hpricot 0.7-x86-mswin32 → 0.8-x86-mswin32
Sign up to get free protection for your applications and to get access to all the features.
- data/CHANGELOG +7 -0
- data/Rakefile +1 -1
- data/ext/fast_xs/fast_xs.c +1 -0
- data/ext/hpricot_scan/hpricot_css.c +2109 -2105
- data/ext/hpricot_scan/hpricot_scan.c +944 -880
- data/ext/hpricot_scan/hpricot_scan.rl +255 -191
- data/lib/fast_xs.so +0 -0
- data/lib/hpricot/inspect.rb +2 -2
- data/lib/hpricot/modules.rb +2 -0
- data/lib/hpricot/tag.rb +12 -10
- data/lib/hpricot_scan.so +0 -0
- data/test/test_parser.rb +11 -0
- metadata +2 -2
@@ -19,20 +19,26 @@ VALUE hpricot_css(VALUE, VALUE, VALUE, VALUE, VALUE);
|
|
19
19
|
#define NO_WAY_SERIOUSLY "*** This should not happen, please send a bug report with the HTML you're parsing to why@whytheluckystiff.net. So sorry!"
|
20
20
|
|
21
21
|
static VALUE sym_xmldecl, sym_doctype, sym_procins, sym_stag, sym_etag, sym_emptytag, sym_comment,
|
22
|
-
sym_cdata,
|
22
|
+
sym_cdata, sym_name, sym_parent, sym_raw_attributes, sym_raw_string, sym_tagno,
|
23
|
+
sym_allowed, sym_text, sym_children, sym_EMPTY, sym_CDATA;
|
23
24
|
static VALUE mHpricot, rb_eHpricotParseError;
|
24
|
-
static VALUE
|
25
|
+
static VALUE cBogusETag, cCData, cComment, cDoc, cDocType, cElem, cText,
|
25
26
|
cXMLDecl, cProcIns, symAllow, symDeny;
|
26
27
|
static ID s_ElementContent;
|
27
28
|
static ID s_downcase, s_new, s_parent, s_read, s_to_str;
|
28
|
-
static ID iv_parent;
|
29
29
|
static VALUE reProcInsParse;
|
30
30
|
|
31
|
-
|
32
|
-
|
33
|
-
|
34
|
-
|
35
|
-
|
31
|
+
#define H_ELE_TAG 0
|
32
|
+
#define H_ELE_PARENT 1
|
33
|
+
#define H_ELE_ATTR 2
|
34
|
+
#define H_ELE_ETAG 3
|
35
|
+
#define H_ELE_RAW 4
|
36
|
+
#define H_ELE_EC 5
|
37
|
+
#define H_ELE_HASH 6
|
38
|
+
#define H_ELE_CHILDREN 7
|
39
|
+
|
40
|
+
#define H_ELE_GET(ele, idx) RSTRUCT_PTR(ele)[idx]
|
41
|
+
#define H_ELE_SET(ele, idx, val) RSTRUCT_PTR(ele)[idx] = val
|
36
42
|
|
37
43
|
#define OPT(opts, key) (!NIL_P(opts) && RTEST(rb_hash_aref(opts, ID2SYM(rb_intern("" # key)))))
|
38
44
|
|
@@ -107,7 +113,7 @@ typedef struct {
|
|
107
113
|
action tag { SET(tag, p); }
|
108
114
|
action tagc { SET(tag, p-1); }
|
109
115
|
action aval { SET(aval, p); }
|
110
|
-
action aunq {
|
116
|
+
action aunq {
|
111
117
|
if (*(p-1) == '"' || *(p-1) == '\'') { SET(aval, p-1); }
|
112
118
|
else { SET(aval, p); }
|
113
119
|
}
|
@@ -118,14 +124,14 @@ typedef struct {
|
|
118
124
|
action pubid { SET(aval, p); ATTR(ID2SYM(rb_intern("public_id")), aval); }
|
119
125
|
action sysid { SET(aval, p); ATTR(ID2SYM(rb_intern("system_id")), aval); }
|
120
126
|
|
121
|
-
action new_attr {
|
127
|
+
action new_attr {
|
122
128
|
akey = Qnil;
|
123
129
|
aval = Qnil;
|
124
130
|
mark_akey = NULL;
|
125
131
|
mark_aval = NULL;
|
126
132
|
}
|
127
133
|
|
128
|
-
action save_attr {
|
134
|
+
action save_attr {
|
129
135
|
ATTR(akey, aval);
|
130
136
|
}
|
131
137
|
|
@@ -144,7 +150,7 @@ void rb_yield_tokens(VALUE sym, VALUE tag, VALUE attr, VALUE raw, int taint)
