- For compatibility with the old template, an underscore in a unitcode is
-- replaced with a space so usage like {{convert|350|board_feet}} works.
-- Wikignomes may also put two spaces or " " in combinations, so
-- replace underscore, " ", and multiple spaces with a single space.
utable = utable or parms.unittable or all_units
fails = fails or {}
depth = depth and depth + 1 or 1
if depth > 9 then
-- There are ways to mistakenly define units which result in infinite
-- recursion when lookup() is called. That gives a long delay and very
-- confusing error messages, so the depth parameter is used as a guard.
return false, { 'cvt_lookup', unitcode }
end
if unitcode == nil or unitcode == '' then
return false, { 'cvt_no_unit' }
end
unitcode = unitcode:gsub('_', ' '):gsub(' ', ' '):gsub(' +', ' ')
local function call_make_per(t)
return make_per(unitcode, t,
function (ucode) return lookup(parms, ucode, 'no_combination', utable, fails, depth) end
)
end
local t = utable[unitcode]
if t then
if t.shouldbe then
return false, { 'cvt_should_be', t.shouldbe }
end
if t.sp_us then
parms.opt_sp_us = true
end
local target = t.target -- nil, or unitcode is an alias for this target
if target then
local success, result = lookup(parms, target, what, utable, fails, depth)
if not success then return false, result end
override_from(result, t, { 'customary', 'default', 'link', 'symbol', 'symlink' })
local multiplier = t.multiplier
if multiplier then
result.multiplier = tostring(multiplier)
result.scale = result.scale * multiplier
end
return true, result
end
if t.per then
return call_make_per(t)
end
local combo = t.combination -- nil or a table of unitcodes
if combo then
local multiple = t.multiple
if what == 'no_combination' or (what == 'only_multiple' and not multiple) then
return false, { 'cvt_bad_unit', unitcode }
end
-- Recursively create a combination table containing the
-- converter table of each unitcode.
local result = { utype = t.utype, multiple = multiple, combination = {} }
local cvt = result.combination
for i, v in ipairs(combo) do
local success, t = lookup(parms, v, multiple and 'no_combination' or 'only_multiple', utable, fails, depth)
if not success then return false, t end
cvt[i] = t
end
return true, result
end
local result = shallow_copy(t)
result.unitcode = unitcode
if result.prefixes then
result.si_name = ''
result.si_prefix = ''
return true, setmetatable(result, unit_prefixed_mt)
end
return true, setmetatable(result, unit_mt)
end
local SIprefixes = text_code.SIprefixes
for plen = SIprefixes[1] or 2, 1, -1 do
-- Look for an SI prefix; should never occur with an alias.
-- Check for longer prefix first ('dam' is decametre).
-- SIprefixes[1] = prefix maximum #characters (as seen by mw.ustring.sub).
local prefix = usub(unitcode, 1, plen)
local si = SIprefixes[prefix]
if si then
local t = utable[usub(unitcode, plen+1)]
if t and t.prefixes then
local result = shallow_copy(t)
result.unitcode = unitcode
result.si_name = parms.opt_sp_us and si.name_us or si.name
result.si_prefix = si.prefix or prefix
result.scale = t.scale * 10 ^ (si.exponent * t.prefixes)
return true, setmetatable(result, unit_prefixed_mt)
end
end
end
-- Accept user-defined combinations like "acre+m2+ha" or "acre m2 ha" for output.
-- If '+' is used, each unit code can include a space, and any error is fatal.
-- If ' ' is used and if each space-separated word is a unit code, it is a combo,
-- but errors are not fatal so the unit code can be looked up as an extra unit.
local err_is_fatal
local combo = collection()
if unitcode:find('+', 1, true) then
err_is_fatal = true
for item in (unitcode .. '+'):gmatch('%s*(.-)%s*%+') do
if item ~= '' then
combo:add(item)
end
end
elseif unitcode:find('%s') then
for item in unitcode:gmatch('%S+') do
combo:add(item)
end
end
if combo.n > 1 then
local function lookup_combo()
if what == 'no_combination' or what == 'only_multiple' then
return false, { 'cvt_bad_unit', unitcode }
end
local result = { combination = {} }
local cvt = result.combination
for i, v in ipairs(combo) do
local success, t = lookup(parms, v, 'only_multiple', utable, fails, depth)
if not success then return false, t end
if i == 1 then
result.utype = t.utype
else
local mismatch = check_mismatch(result, t)
if mismatch then
return false, mismatch
end
end
cvt[i] = t
end
return true, result
end
local success, result = lookup_combo()
if success or err_is_fatal then
return success, result
end
end
-- Accept any unit with an engineering notation prefix like "e6cuft"
-- (million cubic feet), but not chained prefixes like "e3e6cuft",
-- and not if the unit is a combination or multiple,
-- and not if the unit has an offset or is a built-in.
