mathml/luamml-lr.lua

72 lines
2.2 KiB
Lua
Raw Permalink Normal View History

2021-04-30 19:52:32 +02:00
local properties = node.get_properties_table()
local function to_unicode(head, tail)
2021-06-23 18:49:13 +02:00
local result, subresult, i = {[0] = 'mtext'}, {}, 0
2021-04-30 19:52:32 +02:00
local characters, last_fid
local iter, state, n = node.traverse(head)
while true do
local id, sub n, id, sub = iter(state, n)
if not n or n == tail then break end
local props = properties[n]
if props and props.glyph_info then
i = i+1
2021-06-23 18:49:13 +02:00
result[i] = glyph_info
2021-04-30 19:52:32 +02:00
else
local char, fid = node.is_glyph(n)
if char then
if fid ~= last_fid then
local fontdir = font.getfont(fid)
characters, last_fid = fontdir.characters, fid
end
local uni = characters[char]
local uni = uni and uni.unicode
i = i+1
if uni then
if type(uni) == 'number' then
2021-06-23 18:49:13 +02:00
result[i] = utf.char(uni)
2021-04-30 19:52:32 +02:00
else
2021-06-23 18:49:13 +02:00
result[i] = utf.char(table.unpack(uni))
2021-04-30 19:52:32 +02:00
end
else
if char < 0x110000 then
2021-06-23 18:49:13 +02:00
result[i] = utf.char(char)
2021-04-30 19:52:32 +02:00
else
2021-06-23 18:49:13 +02:00
result[i] = '\u{FFFD}'
2021-04-30 19:52:32 +02:00
end
end
elseif node.id'math' == id then
if props then
2021-05-31 01:54:21 +02:00
local mml = props.saved_mathml_table or props.saved_mathml_core
if mml then
2021-06-23 18:49:13 +02:00
i = i+1
result[i] = mml
n = node.end_of_math(n)
end
end
2021-04-30 19:52:32 +02:00
-- elseif node.id'whatsit' == id then
-- TODO(?)
elseif node.id'glue' == id then
if n.width > 1000 then -- FIXME: Coordinate constant with tagpdf
i = i+1
2021-06-23 18:49:13 +02:00
result[i] = '\u{00A0}' -- non breaking space... There is no real reason why it has to be non breaking, except that MathML often ignore other spaces
2021-04-30 19:52:32 +02:00
end
elseif node.id'hlist' == id then
2021-05-01 08:27:52 +02:00
local nested = to_unicode(n.head)
2021-06-23 18:49:13 +02:00
table.move(nested, 1, #nested, i+1, result)
i = i+#nested
2021-04-30 19:52:32 +02:00
elseif node.id'vlist' == id then
i = i+1
2021-06-23 18:49:13 +02:00
result[i] = '\u{FFFD}'
2021-04-30 19:52:32 +02:00
elseif node.id'rule' == id then
if n.width ~= 0 then
i = i+1
2021-06-23 18:49:13 +02:00
result[i] = '\u{FFFD}'
2021-04-30 19:52:32 +02:00
end
end -- CHECK: Everything else can probably be ignored, otherwise shout at me
end
end
2021-05-01 08:27:52 +02:00
return result
2021-04-30 19:52:32 +02:00
end
return to_unicode