|
|
144
150
|
raw = tag;
|
145
151
|
}
|
146
152
|
ary = rb_ary_new3(4, sym, tag, attr, raw);
|
147
|
-
if (taint) {
|
153
|
+
if (taint) {
|
148
154
|
OBJ_TAINT(ary);
|
149
155
|
OBJ_TAINT(tag);
|
150
156
|
OBJ_TAINT(attr);
|
@@ -153,6 +159,7 @@ void rb_yield_tokens(VALUE sym, VALUE tag, VALUE attr, VALUE raw, int taint)
|
|
153
159
|
rb_yield(ary);
|
154
160
|
}
|
155
161
|
|
162
|
+
#ifndef RHASH_TBL
|
156
163
|
/* rb_hash_lookup() is only in Ruby 1.8.7 */
|
157
164
|
static VALUE
|
158
165
|
our_rb_hash_lookup(VALUE hash, VALUE key)
|
@@ -165,17 +172,17 @@ our_rb_hash_lookup(VALUE hash, VALUE key)
|
|
165
172
|
|
166
173
|
return val;
|
167
174
|
}
|
175
|
+
#define rb_hash_lookup our_rb_hash_lookup
|
176
|
+
#endif
|
168
177
|
|
169
178
|
static void
|
170
179
|
rb_hpricot_add(VALUE focus, VALUE ele)
|
171
180
|
{
|
172
|
-
|
173
|
-
|
174
|
-
|
175
|
-
|
176
|
-
|
177
|
-
rb_ary_push(he->children, ele);
|
178
|
-
he2->parent = focus;
|
181
|
+
VALUE children = H_ELE_GET(focus, H_ELE_CHILDREN);
|
182
|
+
if (NIL_P(children))
|
183
|
+
H_ELE_SET(focus, H_ELE_CHILDREN, (children = rb_ary_new2(1)));
|
184
|
+
rb_ary_push(children, ele);
|
185
|
+
H_ELE_SET(ele, H_ELE_PARENT, focus);
|
179
186
|
}
|
180
187
|
|
181
188
|
typedef struct {
|
@@ -186,102 +193,70 @@ typedef struct {
|
|
186
193
|
unsigned char xml, strict, fixup;
|
187
194
|
} hpricot_state;
|
188
195
|
|
189
|
-
|
190
|
-
hpricot_ele_mark(hpricot_ele *he)
|
191
|
-
{
|
192
|
-
rb_gc_mark(he->tag);
|
193
|
-
rb_gc_mark(he->attr);
|
194
|
-
rb_gc_mark(he->etag);
|
195
|
-
rb_gc_mark(he->raw);
|
196
|
-
rb_gc_mark(he->parent);
|
197
|
-
rb_gc_mark(he->children);
|
198
|
-
}
|
199
|
-
|
200
|
-
static void
|
201
|
-
hpricot_ele_free(hpricot_ele *he)
|
202
|
-
{
|
203
|
-
free(he);
|
204
|
-
}
|
205
|
-
|
206
|
-
#define H_PROP(prop) \
|
196
|
+
#define H_PROP(prop, idx) \
|
207
197
|
static VALUE hpricot_ele_set_##prop(VALUE self, VALUE x) { \
|
208
|
-
|
209
|
-
Data_Get_Struct(self, hpricot_ele, he); \
|
210
|
-
he->prop = x; \
|
198
|
+
H_ELE_SET(self, idx, x); \
|
211
199
|
return self; \
|
212
200
|
} \
|
201
|
+
static VALUE hpricot_ele_clear_##prop(VALUE self) { \
|
202
|
+
H_ELE_SET(self, idx, Qnil); \
|
203
|
+
return Qtrue; \
|
204
|
+
} \
|
213
205
|
static VALUE hpricot_ele_get_##prop(VALUE self) { \
|
214
|
-
|
215
|
-
Data_Get_Struct(self, hpricot_ele, he); \
|
216
|
-
return he->prop; \
|
206
|
+
return H_ELE_GET(self, idx); \
|
217
207
|
}
|
218
208
|
|
219
209
|
#define H_ATTR(prop) \
|
220
210
|
static VALUE hpricot_ele_set_##prop(VALUE self, VALUE x) { \
|
221
|
-
|
222
|
-
Data_Get_Struct(self, hpricot_ele, he); \
|
223
|
-
rb_hash_aset(he->attr, ID2SYM(rb_intern("" # prop)), x); \
|
211
|
+
rb_hash_aset(H_ELE_GET(self, H_ELE_ATTR), ID2SYM(rb_intern("" # prop)), x); \
|
224
212
|
return self; \
|
225
213
|
} \
|
226
214
|
static VALUE hpricot_ele_get_##prop(VALUE self) { \
|
227
|
-
|
228
|
-
Data_Get_Struct(self, hpricot_ele, he); \
|
229
|
-