-- Only en digits are accepted.
local exponent, baseunit = unitcode:match('^e(%d+)(.*)')
if exponent then
local engscale = text_code.eng_scales[exponent]
if engscale then
local success, result = lookup(parms, baseunit, 'no_combination', utable, fails, depth)
if success and not (result.offset or result.builtin or result.engscale) then
result.unitcode = unitcode -- 'e6cuft' not 'cuft'
result.defkey = unitcode -- key to lookup default exception
result.engscale = engscale
result.scale = result.scale * 10 ^ tonumber(exponent)
return true, result
end
end
end
-- Look for x/y; split on right-most slash to get scale correct (x/y/z is x/y per z).
local top, bottom = unitcode:match('^(.-)/([^/]+)$')
if top and not unitcode:find('e%d') then
-- If valid, create an automatic per unit for an "x/y" unit code.
-- The unitcode must not include extraneous spaces.
-- Engineering notation (apart from at start and which has been stripped before here),
-- is not supported so do not make a per unit if find text like 'e3' in unitcode.
local success, result = call_make_per({ per = {top, bottom} })
if success then
return true, result
end
end
if not parms.opt_ignore_error and not get_range(unitcode) then
-- Want the "what links here" list for the extra_module to show only cases
-- where an extra unit is used, so do not require it if invoked from {{val}}
-- or if looking up a range word which cannot be a unit.
if not extra_units then
local success, extra = pcall(function () return require(extra_module).extra_units end)
if success and type(extra) == 'table' then
extra_units = extra
end
end
if extra_units then
-- A unit in one data table might refer to a unit in the other table, so
-- switch between them, relying on fails or depth to terminate loops.
if not fails[unitcode] then
fails[unitcode] = true
local other = (utable == all_units) and extra_units or all_units
local success, result = lookup(parms, unitcode, what, other, fails, depth)
if success then
return true, result
end
end
end
end
if to_en_table then
-- At fawiki it is common to translate all digits so a unit like "km2" becomes "km۲".
local en_code = ustring.gsub(unitcode, '%d', to_en_table)
if en_code ~= unitcode then
return lookup(parms, en_code, what, utable, fails, depth)
end
end
return false, { 'cvt_unknown', unitcode }
end
local function valid_number(num)
-- Return true if num is a valid number.
-- In Scribunto (different from some standard Lua), when expressed as a string,
-- overflow or other problems are indicated with text like "inf" or "nan"
-- which are regarded as invalid here (each contains "n").
if type(num) == 'number' and tostring(num):find('n', 1, true) == nil then
return true
end
end
local function hyphenated(name, parts)
-- Return a hyphenated form of given name (for adjectival usage).
-- The name may be linked and the target of the link must not be changed.
-- Hypothetical examples:
-- [[long ton|ton]] → [[long ton|ton]] (no change)
-- [[tonne|long ton]] → [[tonne|long-ton]]
-- [[metric ton|long ton]] → [[metric ton|long-ton]]
-- [[long ton]] → [[long ton|long-ton]]
-- Input can also have multiple links in a single name like:
-- [[United States customary units|U.S.]] [[US gallon|gallon]]
-- [[mile]]s per [[United States customary units|U.S.]] [[quart]]
-- [[long ton]]s per [[short ton]]
-- Assume that links cannot be nested (never like "[[abc[[def]]ghi]]").
-- This uses a simple and efficient procedure that works for most cases.
-- Some units (if used) would require more, and can later think about
-- adding a method to handle exceptions.
-- The procedure is to replace each space with a hyphen, but
-- not a space after ')' [for "(pre-1954 US) nautical mile"], and
-- not spaces immediately before '(' or in '(...)' [for cases like
-- "British thermal unit (ISO)" and "Calorie (International Steam Table)"].
if name:find(' ', 1, true) then
if parts then
local pos
if name:sub(1, 1) == '(' then
pos = name:find(')', 1, true)
if pos then
return name:sub(1, pos+1) .. name:sub(pos+2):gsub(' ', '-')
end
elseif name:sub(-1) == ')' then
pos = name:find('(', 1, true)
if pos then
return name:sub(1, pos-2):gsub(' ', '-') .. name:sub(pos-1)
end
end
return name:gsub(' ', '-')
end
parts = collection()
for before, item, after in name:gmatch('([^[]*)(%[%[[^[]*%]%])([^[]*)') do
if item:find(' ', 1, true) then
local prefix
local plen = item:find('|', 1, true)
if plen then
prefix = item:sub(1, plen)
item = item:sub(plen + 1, -3)
else
prefix = item:sub(1, -3) .. '|'
item = item:sub(3, -3)
end
item = prefix .. hyphenated(item, parts) .. ']]'
end
parts:add(before:gsub(' ', '-') .. item .. after:gsub(' ', '-'))
end
if parts.n == 0 then
-- No link like "[[...]]" was found in the original name.
