local p = {}
local data = mw.loadData( 'Module:Interlinear/data' )
local gloss_override = {} -- for custom gloss abbreviations
local getArgs = require('Module:Arguments').getArgs
local yesno = require('Module:Yesno')
local lang_data = mw.loadData( 'Module:Lang/data' )
--------------------------
-- Almost-global variables
--------------------------
local glossing_type, displaying_messages, free_translation, msg, buffer
-------------------
-- General settings
-------------------
local conf = { --settings
WordSeparator = " \n\r\t", -- Don't replace with %s as this would include non-breaking spaces
GlossAbbrPattern = "^([Ø0-9A-Z]+)$", -- this isn't a full regex, but a Lua pattern
-- NOTE: The following characters must be formatted for use in a pattern set.
GlossAbbrBoundary = "-.,;:<>‹›/\\~+=%?%s%[%]()%_\127'",
GlossExcludeTable = {I = true,}, --strings not be treated as glossing abbreviations
GlossExcludePattern = '^[0-9][0-9]+$', -- excludes strings consisting entirely of digits
GlossSmallCapsExclude = "^[AOPS]$", -- glossing abbreviations matching this pattern will not be rendered in small caps
GlossingType = "label", -- if set to "label" gloss abbreviations are formatted as an <abbr> with the "label" appearing in a tooltip
-- if set to "wikilink" the abbreviation is formatted as a wikilink to the relevant wikipedia article
-- if set to "none" abbreviations aren't formatted at all
ErrorCategory = "[[Category:Pages with errors in interlinear text]]",
AmbiguousGlossCategory = "[[Category:Articles with ambiguous glossing abbreviations]]",
MessageGlossingError = "Error(s) in interlinear glossing",
combining_gender_numbers = "[0-9][0-9]?$", --e.g. G4 '4th gender' or CL7 'class 7'
combining_gender_prefixes = {G = "gender", CL = "class"},
combining_person = {["1"] = "first person", ["2"] = "second person", ["3"] = "third person"},
combining_number = {S = "singular", SG = "singular", P = "plural", PL = "plural", D = "dual", DU = "dual", TRI = "trial"},
combining_gender = {F = "feminine", M = "masculine", N = "neuter"},
LowerCaseGlosses = {["1sg"] = true, ["2sg"] = true, ["3sg"] = true, ["1du"] = true, ["2du"] = true, ["3du"] = true, ["1pl"] = true, ["2pl"] = true,
["3pl"] = true, ["Fsg"] = true, ["Fpl"] = true, ["Msg"] = true, ["Mpl"] = true,}, -- these are the non-all-upper-case strings that will be recognised as glossing abbreviations
ErrorHelpLocation = "Template:Interlinear",
}
---------------------
-- CSS styles and classes
---------------------
conf.style = { --CSS styles
WordDiv = "float: left; margin-bottom: 0.3em;",
WordMargin = "margin-right: 1em;",
WordP = "margin: 0px;", -- the style for the word <p> elements
GlossAbbr = "font-variant: small-caps; font-variant-numeric: oldstyle-nums; text-transform: lowercase; ", -- won't be applied to gloss abbreviations containing lower-case characters
HiddenText = "display: none;",
EndDiv = "clear: left; display: block;", -- style of the <div> element at the end of the interlinear display
ErrorMessage = "font-size: inherit",
}
conf.class = { --CSS classes
Interlinear = "interlinear",
GlossAbbr = "gloss-abbr",
GlossAbbrAmb = "gloss-abbr-ambiguous",
GlossAbbrError = "gloss-abbr-error",
ErrorMessage = "error",
}
---------------------
-- Section transclusion
---------------------
local page_content = nil -- lazy initilization
local function get_section(frame, section_name)
if page_content == nil then
local current_title = mw.title.getCurrentTitle()
page_content = current_title:getContent()
end
if page_content then
if mw.ustring.find(page_content, section_name, 1, true) then
return frame:preprocess('{{#section:{{FULLPAGENAME}}|' .. section_name .. '}}')
end
end
return ''
end
---------------------
-- Sundry small functions
---------------------
local function normalise(str)
return mw.ustring.gsub(str,"[" .. conf.WordSeparator .. "]+"," ")
end
local function tidyCss(str)
str = mw.ustring.gsub(str, '^[\"\']*(.-)[\"\']*$', "%1") -- trims quotation marks
if mw.ustring.sub(str, -1) ~= ";" then str = str .. ";" end -- appends ";" if missing
return str
end
local function highlight(text)
if text then
return '<span style="color:#C00;font-weight:bold;">' .. text .. '</span>'
else return "" end
end
local function tone_sup(str)
return mw.ustring.gsub(str, "([^%p%s0-9])([0-9]+)", "%1<sup>%2</sup>")
end
local function is_empty(str) -- returns "false" if its argument is a string containing chars other than spaces &c.
