| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
/* |
|
2
|
|
|
|
|
|
|
* LineBreak.xs - Perl XS glue for Sombok package. |
|
3
|
|
|
|
|
|
|
* |
|
4
|
|
|
|
|
|
|
* Copyright (C) 2009-2013 Hatuka*nezumi - IKEDA Soji . |
|
5
|
|
|
|
|
|
|
* |
|
6
|
|
|
|
|
|
|
* This file is part of the Unicode::LineBreak package. This program is |
|
7
|
|
|
|
|
|
|
* free software; you can redistribute it and/or modify it under the same |
|
8
|
|
|
|
|
|
|
* terms as Perl itself. |
|
9
|
|
|
|
|
|
|
* |
|
10
|
|
|
|
|
|
|
* $Id$ |
|
11
|
|
|
|
|
|
|
*/ |
|
12
|
|
|
|
|
|
|
|
|
13
|
|
|
|
|
|
|
#include "EXTERN.h" |
|
14
|
|
|
|
|
|
|
#include "perl.h" |
|
15
|
|
|
|
|
|
|
#include "XSUB.h" |
|
16
|
|
|
|
|
|
|
#define NEED_newRV_noinc |
|
17
|
|
|
|
|
|
|
#define NEED_sv_2pv_flags |
|
18
|
|
|
|
|
|
|
#define NEED_newRV_noinc_GLOBAL |
|
19
|
|
|
|
|
|
|
#define NEED_sv_2pv_flags_GLOBAL |
|
20
|
|
|
|
|
|
|
#define NEED_sv_2pv_nolen |
|
21
|
|
|
|
|
|
|
#include "ppport.h" |
|
22
|
|
|
|
|
|
|
#include "sombok.h" |
|
23
|
|
|
|
|
|
|
|
|
24
|
|
|
|
|
|
|
/* for Win32 with Visual Studio (MSVC) */ |
|
25
|
|
|
|
|
|
|
#ifdef _MSC_VER |
|
26
|
|
|
|
|
|
|
# ifndef snprintf |
|
27
|
|
|
|
|
|
|
# define snprintf _snprintf |
|
28
|
|
|
|
|
|
|
# endif /* snprintf */ |
|
29
|
|
|
|
|
|
|
# define strcasecmp _stricmp |
|
30
|
|
|
|
|
|
|
#endif /* _MSC_VER */ |
|
31
|
|
|
|
|
|
|
|
|
32
|
|
|
|
|
|
|
/* Type synonyms for typemap. */ |
|
33
|
|
|
|
|
|
|
typedef IV swapspec_t; |
|
34
|
|
|
|
|
|
|
typedef gcstring_t *generic_string; |
|
35
|
|
|
|
|
|
|
|
|
36
|
|
|
|
|
|
|
/*** |
|
37
|
|
|
|
|
|
|
*** Data conversion. |
|
38
|
|
|
|
|
|
|
***/ |
|
39
|
|
|
|
|
|
|
|
|
40
|
|
|
|
|
|
|
/* |
|
41
|
|
|
|
|
|
|
* Create Unicode string from Perl utf8-flagged string. |
|
42
|
|
|
|
|
|
|
*/ |
|
43
|
|
|
|
|
|
|
static |
|
44
|
11236
|
|
|
|
|
|
unistr_t *SVtounistr(unistr_t *buf, SV *str) |
|
45
|
|
|
|
|
|
|
{ |
|
46
|
|
|
|
|
|
|
U8 *utf8, *utf8ptr; |
|
47
|
|
|
|
|
|
|
STRLEN utf8len, unilen, len; |
|
48
|
|
|
|
|
|
|
unichar_t *uniptr; |
|
49
|
|
|
|
|
|
|
|
|
50
|
11236
|
50
|
|
|
|
|
if (buf == NULL) { |
|
51
|
0
|
0
|
|
|
|
|
if ((buf = malloc(sizeof(unistr_t))) == NULL) |
|
52
|
0
|
|
|
|
|
|
croak("SVtounistr: %s", strerror(errno)); |
|
53
|
11236
|
50
|
|
|
|
|
} else if (buf->str) |
|
54
|
0
|
|
|
|
|
|
free(buf->str); |
|
55
|
11236
|
|
|
|
|
|
buf->str = NULL; |
|
56
|
11236
|
|
|
|
|
|
buf->len = 0; |
|
57
|
|
|
|
|
|
|
|
|
58
|
11236
|
50
|
|
|
|
|
if (SvOK(str)) |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
59
|
11236
|
50
|
|
|
|
|
utf8 = (U8 *)SvPV(str, utf8len); |
|
60
|
|
|
|
|
|
|
else |
|
61
|
0
|
|
|
|
|
|
return buf; |
|
62
|
11236
|
50
|
|
|
|
|
if (utf8len <= 0) |
|
63
|
0
|
|
|
|
|
|
return buf; |
|
64
|
11236
|
|
|
|
|
|
unilen = utf8_length(utf8, utf8 + utf8len); |
|
65
|
11236
|
50
|
|
|
|
|
if ((buf->str = (unichar_t *)malloc(sizeof(unichar_t) * unilen)) == NULL) |
|
66
|
0
|
|
|
|
|
|
croak("SVtounistr: %s", strerror(errno)); |
|
67
|
|
|
|
|
|
|
|
|
68
|
11236
|
|
|
|
|
|
utf8ptr = utf8; |
|
69
|
11236
|
|
|
|
|
|
uniptr = buf->str; |
|
70
|
98265
|
100
|
|
|
|
|
while (utf8ptr < utf8 + utf8len) { |
|
71
|
|
|
|
|
|
|
#if PERL_VERSION >= 20 || (PERL_VERSION == 19 && PERL_SUBVERSION >= 4) |
|
72
|
87029
|
50
|
|
|
|
|
*uniptr = (unichar_t) NATIVE_TO_UNI( |
|
73
|
|
|
|
|
|
|
utf8_to_uvchr_buf(utf8ptr, utf8 + utf8len, &len)); |
|
74
|
|
|
|
|
|
|
#elif PERL_VERSION >= 16 || (PERL_VERSION == 15 && PERL_SUBVERSION >= 9) |
|
75
|
|
|
|
|
|
|
*uniptr = (unichar_t) utf8_to_uvuni_buf(utf8ptr, utf8 + utf8len, |
|
76
|
|
|
|
|
|
|
&len); |
|
77
|
|
|
|
|
|
|
#else |
|
78
|
|
|
|
|
|
|
*uniptr = (unichar_t) utf8n_to_uvuni(utf8ptr, |
|
79
|
|
|
|
|
|
|
utf8 + utf8len - utf8ptr, &len, |
|
80
|
|
|
|
|
|
|
ckWARN(WARN_UTF8) ? 0 : |
|
81
|
|
|
|
|
|
|
UTF8_ALLOW_ANY); |
|
82
|
|
|
|
|
|
|
#endif |
|
83
|
|
|
|
|
|
|
if (len < 0) { |
|
84
|
|
|
|
|
|
|
free(buf->str); |
|
85
|
|
|
|
|
|
|
buf->str = NULL; |
|
86
|
|
|
|
|
|
|
buf->len = 0; |
|
87
|
|
|
|
|
|
|
croak("SVtounistr: Not well-formed UTF-8"); |
|
88
|
|
|
|
|
|
|
} |
|
89
|
87029
|
50
|
|
|
|
|
if (len == 0) { |
|
90
|
0
|
|
|
|
|
|
free(buf->str); |
|
91
|
0
|
|
|
|
|
|
buf->str = NULL; |
|
92
|
0
|
|
|
|
|
|
buf->len = 0; |
|
93
|
0
|
|
|
|
|
|
croak("SVtounistr: Internal error"); |
|
94
|
|
|
|
|
|
|
} |
|
95
|
87029
|
|
|
|
|
|
utf8ptr += len; |
|
96
|
87029
|
|
|
|
|
|
uniptr++; |
|
97
|
|
|
|
|
|
|
} |
|
98
|
11236
|
|
|
|
|
|
buf->len = unilen; |
|
99
|
11236
|
|
|
|
|
|
return buf; |
|
100
|
|
|
|
|
|
|
} |
|
101
|
|
|
|
|
|
|
|
|
102
|
|
|
|
|
|
|
/* |
|
103
|
|
|
|
|
|
|
* Create Unicode string from Perl string NOT utf8-flagged. |
|
104
|
|
|
|
|
|
|
*/ |
|
105
|
|
|
|
|
|
|
static |
|
106
|
35
|
|
|
|
|
|
unistr_t *SVupgradetounistr(unistr_t *buf, SV *str) |
|
107
|
|
|
|
|
|
|
{ |
|
108
|
|
|
|
|
|
|
char *s; |
|
109
|
|
|
|
|
|
|
size_t len, i; |
|
110
|
|
|
|
|
|
|
|
|
111
|
35
|
50
|
|
|
|
|
if (buf == NULL) { |
|
112
|
0
|
0
|
|
|
|
|
if ((buf = malloc(sizeof(unistr_t))) == NULL) |
|
113
|
0
|
|
|
|
|
|
croak("SVupgradetounistr: %s", strerror(errno)); |
|
114
|
35
|
50
|
|
|
|
|
} else if (buf->str) |
|
115
|
0
|
|
|
|
|
|
free(buf->str); |
|
116
|
35
|
|
|
|
|
|
buf->str = NULL; |
|
117
|
35
|
|
|
|
|
|
buf->len = 0; |
|
118
|
|
|
|
|
|
|
|
|
119
|
35
|
50
|
|
|
|
|
if (SvOK(str)) |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
120
|
35
|
100
|
|
|
|
|
s = SvPV(str, len); |
|
121
|
|
|
|
|
|
|
else |
|
122
|
0
|
|
|
|
|
|
return buf; |
|
123
|
35
|
50
|
|
|
|
|
if (len == 0) |
|
124
|
0
|
|
|
|
|
|
return buf; |
|
125
|
35
|
50
|
|
|
|
|
if ((buf->str = malloc(sizeof(unichar_t) * len)) == NULL) |
|
126
|
0
|
|
|
|
|
|
croak("SVupgradetounistr: %s", strerror(errno)); |
|
127
|
|
|
|
|
|
|
|
|
128
|
97
|
100
|
|
|
|
|
for (i = 0; i < len; i++) |
|
129
|
62
|
|
|
|
|
|
buf->str[i] = (unichar_t)(unsigned char)s[i]; |
|
130
|
35
|
|
|
|
|
|
buf->len = len; |
|
131
|
35
|
|
|
|
|
|
return buf; |
|
132
|
|
|
|
|
|
|
} |
|
133
|
|
|
|
|
|
|
|
|
134
|
|
|
|
|
|
|
/* |
|
135
|
|
|
|
|
|
|
* Create Perl utf8-flagged string from Unicode string. |
|
136
|
|
|
|
|
|
|
*/ |
|
137
|
|
|
|
|
|
|
static |
|
138
|
14560
|
|
|
|
|
|
SV *unistrtoSV(unistr_t *unistr, size_t uniidx, size_t unilen) |
|
139
|
|
|
|
|
|
|
{ |
|
140
|
14560
|
|
|
|
|
|
U8 *buf = NULL, *newbuf; |
|
141
|
|
|
|
|
|
|
STRLEN utf8len; |
|
142
|
|
|
|
|
|
|
unichar_t *uniptr; |
|
143
|
|
|
|
|
|
|
SV *utf8; |
|
144
|
|
|
|
|
|
|
|
|
145
|
14560
|
50
|
|
|
|
|
if (unistr == NULL || unistr->str == NULL || unilen == 0) { |
|
|
|
100
|
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
146
|
2230
|
|
|
|
|
|
utf8 = newSVpvn("", 0); |
|
147
|
2230
|
|
|
|
|
|
SvUTF8_on(utf8); |
|
148
|
2230
|
|
|
|
|
|
return utf8; |
|
149
|
|
|
|
|
|
|
} |
|
150
|
|
|
|
|
|
|
|
|
151
|
12330
|
|
|
|
|
|
utf8len = 0; |
|
152
|
12330
|
|
|
|
|
|
uniptr = unistr->str + uniidx; |
|
153
|
114711
|
100
|
|
|
|
|
while (uniptr < unistr->str + uniidx + unilen && |
|
|
|
50
|
|
|
|
|
|
|
154
|
102381
|
|
|
|
|
|
uniptr < unistr->str + unistr->len) { |
|
155
|
102381
|
50
|
|
|
|
|
if ((newbuf = realloc(buf, |
|
156
|
102381
|
|
|
|
|
|
sizeof(U8) * (utf8len + UTF8_MAXLEN + 1))) |
|
157
|
|
|
|
|
|
|
== NULL) { |
|
158
|
0
|
|
|
|
|
|
free(buf); |
|
159
|
0
|
|
|
|
|
|
croak("unistrtoSV: %s", strerror(errno)); |
|
160
|
|
|
|
|
|
|
} |
|
161
|
102381
|
|
|
|
|
|
buf = newbuf; |
|
162
|
|
|
|
|
|
|
#if PERL_VERSION >= 20 || (PERL_VERSION == 19 && PERL_SUBVERSION >= 4) |
|
163
|
102381
|
|
|
|
|
|
utf8len = uvchr_to_utf8(buf + utf8len, UNI_TO_NATIVE(*uniptr)) - buf; |
|
164
|
|
|
|
|
|
|
#else |
|
165
|
|
|
|
|
|
|
utf8len = uvuni_to_utf8(buf + utf8len, *uniptr) - buf; |
|
166
|
|
|
|
|
|
|
#endif |
|
167
|
102381
|
|
|
|
|
|
uniptr++; |
|
168
|
|
|
|
|
|
|
} |
|
169
|
|
|
|
|
|
|
|
|
170
|
12330
|
|
|
|
|
|
utf8 = newSVpvn((char *)(void *)buf, utf8len); |
|
171
|
12330
|
|
|
|
|
|
SvUTF8_on(utf8); |
|
172
|
12330
|
|
|
|
|
|
free(buf); |
|
173
|
12330
|
|
|
|
|
|
return utf8; |
|
174
|
|
|
|
|
|
|
} |
|
175
|
|
|
|
|
|
|
|
|
176
|
|
|
|
|
|
|
/* |
|
177
|
|
|
|
|
|
|
* Convert Perl object to C object |
|
178
|
|
|
|
|
|
|
*/ |
|
179
|
|
|
|
|
|
|
#define PerltoC(type, arg) \ |
|
180
|
|
|
|
|
|
|
(INT2PTR(type, SvIV((SV *)SvRV(arg)))) |
|
181
|
|
|
|
|
|
|
|
|
182
|
|
|
|
|
|
|
/* |
|
183
|
|
|
|
|
|
|
* Create Perl object from C object |
|
184
|
|
|
|
|
|
|
*/ |
|
185
|
|
|
|
|
|
|
# define setCtoPerl(arg, klass, var) \ |
|
186
|
|
|
|
|
|
|
STMT_START { \ |
|
187
|
|
|
|
|
|
|
sv_setref_iv(arg, klass, (IV)(var)); \ |
|
188
|
|
|
|
|
|
|
SvREADONLY_on(arg); \ |
|
189
|
|
|
|
|
|
|
} STMT_END |
|
190
|
|
|
|
|
|
|
static |
|
191
|
32684
|
|
|
|
|
|
SV *CtoPerl(char *klass, void *obj) |
|
192
|
|
|
|
|
|
|
{ |
|
193
|
|
|
|
|
|
|
SV *sv; |
|
194
|
|
|
|
|
|
|
|
|
195
|
32684
|
|
|
|
|
|
sv = newSViv(0); |
|
196
|
32684
|
|
|
|
|
|
setCtoPerl(sv, klass, obj); |
|
197
|
32684
|
|
|
|
|
|
return sv; |
|
198
|
|
|
|
|
|
|
} |
|
199
|
|
|
|
|
|
|
|
|
200
|
|
|
|
|
|
|
/* |
|
201
|
|
|
|
|
|
|
* Convert Perl utf8-flagged string (GCString) to grapheme cluster string. |
|
202
|
|
|
|
|
|
|
*/ |
|
203
|
|
|
|
|
|
|
static |
|
204
|
575
|
|
|
|
|
|
gcstring_t *SVtogcstring(SV *sv, linebreak_t *lbobj) |
|
205
|
|
|
|
|
|
|
{ |
|
206
|
575
|
|
|
|
|
|
unistr_t unistr = {NULL, 0}; |