return rb_hash_aref(he->attr, ID2SYM(rb_intern("" # prop))); \
|
215
|
+
return rb_hash_aref(H_ELE_GET(self, H_ELE_ATTR), ID2SYM(rb_intern("" # prop))); \
|
230
216
|
}
|
231
217
|
|
232
|
-
H_PROP(
|
233
|
-
H_PROP(
|
234
|
-
H_PROP(
|
235
|
-
H_PROP(
|
236
|
-
H_PROP(
|
218
|
+
H_PROP(name, H_ELE_TAG);
|
219
|
+
H_PROP(raw, H_ELE_RAW);
|
220
|
+
H_PROP(parent, H_ELE_PARENT);
|
221
|
+
H_PROP(attr, H_ELE_ATTR);
|
222
|
+
H_PROP(etag, H_ELE_ETAG);
|
223
|
+
H_PROP(children, H_ELE_CHILDREN);
|
224
|
+
H_ATTR(target);
|
237
225
|
H_ATTR(encoding);
|
238
226
|
H_ATTR(version);
|
239
227
|
H_ATTR(standalone);
|
240
228
|
H_ATTR(system_id);
|
241
229
|
H_ATTR(public_id);
|
242
230
|
|
243
|
-
static VALUE
|
244
|
-
hpricot_ele_get_raw(VALUE self, VALUE x) {
|
245
|
-
hpricot_ele *he;
|
246
|
-
Data_Get_Struct(self, hpricot_ele, he);
|
247
|
-
return he->raw;
|
248
|
-
}
|
249
|
-
|
250
|
-
static VALUE
|
251
|
-
hpricot_ele_clear_raw(VALUE self)
|
252
|
-
{
|
253
|
-
hpricot_ele *he;
|
254
|
-
Data_Get_Struct(self, hpricot_ele, he);
|
255
|
-
he->raw = Qnil;
|
256
|
-
return Qtrue;
|
257
|
-
}
|
258
|
-
|
259
231
|
#define H_ELE(klass) \
|
260
|
-
|
261
|
-
|
262
|
-
|
263
|
-
|
264
|
-
|
265
|
-
|
266
|
-
|
267
|
-
|
268
|
-
|
232
|
+
ele = rb_obj_alloc(klass); \
|
233
|
+
if (klass == cElem) { \
|
234
|
+
H_ELE_SET(ele, H_ELE_TAG, tag); \
|
235
|
+
H_ELE_SET(ele, H_ELE_ATTR, attr); \
|
236
|
+
H_ELE_SET(ele, H_ELE_EC, ec); \
|
237
|
+
if (raw != NULL && (sym == sym_emptytag || sym == sym_stag || sym == sym_doctype)) { \
|
238
|
+
H_ELE_SET(ele, H_ELE_RAW, rb_str_new(raw, rawlen)); \
|
239
|
+
} \
|
240
|
+
} else if (klass == cDocType || klass == cProcIns || klass == cXMLDecl || klass == cBogusETag) { \
|
241
|
+
if (klass == cBogusETag) { \
|
242
|
+
H_ELE_SET(ele, H_ELE_TAG, tag); \
|
243
|
+
if (raw != NULL) \
|
244
|
+
H_ELE_SET(ele, H_ELE_ATTR, rb_str_new(raw, rawlen)); \
|
245
|
+
} else { \
|
246
|
+
if (klass == cDocType) \
|
247
|
+
ATTR(ID2SYM(rb_intern("target")), tag); \
|
248
|
+
H_ELE_SET(ele, H_ELE_ATTR, attr); \
|
249
|
+
if (klass != cProcIns) { \
|
250
|
+
tag = Qnil; \
|
251
|
+
if (raw != NULL) tag = rb_str_new(raw, rawlen); \
|
252
|
+
} \
|
253
|
+
H_ELE_SET(ele, H_ELE_TAG, tag); \
|
254
|
+
} \
|
255
|
+
} else { \
|
256
|
+
H_ELE_SET(ele, H_ELE_TAG, tag); \
|
269
257
|
} \
|
270
|
-
ele = Data_Wrap_Struct(klass, hpricot_ele_mark, hpricot_ele_free, he); \
|
271
258
|
S->last = ele
|
272
259
|
|
273
|
-
VALUE
|
274
|
-
hpricot_ele_alloc(VALUE klass)
|
275
|
-
{
|
276
|
-
VALUE ele;
|
277
|
-
hpricot_ele *he = ALLOC(hpricot_ele);
|
278
|
-
he->name = 0;
|
279
|
-
he->tag = he->attr = he->raw = he->EC = Qnil;
|
280
|
-
he->etag = he->parent = he->children = Qnil;
|
281
|
-
ele = Data_Wrap_Struct(klass, hpricot_ele_mark, hpricot_ele_free, he);
|
282
|
-
return ele;
|
283
|
-
}
|
284
|
-
|
285
260
|
//
|
286
261
|
// the swift, compact parser logic. most of the complicated stuff is done
|
287
262
|
// in the lexer. this step just pairs up the start and end tags.
