On 1/27/2014 9:23 PM, Stephan Hennig wrote:
Am 27.01.2014 19:53, schrieb Hans Hagen:
in most cases writing node list manipulation code in lua is efficient
enough (the average page has only a couple of thousands nodes) so then
you have control over that; the more is hardcoded, the more you have to
fight (complex macros at the tex end can be more of a bottleneck)
Fair enough! Still, TeX already has a notion of what a word subject to
hyphenation is. If that could be exposed somehow, that would not be a
bad thing. Anyway, I don't want to push you.
as i assume that you want to loop over the resulting word (range)
anyway, there is not much gain (in fact, you have then a iterator pass
as well as a follow up pass over the word);
btw, i occasionally need work iterators but never felt the need for an
iterator, but here is a quick and dirty one:
1000 * tufte (761 nodes, 544 chars, 93 words)
1874384 nodes/second
1339901 chars/second
229064 chars/second
for first, last, n in node.words(head,true) do
..... -- true checks for lccode
end
so, the time spent in the iterator is much less than the time
typesetting them (esp if the iterator is meant for something high-end)
local glyph = node.id("glyph")
local disc = node.id("disc")
local traverse = node.traverse
local lccode = tex.lccode
function node.words(head,strict)
if not head then
return function() end
end
local next = head
return function()
if not next then
return nil, nil, 0
end
local first
for n in traverse(next) do
local id = n.id
if id == glyph then
if not strict then
first = n
break
elseif lccode[n.char] > 0 then
first = n
break
end
elseif id == disc then
first = n
break
end
end
if not first then
return nil, nil, 0
end
next = first.next
if not next then
return first, first, 1
end
local last = first
local size = 1
for n in traverse(next) do
local id = n.id
if id == glyph then
if not strict then
last = n
size = size + 1
elseif lccode[n.char] > 0 then
last = n
size = size + 1
else
break
end
elseif id == disc then
last = n
else
break
end
end
next = last.next
return first, last, size
end
end
-----------------------------------------------------------------
Hans Hagen | PRAGMA ADE
Ridderstraat 27 | 8061 GH Hasselt | The Netherlands
tel: 038 477 53 69 | voip: 087 875 68 74 | www.pragma-ade.com
| www.pragma-pod.nl
-----------------------------------------------------------------
