hpricot-hpricot 0.7.229

Sign up to get free protection for your applications and to get access to all the features.
Files changed (46) hide show
  1. data/CHANGELOG +75 -0
  2. data/COPYING +18 -0
  3. data/README +284 -0
  4. data/Rakefile +264 -0
  5. data/ext/fast_xs/FastXsService.java +1018 -0
  6. data/ext/fast_xs/extconf.rb +4 -0
  7. data/ext/fast_xs/fast_xs.c +201 -0
  8. data/ext/hpricot_scan/HpricotScanService.java +2085 -0
  9. data/ext/hpricot_scan/extconf.rb +6 -0
  10. data/ext/hpricot_scan/hpricot_common.rl +76 -0
  11. data/ext/hpricot_scan/hpricot_css.c +3502 -0
  12. data/ext/hpricot_scan/hpricot_scan.c +6926 -0
  13. data/ext/hpricot_scan/hpricot_scan.h +79 -0
  14. data/ext/hpricot_scan/hpricot_scan.java.rl +1152 -0
  15. data/ext/hpricot_scan/hpricot_scan.rl +788 -0
  16. data/extras/mingw-rbconfig.rb +176 -0
  17. data/lib/hpricot.rb +26 -0
  18. data/lib/hpricot/blankslate.rb +63 -0
  19. data/lib/hpricot/builder.rb +216 -0
  20. data/lib/hpricot/elements.rb +510 -0
  21. data/lib/hpricot/htmlinfo.rb +691 -0
  22. data/lib/hpricot/inspect.rb +103 -0
  23. data/lib/hpricot/modules.rb +40 -0
  24. data/lib/hpricot/parse.rb +38 -0
  25. data/lib/hpricot/tag.rb +219 -0
  26. data/lib/hpricot/tags.rb +164 -0
  27. data/lib/hpricot/traverse.rb +839 -0
  28. data/lib/hpricot/xchar.rb +94 -0
  29. data/test/files/basic.xhtml +17 -0
  30. data/test/files/boingboing.html +2266 -0
  31. data/test/files/cy0.html +3653 -0
  32. data/test/files/immob.html +400 -0
  33. data/test/files/pace_application.html +1320 -0
  34. data/test/files/tenderlove.html +16 -0
  35. data/test/files/uswebgen.html +220 -0
  36. data/test/files/utf8.html +1054 -0
  37. data/test/files/week9.html +1723 -0
  38. data/test/files/why.xml +19 -0
  39. data/test/load_files.rb +7 -0
  40. data/test/test_alter.rb +95 -0
  41. data/test/test_builder.rb +37 -0
  42. data/test/test_parser.rb +428 -0
  43. data/test/test_paths.rb +25 -0
  44. data/test/test_preserved.rb +79 -0
  45. data/test/test_xml.rb +28 -0
  46. metadata +108 -0
@@ -0,0 +1,4 @@
1
+ require 'mkmf'
2
+ have_header('stdio.h') or exit
3
+ dir_config('fast_xs')
4
+ create_makefile('fast_xs')
@@ -0,0 +1,201 @@
1
+ #define VERSION "0.1"
2
+
3
+ #include <ruby.h>
4
+ #include <assert.h>
5
+ /* #include <stdio.h> */
6
+
7
+ #ifndef RARRAY_LEN
8
+ #define RARRAY_LEN(arr) RARRAY(arr)->len
9
+ #define RARRAY_PTR(arr) RARRAY(arr)->ptr
10
+ #define RSTRING_LEN(str) RSTRING(str)->len
11
+ #define RSTRING_PTR(str) RSTRING(str)->ptr
12
+ #endif
13
+
14
+ static ID unpack_id;
15
+ static VALUE U_fmt, C_fmt;
16
+
17
+ /* give GCC hints for better branch prediction
18
+ * (we layout branches so that ASCII characters are handled faster) */
19
+ #if defined(__GNUC__) && (__GNUC__ >= 3)
20