|
@@ -295,22 +270,23 @@ rb_hpricot_token(hpricot_state *S, VALUE sym, VALUE tag, VALUE attr, char *raw,
|
|
295
270
|
// in html mode, fix up start tags incorrectly formed as empty tags
|
296
271
|
//
|
297
272
|
if (!S->xml) {
|
298
|
-
hpricot_ele *last;
|
299
|
-
Data_Get_Struct(S->focus, hpricot_ele, last);
|
300
|
-
if (last->EC == sym_CDATA &&
|
301
|
-
(sym != sym_procins && sym != sym_comment && sym != sym_cdata && sym != sym_text) &&
|
302
|
-
!(sym == sym_etag && rb_str_hash(tag) == last->name))
|
303
|
-
{
|
304
|
-
sym = sym_text;
|
305
|
-
tag = rb_str_new(raw, rawlen);
|
306
|
-
}
|
307
|
-
|
308
273
|
if (sym == sym_emptytag || sym == sym_stag || sym == sym_etag) {
|
309
274
|
ec = rb_hash_aref(S->EC, tag);
|
310
275
|
if (NIL_P(ec)) {
|
311
276
|
tag = rb_funcall(tag, s_downcase, 0);
|
312
277
|
ec = rb_hash_aref(S->EC, tag);
|
313
278
|
}
|
279
|
+
}
|
280
|
+
|
281
|
+
if (H_ELE_GET(S->focus, H_ELE_EC) == sym_CDATA &&
|
282
|
+
(sym != sym_procins && sym != sym_comment && sym != sym_cdata && sym != sym_text) &&
|
283
|
+
!(sym == sym_etag && INT2NUM(rb_str_hash(tag)) == H_ELE_GET(S->focus, H_ELE_HASH)))
|
284
|
+
{
|
285
|
+
sym = sym_text;
|
286
|
+
tag = rb_str_new(raw, rawlen);
|
287
|
+
}
|
288
|
+
|
289
|
+
if (!NIL_P(ec)) {
|
314
290
|
if (sym == sym_emptytag) {
|
315
291
|
if (ec != sym_EMPTY)
|
316
292
|
sym = sym_stag;
|
@@ -322,19 +298,19 @@ rb_hpricot_token(hpricot_state *S, VALUE sym, VALUE tag, VALUE attr, char *raw,
|
|
322
298
|
}
|
323
299
|
|
324
300
|
if (sym == sym_emptytag || sym == sym_stag) {
|
301
|
+
VALUE name = INT2NUM(rb_str_hash(tag));
|
325
302
|
H_ELE(cElem);
|
326
|
-
|
303
|
+
H_ELE_SET(ele, H_ELE_HASH, name);
|
327
304
|
|
328
305
|
if (!S->xml) {
|
329
306
|
VALUE match = Qnil, e = S->focus;
|
330
307
|
while (e != S->doc)
|
331
308
|
{
|
332
|
-
|
333
|
-
Data_Get_Struct(e, hpricot_ele, hee);
|
309
|
+
VALUE hEC = H_ELE_GET(e, H_ELE_EC);
|
334
310
|
|
335
|
-
if (TYPE(
|
311
|
+
if (TYPE(hEC) == T_HASH)
|
336
312
|
{
|
337
|
-
VALUE has =
|
313
|
+
VALUE has = rb_hash_lookup(hEC, name);
|
338
314
|
if (has != Qnil) {
|
339
315
|
if (has == Qtrue) {
|
340
316
|
if (match == Qnil)
|
@@ -347,7 +323,7 @@ rb_hpricot_token(hpricot_state *S, VALUE sym, VALUE tag, VALUE attr, char *raw,
|
|
347
323
|
}
|
348
324
|
}
|
349
325
|
|
350
|
-
e =
|
326
|
+
e = H_ELE_GET(e, H_ELE_PARENT);
|
351
327
|
}
|
352
328
|
|
353
329
|
if (match == Qnil)
|
@@ -369,8 +345,7 @@ rb_hpricot_token(hpricot_state *S, VALUE sym, VALUE tag, VALUE attr, char *raw,
|
|
369
345
|
}
|
370
346
|
}
|
371
347
|
} else if (sym == sym_etag) {
|
372
|
-
|
373
|
-
VALUE match = Qnil, e = S->focus;
|
348
|
+
VALUE name, match = Qnil, e = S->focus;
|
374
349
|
if (S->strict) {
|
375
350
|
if (NIL_P(rb_hash_aref(S->EC, tag))) {
|
376
351
|
tag = rb_str_new2("div");
|
@@ -383,19 +358,16 @@ rb_hpricot_token(hpricot_state *S, VALUE sym, VALUE tag, VALUE attr, char *raw,
|
|
383
358
|
//
|
384
359
|
// (see also: the search above for fixups)
|
385
360
|
//
|
386
|
-
name = rb_str_hash(tag);
|
361
|
+
name = INT2NUM(rb_str_hash(tag));
|
387
362
|
while (e != S->doc)
|
388
363
|
{
|
389
|
-
|
390
|
-
Data_Get_Struct(e, hpricot_ele, he);
|
391
|
-
|
392
|
-
if (he->name == name)
|
364
|
+