parts:add(hyphenated(name, parts))
end
return table.concat(parts)
end
return name
end
local function hyphenated_maybe(parms, want_name, sep, id, inout)
-- Return s, f where
-- s = id, possibly modified
-- f = true if hyphenated
-- Possible modifications: hyphenate; prepend '-'; append mid text.
if id == nil or id == '' then
return ''
end
local mid = (inout == (parms.opt_flip and 'out' or 'in')) and parms.mid or ''
if want_name then
if parms.opt_adjectival then
return '-' .. hyphenated(id) .. mid, true
end
if parms.opt_add_s and id:sub(-1) ~= 's' then
id = id .. 's' -- for nowiki
end
end
return sep .. id .. mid
end
local function use_minus(text)
-- Return text with Unicode minus instead of '-', if present.
if text:sub(1, 1) == '-' then
return MINUS .. text:sub(2)
end
return text
end
local function digit_groups(parms, text, method)
-- Return a numbered table of groups of digits (left-to-right, in local language).
-- Parameter method is a number or nil:
-- 3 for 3-digit grouping (default), or
-- 2 for 3-then-2 grouping (only for digits before decimal mark).
local len_right
local len_left = text:find('.', 1, true)
if len_left then
len_right = #text - len_left
len_left = len_left - 1
else
len_left = #text
end
local twos = method == 2 and len_left > 5
local groups = collection()
local run = len_left
local n
if run < 4 or (run == 4 and parms.opt_comma5) then
if parms.opt_gaps then
n = run
else
n = #text
end
elseif twos then
n = run % 2 == 0 and 1 or 2
else
n = run % 3 == 0 and 3 or run % 3
end
while run > 0 do
groups:add(n)
run = run - n
n = (twos and run > 3) and 2 or 3
end
if len_right then
if groups.n == 0 then
groups:add(0)
end
if parms.opt_gaps and len_right > 3 then
local want4 = not parms.opt_gaps3 -- true gives no gap before trailing single digit
local isfirst = true
run = len_right
while run > 0 do
n = (want4 and run == 4) and 4 or (run > 3 and 3 or run)
if isfirst then
isfirst = false
groups[groups.n] = groups[groups.n] + 1 + n
else
groups:add(n)
end
run = run - n
end
else
groups[groups.n] = groups[groups.n] + 1 + len_right
end
end
local pos = 1
for i, length in ipairs(groups) do
groups[i] = from_en(text:sub(pos, pos + length - 1))
pos = pos + length
end
return groups
end
function with_separator(parms, text) -- for forward declaration above
-- Input text is a number in en digits with optional '.' decimal mark.
-- Return an equivalent, formatted for display:
-- with a custom decimal mark instead of '.', if wanted
-- with thousand separators inserted, if wanted
-- digits in local language
-- The given text is like '123' or '123.' or '12345.6789'.
-- The text has no sign (caller inserts that later, if necessary).
-- When using gaps, they are inserted before and after the decimal mark.
-- Separators are inserted only before the decimal mark.
-- A trailing dot (as in '123.') is removed because their use appears to
-- be accidental, and such a number should be shown as '123' or '123.0'.
-- It is useful for convert to suppress the dot so, for example, '4000.'
-- is a simple way of indicating that all the digits are significant.
if text:sub(-1) == '.' then
text = text:sub(1, -2)
end
if #text < 4 or parms.opt_nocomma or numsep == '' then
return from_en(text)
end
local groups = digit_groups(parms, text, group_method)
if parms.opt_gaps then
if groups.n <= 1 then
return groups[1] or ''
end
local nowrap = '<span style="white-space: nowrap">'
local gap = '<span style="margin-left: 0.25em">'
local close = '</span>'
return nowrap .. groups[1] .. gap .. table.concat(groups, close .. gap, 2, groups.n) .. close .. close
end
return table.concat(groups, numsep)
end
-- An input value like 1.23e12 is displayed using scientific notation (1.23×10¹²).
-- That also makes the output use scientific notation, except for small values.
-- In addition, very small or very large output values use scientific notation.
-- Use format(fmtpower, significand, '10', exponent) where each argument is a string.
local fmtpower = '%s<span style="margin:0 .15em 0 .25em">×</span>%s<sup>%s</sup>'
local function with_exponent(parms, show, exponent)
-- Return wikitext to display the implied value in scientific notation.