|
207
|
|
|
|
|
|
|
|
|
208
|
575
|
100
|
|
|
|
|
if (!sv_isobject(sv)) { |
|
209
|
465
|
|
|
|
|
|
SVtounistr(&unistr, sv); |
|
210
|
465
|
|
|
|
|
|
return gcstring_new(&unistr, lbobj); |
|
211
|
110
|
50
|
|
|
|
|
} else if (sv_derived_from(sv, "Unicode::GCString")) |
|
212
|
110
|
50
|
|
|
|
|
return PerltoC(gcstring_t *, sv); |
|
213
|
|
|
|
|
|
|
else |
|
214
|
575
|
0
|
|
|
|
|
croak("Unknown object %s", HvNAME(SvSTASH(SvRV(sv)))); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
215
|
|
|
|
|
|
|
} |
|
216
|
|
|
|
|
|
|
|
|
217
|
|
|
|
|
|
|
#if 0 |
|
218
|
|
|
|
|
|
|
/* |
|
219
|
|
|
|
|
|
|
* Convert Perl LineBreak object to C linebreak object. |
|
220
|
|
|
|
|
|
|
*/ |
|
221
|
|
|
|
|
|
|
static |
|
222
|
|
|
|
|
|
|
linebreak_t *SVtolinebreak(SV *sv) |
|
223
|
|
|
|
|
|
|
{ |
|
224
|
|
|
|
|
|
|
if (!sv_isobject(sv)) |
|
225
|
|
|
|
|
|
|
croak("Not object"); |
|
226
|
|
|
|
|
|
|
else if (sv_derived_from(sv, "Unicode::LineBreak")) |
|
227
|
|
|
|
|
|
|
return PerltoC(linebreak_t *, sv); |
|
228
|
|
|
|
|
|
|
else |
|
229
|
|
|
|
|
|
|
croak("Unknown object %s", HvNAME(SvSTASH(SvRV(sv)))); |
|
230
|
|
|
|
|
|
|
} |
|
231
|
|
|
|
|
|
|
#endif /* 0 */ |
|
232
|
|
|
|
|
|
|
|
|
233
|
|
|
|
|
|
|
/* |
|
234
|
|
|
|
|
|
|
* Convert Perl SV to boolean (n.b. string "YES" means true). |
|
235
|
|
|
|
|
|
|
*/ |
|
236
|
|
|
|
|
|
|
static |
|
237
|
2737
|
|
|
|
|
|
int SVtoboolean(SV *sv) |
|
238
|
|
|
|
|
|
|
{ |
|
239
|
|
|
|
|
|
|
char *str; |
|
240
|
|
|
|
|
|
|
|
|
241
|
2737
|
50
|
|
|
|
|
if (!sv || !SvOK(sv)) |
|
|
|
50
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
242
|
0
|
|
|
|
|
|
return 0; |
|
243
|
2737
|
50
|
|
|
|
|
if (SvPOK(sv)) |
|
244
|
3571
|
50
|
|
|
|
|
return strcasecmp((str = SvPV_nolen(sv)), "YES") == 0 || |
|
245
|
834
|
|
|
|
|
|
atof(str) != 0.0; |
|
246
|
0
|
0
|
|
|
|
|
return SvNV(sv) != 0.0; |
|
247
|
|
|
|
|
|
|
} |
|
248
|
|
|
|
|
|
|
|
|
249
|
|
|
|
|
|
|
/*** |
|
250
|
|
|
|
|
|
|
*** Other utilities |
|
251
|
|
|
|
|
|
|
***/ |
|
252
|
|
|
|
|
|
|
|
|
253
|
|
|
|
|
|
|
/* |
|
254
|
|
|
|
|
|
|
* Do regex match once then returns offset and length. |
|
255
|
|
|
|
|
|
|
*/ |
|
256
|
81
|
|
|
|
|
|
void do_pregexec_once(REGEXP *rx, unistr_t *str) |
|
257
|
|
|
|
|
|
|
{ |
|
258
|
|
|
|
|
|
|
SV *screamer; |
|
259
|
|
|
|
|
|
|
char *str_arg, *str_beg, *str_end; |
|
260
|
|
|
|
|
|
|
size_t offs_beg, offs_end; |
|
261
|
|
|
|
|
|
|
|
|
262
|
81
|
|
|
|
|
|
screamer = unistrtoSV(str, 0, str->len); |
|
263
|
81
|
|
|
|
|
|
SvREADONLY_on(screamer); |
|
264
|
81
|
|
|
|
|
|
str_beg = str_arg = SvPVX(screamer); |
|
265
|
81
|
|
|
|
|
|
str_end = SvEND(screamer); |
|
266
|
|
|
|
|
|
|
|
|
267
|
81
|
100
|
|
|
|
|
if (pregexec(rx, str_arg, str_end, str_beg, 0, screamer, 1)) { |
|
268
|
|
|
|
|
|
|
#if PERL_VERSION >= 11 |
|
269
|
49
|
|
|
|
|
|
offs_beg = ((regexp *)SvANY(rx))->offs[0].start; |
|
270
|
49
|
|
|
|
|
|
offs_end = ((regexp *)SvANY(rx))->offs[0].end; |
|
271
|
|
|
|
|
|
|
#elif ((PERL_VERSION == 10) || (PERL_VERSION == 9 && PERL_SUBVERSION >= 5)) |
|
272
|
|
|
|
|
|
|
offs_beg = rx->offs[0].start; |
|
273
|
|
|
|
|
|
|
offs_end = rx->offs[0].end; |
|
274
|
|
|
|
|
|
|
#else /* PERL_VERSION */ |
|
275
|
|
|
|
|
|
|
offs_beg = rx->startp[0]; |
|
276
|
|
|
|
|
|
|
offs_end = rx->endp[0]; |
|
277
|
|
|
|
|
|
|
#endif |
|
278
|
49
|
|
|
|
|
|
str->str += utf8_length((U8 *)str_beg, (U8 *)(str_beg + offs_beg)); |
|
279
|
49
|
|
|
|
|
|
str->len = utf8_length((U8 *)(str_beg + offs_beg), |
|
280
|
|
|
|
|
|
|
(U8 *)(str_beg + offs_end)); |
|
281
|
|
|
|
|
|
|
} else |
|
282
|
32
|
|
|
|
|
|
str->str = NULL; |
|
283
|
|
|
|
|
|
|
|
|
284
|
81
|
|
|
|
|
|
SvREFCNT_dec(screamer); |
|
285
|
81
|
|
|
|
|
|
} |
|
286
|
|
|
|
|
|
|
|
|
287
|
|
|
|
|
|
|
/*** |
|
288
|
|
|
|
|
|
|
*** Callbacks for Sombok library. |
|
289
|
|
|
|
|
|
|
***/ |
|
290
|
|
|
|
|
|
|
|
|
291
|
|
|
|
|
|
|
/* |
|
292
|
|
|
|
|
|
|
* Increment/decrement reference count |
|
293
|
|
|
|
|
|
|
*/ |
|
294
|
1040
|
|
|
|
|
|
void ref_func(void *sv, int datatype, int d) |
|
295
|
|
|
|
|
|
|
{ |
|
296
|
1040
|
50
|
|
|
|
|
if (sv == NULL) |
|
297
|
0
|
|
|
|
|
|
return; |
|
298
|
1040
|
100
|
|
|
|
|
if (0 < d) |
|
299
|
520
|
|
|
|
|
|
SvREFCNT_inc((SV *)sv); |
|
300
|
520
|
50
|
|
|
|
|
else if (d < 0) |
|
301
|
520
|
|
|
|
|
|
SvREFCNT_dec((SV *)sv); |
|
302
|
|
|
|
|
|
|
} |
|
303
|
|
|
|
|
|
|
|
|
304
|
|
|
|
|
|
|
/* |
|
305
|
|
|
|
|
|
|
* Call preprocessing function |
|
306
|
|
|
|
|
|
|
*/ |
|
307
|
|
|
|
|
|
|
static |
|
308
|
130
|
|
|
|
|
|
gcstring_t *prep_func(linebreak_t *lbobj, void *dataref, unistr_t *str, |
|
309
|
|
|
|
|
|
|
unistr_t *text) |
|
310
|
|
|
|
|
|
|
{ |
|
311
|
|
|
|
|
|
|
AV *data; |
|
312
|
130
|
|
|
|
|
|
SV *sv, **pp, *func = NULL; |
|
313
|
130
|
|
|
|
|
|
REGEXP *rx = NULL; |
|
314
|
|
|
|
|
|
|
size_t count, i, j; |
|
315
|
|
|
|
|
|
|
gcstring_t *gcstr, *ret; |
|
316
|
|
|
|
|
|
|
|
|
317
|
130
|
50
|
|
|
|
|
if (dataref == NULL || |
|
|
|
50
|
|
|
|
|
|
|
318
|
130
|
|
|
|
|
|
(data = (AV *)SvRV((SV *)dataref)) == NULL) |
|
319
|
0
|
|
|
|
|
|
return (lbobj->errnum = EINVAL), NULL; |
|
320
|
|
|
|
|
|
|
|
|
321
|
|
|
|
|
|
|
/* Pass I */ |
|
322
|
|
|
|
|
|
|
|
|
323
|
130
|
100
|
|
|
|
|
if (text != NULL) { |
|
324
|
81
|
50
|
|
|
|
|
if ((pp = av_fetch(data, 0, 0)) == NULL) |
|
325
|
0
|
|
|
|
|
|
return (lbobj->errnum = EINVAL), NULL; |
|
326
|
|
|
|
|
|
|
|
|
327
|
|
|
|
|
|
|
#if ((PERL_VERSION >= 10) || (PERL_VERSION >= 9 && PERL_SUBVERSION >= 5)) |
|
328
|
81
|
50
|
|
|
|
|
if (SvRXOK(*pp)) |
|
329
|
81
|
|
|
|
|
|
rx = SvRX(*pp); |
|
330
|
|
|
|
|
|
|
#else /* PERL_VERSION */ |
|
331
|
|
|
|
|
|
|
if (SvROK(*pp) && SvMAGICAL(sv = SvRV(*pp))) { |
|
332
|
|
|
|
|
|
|
MAGIC *mg; |
|
333
|
|
|
|
|
|
|
if ((mg = mg_find(sv, PERL_MAGIC_qr)) != NULL) |
|
334
|
|
|
|
|
|
|
rx = (REGEXP *)mg->mg_obj; |
|
335
|
|
|
|
|
|
|
} |
|
336
|
|
|
|
|
|
|
#endif /* PERL_VERSION */ |
|
337
|
81
|
50
|
|
|
|
|
if (rx == NULL) |
|
338
|
0
|
|
|
|
|
|
return (lbobj->errnum = EINVAL), NULL; |
|
339
|
|
|
|
|
|
|
|
|
340
|
81
|
|
|
|
|
|
do_pregexec_once(rx, str); |
|
341
|
81
|
|
|
|
|
|
return NULL; |
|
342
|
|
|
|
|
|
|
} |
|
343
|
|
|
|
|
|
|
|
|
344
|
|
|
|
|
|
|
/* Pass II */ |
|
345
|
|
|
|
|
|
|
|
|
346
|
49
|
50
|
|
|
|
|
if ((pp = av_fetch(data, 1, 0)) == NULL) |
|
347
|
0
|
|
|
|
|
|
func = NULL; |
|
348
|
49
|
50
|
|
|
|
|
else if (SvOK(*pp)) |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
349
|
49
|
|
|
|
|
|
func = *pp; |
|
350
|
|
|
|
|
|
|
else |
|
351
|
0
|
|
|
|
|
|
func = NULL; |
|
352
|
|
|
|
|
|
|
|
|
353
|
49
|
50
|
|
|
|
|
if (func == NULL) { |
|
354
|
0
|
0
|
|
|
|
|
if ((ret = gcstring_newcopy(str, lbobj)) == NULL) |
|
355
|
0
|
0
|
|
|
|
|
return (lbobj->errnum = errno ? errno : ENOMEM), NULL; |
|
356
|
|
|
|
|
|
|
} else { |
|
357
|
49
|
|
|
|
|
|
dSP; |
|
358
|
49
|
|
|
|
|
|
ENTER; |
|
359
|
49
|
|
|
|
|
|
SAVETMPS; |
|
360
|
49
|
50
|
|
|
|
|
PUSHMARK(SP); |
|
361
|
49
|
|
|
|
|
|
linebreak_incref(lbobj); /* mortal but should not be destroyed.*/ |
|
362
|
49
|
50
|
|
|
|
|
XPUSHs(sv_2mortal(CtoPerl("Unicode::LineBreak", lbobj))); |
|
363
|
49
|
50
|
|
|
|
|
XPUSHs(sv_2mortal(unistrtoSV(str, 0, str->len))); |
|
364
|
49
|
|
|
|
|
|
PUTBACK; |
|
365
|
49
|
|
|
|
|
|
count = call_sv(func, G_ARRAY | G_EVAL); |
|
366
|
|
|
|
|
|
|
|
|
367
|
49
|
|
|
|
|
|
SPAGAIN; |
|
368
|
49
|
50
|
|
|
|
|
if (SvTRUE(ERRSV)) { |
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
369
|
0
|
0
|
|
|
|
|
if (!lbobj->errnum) |
|
370
|
0
|
|
|
|
|
|
lbobj->errnum = LINEBREAK_EEXTN; |
|
371
|
0
|
|
|
|
|
|
return NULL; |
|
372
|
|
|
|
|
|
|
} |
|
373
|
|
|
|
|
|
|
|
|
374
|
49
|
50
|
|
|
|
|
if ((ret = gcstring_new(NULL, lbobj)) == NULL) |
|
375
|
0
|
0
|
|
|
|
|
return (lbobj->errnum = errno ? errno : ENOMEM), NULL; |
|
376
|
|
|
|
|
|
|
|
|
377
|
270
|
100
|
|
|
|
|
for (i = 0; i < count; i++) { |
|
378
|
221
|
|
|
|
|
|
sv = POPs; |
|
379
|
221
|
50
|
|
|
|
|
if (!SvOK(sv)) |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
380
|
0
|
|
|
|
|
|
continue; |
|
381
|
221
|
|
|
|
|
|
gcstr = SVtogcstring(sv, lbobj); |
|
382
|
|
|
|
|
|
|
|
|
383
|
2550
|
100
|
|
|
|
|
for (j = 0; j < gcstr->gclen; j++) { |
|
384
|
2329
|
50
|
|
|
|
|
if (gcstr->gcstr[j].flag & |
|
385
|
|
|
|
|
|
|
(LINEBREAK_FLAG_ALLOW_BEFORE | |
|
386
|
|
|
|
|
|
|
LINEBREAK_FLAG_PROHIBIT_BEFORE)) |
|
387
|
0
|
|
|
|
|
|
continue; |
|
388
|
2329
|
100
|
|
|
|
|
if (i < count - 1 && j == 0) |
|
|
|
100
|
|
|
|
|
|
|
389
|
172
|
|
|
|
|
|
gcstr->gcstr[j].flag |= LINEBREAK_FLAG_ALLOW_BEFORE; |
|
390
|
2157
|
100
|
|
|
|
|
else if (0 < j) |
|
391
|
2108
|
|
|
|
|
|
gcstr->gcstr[j].flag |= LINEBREAK_FLAG_PROHIBIT_BEFORE; |
|
392
|
|
|
|
|
|
|
} |
|
393
|
|
|
|
|
|
|
|
|
394
|
221
|
|
|
|
|
|
gcstring_replace(ret, 0, 0, gcstr); |
|
395
|
221
|
50
|
|
|
|
|
if (!sv_isobject(sv)) |
|
396
|
221
|
|
|
|
|
|
gcstring_destroy(gcstr); |
|
397
|
|
|
|
|
|
|
} |
|
398
|
|
|
|
|
|
|
|
|
399
|
49
|
|
|
|
|
|
PUTBACK; |
|
400
|
49
|
50
|
|
|
|
|
FREETMPS; |
|
401
|
49
|
|
|
|
|
|
LEAVE; |
|
402
|
|
|
|
|
|
|
} |
|
403
|
|
|
|
|
|
|
|
|
404
|
49
|
|
|
|
|
|
return ret; |
|
405
|
|
|
|
|
|
|
} |
|
406
|
|
|
|
|
|
|
|
|
407
|
|
|
|
|
|
|
/* |
|
408
|
|
|
|
|
|
|
* Call format function |
|
409
|
|
|
|
|
|
|
*/ |
|
410
|
|
|
|
|
|
|
static |
|
411
|
|
|
|
|
|
|
