Module:character info
Jump to navigation
Jump to search
- The following documentation is located at Module:character info/documentation. [edit]
- Useful links: subpage list • links • transclusions • testcases • sandbox (diff)
This module generates content for {{character info}} and determines the condition under which {{editnotice-exotic symbols}} is displayed when in edit mode in the main namespace (through MediaWiki:Editnotice-0).
To be fixed:
- Code points with labels beginning in
<are given as unassigned (see box for U+007E in ~ and box for U+F900 in 豈).
local m_str_utils = require("Module:string utilities")
local cp = m_str_utils.codepoint
local decode_entities = m_str_utils.decode_entities
local encode_entities = m_str_utils.encode_entities
local floor = math.floor
local gcodepoint = m_str_utils.gcodepoint
local toNFD = mw.ustring.toNFD
local u = m_str_utils.char
local ulen = m_str_utils.len
local m_unicode = require("Module:Unicode data")
local char_to_script = require("Module:scripts").charToScript
local export = {}
local dingbat_scripts = {
["Zsym"] = true;
["Zmth"] = true;
["Zyyy"] = true;
}
local function page_exists(title)
local ok, title_obj = pcall(mw.title.new, title)
if ok and title_obj then
local ok, exists = pcall(function() return title_obj.exists end)
return ok and exists
else
return false
end
end
function export.exotic_symbol_warning(frame)
local title = mw.title.getCurrentTitle()
if title.exists then
return ""
end
if ulen(title.fullText) ~= 1 then
return ""
end
local codepoint = cp(title.fullText)
local script_code = char_to_script(codepoint)
if dingbat_scripts[script_code] then
return frame:expandTemplate { title = "editnotice-exotic symbols" }
end
return ""
end
local function get_codepoint(codepoint, param_name)
codepoint = tonumber(codepoint) or decode_entities(codepoint)
if type(codepoint) == "string" and ulen(codepoint) == 1 then
codepoint = cp(codepoint)
elseif type(codepoint) ~= "number" then
error("Unrecognised string given for the " .. param_name .. " parameter")
end
return codepoint
end
function export._show(args, parent_title)
local codepoint = args.codepoint
local title = mw.title.getCurrentTitle()
local pagename = args.pagename or mw.loadData("Module:headword/data").pagename
local namespace = mw.title.getCurrentTitle().nsText
if codepoint then
codepoint = get_codepoint(codepoint, "codepoint")
else
if not args.pagename and title.fullText == parent_title then
codepoint = 0xfffd
elseif ulen(pagename) == 1 then
codepoint = cp(pagename)
else
error("Page title is not a single Unicode character")
end
end
local image
if args.image == "-" then
image = nil
else
image = args.image or m_unicode.lookup_image(codepoint)
end
local table_markup = {}
table.insert(table_markup,
'{| class="wikitable floatright" style="width:25em;"\n')
if image then
if not image:match("\127") then -- <hiero> tags generate these; pass them through
if image:match("^%[?%[?[Ff]ile:") or image:match("^%[?%[?[Ii]mage:") then
image = image:gsub("^%[%[", ""):gsub("^[Ff]ile:", ""):gsub("^[Ii]mage:", ""):gsub("|.*", ""):gsub("]]", "")
end
local category = "[[Category:Character boxes with images|*" .. string.format("%010d", codepoint) .. "]]"
image = "[[File:" .. image .. "|120x140px]]"
if namespace == "" then
image = image .. category
end
end
table.insert(table_markup,
('|-\n| colspan="2" style="text-align: center;" | %s<br/>%s\n'):format(
image, args.caption or ""
)
)
elseif args.caption then
table.insert(table_markup,
('|-\n| colspan="2" style="text-align: center;" | %s\n'):format(
args.caption
)
)
end
local script_code = args.sc and args.sc:getCode() or char_to_script(codepoint)
local script_data = mw.loadData("Module:scripts/data")[script_code]
or error("No data for script code " .. script_code .. ".")