if (H_ELE_GET(e, H_ELE_HASH) == name)
|
393
365
|
{
|
394
366
|
match = e;
|
395
367
|
break;
|
396
368
|
}
|
397
369
|
|
398
|
-
e =
|
370
|
+
e = H_ELE_GET(e, H_ELE_PARENT);
|
399
371
|
}
|
400
372
|
|
401
373
|
if (NIL_P(match))
|
@@ -405,10 +377,11 @@ rb_hpricot_token(hpricot_state *S, VALUE sym, VALUE tag, VALUE attr, char *raw,
|
|
405
377
|
}
|
406
378
|
else
|
407
379
|
{
|
408
|
-
|
409
|
-
|
410
|
-
|
411
|
-
|
380
|
+
VALUE ele = Qnil;
|
381
|
+
if (raw != NULL)
|
382
|
+
ele = rb_str_new(raw, rawlen);
|
383
|
+
H_ELE_SET(match, H_ELE_ETAG, ele);
|
384
|
+
S->focus = H_ELE_GET(match, H_ELE_PARENT);
|
412
385
|
S->last = Qnil;
|
413
386
|
}
|
414
387
|
} else if (sym == sym_cdata) {
|
@@ -429,15 +402,13 @@ rb_hpricot_token(hpricot_state *S, VALUE sym, VALUE tag, VALUE attr, char *raw,
|
|
429
402
|
tag = rb_reg_nth_match(1, match);
|
430
403
|
attr = rb_reg_nth_match(2, match);
|
431
404
|
{
|
432
|
-
|
433
|
-
|
405
|
+
H_ELE(cProcIns);
|
406
|
+
rb_hpricot_add(S->focus, ele);
|
434
407
|
}
|
435
408
|
} else if (sym == sym_text) {
|
436
409
|
// TODO: add raw_string as well?
|
437
410
|
if (!NIL_P(S->last) && RBASIC(S->last)->klass == cText) {
|
438
|
-
|
439
|
-
Data_Get_Struct(S->last, hpricot_ele, he);
|
440
|
-
rb_str_append(he->tag, tag);
|
411
|
+
rb_str_append(H_ELE_GET(S->last, H_ELE_TAG), tag);
|
441
412
|
} else {
|
442
413
|
H_ELE(cText);
|
443
414
|
rb_hpricot_add(S->focus, ele);
|
@@ -480,11 +451,8 @@ VALUE hpricot_scan(int argc, VALUE *argv, VALUE self)
|
|
480
451
|
|
481
452
|
if (!rb_block_given_p())
|
482
453
|
{
|
483
|
-
hpricot_ele *he = ALLOC(hpricot_ele);
|
484
454
|
S = ALLOC(hpricot_state);
|
485
|
-
|
486
|
-
he->tag = he->attr = he->etag = he->parent = he->children = Qnil;
|
487
|
-
S->doc = Data_Wrap_Struct(cDoc, hpricot_ele_mark, hpricot_ele_free, he);
|
455
|
+
S->doc = rb_obj_alloc(cDoc);
|
488
456
|
rb_gc_register_address(&S->doc);
|
489
457
|
S->focus = S->doc;
|
490
458
|
S->last = Qnil;
|
@@ -509,7 +477,7 @@ VALUE hpricot_scan(int argc, VALUE *argv, VALUE self)
|
|
509
477
|
buf = ALLOC_N(char, buffer_size);
|
510
478
|
|
511
479
|
%% write init;
|
512
|
-
|
480
|
+
|
513
481
|
while (!done) {
|
514
482
|
VALUE str;
|
515
483
|
char *p, *pe;
|
@@ -561,7 +529,7 @@ VALUE hpricot_scan(int argc, VALUE *argv, VALUE self)
|
|
561
529
|
|
562
530
|
pe = p + len;
|
563
531
|
%% write exec;
|
564
|
-
|
532
|
+
|
565
533
|
if (cs == hpricot_scan_error) {
|
566
534
|
if (buf != NULL)
|
567
535
|
free(buf);
|
@@ -574,7 +542,7 @@ VALUE hpricot_scan(int argc, VALUE *argv, VALUE self)
|
|
574
542
|
rb_raise(rb_eHpricotParseError, "parse error on line %d.\n" NO_WAY_SERIOUSLY, curline);
|
575
543
|
}
|
576
544
|
}
|
577
|
-
|
545
|
+
|
578
546
|
if (done && ele_open)
|
579
547
|
{
|
580
548
|
ele_open = 0;
|
@@ -635,66 +603,103 @@ VALUE hpricot_scan(int argc, VALUE *argv, VALUE self)
|
|
635
603
|
return Qnil;
|
636
604
|
}
|
637
605
|
|
638
|
-
|
606
|
+
static VALUE
|
607
|
+
alloc_hpricot_struct(VALUE klass)
|
639
608
|
{
|
640
|
-
|
641
|
-
|
642
|
-
|
643
|
-
|
644
|
-
|
609
|
+
VALUE size;
|
610
|
+
long n;
|
611
|
+
NEWOBJ(st, struct RStruct);
|
612
|
+
OBJSETUP(st, klass, T_STRUCT);
|
613
|
+
|
614
|
+
size = rb_struct_iv_get(klass, "__size__");
|
615
|
+
n = FIX2LONG(size);