char *linebreak_states[] = { |
|
412
|
|
|
|
|
|
|
NULL, "sot", "sop", "sol", "", "eol", "eop", "eot", NULL |
|
413
|
|
|
|
|
|
|
}; |
|
414
|
|
|
|
|
|
|
static |
|
415
|
756
|
|
|
|
|
|
gcstring_t *format_func(linebreak_t *lbobj, linebreak_state_t action, |
|
416
|
|
|
|
|
|
|
gcstring_t *str) |
|
417
|
|
|
|
|
|
|
{ |
|
418
|
|
|
|
|
|
|
SV *sv; |
|
419
|
|
|
|
|
|
|
char *actionstr; |
|
420
|
|
|
|
|
|
|
int count; |
|
421
|
|
|
|
|
|
|
gcstring_t *ret; |
|
422
|
|
|
|
|
|
|
|
|
423
|
756
|
|
|
|
|
|
dSP; |
|
424
|
756
|
50
|
|
|
|
|
if (action <= LINEBREAK_STATE_NONE || LINEBREAK_STATE_MAX <= action) |
|
|
|
50
|
|
|
|
|
|
|
425
|
0
|
|
|
|
|
|
return NULL; |
|
426
|
756
|
|
|
|
|
|
actionstr = linebreak_states[(size_t)action]; |
|
427
|
756
|
|
|
|
|
|
ENTER; |
|
428
|
756
|
|
|
|
|
|
SAVETMPS; |
|
429
|
756
|
50
|
|
|
|
|
PUSHMARK(SP); |
|
430
|
756
|
|
|
|
|
|
linebreak_incref(lbobj); /* mortal but should not be destroyed. */ |
|
431
|
756
|
50
|
|
|
|
|
XPUSHs(sv_2mortal(CtoPerl("Unicode::LineBreak", lbobj))); |
|
432
|
756
|
50
|
|
|
|
|
XPUSHs(sv_2mortal(newSVpv(actionstr, 0))); |
|
433
|
756
|
50
|
|
|
|
|
XPUSHs(sv_2mortal(CtoPerl("Unicode::GCString", gcstring_copy(str)))); |
|
434
|
756
|
|
|
|
|
|
PUTBACK; |
|
435
|
756
|
|
|
|
|
|
count = call_sv(lbobj->format_data, G_SCALAR | G_EVAL); |
|
436
|
|
|
|
|
|
|
|
|
437
|
756
|
|
|
|
|
|
SPAGAIN; |
|
438
|
756
|
50
|
|
|
|
|
if (SvTRUE(ERRSV)) { |
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
439
|
0
|
0
|
|
|
|
|
if (!lbobj->errnum) |
|
440
|
0
|
|
|
|
|
|
lbobj->errnum = LINEBREAK_EEXTN; |
|
441
|
0
|
|
|
|
|
|
POPs; |
|
442
|
0
|
|
|
|
|
|
return NULL; |
|
443
|
756
|
50
|
|
|
|
|
} else if (count != 1) |
|
444
|
0
|
|
|
|
|
|
croak("format_func: internal error"); |
|
445
|
|
|
|
|
|
|
else |
|
446
|
756
|
|
|
|
|
|
sv = POPs; |
|
447
|
756
|
100
|
|
|
|
|
if (!SvOK(sv)) |
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
448
|
402
|
|
|
|
|
|
ret = NULL; |
|
449
|
|
|
|
|
|
|
else |
|
450
|
354
|
|
|
|
|
|
ret = SVtogcstring(sv, lbobj); |
|
451
|
756
|
100
|
|
|
|
|
if (sv_isobject(sv)) |
|
452
|
110
|
|
|
|
|
|
ret = gcstring_copy(ret); |
|
453
|
|
|
|
|
|
|
|
|
454
|
756
|
|
|
|
|
|
PUTBACK; |
|
455
|
756
|
50
|
|
|
|
|
FREETMPS; |
|
456
|
756
|
|
|
|
|
|
LEAVE; |
|
457
|
|
|
|
|
|
|
|
|
458
|
756
|
|
|
|
|
|
return ret; |
|
459
|
|
|
|
|
|
|
} |
|
460
|
|
|
|
|
|
|
|
|
461
|
|
|
|
|
|
|
/* |
|
462
|
|
|
|
|
|
|
* Call sizing function |
|
463
|
|
|
|
|
|
|
*/ |
|
464
|
|
|
|
|
|
|
static |
|
465
|
2328
|
|
|
|
|
|
double sizing_func(linebreak_t *lbobj, double len, |
|
466
|
|
|
|
|
|
|
gcstring_t *pre, gcstring_t *spc, gcstring_t *str) |
|
467
|
|
|
|
|
|
|
{ |
|
468
|
|
|
|
|
|
|
int count; |
|
469
|
|
|
|
|
|
|
double ret; |
|
470
|
|
|
|
|
|
|
|
|
471
|
2328
|
|
|
|
|
|
dSP; |
|
472
|
2328
|
|
|
|
|
|
ENTER; |
|
473
|
2328
|
|
|
|
|
|
SAVETMPS; |
|
474
|
2328
|
50
|
|
|
|
|
PUSHMARK(SP); |
|
475
|
2328
|
|
|
|
|
|
linebreak_incref(lbobj); /* mortal but should not be destroyed. */ |
|
476
|
2328
|
50
|
|
|
|
|
XPUSHs(sv_2mortal(CtoPerl("Unicode::LineBreak", lbobj))); |
|
477
|
2328
|
50
|
|
|
|
|
XPUSHs(sv_2mortal(newSVnv(len))); |
|
478
|
2328
|
50
|
|
|
|
|
XPUSHs(sv_2mortal(CtoPerl("Unicode::GCString", gcstring_copy(pre)))); |
|
479
|
2328
|
50
|
|
|
|
|
XPUSHs(sv_2mortal(CtoPerl("Unicode::GCString", gcstring_copy(spc)))); |
|
480
|
2328
|
50
|
|
|
|
|
XPUSHs(sv_2mortal(CtoPerl("Unicode::GCString", gcstring_copy(str)))); |
|
481
|
2328
|
|
|
|
|
|
PUTBACK; |
|
482
|
2328
|
|
|
|
|
|
count = call_sv(lbobj->sizing_data, G_SCALAR | G_EVAL); |
|
483
|
|
|
|
|
|
|
|
|
484
|
2328
|
|
|
|
|
|
SPAGAIN; |
|
485
|
2328
|
50
|
|
|
|
|
if (SvTRUE(ERRSV)) { |
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
486
|
0
|
0
|
|
|
|
|
if (!lbobj->errnum) |
|
487
|
0
|
|
|
|
|
|
lbobj->errnum = LINEBREAK_EEXTN; |
|
488
|
0
|
|
|
|
|
|
POPs; |
|
489
|
0
|
|
|
|
|
|
return -1; |
|
490
|
2328
|
50
|
|
|
|
|
} else if (count != 1) |
|
491
|
0
|
|
|
|
|
|
croak("sizing_func: internal error"); |
|
492
|
|
|
|
|
|
|
else |
|
493
|
2328
|
50
|
|
|
|
|
ret = POPn; |
|
494
|
|
|
|
|
|
|
|
|
495
|
2328
|
|
|
|
|
|
PUTBACK; |
|
496
|
2328
|
50
|
|
|
|
|
FREETMPS; |
|
497
|
2328
|
|
|
|
|
|
LEAVE; |
|
498
|
|
|
|
|
|
|
|
|
499
|
2328
|
|
|
|
|
|
return ret; |
|
500
|
|
|
|
|
|
|
} |
|
501
|
|
|
|
|
|
|
|
|
502
|
|
|
|
|
|
|
/* |
|
503
|
|
|
|
|
|
|
* Call urgent breaking function |
|
504
|
|
|
|
|
|
|
*/ |
|
505
|
|
|
|
|
|
|
static |
|
506
|
0
|
|
|
|
|
|
gcstring_t *urgent_func(linebreak_t *lbobj, gcstring_t *str) |
|
507
|
|
|
|
|
|
|
{ |
|
508
|
|
|
|
|
|
|
SV *sv; |
|
509
|
|
|
|
|
|
|
int count; |
|
510
|
|
|
|
|
|
|
size_t i; |
|
511
|
|
|
|
|
|
|
gcstring_t *gcstr, *ret; |
|
512
|
|
|
|
|
|
|
|
|
513
|
0
|
|
|
|
|
|
dSP; |
|
514
|
0
|
|
|
|
|
|
ENTER; |
|
515
|
0
|
|
|
|
|
|
SAVETMPS; |
|
516
|
0
|
0
|
|
|
|
|
PUSHMARK(SP); |
|
517
|
0
|
|
|
|
|
|
linebreak_incref(lbobj); /* mortal but should not be destroyed. */ |
|
518
|
0
|
0
|
|
|
|
|
XPUSHs(sv_2mortal(CtoPerl("Unicode::LineBreak", lbobj))); |
|
519
|
0
|
0
|
|
|
|
|
XPUSHs(sv_2mortal(CtoPerl("Unicode::GCString", gcstring_copy(str)))); |
|
520
|
0
|
|
|
|
|
|
PUTBACK; |
|
521
|
0
|
|
|
|
|
|
count = call_sv(lbobj->urgent_data, G_ARRAY | G_EVAL); |
|
522
|
|
|
|
|
|
|
|
|
523
|
0
|
|
|
|
|
|
SPAGAIN; |
|
524
|
0
|
0
|
|
|
|
|
if (SvTRUE(ERRSV)) { |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
525
|
0
|
0
|
|
|
|
|
if (!lbobj->errnum) |
|
526
|
0
|
|
|
|
|
|
lbobj->errnum = LINEBREAK_EEXTN; |
|
527
|
0
|
|
|
|
|
|
return NULL; |
|
528
|
0
|
0
|
|
|
|
|
} if (count == 0) |
|
529
|
0
|
|
|
|
|
|
return NULL; |
|
530
|
|
|
|
|
|
|
|
|
531
|
0
|
|
|
|
|
|
ret = gcstring_new(NULL, lbobj); |
|
532
|
0
|
0
|
|
|
|
|
for (i = count; i; i--) { |
|
533
|
0
|
|
|
|
|
|
sv = POPs; |
|
534
|
0
|
0
|
|
|
|
|
if (SvOK(sv)) { |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
535
|
0
|
|
|
|
|
|
gcstr = SVtogcstring(sv, lbobj); |
|
536
|
0
|
0
|
|
|
|
|
if (gcstr->gclen) |
|
537
|
0
|
|
|
|
|
|
gcstr->gcstr[0].flag = LINEBREAK_FLAG_ALLOW_BEFORE; |
|
538
|
0
|
|
|
|
|
|
gcstring_replace(ret, 0, 0, gcstr); |
|
539
|
0
|
0
|
|
|
|
|
if (!sv_isobject(sv)) |
|
540
|
0
|
|
|
|
|
|
gcstring_destroy(gcstr); |
|
541
|
|
|
|
|
|
|
} |
|
542
|
|
|
|
|
|
|
} |
|
543
|
|
|
|
|
|
|
|
|
544
|
0
|
|
|
|
|
|
PUTBACK; |
|
545
|
0
|
0
|
|
|
|
|
FREETMPS; |
|
546
|
0
|
|
|
|
|
|
LEAVE; |
|
547
|
|
|
|
|
|
|
|
|
548
|
0
|
|
|
|
|
|
return ret; |
|
549
|
|
|
|
|
|
|
} |
|
550
|
|
|
|
|
|
|
|
|
551
|
|
|
|
|
|
|
|
|
552
|
|
|
|
|
|
|
MODULE = Unicode::LineBreak PACKAGE = Unicode::LineBreak |
|
553
|
|
|
|
|
|
|
|
|
554
|
|
|
|
|
|
|
void |
|
555
|
|
|
|
|
|
|
EAWidths() |
|
556
|
|
|
|
|
|
|
INIT: |
|
557
|
|
|
|
|
|
|
char **p; |
|
558
|
|
|
|
|
|
|
PPCODE: |
|
559
|
210
|
100
|
|
|
|
|
for (p = (char **)linebreak_propvals_EA; *p != NULL; p++) |
|
560
|
189
|
50
|
|
|
|
|
XPUSHs(sv_2mortal(newSVpv(*p, 0))); |
|
561
|
|
|
|
|
|
|
|
|
562
|
|
|
|
|
|
|
void |
|
563
|
|
|
|
|
|
|
LBClasses() |
|
564
|
|
|
|
|
|
|
INIT: |
|
565
|
|
|
|
|
|
|
char **p; |
|
566
|
|
|
|
|
|
|
PPCODE: |
|
567
|
861
|
100
|
|
|
|
|
for (p = (char **)linebreak_propvals_LB; *p != NULL; p++) |
|
568
|
840
|
50
|
|
|
|
|
XPUSHs(sv_2mortal(newSVpv(*p, 0))); |
|
569
|
|
|
|
|
|
|
|
|
570
|
|
|
|
|
|
|
linebreak_t * |
|
571
|
|
|
|
|
|
|
_new(klass) |
|
572
|
|
|
|
|
|
|
char *klass; |
|
573
|
|
|
|
|
|
|
PROTOTYPE: $ |
|
574
|
|
|
|
|
|
|
CODE: |
|
575
|
483
|
50
|
|
|
|
|
if ((RETVAL = linebreak_new(ref_func)) == NULL) |
|
576
|
0
|
|
|
|
|
|
croak("%s->_new: %s", klass, strerror(errno)); |
|
577
|
483
|
|
|
|
|
|
linebreak_set_stash(RETVAL, newRV_noinc((SV *)newHV())); |
|
578
|
483
|
|
|
|
|
|
SvREFCNT_dec(RETVAL->stash); /* fixup */ |
|
579
|
|
|
|
|
|
|
OUTPUT: |
|
580
|
|
|
|
|
|
|
RETVAL |
|
581
|
|
|
|
|
|
|
|
|
582
|
|
|
|
|
|
|
linebreak_t * |
|
583
|
|
|
|
|
|
|
copy(self) |
|
584
|
|
|
|
|
|
|
linebreak_t *self; |
|
585
|
|
|
|
|
|
|
PROTOTYPE: $ |
|
586
|
|
|
|
|
|
|
CODE: |
|
587
|
0
|
|
|
|
|
|
RETVAL = linebreak_copy(self); |
|
588
|
|
|
|
|
|
|
OUTPUT: |
|
589
|
|
|
|
|
|
|
RETVAL |
|
590
|
|
|
|
|
|
|
|
|
591
|
|
|
|
|
|
|
void |
|
592
|
|
|
|
|
|
|
DESTROY(self) |
|
593
|
|
|
|
|
|
|
linebreak_t *self; |
|
594
|
|
|
|
|
|
|
PROTOTYPE: $ |
|
595
|
|
|
|
|
|
|
CODE: |
|
596
|
3616
|
|
|
|
|
|
linebreak_destroy(self); |
|
597
|
|
|
|
|
|
|
|
|
598
|
|
|
|
|
|
|
SV * |
|
599
|
|
|
|
|
|
|
_config(self, ...) |
|
600
|
|
|
|
|
|
|
linebreak_t *self; |
|
601
|
|
|
|
|
|
|
PREINIT: |
|
602
|
|
|
|
|
|
|
size_t i; |
|
603
|
|
|
|
|
|
|
char *key; |
|
604
|
|
|
|
|
|
|
void *func; |
|
605
|
|
|
|
|
|
|
SV *val; |
|
606
|
|
|
|
|
|
|
char *opt; |
|
607
|
|
|
|
|
|
|
CODE: |
|
608
|
1351
|
|
|
|
|
|
RETVAL = NULL; |
|
609
|
1351
|
50
|
|
|
|
|
if (items < 2) |
|
610
|
0
|
|
|
|
|
|
croak("_config: Too few arguments"); |
|
611
|
1351
|
100
|
|
|
|
|
else if (items < 3) { |
|
612
|
353
|
50
|
|
|
|
|
key = (char *)SvPV_nolen(ST(1)); |
|
613
|
|
|
|
|
|
|
|
|
614
|
353
|
50
|
|
|
|
|
if (strcasecmp(key, "BreakIndent") == 0) |
|
615
|
0
|
|
|
|
|
|
RETVAL = newSVuv(self->options & |
|
616
|
|
|
|
|
|
|
LINEBREAK_OPTION_BREAK_INDENT); |
|
617
|
353
|
50
|
|
|
|
|
else if (strcasecmp(key, "CharMax") == 0) |
|
618
|
0
|
|
|
|
|
|
RETVAL = newSVuv(self->charmax); |
|
619
|
353
|
100
|
|
|
|
|
