1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
|
if not modules then modules = { } end modules ['char-utf'] = {
version = 1.001,
comment = "companion to char-utf.mkiv",
author = "Hans Hagen, PRAGMA-ADE, Hasselt NL",
copyright = "PRAGMA ADE / ConTeXt Development Team",
license = "see context related readme files"
}
--[[ldx--
<p>When a sequence of <l n='utf'/> characters enters the application, it may
be neccessary to collapse subsequences into their composed variant.</p>
<p>This module implements methods for collapsing and expanding <l n='utf'/>
sequences. We also provide means to deal with characters that are
special to <l n='tex'/> as well as 8-bit characters that need to end up
in special kinds of output (for instance <l n='pdf'/>).</p>
<p>We implement these manipulations as filters. One can run multiple filters
over a string.</p>
--ldx]]--
local utfchar, utfbyte, utfgsub = utf.char, utf.byte, utf.gsub
local concat, gmatch, gsub = table.concat, string.gmatch, string.gsub
local utfcharacters, utfvalues = string.utfcharacters, string.utfvalues
local allocate = utilities.storage.allocate
-- todo: trackers
characters = characters or { }
local characters = characters
characters.graphemes = allocate()
local graphemes = characters.graphemes
characters.mathpairs = allocate()
local mathpairs = characters.mathpairs
characters.filters = allocate()
local filters = characters.filters
filters.utf = filters.utf or { }
local utffilters = characters.filters.utf
utffilters.collapsing = true
utffilters.expanding = true
--[[ldx--
<p>It only makes sense to collapse at runtime, since we don't expect
source code to depend on collapsing.</p>
--ldx]]--
local function initialize()
for k,v in next, characters.data do
-- using vs and first testing for length is faster (.02->.01 s)
local vs = v.specials
if vs and #vs == 3 and vs[1] == 'char' then
local one, two = vs[2], vs[3]
local first, second, combined = utfchar(one), utfchar(two), utfchar(k)
local cgf = graphemes[first]
if not cgf then
cgf = { }
graphemes[first] = cgf
end
cgf[second] = combined
if v.mathclass or v.mathspec then
local mps = mathpairs[two]
if not mps then
mps = { }
mathpairs[two] = mps
end
mps[one] = k
local mps = mathpairs[second]
if not mps then
mps = { }
mathpairs[second] = mps
end
mps[first] = combined
end
end
end
initialize = false
end
-- utffilters.add_grapheme(utfchar(318),'l','\string~')
-- utffilters.add_grapheme('c','a','b')
function utffilters.add_grapheme(result,first,second)
local r, f, s = tonumber(result), tonumber(first), tonumber(second)
if r then result = utfchar(r) end
if f then first = utfchar(f) end
if s then second = utfchar(s) end
if not graphemes[first] then
graphemes[first] = { [second] = result }
else
graphemes[first][second] = result
end
end
function utffilters.collapse(str) -- old one
if utffilters.collapsing and str and #str > 1 then
if initialize then -- saves a call
initialize()
end
local tokens, first, done = { }, false, false
for second in utfcharacters(str) do
local cgf = graphemes[first]
if cgf and cgf[second] then
first, done = cgf[second], true
elseif first then
tokens[#tokens+1] = first
first = second
else
first = second
end
end
if done then
tokens[#tokens+1] = first
return concat(tokens)
end
end
return str
end
--[[ldx--
<p>In order to deal with 8-bit output, we need to find a way to
go from <l n='utf'/> to 8-bit. This is handled in the
<l n='luatex'/> engine itself.</p>
<p>This leaves us problems with characters that are specific to
<l n='tex'/> like <type>{}</type>, <type>$</type> and alike.</p>
<p>We can remap some chars that tex input files are sensitive for to
a private area (while writing to a utility file) and revert then
to their original slot when we read in such a file. Instead of
reverting, we can (when we resolve characters to glyphs) map them
to their right glyph there.</p>
<p>For this purpose we can use the private planes 0x0F0000 and
0x100000.</p>
--ldx]]--
local low = allocate({ })
local high = allocate({ })
local escapes = allocate({ })
local special = "~#$%^&_{}\\|"
local private = {
low = low,
high = high,
escapes = escapes,
}
utffilters.private = private
local function set(ch)
local cb
if type(ch) == "number" then
cb, ch = ch, utfchar(ch)
else
cb = utfbyte(ch)
end
if cb < 256 then
low[ch] = utfchar(0x0F0000 + cb)
high[utfchar(0x0F0000 + cb)] = ch
escapes[ch] = "\\" .. ch
end
end
private.set = set
function private.escape (str) return gsub(str,"(.)", escapes) end
function private.replace(str) return utfgsub(str,"(.)", low ) end
function private.revert (str) return utfgsub(str,"(.)", high ) end
for ch in gmatch(special,".") do set(ch) end
--[[ldx--
<p>We get a more efficient variant of this when we integrate
replacements in collapser. This more or less renders the previous
private code redundant. The following code is equivalent but the
first snippet uses the relocated dollars.</p>
<typing>
[x] [$x$]
</typing>
<p>The next variant has lazy token collecting, on a 140 page mk.tex this saves
about .25 seconds, which is understandable because we have no graphmes and
not collecting tokens is not only faster but also saves garbage collecting.
</p>
--ldx]]--
-- lpeg variant is not faster
function utffilters.collapse(str) -- not really tested (we could preallocate a table)
if utffilters.collapsing and str then
if #str > 1 then
if initialize then -- saves a call
initialize()
end
local tokens, first, done, n = { }, false, false, 0
for second in utfcharacters(str) do
if done then
local crs = high[second]
if crs then
if first then
tokens[#tokens+1] = first
end
first = crs
else
local cgf = graphemes[first]
if cgf and cgf[second] then
first = cgf[second]
elseif first then
tokens[#tokens+1] = first
first = second
else
first = second
end
end
else
local crs = high[second]
if crs then
for s in utfcharacters(str) do
if n == 1 then
break
else
tokens[#tokens+1], n = s, n - 1
end
end
if first then
tokens[#tokens+1] = first
end
first, done = crs, true
else
local cgf = graphemes[first]
if cgf and cgf[second] then
for s in utfcharacters(str) do
if n == 1 then
break
else
tokens[#tokens+1], n = s, n -1
end
end
first, done = cgf[second], true
else
first, n = second, n + 1
end
end
end
end
if done then
tokens[#tokens+1] = first
return concat(tokens) -- seldom called
end
elseif #str > 0 then
return high[str] or str
end
end
return str
end
--[[ldx--
<p>Next we implement some commands that are used in the user interface.</p>
--ldx]]--
commands = commands or { }
--~ function commands.uchar(first,second)
--~ context(utfchar(first*256+second))
--~ end
--[[ldx--
<p>A few helpers (used to be <t>luat-uni<t/>).</p>
--ldx]]--
function utf.split(str)
local t = { }
for snippet in utfcharacters(str) do
t[#t+1] = snippet
end
return t
end
function utf.each(str,fnc)
for snippet in utfcharacters(str) do
fnc(snippet)
end
end
|