local p = local data = mw.loadData('Module:Interlinear/data')local gloss_override = -- for custom gloss abbreviationslocal getArgs = require('Module:Arguments').getArgslocal yesno = require('Module:Yesno')local lang_data = mw.loadData('Module:Lang/data')
local gcltest = require('Module:Interlinear/sandbox2/gcl').gcl
---------------------------- Almost-global variables--------------------------local glossing_type, displaying_messages, free_translation, msg, buffer
--------------------- General settings-------------------local conf =
----------------------- CSS styles and classes---------------------conf.style = conf.class = ----------------------- Section transclusion---------------------local page_content = nil -- lazy initilizationlocal function get_section(frame, section_name) if page_content
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 strend
local function highlight(text) if text then return '' .. text .. '' else return "" endend
local function tone_sup(str) return mw.ustring.gsub(str, "([^%p%s0-9])([0-9]+)", "%1%2")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 endend
local function help_link (anchor) if anchor then return " (help)" else return "" endend
---------------------- 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 .. "font-variant:normal; text-transform: none;" end --if style ~= "" then conf.style.GlossAbbr = conf.style.GlossAbbr .. style endend
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)
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 endend
----------------------- The UserMessages object contains and processes error messages and warnings---------------------local UserMessages = function UserMessages:add(msgtype, text, gloss) if msgtype
"warning" then table.insert(self.warnings, text) elseif msgtype
"ambiguous gloss" then self.if_ambiguous_glosses = true elseif msgtype
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
To change any of the following default expansions, see the template's documentation:
") end for _,v in pairs(self.gloss_messages) do div:wikitext("" .. v .. "
") end out = out .. "\n\n" .. tostring(div) end return outend----------------------- 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
"" then _label = nil end if _wikilink
----------------------- 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)
"label") or (not wikilink and glossing_type
"no abbr" then gloss_node = mw.html.create("span") else gloss_node = mw.html.create("abbr") end gloss_node:addClass("gloss-abbr") 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:addClass("gloss_node") --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 message = message .. " linking to " .. wikilink .. ";" end msg:add("gloss_message", message, gloss) end elseif glossing_type
----------------------- 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) -- frame:extensionTag('gcl', gloss_abbr) end return boundary .. gloss_abbr end local word = mw.text.decode(word, true) if word
----------------------- The main purpose of the bletcherous parse is to split a line into words and and then for each eligible word-- to call find_gloss. The parser outputs the individual words (with any gloss abbreviation formatting applied).-- The simple job of splitting at whitespaces has been made complicated by a) the fact that the input can contain-- whitespaces inside the various html elements that are the result of the application of various formatting templates;-- and b) the need to be able to recognise the output of the template that formats custom gloss abbreviations-- (and hence skip passing it on to find_gloss). See talk for a suggestion about its future.---------------------local function parse(cline, i, tags_found,ifglossing)
local function issue_error(message, culprit) UserMessages:add("error", message .. ": " .. mw.ustring.sub(cline.whole, 1, i-1) .. "" .. culprit .. "") end if i > cline.length then return i end --this will only be triggered if the current line has less words than line 1 local next_step, j, _, chunk local probe = mw.ustring.sub(cline.whole,i,i) if mw.ustring.match(probe,"[" .. conf.WordSeparator .. "]") and tags_found
"[" then --Wikilink? if mw.ustring.sub(cline.whole,i+1,i+1) == "[" then _,j,chunk = mw.ustring.find(cline.whole,"(%[%[.-%]%])", i) else chunk = "["; j = i end --not a wikilink then buffer = buffer .. chunk next_step = parse(cline, j+1,tags_found,ifglossing) elseif probe == "{" and tags_found == 0 then --curly brackets enclose a sequence of words to be treated as a single unit _,j,chunk = mw.ustring.find(cline.whole,"(.-)(})", i+1) if not chunk then issue_error("Unclosed curly bracket", "{") chunk = highlight("{"); j = i elseif ifglossing==true then chunk = find_gloss(chunk) else if cline.tone_sup then chunk = tone_sup(chunk) end end buffer = buffer .. chunk next_step = parse(cline, j+1,tags_found,ifglossing) elseif probe == "<" then -- We've encountered an HTML tag. What do we do now? local _,j,chunk = mw.ustring.find(cline.whole,"(<.