else if (strcasecmp(key, "ColMax") == 0) |
|
620
|
56
|
|
|
|
|
|
RETVAL = newSVnv((NV)self->colmax); |
|
621
|
297
|
50
|
|
|
|
|
else if (strcasecmp(key, "ColMin") == 0) |
|
622
|
0
|
|
|
|
|
|
RETVAL = newSVnv((NV)self->colmin); |
|
623
|
297
|
50
|
|
|
|
|
else if (strcasecmp(key, "ComplexBreaking") == 0) |
|
624
|
0
|
|
|
|
|
|
RETVAL = newSVuv(self->options & |
|
625
|
|
|
|
|
|
|
LINEBREAK_OPTION_COMPLEX_BREAKING); |
|
626
|
297
|
50
|
|
|
|
|
else if (strcasecmp(key, "Context") == 0) { |
|
627
|
0
|
0
|
|
|
|
|
if (self->options & LINEBREAK_OPTION_EASTASIAN_CONTEXT) |
|
628
|
0
|
|
|
|
|
|
RETVAL = newSVpvn("EASTASIAN", 9); |
|
629
|
|
|
|
|
|
|
else |
|
630
|
0
|
|
|
|
|
|
RETVAL = newSVpvn("NONEASTASIAN", 12); |
|
631
|
297
|
50
|
|
|
|
|
} else if (strcasecmp(key, "EAWidth") == 0) { |
|
632
|
0
|
|
|
|
|
|
AV *av, *codes = NULL, *ret = NULL; |
|
633
|
0
|
|
|
|
|
|
propval_t p = PROP_UNKNOWN; |
|
634
|
|
|
|
|
|
|
unichar_t c; |
|
635
|
|
|
|
|
|
|
size_t i; |
|
636
|
|
|
|
|
|
|
|
|
637
|
0
|
0
|
|
|
|
|
if (self->map == NULL || self->mapsiz == 0) |
|
|
|
0
|
|
|
|
|
|
|
638
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
639
|
|
|
|
|
|
|
|
|
640
|
0
|
0
|
|
|
|
|
for (i = 0; i < self->mapsiz; i++) |
|
641
|
0
|
0
|
|
|
|
|
if (self->map[i].eaw != PROP_UNKNOWN) { |
|
642
|
0
|
0
|
|
|
|
|
if (p != self->map[i].eaw){ |
|
643
|
0
|
|
|
|
|
|
p = self->map[i].eaw; |
|
644
|
0
|
|
|
|
|
|
codes = newAV(); |
|
645
|
0
|
|
|
|
|
|
av = newAV(); |
|
646
|
0
|
|
|
|
|
|
av_push(av, newRV_noinc((SV *)codes)); |
|
647
|
0
|
|
|
|
|
|
av_push(av, newSViv((IV)p)); |
|
648
|
0
|
0
|
|
|
|
|
if (ret == NULL) |
|
649
|
0
|
|
|
|
|
|
ret = newAV(); |
|
650
|
0
|
|
|
|
|
|
av_push(ret, newRV_noinc((SV *)av)); |
|
651
|
|
|
|
|
|
|
} |
|
652
|
0
|
0
|
|
|
|
|
for (c = self->map[i].beg; c <= self->map[i].end; c++) |
|
653
|
0
|
|
|
|
|
|
av_push(codes, newSVuv(c)); |
|
654
|
|
|
|
|
|
|
} |
|
655
|
|
|
|
|
|
|
|
|
656
|
0
|
0
|
|
|
|
|
if (ret == NULL) |
|
657
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
658
|
0
|
|
|
|
|
|
RETVAL = newRV_noinc((SV *)ret); |
|
659
|
297
|
50
|
|
|
|
|
} else if (strcasecmp(key, "Format") == 0) { |
|
660
|
0
|
|
|
|
|
|
func = self->format_func; |
|
661
|
0
|
0
|
|
|
|
|
if (func == NULL) |
|
662
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
663
|
0
|
0
|
|
|
|
|
else if (func == linebreak_format_NEWLINE) |
|
664
|
0
|
|
|
|
|
|
RETVAL = newSVpvn("NEWLINE", 7); |
|
665
|
0
|
0
|
|
|
|
|
else if (func == linebreak_format_SIMPLE) |
|
666
|
0
|
|
|
|
|
|
RETVAL = newSVpvn("SIMPLE", 6); |
|
667
|
0
|
0
|
|
|
|
|
else if (func == linebreak_format_TRIM) |
|
668
|
0
|
|
|
|
|
|
RETVAL = newSVpvn("TRIM", 4); |
|
669
|
0
|
0
|
|
|
|
|
else if (func == format_func) { |
|
670
|
0
|
0
|
|
|
|
|
if ((val = (SV *)self->format_data) == NULL) |
|
671
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
672
|
0
|
|
|
|
|
|
ST(0) = val; /* should not be mortal. */ |
|
673
|
0
|
|
|
|
|
|
XSRETURN(1); |
|
674
|
|
|
|
|
|
|
} else |
|
675
|
0
|
|
|
|
|
|
croak("_config: internal error"); |
|
676
|
297
|
50
|
|
|
|
|
} else if (strcasecmp(key, "HangulAsAL") == 0) |
|
677
|
0
|
|
|
|
|
|
RETVAL = newSVuv(self->options & |
|
678
|
|
|
|
|
|
|
LINEBREAK_OPTION_HANGUL_AS_AL); |
|
679
|
297
|
50
|
|
|
|
|
else if (strcasecmp(key, "LBClass") == 0) { |
|
680
|
0
|
|
|
|
|
|
AV *av, *codes = NULL, *ret = NULL; |
|
681
|
0
|
|
|
|
|
|
propval_t p = PROP_UNKNOWN; |
|
682
|
|
|
|
|
|
|
unichar_t c; |
|
683
|
|
|
|
|
|
|
size_t i; |
|
684
|
|
|
|
|
|
|
|
|
685
|
0
|
0
|
|
|
|
|
if (self->map == NULL || self->mapsiz == 0) |
|
|
|
0
|
|
|
|
|
|
|
686
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
687
|
|
|
|
|
|
|
|
|
688
|
0
|
0
|
|
|
|
|
for (i = 0; i < self->mapsiz; i++) |
|
689
|
0
|
0
|
|
|
|
|
if (self->map[i].lbc != PROP_UNKNOWN) { |
|
690
|
0
|
0
|
|
|
|
|
if (p != self->map[i].lbc){ |
|
691
|
0
|
|
|
|
|
|
p = self->map[i].lbc; |
|
692
|
0
|
|
|
|
|
|
codes = newAV(); |
|
693
|
0
|
|
|
|
|
|
av = newAV(); |
|
694
|
0
|
|
|
|
|
|
av_push(av, newRV_noinc((SV *)codes)); |
|
695
|
0
|
|
|
|
|
|
av_push(av, newSViv((IV)p)); |
|
696
|
0
|
0
|
|
|
|
|
if (ret == NULL) |
|
697
|
0
|
|
|
|
|
|
ret = newAV(); |
|
698
|
0
|
|
|
|
|
|
av_push(ret, newRV_noinc((SV *)av)); |
|
699
|
|
|
|
|
|
|
} |
|
700
|
0
|
0
|
|
|
|
|
for (c = self->map[i].beg; c <= self->map[i].end; c++) |
|
701
|
0
|
|
|
|
|
|
av_push(codes, newSVuv(c)); |
|
702
|
|
|
|
|
|
|
} |
|
703
|
|
|
|
|
|
|
|
|
704
|
0
|
0
|
|
|
|
|
if (ret == NULL) |
|
705
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
706
|
0
|
|
|
|
|
|
RETVAL = newRV_noinc((SV *)ret); |
|
707
|
297
|
50
|
|
|
|
|
} else if (strcasecmp(key, "LegacyCM") == 0) |
|
708
|
0
|
|
|
|
|
|
RETVAL = newSVuv(self->options & LINEBREAK_OPTION_LEGACY_CM); |
|
709
|
297
|
50
|
|
|
|
|
else if (strcasecmp(key, "Newline") == 0) { |
|
710
|
297
|
|
|
|
|
|
unistr_t unistr = {self->newline.str, self->newline.len}; |
|
711
|
297
|
50
|
|
|
|
|
if (self->newline.str == NULL || self->newline.len == 0) |
|
|
|
50
|
|
|
|
|
|
|
712
|
0
|
|
|
|
|
|
RETVAL = unistrtoSV(&unistr, 0, 0); |
|
713
|
|
|
|
|
|
|
else |
|
714
|
297
|
|
|
|
|
|
RETVAL = unistrtoSV(&unistr, 0, self->newline.len); |
|
715
|
0
|
0
|
|
|
|
|
} else if (strcasecmp(key, "Prep") == 0) { |
|
716
|
|
|
|
|
|
|
AV *av; |
|
717
|
0
|
0
|
|
|
|
|
if (self->prep_func == NULL || self->prep_func[0] == NULL) |
|
|
|
0
|
|
|
|
|
|
|
718
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
719
|
0
|
|
|
|
|
|
av = newAV(); |
|
720
|
0
|
0
|
|
|
|
|
for (i = 0; (func = self->prep_func[i]) != NULL; i++) |
|
721
|
0
|
0
|
|
|
|
|
if (func == linebreak_prep_URIBREAK) { |
|
722
|
0
|
0
|
|
|
|
|
if (self->prep_data == NULL || |
|
|
|
0
|
|
|
|
|
|
|
723
|
0
|
|
|
|
|
|
self->prep_data[i] == NULL) |
|
724
|
0
|
|
|
|
|
|
av_push(av, newSVpvn("NONBREAKURI", 11)); |
|
725
|
|
|
|
|
|
|
else |
|
726
|
0
|
|
|
|
|
|
av_push(av, newSVpvn("BREAKURI", 8)); |
|
727
|
0
|
0
|
|
|
|
|
} else if (func == prep_func) { |
|
728
|
0
|
0
|
|
|
|
|
if (self->prep_data == NULL || |
|
|
|
0
|
|
|
|
|
|
|
729
|
0
|
|
|
|
|
|
self->prep_data[i] == NULL) |
|
730
|
0
|
|
|
|
|
|
croak("_config: internal error"); |
|
731
|
0
|
|
|
|
|
|
SvREFCNT_inc(self->prep_data[i]); /* avoid freed */ |
|
732
|
0
|
|
|
|
|
|
av_push(av, self->prep_data[i]); |
|
733
|
|
|
|
|
|
|
} else |
|
734
|
0
|
|
|
|
|
|
croak("_config: internal error"); |
|
735
|
0
|
|
|
|
|
|
RETVAL = newRV_noinc((SV *)av); |
|
736
|
0
|
0
|
|
|
|
|
} else if (strcasecmp(key, "Sizing") == 0) { |
|
737
|
0
|
|
|
|
|
|
func = self->sizing_func; |
|
738
|
0
|
0
|
|
|
|
|
if (func == NULL) |
|
739
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
740
|
0
|
0
|
|
|
|
|
else if (func == linebreak_sizing_UAX11) |
|
741
|
0
|
|
|
|
|
|
RETVAL = newSVpvn("UAX11", 5); |
|
742
|
0
|
0
|
|
|
|
|
else if (func == sizing_func) { |
|
743
|
0
|
0
|
|
|
|
|
if ((val = (SV *)self->sizing_data) == NULL) |
|
744
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
745
|
0
|
|
|
|
|
|
ST(0) = val; /* should not be mortal. */ |
|
746
|
0
|
|
|
|
|
|
XSRETURN(1); |
|
747
|
|
|
|
|
|
|
} else |
|
748
|
0
|
|
|
|
|
|
croak("_config: internal error"); |
|
749
|
0
|
0
|
|
|
|
|
} else if (strcasecmp(key, "Urgent") == 0) { |
|
750
|
0
|
|
|
|
|
|
func = self->urgent_func; |
|
751
|
0
|
0
|
|
|
|
|
if (func == NULL) |
|
752
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
753
|
0
|
0
|
|
|
|
|
else if (func == linebreak_urgent_ABORT) |
|
754
|
0
|
|
|
|
|
|
RETVAL = newSVpvn("CROAK", 5); |
|
755
|
0
|
0
|
|
|
|
|
else if (func == linebreak_urgent_FORCE) |
|
756
|
0
|
|
|
|
|
|
RETVAL = newSVpvn("FORCE", 5); |
|
757
|
0
|
0
|
|
|
|
|
else if (func == urgent_func) { |
|
758
|
0
|
0
|
|
|
|
|
if ((val = (SV *)self->urgent_data) == NULL) |
|
759
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
760
|
0
|
|
|
|
|
|
ST(0) = val; /* should not be mortal. */ |
|
761
|
0
|
|
|
|
|
|
XSRETURN(1); |
|
762
|
|
|
|
|
|
|
} else |
|
763
|
0
|
|
|
|
|
|
croak("_config: internal error"); |
|
764
|
0
|
0
|
|
|
|
|
} else if (strcasecmp(key, "ViramaAsJoiner") == 0) |
|
765
|
0
|
|
|
|
|
|
RETVAL = newSVuv(self->options & LINEBREAK_OPTION_VIRAMA_AS_JOINER); |
|
766
|
|
|
|
|
|
|
else { |
|
767
|
0
|
|
|
|
|
|
warn("_config: Getting unknown option %s", key); |
|
768
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
769
|
|
|
|
|
|
|
} |
|
770
|
998
|
50
|
|
|
|
|
} else if (!(items % 2)) |
|
771
|
0
|
|
|
|
|
|
croak("_config: Argument size mismatch"); |
|
772
|
9235
|
100
|
|
|
|
|
else for (RETVAL = NULL, i = 1; i < items; i += 2) { |
|
773
|
8237
|
50
|
|
|
|
|
if (!SvPOK(ST(i))) |
|
774
|
0
|
|
|
|
|
|
croak("_config: Illegal argument"); |
|
775
|
8237
|
50
|
|
|
|
|
key = (char *)SvPV_nolen(ST(i)); |
|
776
|
8237
|
|
|
|
|
|
val = ST(i + 1); |
|
777
|
|
|
|
|
|
|
|
|
778
|
8237
|
100
|
|
|
|
|
if (strcasecmp(key, "Prep") == 0) { |
|
779
|
|
|
|
|
|
|
SV *sv, *pattern, *func; |
|
780
|
|
|
|
|
|
|
AV *av; |
|
781
|
497
|
|
|
|
|
|
REGEXP *rx = NULL; |
|
782
|
|
|
|
|
|
|
|
|
783
|
497
|
100
|
|
|
|
|
if (! SvOK(val)) |
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
784
|
483
|
|
|
|
|
|
linebreak_add_prep(self, NULL, NULL); |
|
785
|
14
|
100
|
|
|
|
|
else if (SvROK(val) && |
|
|
|
50
|
|
|
|
|
|
|
786
|
11
|
50
|
|
|
|
|
SvTYPE(av = (AV *)SvRV(val)) == SVt_PVAV && |
|
787
|
11
|
|
|
|