+ # define likely(x) __builtin_expect (!!(x), 1)
21
+ # define unlikely(x) __builtin_expect (!!(x), 0)
22
+ #else
23
+ # define unlikely(x) (x)
24
+ # define likely(x) (x)
25
+ #endif
26
+
27
+ /* pass-through certain characters for CP-1252 */
28
+ #define p(x) (x-128)
29
+
30
+ static const int cp_1252[] = {
31
+ 8364, /* 128 => 8364, euro sign */
32
+ p(129), /* 129 => 129, pass-through */
33
+ 8218, /* 130 => 8218, single low-9 quotation mark */
34
+ 402, /* 131 => 402, latin small letter f with hook */
35
+ 8222, /* 132 => 8222, double low-9 quotation mark */
36
+ 8230, /* 133 => 8230, horizontal ellipsis */
37
+ 8224, /* 134 => 8224, dagger */
38
+ 8225, /* 135 => 8225, double dagger */
39
+ 710, /* 136 => 710, modifier letter circumflex accent */
40
+ 8240, /* 137 => 8240, per mille sign */
41
+ 352, /* 138 => 352, latin capital letter s with caron */
42
+ 8249, /* 139 => 8249, single left-pointing angle quotation mark */
43
+ 338, /* 140 => 338, latin capital ligature oe */
44
+ p(141), /* 141 => 141, pass-through */
45
+ 381, /* 142 => 381, latin capital letter z with caron */
46
+ p(143), /* 143 => 143, pass-through */
47
+ p(144), /* 144 => 144, pass-through */
48
+ 8216, /* 145 => 8216, left single quotation mark */
49
+ 8217, /* 146 => 8217, right single quotation mark */
50
+ 8220, /* 147 => 8220, left double quotation mark */
51
+ 8221, /* 148 => 8221, right double quotation mark */
52
+ 8226, /* 149 => 8226, bullet */
53
+ 8211, /* 150 => 8211, en dash */
54
+ 8212, /* 151 => 8212, em dash */
55
+ 732, /* 152 => 732, small tilde */
56
+ 8482, /* 153 => 8482, trade mark sign */
57
+ 353, /* 154 => 353, latin small letter s with caron */
58
+ 8250, /* 155 => 8250, single right-pointing angle quotation mark */
59
+ 339, /* 156 => 339, latin small ligature oe */
60
+ p(157), /* 157 => 157, pass-through */
61
+ 382, /* 158 => 382, latin small letter z with caron */
62
+ 376 /* 159 => 376} latin capital letter y with diaeresis */
63
+ };
64
+
65
+ #define VALID_VALUE(n) \
66
+ (n >= 0x20 && n <= 0xD7FF) || \
67
+ (n >= 0xE000 && n <= 0xFFFD) || \
68
+ (n >= 0x10000 && n <= 0x10FFFF)
69
+
70
+ #define CP_1252_ESCAPE(n) do { \
71
+ if (n >= 128 && n <= 159) \
72
+ n = cp_1252[n - 128]; \
73
+ } while(0)
74
+
75
+ #define return_const_len(x) do { \
76
+ memcpy(buf, x, sizeof(x) - 1); \
77
+ return (sizeof(x) - 1); \
78
+ } while (0)
79
+
80
+ static inline size_t bytes_for(int n)
81
+ {
82
+ if (n < 1000)
83
+ return sizeof("&#999;") - 1;
84
+ if (n < 10000)
85
+ return sizeof("&#9999;") - 1;
86
+ if (n < 100000)
87
+ return sizeof("&#99999;") - 1;
88
+ if (n < 1000000)
89
+ return sizeof("&#999999;") - 1;
90
+ /* if (n < 10000000), we won't have cases above 0x10FFFF */
91
+ return sizeof("&#9999999;") - 1;
92
+ }
93
+
94
+ static long escape(char *buf, int n)