local script_name = script_data[1]
local NAMESPACE = title.namespace
local cat_name
if not args.nocat and ((NAMESPACE == 0) or (NAMESPACE == 100)) then -- main and Appendix
if script_data.character_category ~= nil then
-- false means no category, overriding the default below
cat_name = script_data.character_category or nil
elseif script_name then
cat_name = script_name .. " script characters"
end
end
local block_name = encode_entities(args.block or m_unicode.lookup_block(codepoint))
local aliases
if args.aliases == "-" then
aliases = nil
else
aliases = mw.loadData("Module:Unicode data/aliases")[codepoint]
end
local function parse_aliases(aliases)
local result = {}
if aliases then
local classif = {}
for i, alias in ipairs(aliases) do
if not classif[alias[1]] then
classif[alias[1]] = {}
end
table.insert(classif[alias[1]], encode_entities(alias[2]))
end
if classif.correction then
for i, name in ipairs(classif.correction) do
local category = "[[Category:Character boxes with corrected names]]"
if namespace == "" then
table.insert(result,
("[[Category:Character boxes with corrected names]]Corrected: %s"):format(
name
)
)
else
table.insert(result,
("Corrected: %s"):format(
name
)
)
end
end
end
if classif.alternate then
for i, name in ipairs(classif.alternate) do
local category = "[[Category:Character boxes with alternative names]]"
if namespace == "" then
table.insert(result,
("[[Category:Character boxes with alternative names]]Alternative: %s"):format(
name
)
)
else
table.insert(result,
("Alternative: %s"):format(
name
)
)
end
end
end
if classif.abbreviation then
local category = "[[Category:Character boxes with abbreviations]]"
if namespace == "" then
table.insert(result,
("[[Category:Character boxes with abbreviations]]Abbreviation: %s"):format(
table.concat(classif.abbreviation, ", ")
)
)
else
table.insert(result,
("Abbreviation: %s"):format(
table.concat(classif.abbreviation, ", ")
)
)
end
end
local parsed_result = table.concat(result, ", ")
return "<div>(" .. parsed_result .. ")</div>"
end
return ""
end
local li, vi, ti = nil, nil, nil
if block_name == "Hangul Syllables" then
local index = codepoint - 0xAC00
li, vi, ti = floor(index / 588), floor((index % 588) / 28), index % 28
end
local initial_to_letter = { [0] =
0x3131, 0x3132, 0x3134, 0x3137, 0x3138, 0x3139, 0x3141, 0x3142,
0x3143, 0x3145, 0x3146, 0x3147, 0x3148, 0x3149, 0x314A, 0x314B,
0x314C, 0x314D, 0x314E,
}
local vowel_to_letter = { [0] =
0x314F, 0x3150, 0x3151, 0x3152, 0x3153, 0x3154, 0x3155, 0x3156,
0x3157, 0x3158, 0x3159, 0x315A, 0x315B, 0x315C, 0x315D, 0x315E,
0x315F, 0x3160, 0x3161, 0x3162, 0x3163,
}
local final_to_letter = {
0x3131, 0x3132, 0x3133, 0x3134, 0x3135, 0x3136, 0x3137, 0x3139,
0x313A, 0x313B, 0x313C, 0x313D, 0x313E, 0x313F, 0x3140, 0x3141,
0x3142, 0x3144, 0x3145, 0x3146, 0x3147, 0x3148, 0x314A, 0x314B,
0x314C, 0x314D, 0x314E,
}
local function parse_composition()
local result = nil
if block_name == "Hangul Syllables" then
result = ((ti ~= 0) and
'<big class="Kore" lang="">[[&#%u;]] + [[&#%u;]] + [[&#%u;]]</big>' or
'<big class="Kore" lang="">[[&#%u;]] + [[&#%u;]]</big>'):format(
initial_to_letter[li],
vowel_to_letter[vi],
final_to_letter[ti]
)
else
local nfd = toNFD(u(codepoint))
if ulen(nfd) ~= 1 then
local compo = {}
for nfdcp in gcodepoint(nfd) do
local dotted_circle = (m_unicode.is_combining(nfdcp) and "◌" or "")
local link_target = m_unicode.get_entry_title(nfdcp)
if not link_target or not page_exists(link_target) then
link_target = nil
end
local script = char_to_script(nfdcp)
local character_text =
link_target and ('[[%s|<span class="%s">%s&#%u;</span> [U+%04X]]]')
:format(link_target, script, dotted_circle, nfdcp, nfdcp)
or ('<span class="%s">%s&#%u;</span> [U+%04X]')
:format(script, dotted_circle, nfdcp, nfdcp)
table.insert(compo, '<span class="character-sample-secondary">' .. character_text .. "</span> ")
end
result = table.concat(compo, " + ")
