ox 2.13.2 → 2.13.3
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- checksums.yaml +4 -4
- data/CHANGELOG.md +6 -0
- data/README.md +31 -0
- data/ext/ox/builder.c +5 -1
- data/ext/ox/dump.c +6 -18
- data/ext/ox/extconf.rb +12 -34
- data/ext/ox/gen_load.c +18 -96
- data/ext/ox/hash_load.c +2 -10
- data/ext/ox/obj_load.c +8 -45
- data/ext/ox/ox.c +34 -46
- data/ext/ox/ox.h +6 -15
- data/ext/ox/parse.c +1 -20
- data/ext/ox/sax.c +21 -91
- data/ext/ox/sax.h +1 -3
- data/ext/ox/sax_as.c +2 -6
- data/ext/ox/sax_buf.c +1 -1
- data/lib/ox/version.rb +1 -1
- metadata +2 -2
data/ext/ox/sax.c
CHANGED
@@ -9,13 +9,16 @@
|
|
9
9
|
#include <stdio.h>
|
10
10
|
#include <strings.h>
|
11
11
|
#include <sys/types.h>
|
12
|
-
#if
|
12
|
+
#if HAVE_SYS_UIO_H
|
13
13
|
#include <sys/uio.h>
|
14
14
|
#endif
|
15
15
|
#include <unistd.h>
|
16
16
|
#include <time.h>
|
17
17
|
|
18
18
|
#include "ruby.h"
|
19
|
+
#if HAVE_RB_ENC_ASSOCIATE
|
20
|
+
#include "ruby/encoding.h"
|
21
|
+
#endif
|
19
22
|
#include "ox.h"
|
20
23
|
#include "sax.h"
|
21
24
|
#include "sax_stack.h"
|
@@ -68,9 +71,9 @@ static VALUE protect_parse(VALUE drp) {
|
|
68
71
|
return Qnil;
|
69
72
|
}
|
70
73
|
|
71
|
-
#if
|
74
|
+
#if HAVE_RB_ENC_ASSOCIATE
|
72
75
|
static int
|
73
|
-
|
76
|
+
str_is_ascii(const char *s) {
|
74
77
|
for (; '\0' != *s; s++) {
|
75
78
|
if (*s < ' ' || '~' < *s) {
|
76
79
|
return 0;
|
@@ -87,8 +90,8 @@ str2sym(SaxDrive dr, const char *str, const char **strp) {
|
|
87
90
|
|
88
91
|
if (dr->options.symbolize) {
|
89
92
|
if (Qundef == (sym = ox_cache_get(ox_symbol_cache, str, &slot, strp))) {
|
90
|
-
#if
|
91
|
-
if (0 != dr->encoding && !
|
93
|
+
#if HAVE_RB_ENC_ASSOCIATE
|
94
|
+
if (0 != dr->encoding && !str_is_ascii(str)) {
|
92
95
|
VALUE rstr = rb_str_new2(str);
|
93
96
|
|
94
97
|
// TBD if sym can be pinned down then use this all the time
|
@@ -99,20 +102,6 @@ str2sym(SaxDrive dr, const char *str, const char **strp) {
|
|
99
102
|
sym = ID2SYM(rb_intern(str));
|
100
103
|
*slot = sym;
|
101
104
|
}
|
102
|
-
#elif HAS_PRIVATE_ENCODING
|
103
|
-
if (Qnil != dr->encoding && !strIsAscii(str)) {
|
104
|
-
VALUE rstr = rb_str_new2(str);
|
105
|
-
|
106
|
-
rb_funcall(rstr, ox_force_encoding_id, 1, dr->encoding);
|
107
|
-
sym = rb_funcall(rstr, ox_to_sym_id, 0);
|
108
|
-
// Needed for Ruby 2.2 to get around the GC of symbols created
|
109
|
-
// with to_sym which is needed for encoded symbols.