95
+ {
96
+ /* handle ASCII first */
97
+ if (likely(n < 128)) {
98
+ if (likely(n >= 0x20 || n == 0x9 || n == 0xA || n == 0xD)) {
99
+ if (unlikely(n == 34))
100
+ return_const_len("&quot;");
101
+ if (unlikely(n == 38))
102
+ return_const_len("&amp;");
103
+ if (unlikely(n == 60))
104
+ return_const_len("&lt;");
105
+ if (unlikely(n == 62))
106
+ return_const_len("&gt;");
107
+ buf[0] = (char)n;
108
+ return 1;
109
+ }
110
+
111
+ buf[0] = '*';
112
+ return 1;
113
+ }
114
+
115
+ CP_1252_ESCAPE(n);
116
+
117
+ if (VALID_VALUE(n)) {
118
+ /* return snprintf(buf, sizeof("&#1114111;"), "&#%i;", n); */
119
+ RUBY_EXTERN const char ruby_digitmap[];
120
+ int rv = 3; /* &#; */
121
+ buf += bytes_for(n);
122
+ *--buf = ';';
123
+ do {
124
+ *--buf = ruby_digitmap[(int)(n % 10)];
125
+ ++rv;
126
+ } while (n /= 10);
127
+ *--buf = '#';
128
+ *--buf = '&';
129
+ return rv;
130
+ }
131
+ buf[0] = '*';
132
+ return 1;
133
+ }
134
+
135
+ #undef return_const_len
136
+
137
+ static long escaped_len(int n)
138
+ {
139
+ if (likely(n < 128)) {
140
+ if (unlikely(n == 34))
141
+ return (sizeof("&quot;") - 1);
142
+ if (unlikely(n == 38))
143
+ return (sizeof("&amp;") - 1);
144
+ if (unlikely(n == 60 || n == 62))
145
+ return (sizeof("&gt;") - 1);
146
+ return 1;
147
+ }
148
+
149
+ CP_1252_ESCAPE(n);
150
+
151
+ if (VALID_VALUE(n))
152
+ return bytes_for(n);
153
+ return 1;
154
+ }
155
+
156
+ static VALUE unpack_utf8(VALUE self)
157
+ {
158
+ return rb_funcall(self, unpack_id, 1, U_fmt);
159
+ }
160
+
161
+ static VALUE unpack_uchar(VALUE self)
162
+ {
163
+ return rb_funcall(self, unpack_id, 1, C_fmt);
164
+ }
165
+
166
+ VALUE fast_xs(VALUE self)
167
+ {
168
+ long i;
169
+ struct RArray *array;
170
+ char *s, *c;
171
+ long s_len = 0;
172
+ VALUE *tmp;
173
+
174
+ array = RARRAY(rb_rescue(unpack_utf8, self, unpack_uchar, self));
175
+
176
+ tmp = RARRAY_PTR(array);
177
+ for (i = RARRAY_LEN(array); --i >= 0; tmp++)
178
+ s_len += escaped_len(NUM2INT(*tmp));
179
+
180
+ c = s = alloca(s_len + 1);
181
+
182
+ tmp = RARRAY_PTR(array);
183
+ for (i = RARRAY_LEN(array); --i >= 0; tmp++)
184
+ c += escape(c, NUM2INT(*tmp));
185
+
186
+ *c = '\0';
187
+ return rb_str_new(s, s_len);
188
+ }
189
+
190
+ void Init_fast_xs(void)
191
+ {
192
+ assert(cp_1252[159 - 128] == 376); /* just in case I skipped a line */
193
+
194
+ unpack_id = rb_intern("unpack");
195
+ U_fmt = rb_str_new("U*", 2);
196
+ C_fmt = rb_str_new("C*", 2);
197
+ rb_global_variable(&U_fmt);
198
+ rb_global_variable(&C_fmt);
199
+
200
+ rb_define_method(rb_cString, "fast_xs", fast_xs, 0);
201
+ }
@@ -0,0 +1,2085 @@
1
+ // line 1 "ext/hpricot_scan/hpricot_scan.java.rl"
2
+
3
+ import java.io.IOException;
4
+
5
+ import org.jruby.Ruby;
6
+ import org.jruby.RubyArray;
7