end
end
if result then
return "Composition", result, "[[Category:Character boxes with compositions]]"
end
return nil
end
-- [[ Egyptian Hieroglyphs
local function parse_gardiner()
if args.gardiner then
local result = (
"[http://vincent.euverte.free.fr/Rosette/Rosette_410.php?Hiero=%s&Lang=E %s]\n"):format(
args.gardiner, args.gardiner
)
return "Gardiner number", result, "[[Category:Character boxes with additional information for Egyptian Hieroglyphs]]"
end
return nil
end
local function parse_mdc()
if args.mdc then
return "Manuel de Codage", args.mdc, "[[Category:Character boxes with additional information for Egyptian Hieroglyphs]]"
end
return nil
end
local function parse_egpz()
if args.egpz then
return "EGPZ 1.0", args.egpz, "[[Category:Character boxes with additional information for Egyptian Hieroglyphs]]"
end
return nil
end
-- ]]
local function middle_part()
local rows = {}
local function insert_row(row_title, row_contents, row_category)
if row_contents then
table.insert(rows,
('<tr><td style="text-align: left">%s:</td><td>%s%s</td></tr>'):format(row_title, row_contents, row_category))
end
end
insert_row(parse_composition())
insert_row(parse_gardiner())
insert_row(parse_egpz())
insert_row(parse_mdc())
if rows[1] then
return ('<table style="margin: 0 auto;">%s</table>')
:format(table.concat(rows, ""))
end
return ""
end
local function present_codepoint(codepoint, np, script, combining, name, printable, title)
local display
local link_target
if combining == nil then
combining = m_unicode.is_combining(codepoint)
end
if printable == nil then
printable = m_unicode.is_printable(codepoint)
end
local char = u(codepoint)
if title == "self" or page_exists(char) then
link_target = char
elseif title ~= "-" then
link_target = m_unicode.get_entry_title(codepoint)
end
if printable then
display = ('<span class="character-sample-secondary %s">%s&#x%04X;</span>'):format(
script and script:getCode() or char_to_script(codepoint),
combining and "◌" or "", codepoint
)
end
local arrow_and_maybe_char
if np then
arrow_and_maybe_char = (display or "") .. " →"
else
arrow_and_maybe_char = "← " .. (display or "")
end
local text = ('<span title="%s">%s<br><small>[U+%04X]</small></span>')
:format(encode_entities(name or m_unicode.lookup_name(codepoint)),
arrow_and_maybe_char, codepoint)
if link_target then
return ("[[" .. link_target .. "|" .. text .. "]]")
else
return text
end
end
local function get_next(codepoint, step)
-- Skip past noncharacters and reserved characters (Cn), private-use
-- characters (Co), surrogates (Cs), and control characters (Cc), all
-- of which have a label beginning in "<" rather than a proper name.
if step < 0 and 0 < codepoint or step > 0 and codepoint < 0x10FFFF then
repeat
codepoint = codepoint + step
until m_unicode.lookup_name(codepoint):sub(1, 1) ~= "<"
or not (0 < codepoint and codepoint < 0x10FFFF)
end
return codepoint
end
local previous_codepoint =
args.previous_codepoint and get_codepoint(args.previous_codepoint, "previous_codepoint")
or get_next(codepoint, -1)
local next_codepoint =
args.next_codepoint and get_codepoint(args.next_codepoint, "next_codepoint")
or get_next(codepoint, 1)
local combining = args.combining
if combining == nil then
combining = m_unicode.is_combining(codepoint)
end
table.insert(table_markup,
'|-\n| style="width: 70px;" colspan="2" | ' ..
"<table>" ..
"<tr>" ..
"<td>" ..
('<span class="character-sample-primary %s">%s&#%u;</span>')
:format(script_code, combining and "◌" or "", codepoint) ..
"</td>" ..
"<td>" ..
(" [https://util.unicode.org/UnicodeJsps/character.jsp?a=%.4X U+%.4X]"):format(codepoint, codepoint) ..
", [[w:List of XML and HTML character entity references|&#" .. codepoint .. ";]]\n" ..
'<div class="character-sample-name">' ..
encode_entities(args.name or m_unicode.lookup_name(codepoint)) ..
"</div>" ..
parse_aliases(aliases) ..
"</td>" ..
"</tr>" ..
