require('strict')
local m_data = mw.loadData("Module:Wikt-lang/data/sandbox")
local lang_data = m_data.languages or m_data
local p = {}
local error_msgs = {
internal = {
["get_clean_Wiktionary_page_name"] = "The function get_clean_Wiktionary_page_name requires a page_name.",
["language_code_missing"] = "Name for the language code %q could not be retrieved. Add it to [[Module:Wikt-lang/data]].",
["make_Wiktionary_link"] = "The function make_Wiktionary_link needs a Wiktionary page name, display text and language code.",
},
["no_text"] = "A Wiktionary entry is required.",
["invalid_parameters"] = 'Invalid parameter: |%s=',
}
local cfg = {
template = "Wikt-lang",
valid_parameters = {
[1] = true,
[2] = true,
[3] = true,
["italic"] = true,
},
error_msg = '<span style="color: #d33;">Error: {{%s}}: %s</span>[[Category:%s]]',
category = "Wikt-lang template errors",
namespace = {
appendix = {
name = "Appendix:%s/%s",
data_type = "appendix",
},
reconstruction = {
name = "Reconstruction:%s/%s",
data_type = "reconstructed",
},
},
}
--[[--------------------------< M A K E _ E R R O R >--------------------------------------------------
Creates an error span with the supplied error message and attaches the error category.
]]
local function make_error(msg)
return string.format(cfg.error_msg, cfg.template, msg, cfg.category)
end
--[[--------------------------< A D D _ N A M E S P A C E _ T O _ L I N K >--------------------------------------------------
Returns the page_name with a prefix of a Wiktionary namespace, if relevant.
Current namespaces that can be returned: "Appendix:" and "Reconstruction:".
If not one of the above namespaces, returns the unalerted link_text.
]]
local function add_namespace_to_link(data, language_name, page_name)
if page_name:sub(1, 1) == "*" then
return string.format(cfg.namespace.reconstruction.name, language_name, page_name:sub(2))
end
if data and data.type then
if data.type == cfg.namespace.reconstruction.data_type then
return string.format(cfg.namespace.reconstruction.name, language_name, page_name)
elseif data.type == cfg.namespace.appendix.data_type then
return string.format(cfg.namespace.appendix.name, language_name, page_name)
end
end
-- If for any reason this reaches here, return the unaltered page_name.
return page_name
end
--[[--------------------------< G E T _ L A N G U A G E _ N A M E >--------------------------------------------------
Retrieves the language name.
A langauge is first searched in Module:Wikt-lang/data and if found and has a language name set, returns it.
That database is used to override the language names produced by Module:Lang/data.
If no language is found or the language does not have a language name set, returns the language name from Module:Lang/data.
]]
local function get_Wiktionary_language_name(data, language_name)
if data and data.Wiktionary_name then
return data.Wiktionary_name
end
return language_name
end
--[[--------------------------< M A K E _ W I K T I O N A R Y _ L I N K >--------------------------------------------------
Creates a Wiktionary link.
A page_name, display_text and language_code are always needed. Error if they are missing.
A language name can sometimes be nil if the private code is only listed at Module:Wikt-lang/data and not on Module:Lang/data.
If a language name cannot be retrieved, an erorr is returned.
]]
local function make_Wiktionary_link(page_name, display_text, language_code, language_name)
if not page_name and not display_text and not language_code then
return nil, make_error(error_msgs.internal.make_Wiktionary_link)
end
local data = lang_data[language_code]
language_name = get_Wiktionary_language_name(data, language_name)
if not language_name then
return make_error(error_msgs.language_code_missing)
end
page_name = add_namespace_to_link(data, language_name, page_name)
local link = "[[wikt:%s#%s|%s]]"
return string.format(link, page_name, language_name, display_text)
end
--[[--------------------------< R E P L A C E _ C H A R A C T E R S >--------------------------------------------------
Replaces specific characters as defined in Module:Wikt-lang/data in a language's "replacement" value.
]]
local function replace_characters(data, text)
local replacements = data and data["replacements"]
if replacements == nil then
-- No replacements needed; use text as is.
return text
end
-- Decompose so that the diacritics of characters such
-- as á can be removed in one go.
-- No need to compose at the end, because the MediaWiki software
-- will handle that.
if replacements.decompose then
text = mw.ustring.toNFD(text)
for i, from in ipairs(replacements.from) do
text = mw.ustring.gsub(text, from, replacements.to and replacements.to[i] or "")
end
return text
end
for regex, replacement in pairs(replacements) do
text = mw.ustring.gsub(text, regex, replacement)
end
return text
end
--[[--------------------------< R E M O V E _ B O L D _ I T A L I C >--------------------------------------------------
Removes bold and italics, so that words that contain bolding or emphasis can be linked without piping.
]]
local function remove_bold_italic(text)
if not text then
return text
end
text = text:gsub("\'\'\'", "")
text = text:gsub("\'\'", "")
return text
end
--[[--------------------------< G E T _ C L E A N _ W I K T I O N A R Y _ P A G E _ N A M E >--------------------------------------------------
Returns a clean a Wiktionary page name by removing bold and italics, and by replacing specific characters as defined in Module:Wikt-lang/data.
]]
local function get_clean_Wiktionary_page_name(page_name, language_code)
page_name = tostring(page_name)
if page_name == nil or page_name == "" then
return nil, make_error(error_msgs.internal.get_clean_Wiktionary_page_name)
end
page_name = remove_bold_italic(page_name)
local data = lang_data[language_code]
if data == nil then
-- No language specific data in module; use text as is.
return page_name
end
return replace_characters(data, page_name)
end
--[[--------------------------< C H E C K _ F O R _ U N K N O W N _ P A R A M E T E R S >--------------------------------------------------
Checks that all user-used parameters are valid.
]]
local function check_for_unknown_parameters(args)
for param, _ in pairs(args) do
if not cfg.valid_parameters[param] then
return make_error(string.format(error_msgs.invalid_parameters, param))
end
end
end
--[[--------------------------< M A I N >--------------------------------------------------------------------
Entry point for {{Wikt-lang}}.
Parameters are received from the template's frame (parent frame).
* |1= – language code
* |2= – Wiktionary page name
* |3= – display text
* |italic= – "no" to disable
]]
function p.main(frame)
local getArgs = require('Module:Arguments').getArgs
local args = getArgs(frame)
if not args[2] then
-- A Wiktionary page name is required.
return make_error(error_msgs.no_text)
end
local error_msg = check_for_unknown_parameters(args)
if error_msg then
return error_msg
end
-- For the display text, use args[3] if supplied, if not, use the Wiktionary page name (args[2])
args[2] = args[3] or args[2]
-- To allow the errors to be associated with this template.
args.template = cfg.template
args.error_category = cfg.category
-- Handle the display text html tag.
local lang = require("Module:Lang/sandbox")
local result = lang._wikt_lang(args)
-- An error returned, stop here.
if type(result) == "string" and string.find(result, "Error") then
return result
end
--TODO: Do we need the result to return with a <span title=""> tag?
local page_name, error_msg = get_clean_Wiktionary_page_name(args[2], result.code)
if error_msg then
return error_msg
end
local link, error_msg = make_Wiktionary_link(page_name, result.html, result.code, result.name)
if error_msg then
return error_msg
end
return link .. result.language_categories .. result.maintenance
end
return p