+ import org.jruby.RubyClass;
8
+ import org.jruby.RubyHash;
9
+ import org.jruby.RubyModule;
10
+ import org.jruby.RubyNumeric;
11
+ import org.jruby.RubyObject;
12
+ import org.jruby.RubyObjectAdapter;
13
+ import org.jruby.RubyRegexp;
14
+ import org.jruby.RubyString;
15
+ import org.jruby.anno.JRubyMethod;
16
+ import org.jruby.exceptions.RaiseException;
17
+ import org.jruby.javasupport.JavaEmbedUtils;
18
+ import org.jruby.runtime.Arity;
19
+ import org.jruby.runtime.Block;
20
+ import org.jruby.runtime.ObjectAllocator;
21
+ import org.jruby.runtime.ThreadContext;
22
+ import org.jruby.runtime.builtin.IRubyObject;
23
+ import org.jruby.runtime.callback.Callback;
24
+ import org.jruby.exceptions.RaiseException;
25
+ import org.jruby.runtime.load.BasicLibraryService;
26
+ import org.jruby.util.ByteList;
27
+
28
+ public class HpricotScanService implements BasicLibraryService {
29
+ public static byte[] realloc(byte[] input, int size) {
30
+ byte[] newArray = new byte[size];
31
+ System.arraycopy(input, 0, newArray, 0, input.length);
32
+ return newArray;
33
+ }
34
+
35
+ // hpricot_state
36
+ public static class State {
37
+ public IRubyObject doc;
38
+ public IRubyObject focus;
39
+ public IRubyObject last;
40
+ public IRubyObject EC;
41
+ public boolean xml, strict, fixup;
42
+ }
43
+
44
+ static boolean OPT(IRubyObject opts, String key) {
45
+ Ruby runtime = opts.getRuntime();
46
+ return !opts.isNil() && ((RubyHash)opts).op_aref(runtime.getCurrentContext(), runtime.newSymbol(key)).isTrue();
47
+ }
48
+
49
+ // H_PROP(name, H_ELE_TAG)
50
+ public static IRubyObject hpricot_ele_set_name(IRubyObject self, IRubyObject x) {
51
+ H_ELE_SET(self, H_ELE_TAG, x);
52
+ return self;
53
+ }
54
+
55
+ public static IRubyObject hpricot_ele_clear_name(IRubyObject self) {
56
+ H_ELE_SET(self, H_ELE_TAG, self.getRuntime().getNil());
57
+ return self.getRuntime().getTrue();
58
+ }
59
+
60
+ public static IRubyObject hpricot_ele_get_name(IRubyObject self) {
61
+ return H_ELE_GET(self, H_ELE_TAG);
62
+ }
63
+
64
+ // H_PROP(raw, H_ELE_RAW)
65
+ public static IRubyObject hpricot_ele_set_raw(IRubyObject self, IRubyObject x) {
66
+ H_ELE_SET(self, H_ELE_RAW, x);
67
+ return self;
68
+ }
69
+
70
+ public static IRubyObject hpricot_ele_clear_raw(IRubyObject self) {
71
+ H_ELE_SET(self, H_ELE_RAW, self.getRuntime().getNil());
72
+ return self.getRuntime().getTrue();
73
+ }
74
+
75
+ public static IRubyObject hpricot_ele_get_raw(IRubyObject self) {
76
+ return H_ELE_GET(self, H_ELE_RAW);
77
+ }
78
+
79
+ // H_PROP(parent, H_ELE_PARENT)
80
+ public static IRubyObject hpricot_ele_set_parent(IRubyObject self, IRubyObject x) {
81
+ H_ELE_SET(self, H_ELE_PARENT, x);
82
+ return self;
83
+ }
84
+
85
+ public static IRubyObject hpricot_ele_clear_parent(IRubyObject self) {
86