|
616
|
+
|
617
|
+
#ifndef RSTRUCT_EMBED_LEN_MAX
|
618
|
+
st->ptr = ALLOC_N(VALUE, n);
|
619
|
+
rb_mem_clear(st->ptr, n);
|
620
|
+
st->len = n;
|
621
|
+
#else
|
622
|
+
if (0 < n && n <= RSTRUCT_EMBED_LEN_MAX) {
|
623
|
+
RBASIC(st)->flags &= ~RSTRUCT_EMBED_LEN_MASK;
|
624
|
+
RBASIC(st)->flags |= n << RSTRUCT_EMBED_LEN_SHIFT;
|
625
|
+
rb_mem_clear(st->as.ary, n);
|
626
|
+
} else {
|
627
|
+
st->as.heap.ptr = ALLOC_N(VALUE, n);
|
628
|
+
rb_mem_clear(st->as.heap.ptr, n);
|
629
|
+
st->as.heap.len = n;
|
630
|
+
}
|
631
|
+
#endif
|
645
632
|
|
646
|
-
|
647
|
-
|
648
|
-
|
649
|
-
|
650
|
-
|
651
|
-
|
652
|
-
|
653
|
-
|
654
|
-
|
655
|
-
|
656
|
-
|
657
|
-
|
658
|
-
|
659
|
-
|
660
|
-
|
661
|
-
|
662
|
-
|
663
|
-
|
664
|
-
|
665
|
-
|
666
|
-
|
667
|
-
|
668
|
-
|
669
|
-
|
670
|
-
|
671
|
-
|
672
|
-
|
673
|
-
|
674
|
-
|
675
|
-
|
676
|
-
|
677
|
-
|
678
|
-
|
679
|
-
|
680
|
-
|
681
|
-
|
682
|
-
|
683
|
-
|
684
|
-
|
685
|
-
|
686
|
-
|
687
|
-
|
688
|
-
|
689
|
-
|
690
|
-
|
691
|
-
|
692
|
-
|
693
|
-
|
694
|
-
|
695
|
-
|
696
|
-
|
697
|
-
|
633
|
+
return (VALUE)st;
|
634
|
+
}
|
635
|
+
|
636
|
+
static VALUE hpricot_struct_ref0(VALUE obj) {return H_ELE_GET(obj, 0);}
|
637
|
+
static VALUE hpricot_struct_ref1(VALUE obj) {return H_ELE_GET(obj, 1);}
|
638
|
+
static VALUE hpricot_struct_ref2(VALUE obj) {return H_ELE_GET(obj, 2);}
|
639
|
+
static VALUE hpricot_struct_ref3(VALUE obj) {return H_ELE_GET(obj, 3);}
|
640
|
+
static VALUE hpricot_struct_ref4(VALUE obj) {return H_ELE_GET(obj, 4);}
|
641
|
+
static VALUE hpricot_struct_ref5(VALUE obj) {return H_ELE_GET(obj, 5);}
|
642
|
+
static VALUE hpricot_struct_ref6(VALUE obj) {return H_ELE_GET(obj, 6);}
|
643
|
+
static VALUE hpricot_struct_ref7(VALUE obj) {return H_ELE_GET(obj, 7);}
|
644
|
+
static VALUE hpricot_struct_ref8(VALUE obj) {return H_ELE_GET(obj, 8);}
|
645
|
+
static VALUE hpricot_struct_ref9(VALUE obj) {return H_ELE_GET(obj, 9);}
|
646
|
+
|
647
|
+
static VALUE (*ref_func[10])() = {
|
648
|
+
hpricot_struct_ref0,
|
649
|
+
hpricot_struct_ref1,
|
650
|
+
hpricot_struct_ref2,
|
651
|
+
hpricot_struct_ref3,
|
652
|
+
hpricot_struct_ref4,
|
653
|
+
hpricot_struct_ref5,
|
654
|
+
hpricot_struct_ref6,
|
655
|
+
hpricot_struct_ref7,
|
656
|
+
hpricot_struct_ref8,
|
657
|
+
hpricot_struct_ref9,
|
658
|
+
};
|
659
|
+
|
660
|
+
static VALUE hpricot_struct_set0(VALUE obj, VALUE val) {return H_ELE_SET(obj, 0, val);}
|
661
|
+
static VALUE hpricot_struct_set1(VALUE obj, VALUE val) {return H_ELE_SET(obj, 1, val);}
|
662
|
+
static VALUE hpricot_struct_set2(VALUE obj, VALUE val) {return H_ELE_SET(obj, 2, val);}
|
663
|
+
static VALUE hpricot_struct_set3(VALUE obj, VALUE val) {return H_ELE_SET(obj, 3, val);}
|
664
|
+
static VALUE hpricot_struct_set4(VALUE obj, VALUE val) {return H_ELE_SET(obj, 4, val);}
|
665
|
+
static VALUE hpricot_struct_set5(VALUE obj, VALUE val) {return H_ELE_SET(obj, 5, val);}
|
666
|
+
static VALUE hpricot_struct_set6(VALUE obj, VALUE val) {return H_ELE_SET(obj, 6, val);}
|
667
|
+
static VALUE hpricot_struct_set7(VALUE obj, VALUE val) {return H_ELE_SET(obj, 7, val);}
|
668
|
+
static VALUE hpricot_struct_set8(VALUE obj, VALUE val) {return H_ELE_SET(obj, 8, val);}
|
669
|
+