|
|
0 < av_len(av) + 1) { |
|
788
|
11
|
|
|
|
|
|
pattern = *av_fetch(av, 0, 0); |
|
789
|
|
|
|
|
|
|
#if ((PERL_VERSION >= 10) || (PERL_VERSION >= 9 && PERL_SUBVERSION >= 5)) |
|
790
|
11
|
100
|
|
|
|
|
if (SvRXOK(pattern)) |
|
791
|
10
|
|
|
|
|
|
rx = SvRX(pattern); |
|
792
|
|
|
|
|
|
|
#else /* PERL_VERSION */ |
|
793
|
|
|
|
|
|
|
if (SvROK(pattern) && SvMAGICAL(sv = SvRV(pattern))) { |
|
794
|
|
|
|
|
|
|
MAGIC *mg; |
|
795
|
|
|
|
|
|
|
if ((mg = mg_find(sv, PERL_MAGIC_qr)) != NULL) |
|
796
|
|
|
|
|
|
|
rx = (REGEXP *)mg->mg_obj; |
|
797
|
|
|
|
|
|
|
} |
|
798
|
|
|
|
|
|
|
#endif |
|
799
|
11
|
100
|
|
|
|
|
if (rx != NULL) |
|
800
|
10
|
|
|
|
|
|
SvREFCNT_inc(pattern); /* FIXME:avoid freed */ |
|
801
|
1
|
50
|
|
|
|
|
else if (SvOK(pattern)) { |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
802
|
|
|
|
|
|
|
#if ((PERL_VERSION >= 10) || (PERL_VERSION == 9 && PERL_SUBVERSION >= 5)) |
|
803
|
1
|
|
|
|
|
|
rx = pregcomp(pattern, 0); |
|
804
|
|
|
|
|
|
|
#else /* PERL_VERSION */ |
|
805
|
|
|
|
|
|
|
{ |
|
806
|
|
|
|
|
|
|
PMOP *pm; |
|
807
|
|
|
|
|
|
|
New(1, pm, 1, PMOP); |
|
808
|
|
|
|
|
|
|
rx = pregcomp(SvPVX(pattern), SvEND(pattern), pm); |
|
809
|
|
|
|
|
|
|
} |
|
810
|
|
|
|
|
|
|
#endif |
|
811
|
2
|
50
|
|
|
|
|
if (rx != NULL) { |
|
812
|
|
|
|
|
|
|
#if PERL_VERSION >= 11 |
|
813
|
1
|
|
|
|
|
|
pattern = newRV_noinc((SV *)rx); |
|
814
|
1
|
|
|
|
|
|
sv_bless(pattern, gv_stashpv("Regexp", 0)); |
|
815
|
|
|
|
|
|
|
#else /* PERL_VERSION */ |
|
816
|
|
|
|
|
|
|
sv = newSV(0); |
|
817
|
|
|
|
|
|
|
sv_magic(sv, (SV *)rx, PERL_MAGIC_qr, NULL, 0); |
|
818
|
|
|
|
|
|
|
pattern = newRV_noinc(sv); |
|
819
|
|
|
|
|
|
|
sv_bless(pattern, gv_stashpv("Regexp", 0)); |
|
820
|
|
|
|
|
|
|
#endif |
|
821
|
|
|
|
|
|
|
} |
|
822
|
|
|
|
|
|
|
} else |
|
823
|
0
|
|
|
|
|
|
rx = NULL; |
|
824
|
|
|
|
|
|
|
|
|
825
|
11
|
50
|
|
|
|
|
if (rx == NULL) |
|
826
|
0
|
|
|
|
|
|
croak("_config: Not a regex"); |
|
827
|
|
|
|
|
|
|
|
|
828
|
11
|
50
|
|
|
|
|
if (av_fetch(av, 1, 0) == NULL) |
|
829
|
0
|
|
|
|
|
|
func = NULL; |
|
830
|
11
|
50
|
|
|
|
|
else if (SvOK(func = *av_fetch(av, 1, 0))) |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
831
|
11
|
|
|
|
|
|
SvREFCNT_inc(func); /* avoid freed */ |
|
832
|
|
|
|
|
|
|
else |
|
833
|
0
|
|
|
|
|
|
func = NULL; |
|
834
|
|
|
|
|
|
|
|
|
835
|
11
|
|
|
|
|
|
av = newAV(); |
|
836
|
11
|
|
|
|
|
|
av_push(av, pattern); |
|
837
|
11
|
50
|
|
|
|
|
if (func != NULL) |
|
838
|
11
|
|
|
|
|
|
av_push(av, func); |
|
839
|
11
|
|
|
|
|
|
sv = newRV_noinc((SV *)av); |
|
840
|
11
|
|
|
|
|
|
linebreak_add_prep(self, prep_func, (void *)sv); |
|
841
|
11
|
|
|
|
|
|
SvREFCNT_dec(sv); /* fixup */ |
|
842
|
|
|
|
|
|
|
} else { |
|
843
|
3
|
50
|
|
|
|
|
char *s = SvPV_nolen(val); |
|
844
|
|
|
|
|
|
|
|
|
845
|
3
|
100
|
|
|
|
|
if (strcasecmp(s, "BREAKURI") == 0) |
|
846
|
1
|
|
|
|
|
|
linebreak_add_prep(self, linebreak_prep_URIBREAK, val); |
|
847
|
2
|
50
|
|
|
|
|
else if (strcasecmp(s, "NONBREAKURI") == 0) |
|
848
|
2
|
|
|
|
|
|
linebreak_add_prep(self, linebreak_prep_URIBREAK, |
|
849
|
|
|
|
|
|
|
NULL); |
|
850
|
|
|
|
|
|
|
else |
|
851
|
497
|
|
|
|
|
|
croak("_config: Unknown preprocess option: %s", s); |
|
852
|
|
|
|
|
|
|
} |
|
853
|
7740
|
100
|
|
|
|
|
} else if (strcasecmp(key, "Format") == 0) { |
|
854
|
509
|
100
|
|
|
|
|
if (! SvOK(val)) |
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
855
|
1
|
|
|
|
|
|
linebreak_set_format(self, NULL, NULL); |
|
856
|
508
|
100
|
|
|
|
|
else if (sv_derived_from(val, "CODE")) |
|
857
|
17
|
|
|
|
|
|
linebreak_set_format(self, format_func, (void *)val); |
|
858
|
|
|
|
|
|
|
else { |
|
859
|
491
|
50
|
|
|
|
|
char *s = SvPV_nolen(val); |
|
860
|
|
|
|
|
|
|
|
|
861
|
491
|
50
|
|
|
|
|
if (strcasecmp(s, "DEFAULT") == 0) { |
|
862
|
0
|
|
|
|
|
|
warn("_config: " |
|
863
|
|
|
|
|
|
|
"Method name \"DEFAULT\" for Format option was " |
|
864
|
|
|
|
|
|
|
"obsoleted. Use \"SIMPLE\""); |
|
865
|
0
|
|
|
|
|
|
linebreak_set_format(self, linebreak_format_SIMPLE, |
|
866
|
|
|
|
|
|
|
NULL); |
|
867
|
491
|
100
|
|
|
|
|
} else if (strcasecmp(s, "SIMPLE") == 0) |
|
868
|
483
|
|
|
|
|
|
linebreak_set_format(self, linebreak_format_SIMPLE, |
|
869
|
|
|
|
|
|
|
NULL); |
|
870
|
8
|
100
|
|
|
|
|
else if (strcasecmp(s, "NEWLINE") == 0) |
|
871
|
4
|
|
|
|
|
|
linebreak_set_format(self, linebreak_format_NEWLINE, |
|
872
|
|
|
|
|
|
|
NULL); |
|
873
|
4
|
50
|
|
|
|
|
else if (strcasecmp(s, "TRIM") == 0) |
|
874
|
4
|
|
|
|
|
|
linebreak_set_format(self, linebreak_format_TRIM, |
|
875
|
|
|
|
|
|
|
NULL); |
|
876
|
|
|
|
|
|
|
else |
|
877
|
509
|
|
|
|
|
|
croak("_config: Unknown Format option: %s", s); |
|
878
|
|
|
|
|
|
|
} |
|
879
|
7231
|
100
|
|
|
|
|
} else if (strcasecmp(key, "Sizing") == 0) { |
|
880
|
491
|
50
|
|
|
|
|
if (! SvOK(val)) |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
881
|
0
|
|
|
|
|
|
linebreak_set_sizing(self, NULL, NULL); |
|
882
|
491
|
100
|
|
|
|
|
else if (sv_derived_from(val, "CODE")) |
|
883
|
8
|
|
|
|
|
|
linebreak_set_sizing(self, sizing_func, (void *)val); |
|
884
|
|
|
|
|
|
|
else { |
|
885
|
483
|
50
|
|
|
|
|
char *s = SvPV_nolen(val); |
|
886
|
|
|
|
|
|
|
|
|
887
|
483
|
50
|
|
|
|
|
if (strcasecmp(s, "DEFAULT") == 0) { |
|
888
|
0
|
|
|
|
|
|
warn("_config: " |
|
889
|
|
|
|
|
|
|
"Method name \"DEFAULT\" for Sizing option " |
|
890
|
|
|
|
|
|
|
"was obsoleted. Use \"UAX11\""); |
|
891
|
0
|
|
|
|
|
|
linebreak_set_sizing(self, linebreak_sizing_UAX11, |
|
892
|
|
|
|
|
|
|
NULL); |
|
893
|
483
|
50
|
|
|
|
|
} else if (strcasecmp(s, "UAX11") == 0) |
|
894
|
483
|
|
|
|
|
|
linebreak_set_sizing(self, linebreak_sizing_UAX11, |
|
895
|
|
|
|
|
|
|
NULL); |
|
896
|
|
|
|
|
|
|
else |
|
897
|
491
|
|
|
|
|
|
croak("_config: Unknown Sizing option: %s", s); |
|
898
|
|
|
|
|
|
|
} |
|
899
|
6740
|
100
|
|
|
|
|
} else if (strcasecmp(key, "Urgent") == 0) { |
|
900
|
489
|
100
|
|
|
|
|
if (! SvOK(val)) |
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
901
|
483
|
|
|
|
|
|
linebreak_set_urgent(self, NULL, NULL); |
|
902
|
6
|
50
|
|
|
|
|
else if (sv_derived_from(val, "CODE")) |
|
903
|
0
|
|
|
|
|
|
linebreak_set_urgent(self, urgent_func, (void *)val); |
|
904
|
|
|
|
|
|
|
else { |
|
905
|
6
|
50
|
|
|
|
|
char *s = SvPV_nolen(val); |
|
906
|
|
|
|
|
|
|
|
|
907
|
6
|
50
|
|
|
|
|
if (strcasecmp(s, "NONBREAK") == 0) { |
|
908
|
0
|
|
|
|
|
|
warn("_config: " |
|
909
|
|
|
|
|
|
|
"Method name \"NONBREAK\" for Urgent " |
|
910
|
|
|
|
|
|
|
"option was obsoleted. Use undef"); |
|
911
|
0
|
|
|
|
|
|
linebreak_set_urgent(self, NULL, NULL); |
|
912
|
6
|
100
|
|
|
|
|
} else if (strcasecmp(s, "CROAK") == 0) |
|
913
|
2
|
|
|
|
|
|
linebreak_set_urgent(self, linebreak_urgent_ABORT, |
|
914
|
|
|
|
|
|
|
NULL); |
|
915
|
4
|
50
|
|
|
|
|
else if (strcasecmp(s, "FORCE") == 0) |
|
916
|
4
|
|
|
|
|
|
linebreak_set_urgent(self, linebreak_urgent_FORCE, |
|
917
|
|
|
|
|
|
|
NULL); |
|
918
|
|
|
|
|
|
|
else |
|
919
|
489
|
|
|
|
|
|
croak("_config: Unknown Urgent option: %s", s); |
|
920
|
|
|
|
|
|
|
} |
|
921
|
6251
|
100
|
|
|
|
|
} else if (strcasecmp(key, "BreakIndent") == 0) { |
|
922
|
358
|
100
|
|
|
|
|
if (SVtoboolean(val)) |
|
923
|
357
|
|
|
|
|
|
self->options |= LINEBREAK_OPTION_BREAK_INDENT; |
|
924
|
|
|
|
|
|
|
else |
|
925
|
358
|
|
|
|
|
|
self->options &= ~LINEBREAK_OPTION_BREAK_INDENT; |
|
926
|
5893
|
100
|
|
|
|
|
} else if (strcasecmp(key, "CharMax") == 0) |
|
927
|
485
|
50
|
|
|
|
|
self->charmax = SvUV(val); |
|
928
|
5408
|
100
|
|
|
|
|
else if (strcasecmp(key, "ColMax") == 0) |
|
929
|
538
|
100
|
|
|
|
|
self->colmax = (double)SvNV(val); |
|
930
|
4870
|
100
|
|
|
|
|
else if (strcasecmp(key, "ColMin") == 0) |
|
931
|
485
|
50
|
|
|
|
|
self->colmin = (double)SvNV(val); |
|
932
|
4385
|
100
|
|
|
|
|
else if (strcasecmp(key, "ComplexBreaking") == 0) { |
|
933
|
358
|
100
|
|
|
|
|
if (SVtoboolean(val)) |
|
934
|
357
|
|
|
|
|
|
self->options |= LINEBREAK_OPTION_COMPLEX_BREAKING; |
|
935
|
|
|
|
|
|
|
else |
|
936
|
358
|
|
|
|
|
|
self->options &= ~LINEBREAK_OPTION_COMPLEX_BREAKING; |
|
937
|
4027
|
100
|
|
|
|
|
} else if (strcasecmp(key, "Context") == 0) { |
|
938
|
546
|
50
|
|
|
|
|
if (SvOK(val)) |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
939
|
546
|
50
|
|
|
|
|
opt = (char *)SvPV_nolen(val); |
|
940
|
|
|
|
|
|
|
else |
|
941
|
0
|
|
|
|
|
|
opt = NULL; |
|
942
|
546
|
50
|
|
|
|
|
if (opt && strcasecmp(opt, "EASTASIAN") == 0) |
|
|
|
100
|
|
|
|
|
|
|
943
|
55
|
|
|
|
|
|
self->options |= LINEBREAK_OPTION_EASTASIAN_CONTEXT; |
|
944
|
|
|
|
|
|
|
else |
|
945
|
546
|
|
|
|
|
|
self->options &= ~LINEBREAK_OPTION_EASTASIAN_CONTEXT; |
|
946
|
3481
|
100
|
|
|
|
|
} else if (strcasecmp(key, "EAWidth") == 0) { |
|
947
|
|
|
|
|
|
|
AV *av, *codes; |
|
948
|
|
|
|
|
|
|
SV *sv; |
|
949
|
|
|
|
|
|
|
propval_t p; |
|
950
|
|
|
|
|
|
|
size_t i; |
|
951
|
|
|
|
|
|
|
|
|
952
|
485
|
100
|
|
|
|
|
if (! SvOK(val)) |
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
953
|
357
|
|
|
|
|
|
linebreak_clear_eawidth(self); |
|
954
|
128
|
50
|
|
|
|
|
else if (SvROK(val) && |
|
|
|
50
|
|
|
|
|
|
|
955
|
128
|
50
|
|
|
|
|
SvTYPE(av = (AV *)SvRV(val)) == SVt_PVAV && |
|
956
|
256
|
50
|
|
|
|
|
av_len(av) + 1 == 2 && |
|
957
|
256
|
50
|
|
|
|
|