+ H_ELE_SET(self, H_ELE_PARENT, self.getRuntime().getNil());
87
+ return self.getRuntime().getTrue();
88
+ }
89
+
90
+ public static IRubyObject hpricot_ele_get_parent(IRubyObject self) {
91
+ return H_ELE_GET(self, H_ELE_PARENT);
92
+ }
93
+
94
+ // H_PROP(attr, H_ELE_ATTR)
95
+ public static IRubyObject hpricot_ele_set_attr(IRubyObject self, IRubyObject x) {
96
+ H_ELE_SET(self, H_ELE_ATTR, x);
97
+ return self;
98
+ }
99
+
100
+ public static IRubyObject hpricot_ele_clear_attr(IRubyObject self) {
101
+ H_ELE_SET(self, H_ELE_ATTR, self.getRuntime().getNil());
102
+ return self.getRuntime().getTrue();
103
+ }
104
+
105
+ public static IRubyObject hpricot_ele_get_attr(IRubyObject self) {
106
+ return H_ELE_GET(self, H_ELE_ATTR);
107
+ }
108
+
109
+ // H_PROP(etag, H_ELE_ETAG)
110
+ public static IRubyObject hpricot_ele_set_etag(IRubyObject self, IRubyObject x) {
111
+ H_ELE_SET(self, H_ELE_ETAG, x);
112
+ return self;
113
+ }
114
+
115
+ public static IRubyObject hpricot_ele_clear_etag(IRubyObject self) {
116
+ H_ELE_SET(self, H_ELE_ETAG, self.getRuntime().getNil());
117
+ return self.getRuntime().getTrue();
118
+ }
119
+
120
+ public static IRubyObject hpricot_ele_get_etag(IRubyObject self) {
121
+ return H_ELE_GET(self, H_ELE_ETAG);
122
+ }
123
+
124
+ // H_PROP(children, H_ELE_CHILDREN)
125
+ public static IRubyObject hpricot_ele_set_children(IRubyObject self, IRubyObject x) {
126
+ H_ELE_SET(self, H_ELE_CHILDREN, x);
127
+ return self;
128
+ }
129
+
130
+ public static IRubyObject hpricot_ele_clear_children(IRubyObject self) {
131
+ H_ELE_SET(self, H_ELE_CHILDREN, self.getRuntime().getNil());
132
+ return self.getRuntime().getTrue();
133
+ }
134
+
135
+ public static IRubyObject hpricot_ele_get_children(IRubyObject self) {
136
+ return H_ELE_GET(self, H_ELE_CHILDREN);
137
+ }
138
+
139
+ // H_ATTR(target)
140
+ public static IRubyObject hpricot_ele_set_target(IRubyObject self, IRubyObject x) {
141
+ ((RubyHash)H_ELE_GET(self, H_ELE_ATTR)).fastASet(self.getRuntime().newSymbol("target"), x);
142
+ return self;
143
+ }
144
+
145
+ public static IRubyObject hpricot_ele_get_target(IRubyObject self) {
146
+ return ((RubyHash)H_ELE_GET(self, H_ELE_ATTR)).op_aref(self.getRuntime().getCurrentContext(), self.getRuntime().newSymbol("target"));
147
+ }
148
+
149
+ // H_ATTR(encoding)
150
+ public static IRubyObject hpricot_ele_set_encoding(IRubyObject self, IRubyObject x) {
151
+ ((RubyHash)H_ELE_GET(self, H_ELE_ATTR)).fastASet(self.getRuntime().newSymbol("encoding"), x);
152
+ return self;
153
+ }
154
+
155
+ public static IRubyObject hpricot_ele_get_encoding(IRubyObject self) {
156
+ return ((RubyHash)H_ELE_GET(self, H_ELE_ATTR)).op_aref(self.getRuntime().getCurrentContext(), self.getRuntime().newSymbol("encoding"));
157
+ }
158
+
159
