Module:Sandbox/Erutuon: Difference between revisions

Revision as of 20:59, 3 July 2018

Revision as of 21:49, 3 July 2018

Line 354:

return table.concat(m_table.keysToList(language_codes), ', ')

end

local parsed_subtags_mt = {}

parsed_subtags_mt.__index = parsed_subtags_mt

function parsed_subtags_mt:throw(error, index)

self.error = error

self.invalid = table.concat(self.input, "-", index)

-- Only useful internally.

self.input = nil

setmetatable(self, nil)

end

Line 360 ⟶ 371:

-- Based on https://www.w3.org/International/articles/language-tags/.

-- Parse a language tag.

-- Returns nil if tag is not a string or empty.

-- Else returns a table with a map of subtag type to subtag for all subtags that

-- were parsed. If there was an error, returns an "error" field with a

-- description of the error, and an "invalid" field with the suffix of the tag

-- starting at the index where the error occurred.

-- Does not recognize "extension" tags, such as those introduced by "u" and

-- described here, as they are not needed on Wikipedia:

-- https://www.rfc-editor.org/rfc/rfc6067.txt.

function p.parse_IETF(tag)

if not type(tag) or tag =~= "string" or type(tag) ~== "string" then

return nil

end

Line 374 ⟶ 395:

-- hyphen).

local parsed_subtags = {}

setmetatable(parsed_subtags, parsed_subtags_mt)

local matched_count = 0

-- Language tags probably only contain ASCII alphabetic and numerical

-- characters and hyphen-minus, though I am not sure if this is explicitly.

-- stated in the spec.

if not tag:find '^[A-Za-z0-9-]+$' then

parsed_subtags.error = :throw("invalid characters", 1)

parsed_subtags.invalid = tag

return parsed_subtags

end

local subtagspotential_subtags = mw.text.split(tag, "-")

parsed_subtags.errorinput = "invalid subtag"potential_subtags

-- An array of patterns for each subtag, and a "type" field for the name

Line 409 ⟶ 430:

local index = 1

local last_matched_subtag_i = 0

for subtag_i, subtag in ipairs(subtagspotential_subtags) do

local type

local matched = false

Line 439 ⟶ 460:

end

if #subtagspotential_subtags > matched_count then

-- Not all potential subtags were matched. The unmatched tail end of the tag

-- (after the subtag at the index last_matched_subtag_i) is a

-- private-use subtag if it starts with "x". Otherwise, the tag is

-- invalid.

-- Remove the "x-"?

local suffix = table.concat(subtags, "-", last_matched_subtag_i + 1)

-- https://tools.ietf.org/html/bcp47#section-2.2.7

if suffix:sub(1, 1) == "x" then

local private_use_tag_start_indices = {}

parsed_subtags.private_use = suffix

for subtag_i = last_matched_subtag_i + 1, #potential_subtags do

matched_count = matched_count + 1 -- not used after this point

if potential_subtags[subtag_i] == "x" then

else

table.insert(private_use_tag_start_indices, subtag_i)

parsed_subtags.invalid = suffix

end

parsed_subtags.error = "invalid subtag"

end

if not private_use_tag_start_indices[1] then

local suffix = table.concat parsed_subtags:throw(subtags,"invalid "-subtag", last_matched_subtag_i + 1)

return parsed_subtags

end

for i, subtag_index in pairs(private_use_tag_start_indices) do

local next_subtag_index = private_use_tag_start_indices[i + 1] or #potential_subtags

-- Private-use subtags consist of x- followed by a sequence of alphanumeric characters.

if next_subtag_index - subtag_index ~= 2 then

parsed_subtags.:throw("invalid =subtag", suffixsubtag_index)

return parsed_subtags

end

-- Generate private_use table only if needed.

parsed_subtags.private_use = parsed_subtags.private_use or {}

table.insert(parsed_subtags.private_use = suffix,

table.concat(potential_subtags, "-",

subtag_index,

private_use_tag_start_indices[i + 1]

and private_use_tag_start_indices[i + 1] - 1))

end

end

if not (parsed_subtags.language or parsed_subtags.private_use) then

parsed_subtags.error = :throw("no language subtag", 1)

end

-- Only useful internally.

parsed_subtags.invalidinput = tagnil

setmetatable(parsed_subtags, nil)

return parsed_subtags