-- Input uses en digits; output uses digits in local language.
return format(fmtpower, with_separator(parms, show), from_en('10'), use_minus(from_en(tostring(exponent))))
end
local function make_sigfig(value, sigfig)
-- Return show, exponent that are equivalent to the result of
-- converting the number 'value' (where value >= 0) to a string,
-- rounded to 'sigfig' significant figures.
-- The returned items are:
-- show: a string of digits; no sign and no dot;
-- there is an implied dot before show.
-- exponent: a number (an integer) to shift the implied dot.
-- Resulting value = tonumber('.' .. show) * 10^exponent.
-- Examples:
-- make_sigfig(23.456, 3) returns '235', 2 (.235 * 10^2).
-- make_sigfig(0.0023456, 3) returns '235', -2 (.235 * 10^-2).
-- make_sigfig(0, 3) returns '000', 1 (.000 * 10^1).
if sigfig <= 0 then
sigfig = 1
elseif sigfig > maxsigfig then
sigfig = maxsigfig
end
if value == 0 then
return string.rep('0', sigfig), 1
end
local exp, fracpart = math.modf(log10(value))
if fracpart >= 0 then
fracpart = fracpart - 1
exp = exp + 1
end
local digits = format('%.0f', 10^(fracpart + sigfig))
if #digits > sigfig then
-- Overflow (for sigfig=3: like 0.9999 rounding to "1000"; need "100").
digits = digits:sub(1, sigfig)
exp = exp + 1
end
assert(#digits == sigfig, 'Bug: rounded number has wrong length')
return digits, exp
end
-- Fraction output format.
local fracfmt = {
{ -- Like {{frac}} (fraction slash).
'<span class="frac" role="math">{SIGN}<span class="num">{NUM}</span>⁄<span class="den">{DEN}</span></span>', -- 1/2
'<span class="frac" role="math">{SIGN}{WHOLE}<span class="sr-only">+</span><span class="num">{NUM}</span>⁄<span class="den">{DEN}</span></span>', -- 1+2/3
style = 'frac',
},
{ -- Like {{sfrac}} (stacked fraction, that is, horizontal bar).
'<span class="sfrac tion" role="math">{SIGN}<span class="num">{NUM}</span><span class="sr-only">/</span><span class="den">{DEN}</span></span>', -- 1//2
'<span class="sfrac" role="math">{SIGN}{WHOLE}<span class="sr-only">+</span><span class="tion"><span class="num">{NUM}</span><span class="sr-only">/</span><span class="den">{DEN}</span></span></span>', -- 1+2//3
style = 'sfrac',
},
}
local function format_fraction(parms, inout, negative, wholestr, numstr, denstr, do_spell, style)
-- Return wikitext for a fraction, possibly spelled.
-- Inputs use en digits and have no sign; output uses digits in local language.
local wikitext
if not style then
style = parms.opt_fraction_horizontal and 2 or 1
end
if wholestr == '' then
wholestr = nil
end
local substitute = {
SIGN = negative and MINUS or '',
WHOLE = wholestr and with_separator(parms, wholestr),
NUM = from_en(numstr),
DEN = from_en(denstr),
}
wikitext = fracfmt[style][wholestr and 2 or 1]:gsub('{(%u+)}', substitute)
if do_spell then
if negative then
if wholestr then
wholestr = '-' .. wholestr
else
numstr = '-' .. numstr
end
end
local s = spell_number(parms, inout, wholestr, numstr, denstr)
if s then
return s
end
end
add_style(parms, fracfmt[style].style)
return wikitext
end
local function format_number(parms, show, exponent, isnegative)
-- Parameter show is a string or a table containing strings.
-- Each string is a formatted number in en digits and optional '.' decimal mark.
-- A table represents a fraction: integer, numerator, denominator;
-- if a table is given, exponent must be nil.
-- Return t where t is a table with fields:
-- show = wikitext formatted to display implied value
-- (digits in local language)
-- is_scientific = true if show uses scientific notation
-- clean = unformatted show (possibly adjusted and with inserted '.')
-- (en digits)
-- sign = '' or MINUS
-- exponent = exponent (possibly adjusted)
-- The clean and exponent fields can be used to calculate the
-- rounded absolute value, if needed.
--
-- The value implied by the arguments is found from:
-- exponent is nil; and
-- show is a string of digits (no sign), with an optional dot;
-- show = '123.4' is value 123.4, '1234' is value 1234.0;
-- or:
-- exponent is an integer indicating where dot should be;
-- show is a string of digits (no sign and no dot);
-- there is an implied dot before show;
-- show does not start with '0';
-- show = '1234', exponent = 3 is value 0.1234*10^3 = 123.4.