+ // H_ATTR(version)
160
+ public static IRubyObject hpricot_ele_set_version(IRubyObject self, IRubyObject x) {
161
+ ((RubyHash)H_ELE_GET(self, H_ELE_ATTR)).fastASet(self.getRuntime().newSymbol("version"), x);
162
+ return self;
163
+ }
164
+
165
+ public static IRubyObject hpricot_ele_get_version(IRubyObject self) {
166
+ return ((RubyHash)H_ELE_GET(self, H_ELE_ATTR)).op_aref(self.getRuntime().getCurrentContext(), self.getRuntime().newSymbol("version"));
167
+ }
168
+
169
+ // H_ATTR(standalone)
170
+ public static IRubyObject hpricot_ele_set_standalone(IRubyObject self, IRubyObject x) {
171
+ ((RubyHash)H_ELE_GET(self, H_ELE_ATTR)).fastASet(self.getRuntime().newSymbol("standalone"), x);
172
+ return self;
173
+ }
174
+
175
+ public static IRubyObject hpricot_ele_get_standalone(IRubyObject self) {
176
+ return ((RubyHash)H_ELE_GET(self, H_ELE_ATTR)).op_aref(self.getRuntime().getCurrentContext(), self.getRuntime().newSymbol("standalone"));
177
+ }
178
+
179
+ // H_ATTR(system_id)
180
+ public static IRubyObject hpricot_ele_set_system_id(IRubyObject self, IRubyObject x) {
181
+ ((RubyHash)H_ELE_GET(self, H_ELE_ATTR)).fastASet(self.getRuntime().newSymbol("system_id"), x);
182
+ return self;
183
+ }
184
+
185
+ public static IRubyObject hpricot_ele_get_system_id(IRubyObject self) {
186
+ return ((RubyHash)H_ELE_GET(self, H_ELE_ATTR)).op_aref(self.getRuntime().getCurrentContext(), self.getRuntime().newSymbol("system_id"));
187
+ }
188
+
189
+ // H_ATTR(public_id)
190
+ public static IRubyObject hpricot_ele_set_public_id(IRubyObject self, IRubyObject x) {
191
+ ((RubyHash)H_ELE_GET(self, H_ELE_ATTR)).fastASet(self.getRuntime().newSymbol("public_id"), x);
192
+ return self;
193
+ }
194
+
195
+ public static IRubyObject hpricot_ele_get_public_id(IRubyObject self) {
196
+ return ((RubyHash)H_ELE_GET(self, H_ELE_ATTR)).op_aref(self.getRuntime().getCurrentContext(), self.getRuntime().newSymbol("public_id"));
197
+ }
198
+
199
+ public static class Scanner {
200
+ public IRubyObject SET(int mark, int E, IRubyObject org) {
201
+ if(mark == -1 || E == mark) {
202
+ return runtime.newString("");
203
+ } else if(E > mark) {
204
+ return RubyString.newString(runtime, data, mark, E-mark);
205
+ } else {
206
+ return org;
207
+ }
208
+ }
209
+
210
+ public int SLIDE(int N) {
211
+ if(N > ts) {
212
+ return N - ts;
213
+ } else {
214
+ return N;
215
+ }
216
+ }
217
+
218
+ public IRubyObject CAT(IRubyObject N, int mark, int E) {
219
+ if(N.isNil()) {
220
+ return SET(mark, E, N);
221
+ } else {
222
+ ((RubyString)N).cat(data, mark, E-mark);
223
+ return N;
224
+ }
225
+ }
226
+
227
+ public void ATTR(IRubyObject K, IRubyObject V) {
228
+ if(!K.isNil()) {
229
+ if(attr.isNil()) {
230
+ attr = RubyHash.newHash(runtime);
231
+ }
232
+ ((RubyHash)attr).fastASet(K, V);
