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
|
// Copyright 2012 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package colltab
import (
"unicode/utf8"
"golang.org/x/text/unicode/norm"
)
// Table holds all collation data for a given collation ordering.
type Table struct {
Index Trie // main trie
// expansion info
ExpandElem []uint32
// contraction info
ContractTries ContractTrieSet
ContractElem []uint32
MaxContractLen int
VariableTop uint32
}
func (t *Table) AppendNext(w []Elem, b []byte) (res []Elem, n int) {
return t.appendNext(w, source{bytes: b})
}
func (t *Table) AppendNextString(w []Elem, s string) (res []Elem, n int) {
return t.appendNext(w, source{str: s})
}
func (t *Table) Start(p int, b []byte) int {
// TODO: implement
panic("not implemented")
}
func (t *Table) StartString(p int, s string) int {
// TODO: implement
panic("not implemented")
}
func (t *Table) Domain() []string {
// TODO: implement
panic("not implemented")
}
func (t *Table) Top() uint32 {
return t.VariableTop
}
type source struct {
str string
bytes []byte
}
func (src *source) lookup(t *Table) (ce Elem, sz int) {
if src.bytes == nil {
return t.Index.lookupString(src.str)
}
return t.Index.lookup(src.bytes)
}
func (src *source) tail(sz int) {
if src.bytes == nil {
src.str = src.str[sz:]
} else {
src.bytes = src.bytes[sz:]
}
}
func (src *source) nfd(buf []byte, end int) []byte {
if src.bytes == nil {
return norm.NFD.AppendString(buf[:0], src.str[:end])
}
return norm.NFD.Append(buf[:0], src.bytes[:end]...)
}
func (src *source) rune() (r rune, sz int) {
if src.bytes == nil {
return utf8.DecodeRuneInString(src.str)
}
return utf8.DecodeRune(src.bytes)
}
func (src *source) properties(f norm.Form) norm.Properties {
if src.bytes == nil {
return f.PropertiesString(src.str)
}
return f.Properties(src.bytes)
}
// appendNext appends the weights corresponding to the next rune or
// contraction in s. If a contraction is matched to a discontinuous
// sequence of runes, the weights for the interstitial runes are
// appended as well. It returns a new slice that includes the appended
// weights and the number of bytes consumed from s.
func (t *Table) appendNext(w []Elem, src source) (res []Elem, n int) {
ce, sz := src.lookup(t)
tp := ce.ctype()
if tp == ceNormal {
if ce == 0 {
r, _ := src.rune()
const (
hangulSize = 3
firstHangul = 0xAC00
lastHangul = 0xD7A3
)
if r >= firstHangul && r <= lastHangul {
// TODO: performance can be considerably improved here.
n = sz
var buf [16]byte // Used for decomposing Hangul.
for b := src.nfd(buf[:0], hangulSize); len(b) > 0; b = b[sz:] {
ce, sz = t.Index.lookup(b)
w = append(w, ce)
}
return w, n
}
ce = makeImplicitCE(implicitPrimary(r))
}
w = append(w, ce)
} else if tp == ceExpansionIndex {
w = t.appendExpansion(w, ce)
} else if tp == ceContractionIndex {
n := 0
src.tail(sz)
if src.bytes == nil {
w, n = t.matchContractionString(w, ce, src.str)
} else {
w, n = t.matchContraction(w, ce, src.bytes)
}
sz += n
} else if tp == ceDecompose {
// Decompose using NFKD and replace tertiary weights.
t1, t2 := splitDecompose(ce)
i := len(w)
nfkd := src.properties(norm.NFKD).Decomposition()
for p := 0; len(nfkd) > 0; nfkd = nfkd[p:] {
w, p = t.appendNext(w, source{bytes: nfkd})
}
w[i] = w[i].updateTertiary(t1)
if i++; i < len(w) {
w[i] = w[i].updateTertiary(t2)
for i++; i < len(w); i++ {
w[i] = w[i].updateTertiary(maxTertiary)
}
}
}
return w, sz
}
func (t *Table) appendExpansion(w []Elem, ce Elem) []Elem {
i := splitExpandIndex(ce)
n := int(t.ExpandElem[i])
i++
for _, ce := range t.ExpandElem[i : i+n] {
w = append(w, Elem(ce))
}
return w
}
func (t *Table) matchContraction(w []Elem, ce Elem, suffix []byte) ([]Elem, int) {
index, n, offset := splitContractIndex(ce)
scan := t.ContractTries.scanner(index, n, suffix)
buf := [norm.MaxSegmentSize]byte{}
bufp := 0
p := scan.scan(0)
if !scan.done && p < len(suffix) && suffix[p] >= utf8.RuneSelf {
// By now we should have filtered most cases.