--
-- The formatted result:
-- * Is for an output value and is spelled if wanted and possible.
-- * Includes a Unicode minus if isnegative and not spelled.
-- * Uses a custom decimal mark, if wanted.
-- * Has digits grouped where necessary, if wanted.
-- * Uses scientific notation if requested, or for very small or large values
-- (which forces result to not be spelled).
-- * Has no more than maxsigfig significant digits
-- (same as old template and {{#expr}}).
local xhi, xlo -- these control when scientific notation (exponent) is used
if parms.opt_scientific then
xhi, xlo = 4, 2 -- default for output if input uses e-notation
elseif parms.opt_scientific_always then
xhi, xlo = 0, 0 -- always use scientific notation (experimental)
else
xhi, xlo = 10, 4 -- default
end
local sign = isnegative and MINUS or ''
local maxlen = maxsigfig
local tfrac
if type(show) == 'table' then
tfrac = show
show = tfrac.wholestr
assert(exponent == nil, 'Bug: exponent given with fraction')
end
if not tfrac and not exponent then
local integer, dot, decimals = show:match('^(%d*)(%.?)(.*)')
if integer == '0' or integer == '' then
local zeros, figs = decimals:match('^(0*)([^0]?.*)')
if #figs == 0 then
if #zeros > maxlen then
show = '0.' .. zeros:sub(1, maxlen)
end
elseif #zeros >= xlo then
show = figs
exponent = -#zeros
elseif #figs > maxlen then
show = '0.' .. zeros .. figs:sub(1, maxlen)
end
elseif #integer >= xhi then
show = integer .. decimals
exponent = #integer
else
maxlen = maxlen + #dot
if #show > maxlen then
show = show:sub(1, maxlen)
end
end
end
if exponent then
local function zeros(n)
return string.rep('0', n)
end
if #show > maxlen then
show = show:sub(1, maxlen)
end
if exponent > xhi or exponent <= -xlo or (exponent == xhi and show ~= '1' .. zeros(xhi - 1)) then
-- When xhi, xlo = 10, 4 (the default), scientific notation is used if the
-- rounded value satisfies: value >= 1e9 or value < 1e-4 (1e9 = 0.1e10),
-- except if show is '1000000000' (1e9), for example:
-- {{convert|1000000000|m|m|sigfig=10}} → 1,000,000,000 metres (1,000,000,000 m)
local significand
if #show > 1 then
significand = show:sub(1, 1) .. '.' .. show:sub(2)
else
significand = show
end
return {
clean = '.' .. show,
exponent = exponent,
sign = sign,
show = sign .. with_exponent(parms, significand, exponent-1),
is_scientific = true,
}
end
if exponent >= #show then
show = show .. zeros(exponent - #show) -- result has no dot
elseif exponent <= 0 then
show = '0.' .. zeros(-exponent) .. show
else
show = show:sub(1, exponent) .. '.' .. show:sub(exponent+1)
end
end
local formatted_show
if tfrac then
show = tostring(tfrac.value) -- to set clean in returned table
formatted_show = format_fraction(parms, 'out', isnegative, tfrac.wholestr, tfrac.numstr, tfrac.denstr, parms.opt_spell_out)
else
if isnegative and show:match('^0.?0*$') then
sign = '' -- don't show minus if result is negative but rounds to zero
end
formatted_show = sign .. with_separator(parms, show)
if parms.opt_spell_out then
formatted_show = spell_number(parms, 'out', sign .. show) or formatted_show
end
end
return {
clean = show,
sign = sign,
show = formatted_show,
is_scientific = false, -- to avoid calling __index
}
end
local function extract_fraction(parms, text, negative)
-- If text represents a fraction, return
-- value, altvalue, show, denominator
-- where
-- value is a number (value of the fraction in argument text)
-- altvalue is an alternate interpretation of any fraction for the hands
-- unit where "12.1+3/4" means 12 hands 1.75 inches
-- show is a string (formatted text for display of an input value,
-- and is spelled if wanted and possible)
-- denominator is value of the denominator in the fraction
-- Otherwise, return nil.
-- Input uses en digits and '.' decimal mark (input has been translated).
-- Output uses digits in local language and local decimal mark, if any.
------------------------------------------------------------------------
-- Originally this function accepted x+y/z where x, y, z were any valid
-- numbers, possibly with a sign. For example '1.23e+2+1.2/2.4' = 123.5,
-- and '2-3/8' = 1.625. However, such usages were found to be errors or
-- misunderstandings, so since August 2014 the following restrictions apply:
-- x (if present) is an integer or has a single digit after decimal mark
-- y and z are unsigned integers
-- e-notation is not accepted
-- The overall number can start with '+' or '-' (so '12+3/4' and '+12+3/4'
-- and '-12-3/4' are valid).