->)",i) if not chunk then issue_error("Unclosed angle bracket", "<") chunk = highlight("<"); j = i elseif mw.ustring.sub(cline.whole,i,i+1) == "</" then -- It's a CLOSING tag if cline.glossing and ifglossing==false and mw.ustring.match(chunk,"</abbr>") then ifglossing=true end tags_found = tags_found - 1 elseif not mw.ustring.match(chunk, "/>$") -- It's an OPENING tag, unless it opens a self-closing element (in which case the element is ignored) then if ifglossing == true -- the following checks for the output of {{ggl}}: and mw.ustring.find(chunk, conf.class.GlossAbbr, 1, true) -- it's important that the "find" function uses literal strings and not patterns then ifglossing = false end tags_found = tags_found + 1 end buffer = buffer .. chunk next_step = parse(cline, j+1,tags_found,ifglossing) else -- No HTML tags, so we only need to find where the word ends local _,k,chunk = mw.ustring.find(cline.whole,"(..-)([ <[])",i) if k then --ordinary text if ifglossing
true then chunk = find_gloss(mw.ustring.sub(cline.whole,i)) else chunk = mw.ustring.sub(cline.whole,i) if cline.tone_sup then chunk = tone_sup(chunk) end end buffer = buffer .. chunk next_step = cline.length end end return next_stepend---------------------- 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,) msg = UserMessages set_global_glossing_settings 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 wikilink = nil end local result = format_gloss(gloss, label, wikilink) return resultend
----------------------- HTML stuff---------------------local function build_interlinear_html(args, number_of_words, line) local interlinear_wrapper = mw.html.create("div") interlinear_wrapper:addClass("interlinear") -- right-to-left script if yesno(args.rtl)
true then interlinear_wrapper:addClass("box") end
-- numbering and/or indent in the left margin local number, indent = nil, nil if args.number and args.number ~= "" then number = args.number end if args.indent and args.indent ~="" then indent = args.indent end if indent or number then if not indent then indent = "4" end --default value interlinear_wrapper:css("margin-left", indent .. 'em') if number then interlinear_wrapper:tag("div") :addClass("number") :wikitext(args.number) end end --lines to display above the interlinear block if args.top and args.top ~= "" then interlinear_wrapper:tag("div") :addClass("top") :wikitext(args.top) end -- Producing the interlinear block local block_wrapper = interlinear_wrapper:tag("div") :addClass("block_wrapper")
-- non-standard spacing local _spacing = tonumber(args.spacing) if _spacing and _spacing <= 20 then block_wrapper:css('column-gap', _spacing .. 'em') --block_wrapper:css('margin-right', _spacing .. 'em') end
for wi = 1, number_of_words do local block = block_wrapper:tag("div") :addClass("word_block") for i,_ in ipairs (line) do if line[i].whole ~= "" then -- skipping empty lines local p = block:tag("p") p:attr(line[i].attr) if line[i].class then p:addClass(line[i].class) end local _text = line[i].words[wi] if _text
" " then _text = " " end --
elements without content mess up the interlinear display p:wikitext(_text) end end end
--- "comments", added at the end of each line if line.hasComments then local comment_block = block_wrapper:tag("div") :addClass("comment_block") for i,_ in ipairs (line) do local p = comment_block:tag("p") if line[i].c then p:wikitext(line[i].c) else p:wikitext(" ") end end end
--Add hidden lines containing the content of each line of interlinear text -- this is for accessibility for i,v in ipairs(line) do local hidden_line = interlinear_wrapper:tag("p") hidden_line:addClass("hidden_text") :wikitext(v.whole) end
-- Free translation local ft_line = interlinear_wrapper:tag("p") if free_translation and free_translation ~= "" then ft_line:addClass("free_translation") ft_line:wikitext(free_translation) end ft_line:node(msg:print_errors) -- for error messages -- bottom if args.bottom and args.bottom ~= "" then local bottom = interlinear_wrapper:tag('p') :addClass('bottom') :wikitext(args.bottom) end return interlinear_wrapperend
---------------------- The following is the function called by Template:Interlinear.-- It processes the template arguments, then calls parse to split the input lines into words-- and it then builds the output html.--------------------function p.interlinearise(frame)----------------------- Prepare arguments--------------------- local if_auto_translit = false local args = getArgs(frame,) local template_name = frame:getParent:getTitle if template_name
end if args.wordseparator and (args.wordseparator ~= "") then conf.WordSeparator = conf.WordSeparator .. args.wordseparator end local revid = frame:preprocess("") if revid
local function set_italics(n) line[n].class = "italics" line[n].tone_sup = true -- single digits are assumed to be tone markers and will hence be superscripted if args['tone-superscripting'] and not yesno(args['tone-superscripting']) then line[n].tone_sup = false end end