av_fetch(av, 0, 0) != NULL && av_fetch(av, 1, 0) != NULL) { |
|
958
|
128
|
|
|
|
|
|
sv = *av_fetch(av, 1, 0); |
|
959
|
128
|
50
|
|
|
|
|
if (SvIOK(sv)) |
|
960
|
128
|
50
|
|
|
|
|
p = (propval_t) SvIV(sv); |
|
961
|
|
|
|
|
|
|
else |
|
962
|
0
|
|
|
|
|
|
croak("_config: Invalid argument"); |
|
963
|
|
|
|
|
|
|
|
|
964
|
128
|
|
|
|
|
|
sv = *av_fetch(av, 0, 0); |
|
965
|
256
|
50
|
|
|
|
|
if (SvROK(sv) && |
|
|
|
50
|
|
|
|
|
|
|
966
|
128
|
|
|
|
|
|
SvTYPE(codes = (AV *)SvRV(sv)) == SVt_PVAV) { |
|
967
|
66084
|
100
|
|
|
|
|
for (i = 0; i < av_len(codes) + 1; i++) { |
|
968
|
65956
|
50
|
|
|
|
|
if (av_fetch(codes, i, 0) == NULL) |
|
969
|
0
|
|
|
|
|
|
continue; |
|
970
|
65956
|
50
|
|
|
|
|
if (! SvIOK(sv = *av_fetch(codes, i, 0))) |
|
971
|
0
|
|
|
|
|
|
croak("_config: Invalid argument"); |
|
972
|
131912
|
50
|
|
|
|
|
linebreak_update_eawidth(self, |
|
973
|
65956
|
|
|
|
|
|
(unichar_t) SvUV(sv), p); |
|
974
|
|
|
|
|
|
|
} |
|
975
|
0
|
0
|
|
|
|
|
} else if (SvIOK(sv)) { |
|
976
|
0
|
0
|
|
|
|
|
linebreak_update_eawidth(self, (unichar_t) SvUV(sv), |
|
977
|
|
|
|
|
|
|
p); |
|
978
|
|
|
|
|
|
|
} else |
|
979
|
0
|
|
|
|
|
|
croak("_config: Invalid argument"); |
|
980
|
|
|
|
|
|
|
} else |
|
981
|
485
|
|
|
|
|
|
croak("_config: Invalid argument"); |
|
982
|
2996
|
100
|
|
|
|
|
} else if (strcasecmp(key, "HangulAsAL") == 0) { |
|
983
|
484
|
100
|
|
|
|
|
if (SVtoboolean(val)) |
|
984
|
1
|
|
|
|
|
|
self->options |= LINEBREAK_OPTION_HANGUL_AS_AL; |
|
985
|
|
|
|
|
|
|
else |
|
986
|
484
|
|
|
|
|
|
self->options &= ~LINEBREAK_OPTION_HANGUL_AS_AL; |
|
987
|
2512
|
100
|
|
|
|
|
} else if (strcasecmp(key, "LBClass") == 0) { |
|
988
|
|
|
|
|
|
|
AV *av, *codes; |
|
989
|
|
|
|
|
|
|
SV *sv; |
|
990
|
|
|
|
|
|
|
propval_t p; |
|
991
|
|
|
|
|
|
|
size_t i; |
|
992
|
|
|
|
|
|
|
|
|
993
|
492
|
100
|
|
|
|
|
if (! SvOK(val)) |
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
994
|
483
|
|
|
|
|
|
linebreak_clear_lbclass(self); |
|
995
|
9
|
50
|
|
|
|
|
else if (SvROK(val) && |
|
|
|
50
|
|
|
|
|
|
|
996
|
9
|
50
|
|
|
|
|
SvTYPE(av = (AV *)SvRV(val)) == SVt_PVAV && |
|
997
|
18
|
50
|
|
|
|
|
av_len(av) + 1 == 2 && |
|
998
|
18
|
50
|
|
|
|
|
av_fetch(av, 0, 0) != NULL && av_fetch(av, 1, 0) != NULL) { |
|
999
|
9
|
|
|
|
|
|
sv = *av_fetch(av, 1, 0); |
|
1000
|
9
|
50
|
|
|
|
|
if (SvIOK(sv)) |
|
1001
|
9
|
50
|
|
|
|
|
p = (propval_t) SvIV(sv); |
|
1002
|
|
|
|
|
|
|
else |
|
1003
|
0
|
|
|
|
|
|
croak("_config: Invalid argument"); |
|
1004
|
|
|
|
|
|
|
|
|
1005
|
9
|
|
|
|
|
|
sv = *av_fetch(av, 0, 0); |
|
1006
|
18
|
100
|
|
|
|
|
if (SvROK(sv) && |
|
|
|
50
|
|
|
|
|
|
|
1007
|
1
|
|
|
|
|
|
SvTYPE(codes = (AV *)SvRV(sv)) == SVt_PVAV) { |
|
1008
|
59
|
100
|
|
|
|
|
for (i = 0; i < av_len(codes) + 1; i++) { |
|
1009
|
58
|
50
|
|
|
|
|
if (av_fetch(codes, i, 0) == NULL) |
|
1010
|
0
|
|
|
|
|
|
continue; |
|
1011
|
58
|
50
|
|
|
|
|
if (! SvIOK(sv = *av_fetch(codes, i, 0))) |
|
1012
|
0
|
|
|
|
|
|
croak("_config: Invalid argument"); |
|
1013
|
116
|
50
|
|
|
|
|
linebreak_update_lbclass(self, |
|
1014
|
58
|
|
|
|
|
|
(unichar_t) SvUV(sv), p); |
|
1015
|
|
|
|
|
|
|
} |
|
1016
|
8
|
50
|
|
|
|
|
} else if (SvIOK(sv)) { |
|
1017
|
8
|
50
|
|
|
|
|
linebreak_update_lbclass(self, (unichar_t) SvUV(sv), |
|
1018
|
|
|
|
|
|
|
p); |
|
1019
|
|
|
|
|
|
|
} else |
|
1020
|
0
|
|
|
|
|
|
croak("_config: Invalid argument"); |
|
1021
|
|
|
|
|
|
|
} else |
|
1022
|
492
|
|
|
|
|
|
croak("_config: Invalid argument"); |
|
1023
|
2020
|
100
|
|
|
|
|
} else if (strcasecmp(key, "LegacyCM") == 0) { |
|
1024
|
832
|
100
|
|
|
|
|
if (SVtoboolean(val)) |
|
1025
|
831
|
|
|
|
|
|
self->options |= LINEBREAK_OPTION_LEGACY_CM; |
|
1026
|
|
|
|
|
|
|
else |
|
1027
|
832
|
|
|
|
|
|
self->options &= ~LINEBREAK_OPTION_LEGACY_CM; |
|
1028
|
1188
|
100
|
|
|
|
|
} else if (strcasecmp(key, "Newline") == 0) { |
|
1029
|
483
|
50
|
|
|
|
|
if (!sv_isobject(val)) { |
|
1030
|
483
|
|
|
|
|
|
unistr_t unistr = {NULL, 0}; |
|
1031
|
483
|
|
|
|
|
|
SVtounistr(&unistr, val); |
|
1032
|
483
|
|
|
|
|
|
linebreak_set_newline(self, &unistr); |
|
1033
|
483
|
|
|
|
|
|
free(unistr.str); |
|
1034
|
0
|
0
|
|
|
|
|
} else if (sv_derived_from(val, "Unicode::GCString")) |
|
1035
|
0
|
0
|
|
|
|
|
linebreak_set_newline(self, |
|
1036
|
0
|
|
|
|
|
|
(unistr_t *)PerltoC(gcstring_t *, |
|
1037
|
|
|
|
|
|
|
val)); |
|
1038
|
|
|
|
|
|
|
else |
|
1039
|
0
|
0
|
|
|
|
|
croak("_config: Unknown object %s", |
|
|
|
0
|
|
|
|
|
|
|
1040
|
0
|
0
|
|
|
|
|
HvNAME(SvSTASH(SvRV(val)))); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
1041
|
705
|
50
|
|
|
|
|
} else if (strcasecmp(key, "ViramaAsJoiner") == 0) { |
|
1042
|
705
|
100
|
|
|
|
|
if (SVtoboolean(val)) |
|
1043
|
357
|
|
|
|
|
|
self->options |= LINEBREAK_OPTION_VIRAMA_AS_JOINER; |
|
1044
|
|
|
|
|
|
|
else |
|
1045
|
705
|
|
|
|
|
|
self->options &= ~LINEBREAK_OPTION_VIRAMA_AS_JOINER; |
|
1046
|
|
|
|
|
|
|
} else |
|
1047
|
0
|
|
|
|
|
|
warn("_config: Setting unknown option %s", key); |
|
1048
|
|
|
|
|
|
|
} |
|
1049
|
|
|
|
|
|
|
OUTPUT: |
|
1050
|
|
|
|
|
|
|
RETVAL |
|
1051
|
|
|
|
|
|
|
|
|
1052
|
|
|
|
|
|
|
void |
|
1053
|
|
|
|
|
|
|
as_hashref(self, ...) |
|
1054
|
|
|
|
|
|
|
linebreak_t *self; |
|
1055
|
|
|
|
|
|
|
CODE: |
|
1056
|
2971
|
50
|
|
|
|
|
if (self->stash == NULL) |
|
1057
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
1058
|
2971
|
|
|
|
|
|
ST(0) = self->stash; /* should not be mortal */ |
|
1059
|
2971
|
|
|
|
|
|
XSRETURN(1); |
|
1060
|
|
|
|
|
|
|
|
|
1061
|
|
|
|
|
|
|
SV* |
|
1062
|
|
|
|
|
|
|
as_scalarref(self, ...) |
|
1063
|
|
|
|
|
|
|
linebreak_t *self; |
|
1064
|
|
|
|
|
|
|
PREINIT: |
|
1065
|
|
|
|
|
|
|
char buf[64]; |
|
1066
|
|
|
|
|
|
|
CODE: |
|
1067
|
0
|
|
|
|
|
|
buf[0] = '\0'; |
|
1068
|
0
|
0
|
|
|
|
|
snprintf(buf, 64, "%s(0x%lx)", HvNAME(SvSTASH(SvRV(ST(0)))), |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
1069
|
|
|
|
|
|
|
(unsigned long)(void *)self); |
|
1070
|
0
|
|
|
|
|
|
RETVAL = newRV_noinc(newSVpv(buf, 0)); |
|
1071
|
|
|
|
|
|
|
OUTPUT: |
|
1072
|
|
|
|
|
|
|
RETVAL |
|
1073
|
|
|
|
|
|
|
|
|
1074
|
|
|
|
|
|
|
SV * |
|
1075
|
|
|
|
|
|
|
as_string(self, ...) |
|
1076
|
|
|
|
|
|
|
linebreak_t *self; |
|
1077
|
|
|
|
|
|
|
PREINIT: |
|
1078
|
|
|
|
|
|
|
char buf[64]; |
|
1079
|
|
|
|
|
|
|
CODE: |
|
1080
|
0
|
|
|
|
|
|
buf[0] = '\0'; |
|
1081
|
0
|
0
|
|
|
|
|
snprintf(buf, 64, "%s(0x%lx)", HvNAME(SvSTASH(SvRV(ST(0)))), |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
1082
|
|
|
|
|
|
|
(unsigned long)(void *)self); |
|
1083
|
0
|
|
|
|
|
|
RETVAL = newSVpv(buf, 0); |
|
1084
|
|
|
|
|
|
|
OUTPUT: |
|
1085
|
|
|
|
|
|
|
RETVAL |
|
1086
|
|
|
|
|
|
|
|
|
1087
|
|
|
|
|
|
|
propval_t |
|
1088
|
|
|
|
|
|
|
lbrule(self, b_idx, a_idx) |
|
1089
|
|
|
|
|
|
|
linebreak_t *self; |
|
1090
|
|
|
|
|
|
|
propval_t b_idx; |
|
1091
|
|
|
|
|
|
|
propval_t a_idx; |
|
1092
|
|
|
|
|
|
|
PROTOTYPE: $$$ |
|
1093
|
|
|
|
|
|
|
CODE: |
|
1094
|
0
|
|
|
|
|
|
warn("lbrule() is obsoleted. Use breakingRule()"); |
|
1095
|
0
|
0
|
|
|
|
|
if (!SvOK(ST(1)) || !SvOK(ST(2))) |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
1096
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
1097
|
0
|
0
|
|
|
|
|
if (self == NULL) |
|
1098
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
1099
|
0
|
|
|
|
|
|
RETVAL = linebreak_get_lbrule(self, b_idx, a_idx); |
|
1100
|
0
|
0
|
|
|
|
|
if (RETVAL == PROP_UNKNOWN) |
|
1101
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
1102
|
|
|
|
|
|
|
OUTPUT: |
|
1103
|
|
|
|
|
|
|
RETVAL |
|
1104
|
|
|
|
|
|
|
|
|
1105
|
|
|
|
|
|
|
propval_t |
|
1106
|
|
|
|
|
|
|
breakingRule(lbobj, bgcstr, agcstr) |
|
1107
|
|
|
|
|
|
|
linebreak_t *lbobj; |
|
1108
|
|
|
|
|
|
|
generic_string bgcstr; |
|
1109
|
|
|
|
|
|
|
generic_string agcstr; |
|
1110
|
|
|
|
|
|
|
PROTOTYPE: $$$ |
|
1111
|
|
|
|
|
|
|
PREINIT: |
|
1112
|
|
|
|
|
|
|
propval_t blbc, albc; |
|
1113
|
|
|
|
|
|
|
CODE: |
|
1114
|
18
|
50
|
|
|
|
|
if (!SvOK(ST(1)) || !SvOK(ST(2))) |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
1115
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
1116
|
18
|
50
|
|
|
|
|
if (lbobj == NULL) |
|
1117
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
1118
|
18
|
50
|
|
|
|
|
if ((blbc = gcstring_lbclass_ext(bgcstr, -1)) == PROP_UNKNOWN) |
|
1119
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
1120
|
18
|
50
|
|
|
|
|
if ((albc = gcstring_lbclass(agcstr, 0)) == PROP_UNKNOWN) |
|
1121
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
1122
|
18
|
|
|
|
|
|
RETVAL = linebreak_get_lbrule(lbobj, blbc, albc); |
|
1123
|
18
|
50
|
|
|
|
|
if (RETVAL == PROP_UNKNOWN) |
|
1124
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
1125
|
|
|
|
|
|
|
OUTPUT: |
|
1126
|
|
|
|
|
|
|
RETVAL |
|
1127
|
|
|
|
|
|
|
|
|
1128
|
|
|
|
|
|
|
void |
|
1129
|
|
|
|
|
|
|
reset(self) |
|
1130
|
|
|
|
|
|
|
linebreak_t *self; |
|
1131
|
|
|
|
|
|
|
PROTOTYPE: $ |
|
1132
|
|
|
|
|
|
|
CODE: |
|
1133
|
0
|
|
|
|
|
|
linebreak_reset(self); |
|
1134
|
|
|
|
|
|
|
|
|
1135
|
|
|
|
|
|
|
double |
|
1136
|
|
|
|
|
|
|
strsize(lbobj, len, pre, spc, str, ...) |
|
1137
|
|
|
|
|
|
|
linebreak_t *lbobj; |
|
1138
|
|
|
|
|
|
|
double len; |
|
1139
|
|
|
|
|
|
|
SV *pre; |
|
1140
|
|
|
|
|
|
|
generic_string spc; |
|
1141
|
|
|
|
|
|
|
generic_string str; |
|