if not str then return true end
if mw.ustring.find(str, "[^" .. conf.WordSeparator .. "]")
then return false
else return true end
end
local function help_link (anchor)
if anchor then
return " ([[" .. conf.ErrorHelpLocation .. "#" .. anchor .. "|help]])"
else return "" end
end
--------------------
-- The following two functions update the glossing settings based on the received
-- template arguments. set_global_glossing_settings() updates the global settings
-- that are valid for all gloss abbreviations. set_glossing_type()
-- returns the glossing type, which can vary between the different lines.
--------------------
local function set_global_glossing_settings(a)
local style = ""
if a.style then style = tidyCss(a.style) end
if a.underline == "no" then
style = style .. "text-decoration: none;" end
if a.small_caps == "no" then
style = style .. "font-variant:normal; text-transform: none;" end
if style ~= "" then conf.style.GlossAbbr = conf.style.GlossAbbr .. style end
end
local function set_glossing_type(glossing)
if glossing then
local GlossingType
glossing = mw.ustring.lower(mw.text.trim(glossing))
if mw.ustring.find(glossing, 'link') then
GlossingType = "wikilink"
elseif mw.ustring.find(glossing, 'label')
or mw.ustring.find(glossing, 'no link') then
GlossingType = 'label'
elseif mw.ustring.find(glossing, 'no abbr') then
GlossingType = "no abbr"
elseif yesno(glossing) == false then
GlossingType = nil
elseif yesno(glossing) then
GlossingType = conf.GlossingType
else
msg:add('error', 'Glossing type "' .. glossing .. '" not recognised') end
return GlossingType
else error("set_glossing_type: 'glossing' is nil or false", 2)
end
end
local function set_custom_glosses(list)
local abbs = mw.text.split(list, '[;\n\t]')
for _,v in pairs(abbs) do
local gloss = mw.text.split(v, ':')
local a = mw.text.trim(gloss[1])
if a and a ~= "" then
gloss_override[a] = {}
gloss_override[a].expansion = gloss[2]
gloss_override[a].wikipage = gloss[3]
end
end
end
---------------------
-- The UserMessages object contains and processes error messages and warnings
---------------------
local UserMessages = {errors = {}, warnings = {}, gloss_messages = {}}
function UserMessages:add(msgtype, text, gloss)
if msgtype == "gloss_message" then
self.gloss_messages[gloss] = text
elseif msgtype == "warning" then
table.insert(self.warnings, text)
elseif msgtype == "non-repeating error" then
self.errors.nre = text
elseif msgtype == "ambiguous gloss" then
self.if_ambiguous_glosses = true
elseif msgtype == "error" then
table.insert(self.errors, text)
else return error("UserMessages:add(): unknown message type", 2)
end
end
function UserMessages:print_errors()
local out = ""
local namespace = mw.title.getCurrentTitle().namespace
if next(self.errors) or self.warnings[1] then
local err_span = mw.html.create("span")
err_span:attr("style", conf.style.ErrorMessage)
err_span:addClass(conf.class.ErrorMessage)
for _,v in pairs(self.errors) do
err_span:wikitext(" " .. v .. ";") end
if namespace % 2 == 0 and namespace ~= 2 -- non-talk namespaces, excluding user pages; if modifying please update the description on the category page
then err_span:wikitext(conf.ErrorCategory)
end
out = tostring(err_span)
mw.addWarning(conf.MessageGlossingError)
end
if self.if_ambiguous_glosses then
if namespace == 0 -- article namespace
then out = out .. conf.AmbiguousGlossCategory -- this category will only track articles
end
end
return out
end
function UserMessages:print_warnings()
local out = ""
-- Messages and warnings get displayed only if the page is being viewed in "preview" mode:
if displaying_messages and (next(self.gloss_messages) or next(self.warnings)) then
local div = mw.html.create("div")
div:addClass("interlinear-preview-warning")
:cssText('border: 1px solid #a2a9b1; background-color: #f8f9fa; width: 80%; padding: 0.2em;')
:wikitext("<i>This message box is shown only in preview:</i>")
:newline()
for _,v in ipairs(self.warnings) do
local p = div:tag("p")
p:addClass(conf.class.ErrorMessage)
p:attr("style", conf.style.ErrorMessage)
p:wikitext(v)
end
if self.gloss_messages then
div:wikitext("<p> To change any of the following default expansions, see [[Template:Interlinear/doc#Custom abbreviations|the template's documentation]]:</p>")
end
for _,v in pairs(self.gloss_messages) do
div:wikitext("<p>" .. v .. "</p>")
end
out = out .. "\n\n" .. tostring(div)
end
return out
end
---------------------
-- gloss_lookup() receives a gloss abbreviation and tries to uncover its meaning.