if args.glossing then -- the glossing= parameter sets the default glossing type local _gl = set_glossing_type(args.glossing) if _gl then conf.GlossingType = _gl end end --this looks for a list of glossing abbreviations on the page that transcludes the template: local _ablist_section = get_section(frame, 'list-of-glossing-abbreviations') if _ablist_section and _ablist_section ~= "" then local _a = mw.ustring.gsub(_ablist_section, '', ) -- strips off the div tags set_custom_glosses(_a) end --and this looks looks for a list of abbreviations set within the template: local _ablist = args.abbreviations if _ablist and _ablist ~= "" then set_custom_glosses(_ablist) end local _ablist = args.ablist if _ablist and _ablist ~= "" then set_custom_glosses(_ablist) end
local _spacing = tonumber(args.spacing) if _spacing and _spacing <= 20 then conf.style.WordDiv = 'margin-right: ' .. _spacing .. 'em;' end
local offset, last_line = 0, 0 for j,v in ipairs(args) do -- iterates over the unnamed parameters from the template last_line = last_line +1 if is_empty(v) then offset = offset + 1 else local i = j - offset line[i] = v = normalise(v)
line[i].whole = v line[i].length = mw.ustring.len(v)
local _c = args["c" .. i] if _c and _c ~= "" then line.hasComments = true line[i].c = _c end
---prepare style arguments---- line[i].class = "" local _style = args["style" .. i] if not _style then _style = "" else _style = tidyCss(_style) end --line[i].attr holds the attributes for the
elements that enclose the words in line i line[i].attr =
local _lang = args["lang" .. i] if _lang and #_lang > 1 then line[i].lang = _lang else _lang = args.lang if _lang and #_lang > 1 and i
if yesno(args["italics" .. i]) then set_italics(i) end
local _transl = args["transl" .. i] if _transl and #_transl > 1 then _transl = mw.ustring.lower(_transl) local _lookup = lang_data.translit_title_table[_transl] if _lookup then if _lang and _lookup[_lang] then _transl = _lookup[_lang] else _transl = _lookup.default end if _transl then line[i].attr.title = _transl end else msg:add("error", "Transliteration scheme '" .. _transl .. "' not recognised") end end
local _glossing = args["glossing" .. i] if _glossing then line[i].glossing = set_glossing_type(_glossing) -- Do not treat default glossing settings as custom. if not ((i
2 and yesno(_glossing))) then line.HasCustomGlossing = true end end
local _ipa = args['ipa' .. i] if yesno(_ipa) then line[i].class = "IPA" end
-- formatting classes that can be applied, like "smallcaps" or "bold" local _class = args['class' .. i] if _class then line[i].class = _class line[i].glossing = false end
if line[i].class
local line_count = #line if line_count
1 then msg:add("error", template_name .. ": only 1 line supplied.") return msg:print_errors end
if line_count > 1 then local _italics = args.italics local n = tonumber(_italics) if n and n > 0 then set_italics(n) elseif not (_italics and not yesno(_italics)) and not (args["italics1"] and not yesno(args["italics1"])) then set_italics(1) -- by default, the first line will get italicised, unless italics=no or italics1=no end -- the last unnamed parameter is assumed to be the free translation: free_translation = args[last_line] if not is_empty(free_translation) then line [line_count] = nil end --... and is thus excluded from interlinearising end
-- If glossing isn't specified for any line, then it's chosen by default to occur-- in the second line, unless only a single line has been supplied, in which case-- the assumption is that it is the one containing grammatical glosses if yesno(args.glossing)
1 then line[1].glossing = conf.GlossingType elseif line[2] and line[2].class ~= "" then line[2].glossing = conf.GlossingType end end set_global_glossing_settings
----------------------- Segment lines into words--------------------- for i,v in ipairs(line) do local ifglossing = false if line[i].glossing then ifglossing = true -- if true the parser will attempt to format gloss abbreviations in the current line glossing_type = line[i].glossing -- neccessarily a global variable end local wc, n = 1, 1 line[i].words = while n <= line[i].length do buffer = "" n = parse(line[i], n, 0, ifglossing)+2 line[i].words[wc] = buffer wc = wc + 1 end end
----Check for mismatches in number of words across lines---- local number_of_words, mismatch_found = 0, false for i,v in ipairs(line) do -- find the maximum number of words in any line local wc = #line[i].words if wc ~= number_of_words then if i ~= 1 and wc ~= 0 then mismatch_found = true end if wc > number_of_words then number_of_words = wc end end end ----Deal with mismatches--- if mismatch_found then local error_text = "Mismatch in the number of words between lines: " for i,v in ipairs(line) do local wc = #line[i].words error_text = error_text .. wc .. " word(s) in line " .. i .. ", " if wc ~= number_of_words then for current_word = wc+1, number_of_words do line[i].words[current_word] = " " end end end if string.sub(error_text, -2)
'Template:Fs interlinear/sandbox' and args.script and line[1].words then for i=0,number_of_words do if args.script then line[1].words[i] = frame:expandTemplate end end end
-- Build the HTML local div = build_interlinear_html(args, number_of_words, line)
-- Add categories local temp_track = "" if last_line
return p