-- Any leading negative sign is removed by the caller, so only inputs
-- like the following are accepted here (may have whitespace):
-- negative = false false true (there was a leading '-')
-- text = '2/3' '+2/3' '2/3'
-- text = '1+2/3' '+1+2/3' '1-2/3'
-- text = '12.3+1/2' '+12.3+1/2' '12.3-1/2'
-- Values like '12.3+1/2' are accepted, but are intended only for use
-- with the hands unit (not worth adding code to enforce that).
------------------------------------------------------------------------
local leading_plus, prefix, numstr, slashes, denstr =
text:match('^%s*(%+?)%s*(.-)%s*(%d+)%s*(/+)%s*(%d+)%s*$')
if not leading_plus then
-- Accept a single U+2044 fraction slash because that may be pasted.
leading_plus, prefix, numstr, denstr =
text:match('^%s*(%+?)%s*(.-)%s*(%d+)%s*⁄%s*(%d+)%s*$')
slashes = '/'
end
local numerator = tonumber(numstr)
local denominator = tonumber(denstr)
if numerator == nil or denominator == nil or (negative and leading_plus ~= '') then
return nil
end
local whole, wholestr
if prefix == '' then
wholestr = ''
whole = 0
else
-- Any prefix must be like '12+' or '12-' (whole number and fraction sign);
-- '12.3+' and '12.3-' are also accepted (single digit after decimal point)
-- because '12.3+1/2 hands' is valid (12 hands 3½ inches).
local num1, num2, frac_sign = prefix:match('^(%d+)(%.?%d?)%s*([+%-])$')
if num1 == nil then return nil end
if num2 == '' then -- num2 must be '' or like '.1' but not '.' or '.12'
wholestr = num1
else
if #num2 ~= 2 then return nil end
wholestr = num1 .. num2
end
if frac_sign ~= (negative and '-' or '+') then return nil end
whole = tonumber(wholestr)
if whole == nil then return nil end
end
local value = whole + numerator / denominator
if not valid_number(value) then return nil end
local altvalue = whole + numerator / (denominator * 10)
local style = #slashes -- kludge: 1 or 2 slashes can be used to select style
if style > 2 then style = 2 end
local wikitext = format_fraction(parms, 'in', negative, leading_plus .. wholestr, numstr, denstr, parms.opt_spell_in, style)
return value, altvalue, wikitext, denominator
end
local function extract_number(parms, text, another, no_fraction)
-- Return true, info if can extract a number from text,
-- where info is a table with the result,
-- or return false, t where t is an error message table.
-- Input can use en digits or digits in local language and can
-- have references at the end. Accepting references is intended
-- for use in infoboxes with a field for a value passed to convert.
-- Parameter another = true if the expected value is not the first.
-- Before processing, the input text is cleaned:
-- * Any thousand separators (valid or not) are removed.
-- * Any sign is replaced with '-' (if negative) or '' (otherwise).
-- That replaces Unicode minus with '-'.
-- If successful, the returned info table contains named fields:
-- value = a valid number
-- altvalue = a valid number, usually same as value but different
-- if fraction used (for hands unit)
-- singular = true if value is 1 or -1 (to use singular form of units)
-- clean = cleaned text with any separators and sign removed
-- (en digits and '.' decimal mark)
-- show = text formatted for output, possibly with ref strip markers
-- (digits in local language and custom decimal mark)
-- The resulting show:
-- * Is for an input value and is spelled if wanted and possible.
-- * Has a rounded value, if wanted.
-- * Has digits grouped where necessary, if wanted.
-- * If negative, a Unicode minus is used; otherwise the sign is
-- '+' (if the input text used '+'), or is '' (if no sign in input).
text = strip(text or '')
local reference
local pos = text:find('\127', 1, true)
if pos then
local before = text:sub(1, pos - 1)
local remainder = text:sub(pos)
local refs = {}
while #remainder > 0 do
local ref, spaces
ref, spaces, remainder = remainder:match('^(\127[^\127]*UNIQ[^\127]*%-ref[^\127]*\127)(%s*)(.*)')
if ref then
table.insert(refs, ref)
else
refs = {}
break
end
end
if #refs > 0 then
text = strip(before)
reference = table.concat(refs)
end
end
local clean = to_en(text, parms)
if clean == '' then
return false, { another and 'cvt_no_num2' or 'cvt_no_num' }
end
local isnegative, propersign = false, '' -- most common case
local singular, show, denominator
local value = tonumber(clean)
local altvalue
if value then
local sign = clean:sub(1, 1)
if sign == '+' or sign == '-' then
propersign = (sign == '+') and '+' or MINUS
clean = clean:sub(2)
end
if value < 0 then
isnegative = true
value = -value
end
else
local valstr
for _, prefix in ipairs({ '-', MINUS, '−' }) do
-- Including '-' sets isnegative in case input is a fraction like '-2-3/4'.