|
110
|
-
rb_ary_push(ox_sym_bank, sym);
|
111
|
-
*slot = Qundef;
|
112
|
-
} else {
|
113
|
-
sym = ID2SYM(rb_intern(str));
|
114
|
-
*slot = sym;
|
115
|
-
}
|
116
105
|
#else
|
117
106
|
sym = ID2SYM(rb_intern(str));
|
118
107
|
*slot = sym;
|
@@ -120,14 +109,10 @@ str2sym(SaxDrive dr, const char *str, const char **strp) {
|
|
120
109
|
}
|
121
110
|
} else {
|
122
111
|
sym = rb_str_new2(str);
|
123
|
-
#if
|
112
|
+
#if HAVE_RB_ENC_ASSOCIATE
|
124
113
|
if (0 != dr->encoding) {
|
125
114
|
rb_enc_associate(sym, dr->encoding);
|
126
115
|
}
|
127
|
-
#elif HAS_PRIVATE_ENCODING
|
128
|
-
if (Qnil != dr->encoding) {
|
129
|
-
rb_funcall(sym, ox_force_encoding_id, 1, dr->encoding);
|
130
|
-
}
|
131
116
|
#endif
|
132
117
|
if (0 != strp) {
|
133
118
|
*strp = StringValuePtr(sym);
|
@@ -182,7 +167,7 @@ sax_drive_init(SaxDrive dr, VALUE handler, VALUE io, SaxOptions options) {
|
|
182
167
|
dr->blocked = 0;
|
183
168
|
dr->abort = false;
|
184
169
|
has_init(&dr->has, handler);
|
185
|
-
#if
|
170
|
+
#if HAVE_RB_ENC_FIND
|
186
171
|
if ('\0' == *ox_default_options.encoding) {
|
187
172
|
VALUE encoding;
|
188
173
|
|
@@ -196,18 +181,6 @@ sax_drive_init(SaxDrive dr, VALUE handler, VALUE io, SaxOptions options) {
|
|
196
181
|
} else {
|
197
182
|
dr->encoding = rb_enc_find(ox_default_options.encoding);
|
198
183
|
}
|
199
|
-
#elif HAS_PRIVATE_ENCODING
|
200
|
-
if ('\0' == *ox_default_options.encoding) {
|
201
|
-
VALUE encoding;
|
202
|
-
|
203
|
-
if (rb_respond_to(io, ox_external_encoding_id) && Qnil != (encoding = rb_funcall(io, ox_external_encoding_id, 0))) {
|
204
|
-
dr->encoding = encoding;
|
205
|
-
} else {
|
206
|
-
dr->encoding = Qnil;
|
207
|
-
}
|
208
|
-
} else {
|
209
|
-
dr->encoding = rb_str_new2(ox_default_options.encoding);
|
210
|
-
}
|
211
184
|
#else
|
212
185
|
dr->encoding = 0;
|
213
186
|
#endif
|
@@ -255,9 +228,7 @@ skipBOM(SaxDrive dr) {
|
|
255
228
|
|
256
229
|
if (0xEF == (uint8_t)c) { /* only UTF8 is supported */
|
257
230
|
if (0xBB == (uint8_t)buf_get(&dr->buf) && 0xBF == (uint8_t)buf_get(&dr->buf)) {
|
258
|
-
#if
|
259
|
-
dr->encoding = ox_utf8_encoding;
|
260
|
-
#elif HAS_PRIVATE_ENCODING
|
231
|
+
#if HAVE_RB_ENC_FIND
|
261
232
|
dr->encoding = ox_utf8_encoding;
|
262
233
|
#else
|
263
234
|
dr->encoding = UTF8_STR;
|
@@ -364,14 +335,10 @@ parse(SaxDrive dr) {
|
|
364
335
|
off_t col = dr->buf.col - 1;
|
365
336
|
|
366
337
|
args[0] = rb_str_new2("");
|
367
|
-
#if
|
338
|
+
#if HAVE_RB_ENC_ASSOCIATE
|
368
339
|
if (0 != dr->encoding) {
|
369
340
|
rb_enc_associate(args[0], dr->encoding);
|
370
341
|
}
|
371
|
-
#elif HAS_PRIVATE_ENCODING
|