"</table>"
)
table.insert(table_markup,
middle_part()
)
local previous_unassigned_first = previous_codepoint + 1
local previous_unassigned_last = codepoint - 1
local next_unassigned_first = codepoint + 1
local next_unassigned_last = next_codepoint - 1
local left_unassigned_text
local right_unassigned_text
if previous_codepoint == 0 then
previous_unassigned_first = 0
end
if previous_unassigned_first <= previous_unassigned_last or next_unassigned_first <= next_unassigned_last then
if previous_unassigned_first < previous_unassigned_last then
left_unassigned_text = ("[unassigned: U+%.4X–U+%.4X]"):format(previous_unassigned_first, previous_unassigned_last)
elseif previous_unassigned_first == previous_unassigned_last then
left_unassigned_text = ("[unassigned: U+%.4X]"):format(previous_unassigned_first)
end
if next_unassigned_first < next_unassigned_last then
right_unassigned_text = ("[unassigned: U+%.4X–U+%.4X]"):format(next_unassigned_first, next_unassigned_last)
elseif next_unassigned_first == next_unassigned_last then
right_unassigned_text = ("[unassigned: U+%.4X]"):format(next_unassigned_first)
end
end
local unassignedsRow =
mw.html.create("table"):css("width", "100%"):css("font-size", "80%"):css("white-space", "nowrap")
:tag("tr")
:tag("td"):css("width", "50%"):css("text-align", "left"):wikitext(left_unassigned_text or ""):done()
:tag("td"):css("width", "50%"):css("text-align", "right"):wikitext(right_unassigned_text or ""):done()
:allDone()
table.insert(table_markup, tostring(unassignedsRow) .."\n")
local previous_codepoint_text = ""
local next_codepoint_text = ("%s\n")
:format(present_codepoint(next_codepoint, true,
args.next_codepoint_sc, args.next_codepoint_combining,
args.next_codepoint_name, args.next_codepoint_printable,
args.next_codepoint_title))
if previous_codepoint > 0 then
previous_codepoint_text = ("%s\n")
:format(present_codepoint(previous_codepoint, false,
args.previous_codepoint_sc, args.previous_codepoint_combining,
args.previous_codepoint_name, args.previous_codepoint_printable,
args.previous_codepoint_title))
end
local block_name_text = ("[[Appendix:Unicode/%s|%s]]")
:format(block_name, block_name)
if namespace == "" then
block_name_text = block_name_text .. ("[[Category:%s block|*%010d]]\n")
:format(block_name, codepoint)
else
block_name_text = block_name_text .. "\n"
end
local lastRow =
mw.html.create("table"):css("width", "100%"):css("text-align", "center")
:tag("tr")
:tag("td"):css("width", "20%"):wikitext(previous_codepoint_text):done()
-- :tag("td"):css("width", "15%")
-- :tag("span"):wikitext(left_unassigned_text and "'''...'''" or ""):attr("title", left_unassigned_text or ""):done():done()
:tag("td"):css("width", "60%"):css("font-size", "110%"):css("font-weight", "bold"):wikitext(block_name_text)
-- :tag("td"):css("width", "15%")
-- :tag("span"):wikitext(right_unassigned_text and "'''...'''" or ""):attr("title", right_unassigned_text or ""):done():done()
:tag("td"):css("width", "20%"):wikitext(next_codepoint_text):done()
:allDone()
table.insert(table_markup, tostring(lastRow) .."\n")
table.insert(table_markup, "|}")
if cat_name and namespace == "" then
table.insert(table_markup, "[[Category:" .. cat_name .. "| " .. u(codepoint) .. "]]")
end
table.insert(table_markup, require("Module:TemplateStyles")("Template:character info/style.css"))
return table.concat(table_markup)
end
function export.show(frame)
local params = {
[1] = {alias_of = "codepoint"},
["codepoint"] = {},
["previous_codepoint"] = {},
["next_codepoint"] = {},
["name"] = {},
["previous_codepoint_name"] = {},
["next_codepoint_name"] = {},
["combining"] = {type = "boolean"},
["previous_codepoint_combining"] = {type = "boolean"},
["next_codepoint_combining"] = {type = "boolean"},
["printable"] = {type = "boolean"},
["previous_codepoint_printable"] = {type = "boolean"},
["next_codepoint_printable"] = {type = "boolean"},
["previous_codepoint_title"] = {},
["next_codepoint_title"] = {},
["sc"] = {type = "script"},
["next_codepoint_sc"] = {type = "script"},
["previous_codepoint_sc"] = {type = "script"},
["caption"] = {},
["aliases"] = {},
["image"] = {},
["block"] = {},
["gardiner"] = {},
["mdc"] = {},
["egpz"] = {},
["nocat"] = {type = "boolean"},
["pagename"] = {}, -- for testing etc.
}
local parent_frame = frame:getParent()
local args = require("Module:parameters").process(parent_frame.args, params)
return export._show(args, parent_frame:getTitle())
end
return export