local plen = #prefix
if clean:sub(1, plen) == prefix then
valstr = clean:sub(plen + 1)
if valstr:match('^%s') then -- "- 1" is invalid but "-1 - 1/2" is ok
return false, { 'cvt_bad_num', text }
end
break
end
end
if valstr then
isnegative = true
propersign = MINUS
clean = valstr
value = tonumber(clean)
end
if value == nil then
if not no_fraction then
value, altvalue, show, denominator = extract_fraction(parms, clean, isnegative)
end
if value == nil then
return false, { 'cvt_bad_num', text }
end
if value <= 1 then
singular = true -- for example, "½ mile" or "one half mile" (singular unit)
end
end
end
if not valid_number(value) then -- for example, "1e310" may overflow
return false, { 'cvt_invalid_num' }
end
if show == nil then
-- clean is a non-empty string with no spaces, and does not represent a fraction,
-- and value = tonumber(clean) is a number >= 0.
-- If the input uses e-notation, show will be displayed using a power of ten, but
-- we use the number as given so it might not be normalized scientific notation.
-- The input value is spelled if specified so any e-notation is ignored;
-- that allows input like 2e6 to be spelled as "two million" which works
-- because the spell module converts '2e6' to '2000000' before spelling.
local function rounded(value, default, exponent)
local precision = parms.opt_ri
if precision then
local fmt = '%.' .. format('%d', precision) .. 'f'
local result = fmt:format(tonumber(value) + 2e-14) -- fudge for some common cases of bad rounding
if not exponent then
singular = (tonumber(result) == 1)
end
return result
end
return default
end
singular = (value == 1)
local scientific
local significand, exponent = clean:match('^([%d.]+)[Ee]([+%-]?%d+)')
if significand then
show = with_exponent(parms, rounded(significand, significand, exponent), exponent)
scientific = true
else
show = with_separator(parms, rounded(value, clean))
end
show = propersign .. show
if parms.opt_spell_in then
show = spell_number(parms, 'in', propersign .. rounded(value, clean)) or show
scientific = false
end
if scientific then
parms.opt_scientific = true
end
end
if isnegative and (value ~= 0) then
value = -value
altvalue = -(altvalue or value)
end
return true, {
value = value,
altvalue = altvalue or value,
singular = singular,
clean = clean,
show = show .. (reference or ''),
denominator = denominator,
}
end
local function get_number(text)
-- Return v, f where:
-- v = nil (text is not a number)
-- or
-- v = value of text (text is a number)
-- f = true if value is an integer
-- Input can use en digits or digits in local language or separators,
-- but no Unicode minus, and no fraction.
if text then
local number = tonumber(to_en(text))
if number then
local _, fracpart = math.modf(number)
return number, (fracpart == 0)
end
end
end
local function gcd(a, b)
-- Return the greatest common denominator for the given values,
-- which are known to be positive integers.
if a > b then
a, b = b, a
end
if a <= 0 then
return b
end
local r = b % a
if r <= 0 then
return a
end
if r == 1 then
return 1
end
return gcd(r, a)
end
local function fraction_table(value, denominator)
-- Return value as a string or a table:
-- * If result is a string, there is no fraction, and the result
-- is value formatted as a string of en digits.
-- * If result is a table, it represents a fraction with named fields:
-- wholestr, numstr, denstr (strings of en digits for integer, numerator, denominator).
-- The result is rounded to the nearest multiple of (1/denominator).
-- If the multiple is zero, no fraction is included.
-- No fraction is included if value is very large as the fraction would
-- be unhelpful, particularly if scientific notation is required.
-- Input value is a non-negative number.