---------------------
local function gloss_lookup(a, label, wikilink)
local _label, _wikilink, _lookup, source = nil, nil, nil, nil
if gloss_override[a] then
_lookup = gloss_override[a]
source = "local"
elseif data.abbreviations[a] then _lookup = data.abbreviations[a] end
if _lookup and _lookup.expansion ~= "" then
_label, _wikilink = _lookup.expansion, _lookup.wikipage
else
local prefix = mw.ustring.sub(a,1,1)
local suffix = mw.ustring.sub(a,2)
if conf.combining_person[prefix] then -- is it of the form 1PL or 3FS?
_label = conf.combining_person[prefix]
local _suffix = conf.combining_number[suffix] or conf.combining_gender[suffix]
if _suffix then
_label = _label .. ", " .. _suffix
else
local suffix1 = mw.ustring.sub(suffix,1,1)
local suffix2 = mw.ustring.sub(suffix,2)
if conf.combining_gender[suffix1]
and conf.combining_number[suffix2] then
_label = _label .. ", " .. conf.combining_gender[suffix1] .. ", " .. conf.combining_number[suffix2]
else _label = nil end
end
elseif mw.ustring.match(suffix,conf.combining_gender_numbers) then -- cases like G4 = gender 4
local _i,_j = mw.ustring.find(a, conf.combining_gender_numbers)
local _pre = mw.ustring.sub(a, 1, _i - 1)
local _suff = mw.ustring.sub(a, _i)
if conf.combining_gender_prefixes[_pre] then
_label = conf.combining_gender_prefixes[_pre] .. " " .. _suff
end
elseif prefix == "N" then -- dealing with cases like NPST = non-past
local s = gloss_override[suffix] or data.abbreviations[suffix]
if s ~= nil and not s.ExcludeNegation then
_label = "non-" .. s.expansion
_wikilink = s.wikipage
end
s = nil
end
end
if _label == "" then _label = nil end
if _wikilink == "" then _wikilink = nil end
if not label then label = _label end
if not wikilink then wikilink = _wikilink end
return label, wikilink, source
end
---------------------
-- format_gloss() calls gloss_lookup() to find the meaning of a gloss
-- abbreviation, which it then proceeds to format
---------------------
local function format_gloss(gloss, label, wikilink)
if string.sub(gloss,1,3) == "000" then -- checks for a common component of exposed strip markers (see [[:mw:Strip marker]])
return gloss
end
local gloss2 = mw.ustring.gsub(gloss,"<.->","") -- remove any html fluff
gloss2 = mw.ustring.gsub(gloss2, "%'%'+", "") -- remove wiki bold/italic formatting
gloss2 = mw.text.trim(mw.ustring.upper(gloss2))
if not (label or wikilink)
or (not label and glossing_type == "label")
or (not wikilink and glossing_type == "wikilink")
then
if glossing_type ~= "no abbr"
then label, wikilink, source = gloss_lookup(gloss2, label, wikilink)
end
end
local gloss_node
if glossing_type == "no abbr"
then gloss_node = mw.html.create("span")
else gloss_node = mw.html.create("abbr") end
gloss_node:addClass(conf.class.GlossAbbr)
if label or wikilink then
if not mw.ustring.match(gloss, "%l") -- excluding glosses that contain lower-case characters
and not mw.ustring.match(gloss,conf.GlossSmallCapsExclude) -- and also excluding A, O etc. from rendering in small caps
then gloss_node:attr("style", conf.style.GlossAbbr)
end
local abbr_label
if label then abbr_label = label
else abbr_label = wikilink end
gloss_node:attr("title", abbr_label)
if source ~= "local" and data.abbreviations[gloss2] then