1142
|
|
|
|
|
|
|
PROTOTYPE: $$$$$;$ |
|
1143
|
|
|
|
|
|
|
CODE: |
|
1144
|
0
|
|
|
|
|
|
warn("strsize() is obsoleted. Use Unicode::GCString::columns"); |
|
1145
|
0
|
0
|
|
|
|
|
if (5 < items) |
|
1146
|
0
|
|
|
|
|
|
warn("``max'' argument of strsize was obsoleted"); |
|
1147
|
|
|
|
|
|
|
|
|
1148
|
0
|
|
|
|
|
|
RETVAL = linebreak_sizing_UAX11(lbobj, len, NULL, spc, str); |
|
1149
|
0
|
0
|
|
|
|
|
if (RETVAL == -1.0) |
|
1150
|
0
|
|
|
|
|
|
croak("strsize: %s", strerror(lbobj->errnum)); |
|
1151
|
|
|
|
|
|
|
OUTPUT: |
|
1152
|
|
|
|
|
|
|
RETVAL |
|
1153
|
|
|
|
|
|
|
|
|
1154
|
|
|
|
|
|
|
void |
|
1155
|
|
|
|
|
|
|
break(self, input) |
|
1156
|
|
|
|
|
|
|
linebreak_t *self; |
|
1157
|
|
|
|
|
|
|
unistr_t *input; |
|
1158
|
|
|
|
|
|
|
PROTOTYPE: $$ |
|
1159
|
|
|
|
|
|
|
PREINIT: |
|
1160
|
|
|
|
|
|
|
gcstring_t **ret, *r; |
|
1161
|
|
|
|
|
|
|
size_t i; |
|
1162
|
|
|
|
|
|
|
PPCODE: |
|
1163
|
6382
|
50
|
|
|
|
|
if (input == NULL) |
|
1164
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
1165
|
6382
|
|
|
|
|
|
ret = linebreak_break(self, input); |
|
1166
|
|
|
|
|
|
|
|
|
1167
|
6382
|
100
|
|
|
|
|
if (ret == NULL) { |
|
1168
|
2
|
50
|
|
|
|
|
if (self->errnum == LINEBREAK_EEXTN) |
|
1169
|
0
|
0
|
|
|
|
|
croak("%s", SvPV_nolen(ERRSV)); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
1170
|
2
|
50
|
|
|
|
|
else if (self->errnum == LINEBREAK_ELONG) |
|
1171
|
2
|
|
|
|
|
|
croak("%s", "Excessive line was found"); |
|
1172
|
0
|
0
|
|
|
|
|
else if (self->errnum) |
|
1173
|
0
|
|
|
|
|
|
croak("%s", strerror(self->errnum)); |
|
1174
|
|
|
|
|
|
|
else |
|
1175
|
0
|
|
|
|
|
|
croak("%s", "Unknown error"); |
|
1176
|
|
|
|
|
|
|
} |
|
1177
|
|
|
|
|
|
|
|
|
1178
|
6380
|
50
|
|
|
|
|
switch (GIMME_V) { |
|
1179
|
|
|
|
|
|
|
case G_SCALAR: |
|
1180
|
51
|
|
|
|
|
|
r = gcstring_new(NULL, self); |
|
1181
|
1170
|
100
|
|
|
|
|
for (i = 0; ret[i] != NULL; i++) |
|
1182
|
1119
|
|
|
|
|
|
gcstring_append(r, ret[i]); |
|
1183
|
51
|
|
|
|
|
|
linebreak_free_result(ret, 1); |
|
1184
|
51
|
50
|
|
|
|
|
XPUSHs(sv_2mortal(unistrtoSV((unistr_t *)r, 0, r->len))); |
|
1185
|
51
|
|
|
|
|
|
gcstring_destroy(r); |
|
1186
|
51
|
|
|
|
|
|
XSRETURN(1); |
|
1187
|
|
|
|
|
|
|
|
|
1188
|
|
|
|
|
|
|
case G_ARRAY: |
|
1189
|
17128
|
100
|
|
|
|
|
for (i = 0; ret[i] != NULL; i++) |
|
1190
|
10799
|
50
|
|
|
|
|
XPUSHs(sv_2mortal(CtoPerl("Unicode::GCString", ret[i]))); |
|
1191
|
6329
|
|
|
|
|
|
linebreak_free_result(ret, 0); |
|
1192
|
6329
|
|
|
|
|
|
XSRETURN(i); |
|
1193
|
|
|
|
|
|
|
|
|
1194
|
|
|
|
|
|
|
default: |
|
1195
|
0
|
|
|
|
|
|
linebreak_free_result(ret, 1); |
|
1196
|
0
|
|
|
|
|
|
XSRETURN_EMPTY; |
|
1197
|
|
|
|
|
|
|
} |
|
1198
|
|
|
|
|
|
|
|
|
1199
|
|
|
|
|
|
|
void |
|
1200
|
|
|
|
|
|
|
break_partial(self, input) |
|
1201
|
|
|
|
|
|
|
linebreak_t *self; |
|
1202
|
|
|
|
|
|
|
unistr_t *input; |
|
1203
|
|
|
|
|
|
|
PROTOTYPE: $$ |
|
1204
|
|
|
|
|
|
|
PREINIT: |
|
1205
|
|
|
|
|
|
|
gcstring_t **ret, *r; |
|
1206
|
|
|
|
|
|
|
size_t i; |
|
1207
|
|
|
|
|
|
|
PPCODE: |
|
1208
|
2346
|
|
|
|
|
|
ret = linebreak_break_partial(self, input); |
|
1209
|
|
|
|
|
|
|
|
|
1210
|
2346
|
50
|
|
|
|
|
if (ret == NULL) { |
|
1211
|
0
|
0
|
|
|
|
|
if (self->errnum == LINEBREAK_EEXTN) |
|
1212
|
0
|
0
|
|
|
|
|
croak("%s", SvPV_nolen(ERRSV)); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
1213
|
0
|
0
|
|
|
|
|
else if (self->errnum == LINEBREAK_ELONG) |
|
1214
|
0
|
|
|
|
|
|
croak("%s", "Excessive line was found"); |
|
1215
|
0
|
0
|
|
|
|
|
else if (self->errnum) |
|
1216
|
0
|
|
|
|
|
|
croak("%s", strerror(self->errnum)); |
|
1217
|
|
|
|
|
|
|
else |
|
1218
|
0
|
|
|
|
|
|
croak("%s", "Unknown error"); |
|
1219
|
|
|
|
|
|
|
} |
|
1220
|
|
|
|
|
|
|
|
|
1221
|
2346
|
50
|
|
|
|
|
switch (GIMME_V) { |
|
1222
|
|
|
|
|
|
|
case G_SCALAR: |
|
1223
|
2346
|
|
|
|
|
|
r = gcstring_new(NULL, self); |
|
1224
|
2640
|
100
|
|
|
|
|
for (i = 0; ret[i] != NULL; i++) |
|
1225
|
294
|
|
|
|
|
|
gcstring_append(r, ret[i]); |
|
1226
|
2346
|
|
|
|
|
|
linebreak_free_result(ret, 1); |
|
1227
|
2346
|
50
|
|
|
|
|
XPUSHs(sv_2mortal(unistrtoSV((unistr_t *)r, 0, r->len))); |
|
1228
|
2346
|
|
|
|
|
|
gcstring_destroy(r); |
|
1229
|
2346
|
|
|
|
|
|
XSRETURN(1); |
|
1230
|
|
|
|
|
|
|
|
|
1231
|
|
|
|
|
|
|
case G_ARRAY: |
|
1232
|
0
|
0
|
|
|
|
|
for (i = 0; ret[i] != NULL; i++) |
|
1233
|
0
|
0
|
|
|
|
|
XPUSHs(sv_2mortal(CtoPerl("Unicode::GCString", ret[i]))); |
|
1234
|
0
|
|
|
|
|
|
linebreak_free_result(ret, 0); |
|
1235
|
0
|
|
|
|
|
|
XSRETURN(i); |
|
1236
|
|
|
|
|
|
|
|
|
1237
|
|
|
|
|
|
|
default: |
|
1238
|
0
|
|
|
|
|
|
linebreak_free_result(ret, 1); |
|
1239
|
0
|
|
|
|
|
|
XSRETURN_EMPTY; |
|
1240
|
|
|
|
|
|
|
} |
|
1241
|
|
|
|
|
|
|
|
|
1242
|
|
|
|
|
|
|
const char * |
|
1243
|
|
|
|
|
|
|
UNICODE_VERSION() |
|
1244
|
|
|
|
|
|
|
CODE: |
|
1245
|
1
|
|
|
|
|
|
RETVAL = linebreak_unicode_version; |
|
1246
|
|
|
|
|
|
|
OUTPUT: |
|
1247
|
|
|
|
|
|
|
RETVAL |
|
1248
|
|
|
|
|
|
|
|
|
1249
|
|
|
|
|
|
|
const char * |
|
1250
|
|
|
|
|
|
|
SOMBOK_VERSION() |
|
1251
|
|
|
|
|
|
|
CODE: |
|
1252
|
1
|
|
|
|
|
|
RETVAL = SOMBOK_VERSION; |
|
1253
|
|
|
|
|
|
|
OUTPUT: |
|
1254
|
|
|
|
|
|
|
RETVAL |
|
1255
|
|
|
|
|
|
|
|
|
1256
|
|
|
|
|
|
|
|
|
1257
|
|
|
|
|
|
|
MODULE = Unicode::LineBreak PACKAGE = Unicode::LineBreak::SouthEastAsian |
|
1258
|
|
|
|
|
|
|
|
|
1259
|
|
|
|
|
|
|
const char * |
|
1260
|
|
|
|
|
|
|
supported() |
|
1261
|
|
|
|
|
|
|
PROTOTYPE: |
|
1262
|
|
|
|
|
|
|
CODE: |
|
1263
|
4
|
|
|
|
|
|
RETVAL = linebreak_southeastasian_supported; |
|
1264
|
4
|
50
|
|
|
|
|
if (RETVAL == NULL) |
|
1265
|
4
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
1266
|
|
|
|
|
|
|
OUTPUT: |
|
1267
|
|
|
|
|
|
|
RETVAL |
|
1268
|
|
|
|
|
|
|
|
|
1269
|
|
|
|
|
|
|
MODULE = Unicode::LineBreak PACKAGE = Unicode::GCString |
|
1270
|
|
|
|
|
|
|
|
|
1271
|
|
|
|
|
|
|
gcstring_t * |
|
1272
|
|
|
|
|
|
|
_new(klass, str, lbobj=NULL) |
|
1273
|
|
|
|
|
|
|
char *klass; |
|
1274
|
|
|
|
|
|
|
unistr_t *str; |
|
1275
|
|
|
|
|
|
|
linebreak_t *lbobj; |
|
1276
|
|
|
|
|
|
|
PROTOTYPE: $$;$ |
|
1277
|
|
|
|
|
|
|
CODE: |
|
1278
|
416
|
50
|
|
|
|
|
if (str == NULL) |
|
1279
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
1280
|
|
|
|
|
|
|
/* FIXME:buffer is copied twice. */ |
|
1281
|
416
|
50
|
|
|
|
|
if ((RETVAL = gcstring_newcopy(str, lbobj)) == NULL) |
|
1282
|
0
|
|
|
|
|
|
croak("%s->_new: %s", klass, strerror(errno)); |
|
1283
|
|
|
|
|
|
|
OUTPUT: |
|
1284
|
|
|
|
|
|
|
RETVAL |
|
1285
|
|
|
|
|
|
|
|
|
1286
|
|
|
|
|
|
|
void |
|
1287
|
|
|
|
|
|
|
DESTROY(self) |
|
1288
|
|
|
|
|
|
|
gcstring_t *self; |
|
1289
|
|
|
|
|
|
|
PROTOTYPE: $ |
|
1290
|
|
|
|
|
|
|
CODE: |
|
1291
|
39084
|
|
|
|
|
|
gcstring_destroy(self); |
|
1292
|
|
|
|
|
|
|
|
|
1293
|
|
|
|
|
|
|
void |
|
1294
|
|
|
|
|
|
|
as_array(self) |
|
1295
|
|
|
|
|
|
|
gcstring_t *self; |
|
1296
|
|
|
|
|
|
|
PROTOTYPE: $ |
|
1297
|
|
|
|
|
|
|
PREINIT: |
|
1298
|
|
|
|
|
|
|
size_t i; |
|
1299
|
|
|
|
|
|
|
PPCODE: |
|
1300
|
356
|
50
|
|
|
|
|
if (self != NULL) |
|
1301
|
1045
|
100
|
|
|
|
|
for (i = 0; i < self->gclen; i++) |
|
1302
|
689
|
50
|
|
|
|
|
XPUSHs(sv_2mortal( |
|
1303
|
|
|
|
|
|
|
CtoPerl("Unicode::GCString", |
|
1304
|
|
|
|
|
|
|
gcstring_substr(self, i, 1)))); |
|
1305
|
|
|
|
|
|
|
|
|
1306
|
|
|
|
|
|
|
SV* |
|
1307
|
|
|
|
|
|
|
as_scalarref(self, ...) |
|
1308
|
|
|
|
|
|
|
gcstring_t *self; |
|
1309
|
|
|
|
|
|
|
PREINIT: |
|
1310
|
|
|
|
|
|
|
char buf[64]; |
|
1311
|
|
|
|
|
|
|
CODE: |
|
1312
|
0
|
|
|
|
|
|
buf[0] = '\0'; |
|
1313
|
0
|
0
|
|
|
|
|
snprintf(buf, 64, "%s(0x%lx)", HvNAME(SvSTASH(SvRV(ST(0)))), |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
1314
|
|
|
|
|
|
|
(unsigned long)(void *)self); |
|
1315
|
0
|
|
|
|
|
|
RETVAL = newRV_noinc(newSVpv(buf, 0)); |
|
1316
|
|
|
|
|
|
|
OUTPUT: |
|
1317
|
|
|
|
|
|
|
RETVAL |
|
1318
|
|
|
|
|
|
|
|
|
1319
|
|
|
|
|
|
|
SV * |
|
1320
|
|
|
|
|
|
|
as_string(self, ...) |
|
1321
|
|
|
|
|
|
|
gcstring_t *self; |
|
1322
|
|
|
|
|
|
|
PROTOTYPE: $;$;$ |
|
1323
|
|
|
|
|
|
|
CODE: |
|
1324
|
11736
|
|
|
|
|
|
RETVAL = unistrtoSV((unistr_t *)self, 0, self->len); |
|
1325
|
|
|
|
|
|
|
OUTPUT: |
|
1326
|
|
|
|
|
|
|
RETVAL |
|
1327
|
|
|
|
|
|
|
|
|
1328
|
|
|
|
|
|
|
size_t |
|
1329
|
|
|
|
|
|
|
chars(self) |
|
1330
|
|
|
|
|
|
|
gcstring_t *self; |
|
1331
|
|
|
|
|
|
|
PROTOTYPE: $ |
|
1332
|
|
|
|
|
|
|
CODE: |
|
1333
|
5
|
|
|
|
|
|
RETVAL = self->len; |
|
1334
|
|
|
|
|
|
|
OUTPUT: |
|
1335
|
|
|
|
|
|
|
RETVAL |
|
1336
|
|
|
|
|
|
|
|
|
1337
|
|
|
|
|
|
|
#define lbobj self->lbobj |
|
1338
|
|
|
|
|
|
|
int |
|
1339
|
|
|
|
|
|
|
cmp(self, str, swap=FALSE) |
|
1340
|
|
|
|
|
|
|
gcstring_t *self; |
|
1341
|
|
|
|
|
|
|
generic_string str; |
|
1342
|
|
|
|
|
|
|
swapspec_t swap; |
|
1343
|
|
|
|
|
|
|
PROTOTYPE: $$;$ |
|
1344
|
|
|
|
|
|
|
CODE: |
|
1345
|
1035
|
50
|
|
|
|
|
if (swap == TRUE) |
|
1346
|
0
|
|
|
|
|
|
RETVAL = gcstring_cmp(str, self); |
|
1347
|
|
|
|
|
|
|
else |
|
1348
|
1035
|
|
|
|
|
|
RETVAL = gcstring_cmp(self, str); |
|
1349
|
|
|
|
|
|
|
OUTPUT: |
|
1350
|
|
|
|
|
|
|
RETVAL |
|
1351
|
|
|
|
|
|
|
|
|
1352
|
|
|
|
|
|
|
size_t |
|
1353
|
|
|
|
|
|
|
columns(self) |
|
1354
|
|
|
|
|
|
|
gcstring_t *self; |
|
1355
|
|
|
|
|
|
|
CODE: |
|
1356
|
3395
|
