Allow more hbox material

This commit is contained in:
Marcel Fabian Krüger 2021-05-01 08:27:52 +02:00
parent 64d5adfb33
commit 912173cab1

View File

@ -1,7 +1,7 @@
local properties = node.get_properties_table() local properties = node.get_properties_table()
local function to_unicode(head, tail) local function to_unicode(head, tail)
local result, i = {}, 0 local result, subresult, i = {[0] = 'mrow'}, {}, 0
local characters, last_fid local characters, last_fid
local iter, state, n = node.traverse(head) local iter, state, n = node.traverse(head)
while true do while true do
@ -10,7 +10,7 @@ local function to_unicode(head, tail)
local props = properties[n] local props = properties[n]
if props and props.glyph_info then if props and props.glyph_info then
i = i+1 i = i+1
result[i] = glyph_info subresult[i] = glyph_info
else else
local char, fid = node.is_glyph(n) local char, fid = node.is_glyph(n)
if char then if char then
@ -23,15 +23,15 @@ local function to_unicode(head, tail)
i = i+1 i = i+1
if uni then if uni then
if type(uni) == 'number' then if type(uni) == 'number' then
result[i] = utf.char(uni) subresult[i] = utf.char(uni)
else else
result[i] = utf.char(table.unpack(uni)) subresult[i] = utf.char(table.unpack(uni))
end end
else else
if char < 0x110000 then if char < 0x110000 then
result[i] = utf.char(char) subresult[i] = utf.char(char)
else else
result[i] = '\u{FFFD}' subresult[i] = '\u{FFFD}'
end end
end end
-- elseif node.id'math' == id then -- elseif node.id'math' == id then
@ -41,23 +41,47 @@ local function to_unicode(head, tail)
elseif node.id'glue' == id then elseif node.id'glue' == id then
if n.width > 1000 then -- FIXME: Coordinate constant with tagpdf if n.width > 1000 then -- FIXME: Coordinate constant with tagpdf
i = i+1 i = i+1
result[i] = ' ' subresult[i] = ' '
end end
elseif node.id'hlist' == id then elseif node.id'hlist' == id then
i = i+1 local nested = to_unicode(n.head)
result[i] = '\u{FFFD}' if nested[0] == 'mtext' and #nested == 1 and type(nested[1]) == 'string' then
i=i+1
subresult[i] = nested[1]
else
if i ~= 0 then
i = 0
result[#result+1] = {[0] = 'mtext', table.concat(subresult)}
end
if nested[0] == 'mrow' then
table.move(nested, 1, #nested, #result+1, result)
else -- should be unreachable
result[#result+1] = nested
end
end
elseif node.id'vlist' == id then elseif node.id'vlist' == id then
i = i+1 i = i+1
result[i] = '\u{FFFD}' subresult[i] = '\u{FFFD}'
elseif node.id'rule' == id then elseif node.id'rule' == id then
if n.width ~= 0 then if n.width ~= 0 then
i = i+1 i = i+1
result[i] = '\u{FFFD}' subresult[i] = '\u{FFFD}'
end end
end -- CHECK: Everything else can probably be ignored, otherwise shout at me end -- CHECK: Everything else can probably be ignored, otherwise shout at me
end end
end end
return {[0] = 'mtext', table.concat(result)} if i ~= 0 then
i = 0
result[#result+1] = {[0] = 'mtext', table.concat(subresult)}
end
if #result == 0 then
local r = {[0] = 'mtext', ''}
return r, r
elseif #result == 1 then
result = result[1]
if result[1] == 'mtext' then return result, result end
end
return result
end end
return to_unicode return to_unicode