if data.abbreviations[gloss2].ambiguous then
gloss_node:addClass(conf.class.GlossAbbrAmb)
msg:add("ambiguous gloss")
end
end
if glossing_type == "wikilink" and wikilink
then gloss_node:wikitext("[[", wikilink, "|" , gloss, "]]")
else gloss_node:wikitext(gloss) end
if source ~= "local" and displaying_messages then -- logging gloss lookups:
local message = ""
if label then
message = "assuming " .. gloss2 .. " means \"" .. abbr_label .. "\";" end
if glossing_type == "wikilink" and wikilink then
message = message .. " linking to [[" .. wikilink .. "]];"
end
msg:add("gloss_message", message, gloss)
end
elseif glossing_type == "no abbr"
then gloss_node
:attr("style", conf.style.GlossAbbr)
:wikitext(gloss)
else
if displaying_messages then
msg:add("warning", "Gloss abbreviation " .. highlight(gloss2) .. " not recognised" .. help_link("gloss abbr"))
end
msg:add("non-repeating error", "Unknown glossing abbreviation(s)" .. help_link("gloss abbr"))
gloss_node
:addClass(conf.class.GlossAbbrError)
:addClass("error")
:css("font-size", "100%")
:attr("title", gloss2 .. ": glossing abbreviation not found")
:attr("style", conf.style.ErrorMessage)
:wikitext(gloss)
end
return tostring(gloss_node)
end
---------------------
-- find_gloss() parses a word into morphemes, and it calls format_gloss()
-- for anything that looks like a glossing abbreviation.
---------------------
local function find_gloss(word)
local function scan_gloss(boundary, gloss_abbr) -- checks a morpheme if it is a gloss abbreviation
if (mw.ustring.match(gloss_abbr, conf.GlossAbbrPattern)
or conf.LowerCaseGlosses[gloss_abbr])
and not (conf.GlossExcludeTable[gloss_abbr]
or mw.ustring.match(gloss_abbr, conf.GlossExcludePattern))
then gloss_abbr = format_gloss(gloss_abbr)
end
return boundary .. gloss_abbr
end
local word = mw.text.decode(word, true)
if word == "I" -- for the case of the English word "I", the 1SG pronoun
then return word end
local pattern = "([" .. conf.GlossAbbrBoundary .. "]?)([^" .. conf.GlossAbbrBoundary .. "]+)"
word = mw.ustring.gsub(word, pattern, scan_gloss) -- splits into morphemes
return word
end
--------------------
-- The following function is called by Template:gcl and is used for formatting an individual glossing abbreviation
--------------------
function p.gcl(frame)
local args = getArgs(frame,{
trim = true,
removeBlanks = false,
parentOnly = true,
wrappers = {'Template:Gcl'},
})
msg = UserMessages
set_global_glossing_settings{style = args.style, underline = args.underline, small_caps = args['small-caps']}
if not args.glossing then
glossing_type = conf.GlossingType -- a global variable
else glossing_type = set_glossing_type(args.glossing)
end
local gloss, label, wikilink = args[1], args[2], args[3]
if not gloss then UserMessages:add("error", "No gloss supplied")
return UserMessages:print() end
if wikilink and not args.glossing then -- if a wikilink is supplied and glossing isn't set to 'label'...
glossing_type = 'wikilink' end -- .. then the wikilink will be formatted as such
if label == "" then label = nil end
if wikilink == "" then wikilink = nil end
local result = format_gloss(gloss, label, wikilink)
return result
end
return p