|
|
|
|
|
RETVAL = gcstring_columns(self); |
|
1357
|
|
|
|
|
|
|
OUTPUT: |
|
1358
|
|
|
|
|
|
|
RETVAL |
|
1359
|
|
|
|
|
|
|
|
|
1360
|
|
|
|
|
|
|
#define lbobj self->lbobj |
|
1361
|
|
|
|
|
|
|
gcstring_t * |
|
1362
|
|
|
|
|
|
|
concat(self, str, swap=FALSE) |
|
1363
|
|
|
|
|
|
|
gcstring_t *self; |
|
1364
|
|
|
|
|
|
|
generic_string str; |
|
1365
|
|
|
|
|
|
|
swapspec_t swap; |
|
1366
|
|
|
|
|
|
|
PROTOTYPE: $$;$ |
|
1367
|
|
|
|
|
|
|
CODE: |
|
1368
|
2457
|
100
|
|
|
|
|
if (swap == TRUE) |
|
1369
|
106
|
|
|
|
|
|
RETVAL = gcstring_concat(str, self); |
|
1370
|
2351
|
100
|
|
|
|
|
else if (swap == -1) { |
|
1371
|
9
|
|
|
|
|
|
gcstring_append(self, str); |
|
1372
|
9
|
|
|
|
|
|
XSRETURN(1); |
|
1373
|
|
|
|
|
|
|
} else |
|
1374
|
2342
|
|
|
|
|
|
RETVAL = gcstring_concat(self, str); |
|
1375
|
|
|
|
|
|
|
OUTPUT: |
|
1376
|
|
|
|
|
|
|
RETVAL |
|
1377
|
|
|
|
|
|
|
|
|
1378
|
|
|
|
|
|
|
gcstring_t * |
|
1379
|
|
|
|
|
|
|
copy(self) |
|
1380
|
|
|
|
|
|
|
gcstring_t *self; |
|
1381
|
|
|
|
|
|
|
PROTOTYPE: $ |
|
1382
|
|
|
|
|
|
|
CODE: |
|
1383
|
1
|
|
|
|
|
|
RETVAL = gcstring_copy(self); |
|
1384
|
|
|
|
|
|
|
OUTPUT: |
|
1385
|
|
|
|
|
|
|
RETVAL |
|
1386
|
|
|
|
|
|
|
|
|
1387
|
|
|
|
|
|
|
int |
|
1388
|
|
|
|
|
|
|
eos(self) |
|
1389
|
|
|
|
|
|
|
gcstring_t *self; |
|
1390
|
|
|
|
|
|
|
CODE: |
|
1391
|
3344
|
|
|
|
|
|
RETVAL = gcstring_eos(self); |
|
1392
|
|
|
|
|
|
|
OUTPUT: |
|
1393
|
|
|
|
|
|
|
RETVAL |
|
1394
|
|
|
|
|
|
|
|
|
1395
|
|
|
|
|
|
|
unsigned int |
|
1396
|
|
|
|
|
|
|
flag(self, ...) |
|
1397
|
|
|
|
|
|
|
gcstring_t *self; |
|
1398
|
|
|
|
|
|
|
PROTOTYPE: $;$;$ |
|
1399
|
|
|
|
|
|
|
PREINIT: |
|
1400
|
|
|
|
|
|
|
int i; |
|
1401
|
|
|
|
|
|
|
unsigned int flag; |
|
1402
|
|
|
|
|
|
|
CODE: |
|
1403
|
0
|
|
|
|
|
|
warn("flag() will be deprecated in near future"); |
|
1404
|
0
|
0
|
|
|
|
|
if (1 < items) |
|
1405
|
0
|
0
|
|
|
|
|
i = SvIV(ST(1)); |
|
1406
|
|
|
|
|
|
|
else |
|
1407
|
0
|
|
|
|
|
|
i = self->pos; |
|
1408
|
0
|
0
|
|
|
|
|
if (i < 0 || self == NULL || self->gclen <= i) |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
1409
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
1410
|
0
|
0
|
|
|
|
|
if (2 < items) { |
|
1411
|
0
|
0
|
|
|
|
|
flag = SvUV(ST(2)); |
|
1412
|
0
|
0
|
|
|
|
|
if (flag == (flag & 255)) |
|
1413
|
0
|
|
|
|
|
|
self->gcstr[i].flag = (unsigned char)flag; |
|
1414
|
|
|
|
|
|
|
else |
|
1415
|
0
|
|
|
|
|
|
warn("flag: unknown flag(s)"); |
|
1416
|
|
|
|
|
|
|
} |
|
1417
|
0
|
|
|
|
|
|
RETVAL = (unsigned int)self->gcstr[i].flag; |
|
1418
|
|
|
|
|
|
|
OUTPUT: |
|
1419
|
|
|
|
|
|
|
RETVAL |
|
1420
|
|
|
|
|
|
|
|
|
1421
|
|
|
|
|
|
|
gcstring_t * |
|
1422
|
|
|
|
|
|
|
item(self, ...) |
|
1423
|
|
|
|
|
|
|
gcstring_t *self; |
|
1424
|
|
|
|
|
|
|
PROTOTYPE: $;$ |
|
1425
|
|
|
|
|
|
|
PREINIT: |
|
1426
|
|
|
|
|
|
|
int i; |
|
1427
|
|
|
|
|
|
|
CODE: |
|
1428
|
3311
|
50
|
|
|
|
|
if (1 < items) |
|
1429
|
0
|
0
|
|
|
|
|
i = SvIV(ST(1)); |
|
1430
|
|
|
|
|
|
|
else |
|
1431
|
3311
|
|
|
|
|
|
i = self->pos; |
|
1432
|
3311
|
50
|
|
|
|
|
if (i < 0 || self == NULL || self->gclen <= i) |
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
1433
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
1434
|
|
|
|
|
|
|
|
|
1435
|
3311
|
|
|
|
|
|
RETVAL = gcstring_substr(self, i, 1); |
|
1436
|
|
|
|
|
|
|
OUTPUT: |
|
1437
|
|
|
|
|
|
|
RETVAL |
|
1438
|
|
|
|
|
|
|
|
|
1439
|
|
|
|
|
|
|
gcstring_t * |
|
1440
|
|
|
|
|
|
|
join(self, ...) |
|
1441
|
|
|
|
|
|
|
gcstring_t *self; |
|
1442
|
|
|
|
|
|
|
PREINIT: |
|
1443
|
|
|
|
|
|
|
size_t i; |
|
1444
|
|
|
|
|
|
|
gcstring_t *str; |
|
1445
|
|
|
|
|
|
|
CODE: |
|
1446
|
0
|
|
|
|
|
|
switch (items) { |
|
1447
|
|
|
|
|
|
|
case 0: |
|
1448
|
0
|
|
|
|
|
|
croak("join: Too few arguments"); |
|
1449
|
|
|
|
|
|
|
case 1: |
|
1450
|
0
|
|
|
|
|
|
RETVAL = gcstring_new(NULL, self->lbobj); |
|
1451
|
0
|
|
|
|
|
|
break; |
|
1452
|
|
|
|
|
|
|
case 2: |
|
1453
|
0
|
|
|
|
|
|
RETVAL = SVtogcstring(ST(1), self->lbobj); |
|
1454
|
0
|
0
|
|
|
|
|
if (sv_isobject(ST(1))) |
|
1455
|
0
|
|
|
|
|
|
RETVAL = gcstring_copy(RETVAL); |
|
1456
|
0
|
|
|
|
|
|
break; |
|
1457
|
|
|
|
|
|
|
default: |
|
1458
|
0
|
|
|
|
|
|
RETVAL = SVtogcstring(ST(1), self->lbobj); |
|
1459
|
0
|
0
|
|
|
|
|
if (sv_isobject(ST(1))) |
|
1460
|
0
|
|
|
|
|
|
RETVAL = gcstring_copy(RETVAL); |
|
1461
|
0
|
0
|
|
|
|
|
for (i = 2; i < items; i++) { |
|
1462
|
0
|
|
|
|
|
|
gcstring_append(RETVAL, self); |
|
1463
|
0
|
|
|
|
|
|
str = SVtogcstring(ST(i), self->lbobj); |
|
1464
|
0
|
|
|
|
|
|
gcstring_append(RETVAL, str); |
|
1465
|
0
|
0
|
|
|
|
|
if (!sv_isobject(ST(i))) |
|
1466
|
0
|
|
|
|
|
|
gcstring_destroy(str); |
|
1467
|
|
|
|
|
|
|
} |
|
1468
|
0
|
|
|
|
|
|
break; |
|
1469
|
|
|
|
|
|
|
} |
|
1470
|
|
|
|
|
|
|
OUTPUT: |
|
1471
|
|
|
|
|
|
|
RETVAL |
|
1472
|
|
|
|
|
|
|
|
|
1473
|
|
|
|
|
|
|
propval_t |
|
1474
|
|
|
|
|
|
|
lbc(self) |
|
1475
|
|
|
|
|
|
|
gcstring_t *self; |
|
1476
|
|
|
|
|
|
|
PROTOTYPE: $ |
|
1477
|
|
|
|
|
|
|
CODE: |
|
1478
|
3365
|
50
|
|
|
|
|
if ((RETVAL = gcstring_lbclass(self, 0)) == PROP_UNKNOWN) |
|
1479
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
1480
|
|
|
|
|
|
|
OUTPUT: |
|
1481
|
|
|
|
|
|
|
RETVAL |
|
1482
|
|
|
|
|
|
|
|
|
1483
|
|
|
|
|
|
|
propval_t |
|
1484
|
|
|
|
|
|
|
lbcext(self) |
|
1485
|
|
|
|
|
|
|
gcstring_t *self; |
|
1486
|
|
|
|
|
|
|
PROTOTYPE: $ |
|
1487
|
|
|
|
|
|
|
CODE: |
|
1488
|
0
|
0
|
|
|
|
|
if ((RETVAL = gcstring_lbclass_ext(self, -1)) == PROP_UNKNOWN) |
|
1489
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
1490
|
|
|
|
|
|
|
OUTPUT: |
|
1491
|
|
|
|
|
|
|
RETVAL |
|
1492
|
|
|
|
|
|
|
|
|
1493
|
|
|
|
|
|
|
propval_t |
|
1494
|
|
|
|
|
|
|
lbclass(self, ...) |
|
1495
|
|
|
|
|
|
|
gcstring_t *self; |
|
1496
|
|
|
|
|
|
|
PROTOTYPE: $;$ |
|
1497
|
|
|
|
|
|
|
PREINIT: |
|
1498
|
|
|
|
|
|
|
int i; |
|
1499
|
|
|
|
|
|
|
CODE: |
|
1500
|
0
|
|
|
|
|
|
warn("lbclass() is obsoleted. Use lbc()"); |
|
1501
|
0
|
0
|
|
|
|
|
if (1 < items) |
|
1502
|
0
|
0
|
|
|
|
|
i = SvIV(ST(1)); |
|
1503
|
|
|
|
|
|
|
else |
|
1504
|
0
|
|
|
|
|
|
i = self->pos; |
|
1505
|
0
|
|
|
|
|
|
RETVAL = gcstring_lbclass(self, i); |
|
1506
|
0
|
0
|
|
|
|
|
if (RETVAL == PROP_UNKNOWN) |
|
1507
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
1508
|
|
|
|
|
|
|
OUTPUT: |
|
1509
|
|
|
|
|
|
|
RETVAL |
|
1510
|
|
|
|
|
|
|
|
|
1511
|
|
|
|
|
|
|
propval_t |
|
1512
|
|
|
|
|
|
|
lbclass_ext(self, ...) |
|
1513
|
|
|
|
|
|
|
gcstring_t *self; |
|
1514
|
|
|
|
|
|
|
PROTOTYPE: $;$ |
|
1515
|
|
|
|
|
|
|
PREINIT: |
|
1516
|
|
|
|
|
|
|
int i; |
|
1517
|
|
|
|
|
|
|
CODE: |
|
1518
|
0
|
|
|
|
|
|
warn("lbclass_ext() is obsoleted. Use lbcext()"); |
|
1519
|
0
|
0
|
|
|
|
|
if (1 < items) |
|
1520
|
0
|
0
|
|
|
|
|
i = SvIV(ST(1)); |
|
1521
|
|
|
|
|
|
|
else |
|
1522
|
0
|
|
|
|
|
|
i = self->pos; |
|
1523
|
0
|
|
|
|
|
|
RETVAL = gcstring_lbclass_ext(self, i); |
|
1524
|
0
|
0
|
|
|
|
|
if (RETVAL == PROP_UNKNOWN) |
|
1525
|
0
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
1526
|
|
|
|
|
|
|
OUTPUT: |
|
1527
|
|
|
|
|
|
|
RETVAL |
|
1528
|
|
|
|
|
|
|
|
|
1529
|
|
|
|
|
|
|
size_t |
|
1530
|
|
|
|
|
|
|
length(self) |
|
1531
|
|
|
|
|
|
|
gcstring_t *self; |
|
1532
|
|
|
|
|
|
|
PROTOTYPE: $ |
|
1533
|
|
|
|
|
|
|
CODE: |
|
1534
|
5
|
|
|
|
|
|
RETVAL = self->gclen; |
|
1535
|
|
|
|
|
|
|
OUTPUT: |
|
1536
|
|
|
|
|
|
|
RETVAL |
|
1537
|
|
|
|
|
|
|
|
|
1538
|
|
|
|
|
|
|
gcstring_t * |
|
1539
|
|
|
|
|
|
|
next(self, ...) |
|
1540
|
|
|
|
|
|
|
gcstring_t *self; |
|
1541
|
|
|
|
|
|
|
PROTOTYPE: $;$;$ |
|
1542
|
|
|
|
|
|
|
PREINIT: |
|
1543
|
|
|
|
|
|
|
gcchar_t *gc; |
|
1544
|
|
|
|
|
|
|
CODE: |
|
1545
|
1024
|
100
|
|
|
|
|
if (gcstring_eos(self)) |
|
1546
|
1
|
|
|
|
|
|
XSRETURN_UNDEF; |
|
1547
|
1023
|
|
|
|
|
|
gc = gcstring_next(self); |
|
1548
|
1023
|
|
|
|
|
|
RETVAL = gcstring_substr(self, gc - self->gcstr, 1); |
|
1549
|
|
|
|
|
|
|
OUTPUT: |
|
1550
|
|
|
|
|
|
|
RETVAL |
|
1551
|
|
|
|
|
|
|
|
|
1552
|
|
|
|
|
|
|
size_t |
|
1553
|
|
|
|
|
|
|
pos(self, ...) |
|
1554
|
|
|
|
|
|
|
gcstring_t *self; |
|
1555
|
|
|
|
|
|
|
PROTOTYPE: $;$ |
|
1556
|
|
|
|
|
|
|
CODE: |
|
1557
|
4656
|
100
|
|
|
|
|
if (1 < items) |
|
1558
|
2328
|
50
|
|
|
|
|
gcstring_setpos(self, SvIV(ST(1))); |
|
1559
|
4656
|
|
|
|
|
|
RETVAL = self->pos; |
|
1560
|
|
|
|
|
|
|
OUTPUT: |
|
1561
|
|
|
|
|
|
|
RETVAL |
|
1562
|
|
|
|
|
|
|
|
|
1563
|
|
|
|
|
|
|
#define lbobj self->lbobj |
|
1564
|
|
|
|
|
|
|
gcstring_t * |
|
1565
|
|
|
|
|
|
|
substr(self, offset, length=self->gclen, replacement=NULL) |
|
1566
|
|
|
|
|
|
|
gcstring_t *self; |
|
1567
|
|
|
|
|
|
|
int offset; |
|
1568
|
|
|
|
|
|
|
int length; |
|
1569
|
|
|
|
|
|
|
generic_string replacement; |
|
1570
|
|
|
|
|
|
|
PROTOTYPE: $$;$;$ |
|
1571
|
|
|
|
|
|
|
CODE: |
|
1572
|
2334
|
|
|
|
|
|
RETVAL = gcstring_substr(self, offset, length); |
|
1573
|
2334
|
100
|
|
|
|
|
if (replacement != NULL) |
|
1574
|
3
|
50
|
|
|
|
|
if (gcstring_replace(self, offset, length, replacement) == NULL) |
|
1575
|
0
|
|
|
|
|
|
croak("substr: %s", strerror(errno)); |
|
1576
|
2334
|
50
|
|
|
|
|
if (RETVAL == NULL) |
|
1577
|
0
|
|
|
|
|
|
croak("substr: %s", strerror(errno)); |
|
1578
|
|
|
|
|
|
|
OUTPUT: |
|
1579
|
|
|
|
|
|
|
RETVAL |
|
1580
|
|
|
|
|
|
|
|