372
|
-
if (Qnil != dr->encoding) {
|
373
|
-
rb_funcall(args[0], ox_force_encoding_id, 1, dr->encoding);
|
374
|
-
}
|
375
342
|
#endif
|
376
343
|
if (dr->has.pos) {
|
377
344
|
rb_ivar_set(dr->handler, ox_at_pos_id, LONG2NUM(pos));
|
@@ -526,14 +493,10 @@ read_instruction(SaxDrive dr) {
|
|
526
493
|
ox_sax_collapse_special(dr, content, (int)pos, (int)line, (int)col);
|
527
494
|
}
|
528
495
|
args[0] = rb_str_new2(content);
|
529
|
-
#if
|
496
|
+
#if HAVE_RB_ENC_ASSOCIATE
|
530
497
|
if (0 != dr->encoding) {
|
531
498
|
rb_enc_associate(args[0], dr->encoding);
|
532
499
|
}
|
533
|
-
#elif HAS_PRIVATE_ENCODING
|
534
|
-
if (Qnil != dr->encoding) {
|
535
|
-
rb_funcall(args[0], ox_force_encoding_id, 1, dr->encoding);
|
536
|
-
}
|
537
500
|
#endif
|
538
501
|
if (dr->has.line) {
|
539
502
|
rb_ivar_set(dr->handler, ox_at_line_id, LONG2NUM(line));
|
@@ -732,14 +695,10 @@ read_cdata(SaxDrive dr) {
|
|
732
695
|
VALUE args[1];
|
733
696
|
|
734
697
|
args[0] = rb_str_new2(dr->buf.str);
|
735
|
-
#if
|
698
|
+
#if HAVE_RB_ENC_ASSOCIATE
|
736
699
|
if (0 != dr->encoding) {
|
737
700
|
rb_enc_associate(args[0], dr->encoding);
|
738
701
|
}
|
739
|
-
#elif HAS_PRIVATE_ENCODING
|
740
|
-
if (Qnil != dr->encoding) {
|
741
|
-
rb_funcall(args[0], ox_force_encoding_id, 1, dr->encoding);
|
742
|
-
}
|
743
702
|
#endif
|
744
703
|
if (dr->has.pos) {
|
745
704
|
rb_ivar_set(dr->handler, ox_at_pos_id, LONG2NUM(pos));
|
@@ -826,14 +785,10 @@ read_comment(SaxDrive dr) {
|
|
826
785
|
(NULL != h && (ActiveOverlay == h->overlay || ActiveOverlay == h->overlay))) {
|
827
786
|
|
828
787
|
args[0] = rb_str_new2(dr->buf.str);
|
829
|
-
#if
|
788
|
+
#if HAVE_RB_ENC_ASSOCIATE
|
830
789
|
if (0 != dr->encoding) {
|
831
790
|
rb_enc_associate(args[0], dr->encoding);
|
832
791
|
}
|
833
|
-
#elif HAS_PRIVATE_ENCODING
|
834
|
-
if (Qnil != dr->encoding) {
|
835
|
-
rb_funcall(args[0], ox_force_encoding_id, 1, dr->encoding);
|
836
|
-
}
|
837
792
|
#endif
|
838
793
|
if (dr->has.pos) {
|
839
794
|
rb_ivar_set(dr->handler, ox_at_pos_id, LONG2NUM(pos));
|
@@ -1158,14 +1113,10 @@ read_text(SaxDrive dr) {
|
|
1158
1113
|
((NoSkip == dr->options.skip && !isEnd) ||
|
1159
1114
|
(OffSkip == dr->options.skip))) {
|
1160
1115
|
args[0] = rb_str_new2(dr->buf.str);
|
1161
|
-
#if
|
1116
|
+
#if HAVE_RB_ENC_ASSOCIATE
|
1162
1117
|
if (0 != dr->encoding) {
|
1163
1118
|
rb_enc_associate(args[0], dr->encoding);
|
1164
1119
|
}
|
1165
|
-
#elif HAS_PRIVATE_ENCODING
|
1166
|
-
if (Qnil != dr->encoding) {
|
1167
|
-
rb_funcall(args[0], ox_force_encoding_id, 1, dr->encoding);
|
1168
|
-
}
|
1169
1120
|
#endif
|
1170
1121
|