233
+ }
234
+ }
235
+
236
+ public void TEXT_PASS() {
237
+ if(!text) {
238
+ if(ele_open) {
239
+ ele_open = false;
240
+ if(ts != -1) {
241
+ mark_tag = ts;
242
+ }
243
+ } else {
244
+ mark_tag = p;
245
+ }
246
+ attr = runtime.getNil();
247
+ tag = runtime.getNil();
248
+ text = true;
249
+ }
250
+ }
251
+
252
+ public void ELE(IRubyObject N) {
253
+ if(te > ts || text) {
254
+ int raw = -1;
255
+ int rawlen = 0;
256
+ ele_open = false;
257
+ text = false;
258
+
259
+ if(ts != -1 && N != x.sym_cdata && N != x.sym_text && N != x.sym_procins && N != x.sym_comment) {
260
+ raw = ts;
261
+ rawlen = te - ts;
262
+ }
263
+
264
+ if(block.isGiven()) {
265
+ IRubyObject raw_string = runtime.getNil();
266
+ if(raw != -1) {
267
+ raw_string = RubyString.newString(runtime, data, raw, rawlen);
268
+ }
269
+ yieldTokens(N, tag, attr, runtime.getNil(), taint);
270
+ } else {
271
+ hpricotToken(S, N, tag, attr, raw, rawlen, taint);
272
+ }
273
+ }
274
+ }
275
+
276
+
277
+ public void EBLK(IRubyObject N, int T) {
278
+ tag = CAT(tag, mark_tag, p - T + 1);
279
+ ELE(N);
280
+ }
281
+
282
+ public void hpricotAdd(IRubyObject focus, IRubyObject ele) {
283
+ IRubyObject children = H_ELE_GET(focus, H_ELE_CHILDREN);
284
+ if(children.isNil()) {
285
+ H_ELE_SET(focus, H_ELE_CHILDREN, children = RubyArray.newArray(runtime, 1));
286
+ }
287
+ ((RubyArray)children).append(ele);
288
+ H_ELE_SET(ele, H_ELE_PARENT, focus);
289
+ }
290
+
291
+ private static class TokenInfo {
292
+ public IRubyObject sym;
293
+ public IRubyObject tag;
294
+ public IRubyObject attr;
295
+ public int raw;
296
+ public int rawlen;
297
+ public IRubyObject ec;
298
+ public IRubyObject ele;
299
+ public Extra x;
300
+ public Ruby runtime;
301
+ public Scanner scanner;
302
+ public State S;
303
+
304
+ public void H_ELE(RubyClass klass) {
305
+ ele = klass.allocate();
306
+ if(klass == x.cElem) {
307
+ H_ELE_SET(ele, H_ELE_TAG, tag);
308
+ H_ELE_SET(ele, H_ELE_ATTR, attr);
309
+ H_ELE_SET(ele, H_ELE_EC, ec);
310
+ if(raw != -1 && (sym == x.sym_emptytag || sym == x.sym_stag || sym == x.sym_doctype)) {
311
+ H_ELE_SET(ele, H_ELE_RAW, RubyString.newString(runtime, scanner.data, raw, rawlen));
312
+ }
313
+ } else if(klass == x.cDocType || klass == x.cProcIns || klass == x.cXMLDecl || klass == x.cBogusETag) {
314
+ if(klass == x.cBogusETag) {
315
+ H_ELE_SET(ele, H_ELE_TAG, tag);
316
+ if(raw != -1) {
317
+ H_ELE_SET(ele, H_ELE_ATTR, RubyString.newString(runtime, scanner.data, raw, rawlen));
318
+ }
319
+ } else {
320
+ if(klass == x.cDocType) {
321
+ scanner.ATTR(runtime.newSymbol("target"), tag);
322
+ }
323
+ H_ELE_SET(ele, H_ELE_ATTR, attr);
324
+ if(klass != x.cProcIns) {
325
+ tag = runtime.getNil();
326
+ if(raw != -1) {
327
+ tag = RubyString.newString(runtime, scanner.data, raw, rawlen);