static VALUE hpricot_struct_set9(VALUE obj, VALUE val) {return H_ELE_SET(obj, 9, val);}
|
670
|
+
|
671
|
+
static VALUE (*set_func[10])() = {
|
672
|
+
hpricot_struct_set0,
|
673
|
+
hpricot_struct_set1,
|
674
|
+
hpricot_struct_set2,
|
675
|
+
hpricot_struct_set3,
|
676
|
+
hpricot_struct_set4,
|
677
|
+
hpricot_struct_set5,
|
678
|
+
hpricot_struct_set6,
|
679
|
+
hpricot_struct_set7,
|
680
|
+
hpricot_struct_set8,
|
681
|
+
hpricot_struct_set9,
|
682
|
+
};
|
683
|
+
|
684
|
+
static VALUE
|
685
|
+
make_hpricot_struct(VALUE members)
|
686
|
+
{
|
687
|
+
int i = 0;
|
688
|
+
VALUE klass = rb_class_new(rb_cObject);
|
689
|
+
rb_iv_set(klass, "__size__", INT2NUM(RARRAY_LEN(members)));
|
690
|
+
rb_define_alloc_func(klass, alloc_hpricot_struct);
|
691
|
+
rb_define_singleton_method(klass, "new", rb_class_new_instance, -1);
|
692
|
+
for (i = 0; i < RARRAY_LEN(members); i++) {
|
693
|
+
ID id = SYM2ID(RARRAY_PTR(members)[i]);
|
694
|
+
rb_define_method_id(klass, id, ref_func[i], 0);
|
695
|
+
rb_define_method_id(klass, rb_id_attrset(id), set_func[i], 1);
|
696
|
+
}
|
697
|
+
return klass;
|
698
|
+
}
|
699
|
+
|
700
|
+
void Init_hpricot_scan()
|
701
|
+
{
|
702
|
+
VALUE structElem, structAttr, structBasic;
|
698
703
|
|
699
704
|
s_ElementContent = rb_intern("ElementContent");
|
700
705
|
symAllow = ID2SYM(rb_intern("allow"));
|
@@ -704,19 +709,78 @@ void Init_hpricot_scan()
|
|
704
709
|
s_parent = rb_intern("parent");
|
705
710
|
s_read = rb_intern("read");
|
706
711
|
s_to_str = rb_intern("to_str");
|
707
|
-
iv_parent = rb_intern("parent");
|
708
712
|
sym_xmldecl = ID2SYM(rb_intern("xmldecl"));
|
709
713
|
sym_doctype = ID2SYM(rb_intern("doctype"));
|
710
714
|
sym_procins = ID2SYM(rb_intern("procins"));
|
711
715
|
sym_stag = ID2SYM(rb_intern("stag"));
|
712
716
|
sym_etag = ID2SYM(rb_intern("etag"));
|
713
717
|
sym_emptytag = ID2SYM(rb_intern("emptytag"));
|
718
|
+
sym_allowed = ID2SYM(rb_intern("allowed"));
|
719
|
+
sym_children = ID2SYM(rb_intern("children"));
|
714
720
|
sym_comment = ID2SYM(rb_intern("comment"));
|
715
721
|
sym_cdata = ID2SYM(rb_intern("cdata"));
|
722
|
+
sym_name = ID2SYM(rb_intern("name"));
|
723
|
+
sym_parent = ID2SYM(rb_intern("parent"));
|
724
|
+
sym_raw_attributes = ID2SYM(rb_intern("raw_attributes"));
|
725
|
+
sym_raw_string = ID2SYM(rb_intern("raw_string"));
|
726
|
+
sym_tagno = ID2SYM(rb_intern("tagno"));
|
716
727
|
sym_text = ID2SYM(rb_intern("text"));
|
717
728
|
sym_EMPTY = ID2SYM(rb_intern("EMPTY"));
|
718
729
|
sym_CDATA = ID2SYM(rb_intern("CDATA"));
|
719
730
|
|
731
|
+
mHpricot = rb_define_module("Hpricot");
|
732
|
+
rb_define_attr(rb_singleton_class(mHpricot), "buffer_size", 1, 1);
|
733
|
+
rb_define_singleton_method(mHpricot, "scan", hpricot_scan, -1);
|
734
|
+
rb_define_singleton_method(mHpricot, "css", hpricot_css, 3);
|
735
|
+
rb_eHpricotParseError = rb_define_class_under(mHpricot, "ParseError", rb_eStandardError);
|
736
|
+
|
737
|
+
structElem = make_hpricot_struct(rb_ary_new3(8, sym_name, sym_parent,
|
738
|
+
sym_raw_attributes, sym_etag, sym_raw_string, sym_allowed,
|
739
|
+
sym_tagno, sym_children));
|
740
|
+
structAttr = make_hpricot_struct(rb_ary_new3(3, sym_name, sym_parent, sym_raw_attributes));
|
741
|
+
structBasic = make_hpricot_struct(rb_ary_new3(2, sym_name, sym_parent));