-- Input denominator is a positive integer for the desired fraction.
if value <= 0 then
return '0'
end
if denominator <= 0 or value > 1e8 then
return format('%.2f', value)
end
local integer, decimals = math.modf(value)
local numerator = floor((decimals * denominator) +
0.5 + 2e-14) -- add fudge for some common cases of bad rounding
if numerator >= denominator then
integer = integer + 1
numerator = 0
end
local wholestr = tostring(integer)
if numerator > 0 then
local div = gcd(numerator, denominator)
if div > 1 then
numerator = numerator / div
denominator = denominator / div
end
return {
wholestr = (integer > 0) and wholestr or '',
numstr = tostring(numerator),
denstr = tostring(denominator),
value = value,
}
end
return wholestr
end
local function preunits(count, preunit1, preunit2)
-- If count is 1:
-- ignore preunit2
-- return p1
-- else:
-- preunit1 is used for preunit2 if the latter is empty
-- return p1, p2
-- where:
-- p1 is text to insert before the input unit
-- p2 is text to insert before the output unit
-- p1 or p2 may be nil to mean "no preunit"
-- Using '+' gives output like "5+ feet" (no space before, but space after).
local function withspace(text, wantboth)
-- Return text with space before and, if wantboth, after.
-- However, no space is added if there is a space or ' ' or '-'
-- at that position ('-' is for adjectival text).
-- There is also no space if text starts with '&'
-- (e.g. '°' would display a degree symbol with no preceding space).
local char = text:sub(1, 1)
if char == '&' then
return text -- an html entity can be used to specify the exact display
end
if not (char == ' ' or char == '-' or char == '+') then
text = ' ' .. text
end
if wantboth then
char = text:sub(-1, -1)
if not (char == ' ' or char == '-' or text:sub(-6, -1) == ' ') then
text = text .. ' '
end
end
return text
end
local PLUS = '+ '
preunit1 = preunit1 or ''
local trim1 = strip(preunit1)
if count == 1 then
if trim1 == '' then
return nil
end
if trim1 == '+' then
return PLUS
end
return withspace(preunit1, true)
end
preunit1 = withspace(preunit1)
preunit2 = preunit2 or ''
local trim2 = strip(preunit2)
if trim1 == '+' then
if trim2 == '' or trim2 == '+' then
return PLUS, PLUS
end
preunit1 = PLUS
end
if trim2 == '' then
if trim1 == '' then
return nil, nil
end
preunit2 = preunit1
elseif trim2 == '+' then
preunit2 = PLUS
elseif trim2 == ' ' then -- trick to make preunit2 empty
preunit2 = nil
else
preunit2 = withspace(preunit2)
end
return preunit1, preunit2
end
local function range_text(range, want_name, parms, before, after, inout, options)
-- Return before .. rtext .. after
-- where rtext is the text that separates two values in a range.
local rtext, adj_text, exception
options = options or {}
if type(range) == 'table' then
-- Table must specify range text for ('off' and 'on') or ('input' and 'output'),
-- and may specify range text for 'adj=on',
-- and may specify exception = true.
rtext = range[want_name and 'off' or 'on'] or
range[((inout == 'in') == (parms.opt_flip == true)) and 'output' or 'input']
adj_text = range['adj']
exception = range['exception']
else
rtext = range
end
if parms.opt_adjectival then
if want_name or (exception and parms.abbr_org == 'on') then
rtext = adj_text or rtext:gsub(' ', '-'):gsub(' ', '-')
end
end
if rtext == '–' and (options.spaced or after:sub(1, #MINUS) == MINUS) then
rtext = ' – '
end
return before .. rtext .. after
end
local function get_composite(parms, iparm, in_unit_table)
-- Look for a composite input unit. For example, {{convert|1|yd|2|ft|3|in}}
-- would result in a call to this function with
-- iparm = 3 (parms[iparm] = "2", just after the first unit)
-- in_unit_table = (unit table for "yd"; contains value 1 for number of yards)
-- Return true, iparm, unit where
-- iparm = index just after the composite units (7 in above example)
-- unit = composite unit table holding all input units,
-- or return true if no composite unit is present in parms,
-- or return false, t where t is an error message table.
local default, subinfo
local composite_units, count = { in_unit_table }, 1
local fixups = {}
local total = in_unit_table.valinfo[1].value
local subunit = in_unit_table
while subunit.subdivs do -- subdivs is nil or a table of allowed subdivisions
local subcode = strip(parms[iparm+1])
local subdiv = subunit.subdivs[subcode] or subunit.subdivs[(all_units[subcode] or {}).target]
if not subdiv then
break
end
local success
success, subunit = lookup(parms, subcode, 'no_combination')
if not success then return false, subunit end -- should never occur
success, subinfo = extract_number(parms, parms[iparm])
if not success then return false, subinfo end
iparm = iparm + 2
subunit.inout = 'in'
subunit.valinfo = { subinfo }
-- Recalculate total as a number of subdivisions.
-- subdiv[1] = number of subdivisions per previous unit (integer > 1).
total = total * subdiv[1] + subinfo.value
if not default then -- set by the first subdiv with a default defined
default = subdiv.default
end
count = count + 1
composite_units[count] = subunit
if su