if (dr->has.pos) {
|
1171
1122
|
rb_ivar_set(dr->handler, ox_at_pos_id, LONG2NUM(pos));
|
@@ -1213,14 +1164,10 @@ read_text(SaxDrive dr) {
|
|
1213
1164
|
break;
|
1214
1165
|
}
|
1215
1166
|
args[0] = rb_str_new2(dr->buf.str);
|
1216
|
-
#if
|
1167
|
+
#if HAVE_RB_ENC_ASSOCIATE
|
1217
1168
|
if (0 != dr->encoding) {
|
1218
1169
|
rb_enc_associate(args[0], dr->encoding);
|
1219
1170
|
}
|
1220
|
-
#elif HAS_PRIVATE_ENCODING
|
1221
|
-
if (Qnil != dr->encoding) {
|
1222
|
-
rb_funcall(args[0], ox_force_encoding_id, 1, dr->encoding);
|
1223
|
-
}
|
1224
1171
|
#endif
|
1225
1172
|
if (dr->has.pos) {
|
1226
1173
|
rb_ivar_set(dr->handler, ox_at_pos_id, LONG2NUM(pos));
|
@@ -1299,14 +1246,10 @@ read_jump(SaxDrive dr, const char *pat) {
|
|
1299
1246
|
// TBD check parent overlay
|
1300
1247
|
if (dr->has.text && !dr->blocked) {
|
1301
1248
|
args[0] = rb_str_new2(dr->buf.str);
|
1302
|
-
#if
|
1249
|
+
#if HAVE_RB_ENC_ASSOCIATE
|
1303
1250
|
if (0 != dr->encoding) {
|
1304
1251
|
rb_enc_associate(args[0], dr->encoding);
|
1305
1252
|
}
|
1306
|
-
#elif HAS_PRIVATE_ENCODING
|
1307
|
-
if (Qnil != dr->encoding) {
|
1308
|
-
rb_funcall(args[0], ox_force_encoding_id, 1, dr->encoding);
|
1309
|
-
}
|
1310
1253
|
#endif
|
1311
1254
|
if (dr->has.pos) {
|
1312
1255
|
rb_ivar_set(dr->handler, ox_at_pos_id, LONG2NUM(pos));
|
@@ -1377,10 +1320,8 @@ read_attrs(SaxDrive dr, char c, char termc, char term2, int is_xml, int eq_req,
|
|
1377
1320
|
c = read_quoted_value(dr);
|
1378
1321
|
attr_value = dr->buf.str;
|
1379
1322
|
if (is_encoding) {
|
1380
|
-
#if
|
1323
|
+
#if HAVE_RB_ENC_FIND
|
1381
1324
|
dr->encoding = rb_enc_find(dr->buf.str);
|
1382
|
-
#elif HAS_PRIVATE_ENCODING
|
1383
|
-
dr->encoding = rb_str_new2(dr->buf.str);
|
1384
1325
|
#else
|
1385
1326
|
dr->encoding = dr->buf.str;
|
1386
1327
|
#endif
|
@@ -1411,14 +1352,10 @@ read_attrs(SaxDrive dr, char c, char termc, char term2, int is_xml, int eq_req,
|
|
1411
1352
|
ox_sax_collapse_special(dr, dr->buf.str, pos, line, col);
|
1412
1353
|
}
|
1413
1354
|
args[1] = rb_str_new2(attr_value);
|
1414
|
-
#if
|
1355
|
+
#if HAVE_RB_ENC_ASSOCIATE
|
1415
1356
|
if (0 != dr->encoding) {
|
1416
1357
|
rb_enc_associate(args[1], dr->encoding);
|
1417
1358
|
}
|
1418
|
-
#elif HAS_PRIVATE_ENCODING
|
1419
|
-
if (Qnil != dr->encoding) {
|
1420
|
-
rb_funcall(args[1], ox_force_encoding_id, 1, dr->encoding);
|
1421
|
-
}
|
1422
1359
|
#endif
|
1423
1360
|
if (dr->has.pos) {
|
1424
1361
|
rb_ivar_set(dr->handler, ox_at_pos_id, LONG2NUM(pos));
|
@@ -1614,19 +1551,12 @@ ox_sax_collapse_special(SaxDrive dr, char *str, long pos, long line, long col) {
|
|
1614