|
742
|
+
|
743
|
+
cDoc = rb_define_class_under(mHpricot, "Doc", structElem);
|
744
|
+
cCData = rb_define_class_under(mHpricot, "CData", structBasic);
|
745
|
+
rb_define_method(cCData, "content", hpricot_ele_get_name, 0);
|
746
|
+
rb_define_method(cCData, "content=", hpricot_ele_set_name, 1);
|
747
|
+
cComment = rb_define_class_under(mHpricot, "Comment", structBasic);
|
748
|
+
rb_define_method(cComment, "content", hpricot_ele_get_name, 0);
|
749
|
+
rb_define_method(cComment, "content=", hpricot_ele_set_name, 1);
|
750
|
+
cDocType = rb_define_class_under(mHpricot, "DocType", structAttr);
|
751
|
+
rb_define_method(cDocType, "raw_string", hpricot_ele_get_name, 0);
|
752
|
+
rb_define_method(cDocType, "clear_raw", hpricot_ele_clear_name, 0);
|
753
|
+
rb_define_method(cDocType, "target", hpricot_ele_get_target, 0);
|
754
|
+
rb_define_method(cDocType, "target=", hpricot_ele_set_target, 1);
|
755
|
+
rb_define_method(cDocType, "public_id", hpricot_ele_get_public_id, 0);
|
756
|
+
rb_define_method(cDocType, "public_id=", hpricot_ele_set_public_id, 1);
|
757
|
+
rb_define_method(cDocType, "system_id", hpricot_ele_get_system_id, 0);
|
758
|
+
rb_define_method(cDocType, "system_id=", hpricot_ele_set_system_id, 1);
|
759
|
+
cElem = rb_define_class_under(mHpricot, "Elem", structElem);
|
760
|
+
rb_define_method(cElem, "clear_raw", hpricot_ele_clear_raw, 0);
|
761
|
+
cBogusETag = rb_define_class_under(mHpricot, "BogusETag", structAttr);
|
762
|
+
rb_define_method(cBogusETag, "raw_string", hpricot_ele_get_attr, 0);
|
763
|
+
rb_define_method(cBogusETag, "clear_raw", hpricot_ele_clear_attr, 0);
|
764
|
+
cText = rb_define_class_under(mHpricot, "Text", structBasic);
|
765
|
+
rb_define_method(cText, "raw_string", hpricot_ele_get_name, 0);
|
766
|
+
rb_define_method(cText, "clear_raw", hpricot_ele_clear_name, 0);
|
767
|
+
rb_define_method(cText, "content", hpricot_ele_get_name, 0);
|
768
|
+
rb_define_method(cText, "content=", hpricot_ele_set_name, 1);
|
769
|
+
cXMLDecl = rb_define_class_under(mHpricot, "XMLDecl", structAttr);
|
770
|
+
rb_define_method(cXMLDecl, "raw_string", hpricot_ele_get_name, 0);
|
771
|
+
rb_define_method(cXMLDecl, "clear_raw", hpricot_ele_clear_name, 0);
|
772
|
+
rb_define_method(cXMLDecl, "encoding", hpricot_ele_get_encoding, 0);
|
773
|
+
rb_define_method(cXMLDecl, "encoding=", hpricot_ele_set_encoding, 1);
|
774
|
+
rb_define_method(cXMLDecl, "standalone", hpricot_ele_get_standalone, 0);
|
775
|
+
rb_define_method(cXMLDecl, "standalone=", hpricot_ele_set_standalone, 1);
|
776
|
+
rb_define_method(cXMLDecl, "version", hpricot_ele_get_version, 0);
|
777
|
+
rb_define_method(cXMLDecl, "version=", hpricot_ele_set_version, 1);
|
778
|
+
cProcIns = rb_define_class_under(mHpricot, "ProcIns", structAttr);
|
779
|
+
rb_define_method(cProcIns, "target", hpricot_ele_get_name, 0);
|
780
|
+
rb_define_method(cProcIns, "target=", hpricot_ele_set_name, 1);
|
781
|
+
rb_define_method(cProcIns, "content", hpricot_ele_get_attr, 0);
|
782
|
+
rb_define_method(cProcIns, "content=", hpricot_ele_set_attr, 1);
|
783
|
+
|
720
784
|
rb_const_set(mHpricot, rb_intern("ProcInsParse"),
|
721
785
|
reProcInsParse = rb_eval_string("/\\A<\\?(\\S+)\\s+(.+)/m"));
|
722
786
|
}
|