p0 := p
bufn := 0
rune := norm.NFD.Properties(suffix[p:])
p += rune.Size()
if rune.LeadCCC() != 0 {
prevCC := rune.TrailCCC()
// A gap may only occur in the last normalization segment.
// This also ensures that len(scan.s) < norm.MaxSegmentSize.
if end := norm.NFD.FirstBoundary(suffix[p:]); end != -1 {
scan.s = suffix[:p+end]
}
for p < len(suffix) && !scan.done && suffix[p] >= utf8.RuneSelf {
rune = norm.NFD.Properties(suffix[p:])
if ccc := rune.LeadCCC(); ccc == 0 || prevCC >= ccc {
break
}
prevCC = rune.TrailCCC()
if pp := scan.scan(p); pp != p {
// Copy the interstitial runes for later processing.
bufn += copy(buf[bufn:], suffix[p0:p])
if scan.pindex == pp {
bufp = bufn
}
p, p0 = pp, pp
} else {
p += rune.Size()
}
}
}
}
// Append weights for the matched contraction, which may be an expansion.
i, n := scan.result()
ce = Elem(t.ContractElem[i+offset])
if ce.ctype() == ceNormal {
w = append(w, ce)
} else {
w = t.appendExpansion(w, ce)
}
// Append weights for the runes in the segment not part of the contraction.
for b, p := buf[:bufp], 0; len(b) > 0; b = b[p:] {
w, p = t.appendNext(w, source{bytes: b})
}
return w, n
}
// TODO: unify the two implementations. This is best done after first simplifying
// the algorithm taking into account the inclusion of both NFC and NFD forms
// in the table.
func (t *Table) matchContractionString(w []Elem, ce Elem, suffix string) ([]Elem, int) {
index, n, offset := splitContractIndex(ce)
scan := t.ContractTries.scannerString(index, n, suffix)
buf := [norm.MaxSegmentSize]byte{}
bufp := 0
p := scan.scan(0)
if !scan.done && p < len(suffix) && suffix[p] >= utf8.RuneSelf {
// By now we should have filtered most cases.
p0 := p
bufn := 0
rune := norm.NFD.PropertiesString(suffix[p:])
p += rune.Size()
if rune.LeadCCC() != 0 {
prevCC := rune.TrailCCC()
// A gap may only occur in the last normalization segment.
// This also ensures that len(scan.s) < norm.MaxSegmentSize.
if end := norm.NFD.FirstBoundaryInString(suffix[p:]); end != -1 {
scan.s = suffix[:p+end]
}
for p < len(suffix) && !scan.done && suffix[p] >= utf8.RuneSelf {
rune = norm.NFD.PropertiesString(suffix[p:])
if ccc := rune.LeadCCC(); ccc == 0 || prevCC >= ccc {
break
}
prevCC = rune.TrailCCC()
if pp := scan.scan(p); pp != p {
// Copy the interstitial runes for later processing.
bufn += copy(buf[bufn:], suffix[p0:p])
if scan.pindex == pp {
bufp = bufn
}
p, p0 = pp, pp
} else {
p += rune.Size()
}
}
}
}
// Append weights for the matched contraction, which may be an expansion.
i, n := scan.result()
ce = Elem(t.ContractElem[i+offset])
if ce.ctype() == ceNormal {
w = append(w, ce)
} else {
w = t.appendExpansion(w, ce)
}
// Append weights for the runes in the segment not part of the contraction.
for b, p := buf[:bufp], 0; len(b) > 0; b = b[p:] {
w, p = t.appendNext(w, source{bytes: b})
}
return w, n
}
|