1551
|
}
|
1615
1552
|
if (u <= 0x000000000000007FULL) {
|
1616
1553
|
*b++ = (char)u;
|
1617
|
-
#if
|
1554
|
+
#if HAVE_RB_ENC_FIND
|
1618
1555
|
} else if (ox_utf8_encoding == dr->encoding) {
|
1619
1556
|
b = ox_ucs_to_utf8_chars(b, u);
|
1620
1557
|
} else if (0 == dr->encoding) {
|
1621
1558
|
dr->encoding = ox_utf8_encoding;
|
1622
1559
|
b = ox_ucs_to_utf8_chars(b, u);
|
1623
|
-
#elif HAS_PRIVATE_ENCODING
|
1624
|
-
} else if (ox_utf8_encoding == dr->encoding ||
|
1625
|
-
0 == strcasecmp(rb_str_ptr(rb_String(ox_utf8_encoding)), rb_str_ptr(rb_String(dr->encoding)))) {
|
1626
|
-
b = ox_ucs_to_utf8_chars(b, u);
|
1627
|
-
} else if (Qnil == dr->encoding) {
|
1628
|
-
dr->encoding = ox_utf8_encoding;
|
1629
|
-
b = ox_ucs_to_utf8_chars(b, u);
|
1630
1560
|
#else
|
1631
1561
|
} else if (0 == dr->encoding) {
|
1632
1562
|
dr->encoding = UTF8_STR;
|
data/ext/ox/sax.h
CHANGED
@@ -33,10 +33,8 @@ typedef struct _saxDrive {
|
|
33
33
|
int blocked;
|
34
34
|
bool abort;
|
35
35
|
struct _has has;
|
36
|
-
#if
|
36
|
+
#if HAVE_RB_ENC_ASSOCIATE
|
37
37
|
rb_encoding *encoding;
|
38
|
-
#elif HAS_PRIVATE_ENCODING
|
39
|
-
VALUE encoding;
|
40
38
|
#else
|
41
39
|
const char *encoding;
|
42
40
|
#endif
|
data/ext/ox/sax_as.c
CHANGED
@@ -8,7 +8,7 @@
|
|
8
8
|
#include <stdio.h>
|
9
9
|
#include <strings.h>
|
10
10
|
#include <sys/types.h>
|
11
|
-
#if
|
11
|
+
#if HAVE_SYS_UIO_H
|
12
12
|
#include <sys/uio.h>
|
13
13
|
#endif
|
14
14
|
#include <unistd.h>
|
@@ -136,14 +136,10 @@ sax_value_as_s(VALUE self) {
|
|
136
136
|
break;
|
137
137
|
}
|
138
138
|
rs = rb_str_new2(dr->buf.str);
|
139
|
-
#if
|
139
|
+
#if HAVE_RB_ENC_ASSOCIATE
|
140
140
|
if (0 != dr->encoding) {
|
141
141
|
rb_enc_associate(rs, dr->encoding);
|
142
142
|
}
|
143
|
-
#elif HAS_PRIVATE_ENCODING
|
144
|
-
if (Qnil != dr->encoding) {
|
145
|
-
rb_funcall(rs, ox_force_encoding_id, 1, dr->encoding);
|
146
|
-
}
|
147
143
|
#endif
|
148
144
|
return rs;
|
149
145
|
}
|
data/ext/ox/sax_buf.c
CHANGED
data/lib/ox/version.rb
CHANGED
metadata
CHANGED
@@ -1,14 +1,14 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: ox
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 2.13.
|
4
|
+
version: 2.13.3
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Peter Ohler
|
8
8
|
autorequire:
|
9
9
|
bindir: bin
|
10
10
|
cert_chain: []
|
11
|
-
date: 2020-
|
11
|
+
date: 2020-09-03 00:00:00.000000000 Z
|
12
12
|
dependencies: []
|
13
13
|
description: "A fast XML parser and object serializer that uses only standard C lib.\n\nOptimized
|
14
14
|
XML (Ox), as the name implies was written to provide speed optimized\nXML handling.
|