328
+ }
329
+ }
330
+ H_ELE_SET(ele, H_ELE_TAG, tag);
331
+ }
332
+ } else {
333
+ H_ELE_SET(ele, H_ELE_TAG, tag);
334
+ }
335
+ S.last = ele;
336
+ }
337
+
338
+ public void hpricotToken(boolean taint) {
339
+ //
340
+ // in html mode, fix up start tags incorrectly formed as empty tags
341
+ //
342
+ if(!S.xml) {
343
+ if(sym == x.sym_emptytag || sym == x.sym_stag || sym == x.sym_etag) {
344
+ ec = ((RubyHash)S.EC).op_aref(scanner.ctx, tag);
345
+ if(ec.isNil()) {
346
+ tag = tag.callMethod(scanner.ctx, "downcase");
347
+ ec = ((RubyHash)S.EC).op_aref(scanner.ctx, tag);
348
+ }
349
+ }
350
+
351
+ if(H_ELE_GET(S.focus, H_ELE_EC) == x.sym_CDATA &&
352
+ (sym != x.sym_procins && sym != x.sym_comment && sym != x.sym_cdata && sym != x.sym_text) &&
353
+ !(sym == x.sym_etag && runtime.newFixnum(tag.hashCode()).equals(H_ELE_GET(S.focus, H_ELE_HASH)))) {
354
+ sym = x.sym_text;
355
+ tag = RubyString.newString(runtime, scanner.data, raw, rawlen);
356
+ }
357
+
358
+ if(!ec.isNil()) {
359
+ if(sym == x.sym_emptytag) {
360
+ if(ec != x.sym_EMPTY) {
361
+ sym = x.sym_stag;
362
+ }
363
+ } else if(sym == x.sym_stag) {
364
+ if(ec == x.sym_EMPTY) {
365
+ sym = x.sym_emptytag;
366
+ }
367
+ }
368
+ }
369
+ }
370
+
371
+ if(sym == x.sym_emptytag || sym == x.sym_stag) {
372
+ IRubyObject name = runtime.newFixnum(tag.hashCode());
373
+ H_ELE(x.cElem);
374
+ H_ELE_SET(ele, H_ELE_HASH, name);
375
+
376
+ if(!S.xml) {
377
+ IRubyObject match = runtime.getNil(), e = S.focus;
378
+ while(e != S.doc) {
379
+ IRubyObject hEC = H_ELE_GET(e, H_ELE_EC);
380
+ if(hEC instanceof RubyHash) {
381
+ IRubyObject has = ((RubyHash)hEC).op_aref(scanner.ctx, name);
382
+ if(!has.isNil()) {
383
+ if(has == runtime.getTrue()) {
384
+ if(match.isNil()) {
385
+ match = e;
386
+ }
387
+ } else if(has == x.symAllow) {
388
+ match = S.focus;
389
+ } else if(has == x.symDeny) {
390
+ match = runtime.getNil();
391
+ }
392
+ }
393
+ }
394
+ e = H_ELE_GET(e, H_ELE_PARENT);
395
+ }
396
+
397
+ if(match.isNil()) {
398
+ match = S.focus;
399
+ }
400
+ S.focus = match;
401
+ }
402
+
403
+ scanner.hpricotAdd(S.focus, ele);
404
+
405
+ //
406
+ // in the case of a start tag that should be empty, just
407
+ // skip the step that focuses the element. focusing moves
408
+ // us deeper into the document.
409
+ //
410
+ if(sym == x.sym_stag) {
411
+ if(S.xml || ec != x.sym_EMPTY) {
412
+ S.focus = ele;
413
+ S.last = runtime.getNil();
414
+ }
415
+ }
416
+ } else if(sym == x.sym_etag) {
417
+ IRubyObject name, match = runtime.getNil(), e = S.focus;
418
+ if(S.strict) {
419
+ if(((RubyHash)S.EC).op_aref(scanner.ctx, tag).isNil()) {
420
+ tag = runtime.newString("div");
421
+ }
422
+ }
423
+
424
+ name = runtime.newFixnum(tag.hashCode());
425
+ while(e != S.doc) {
426
+ if(H_ELE_GET(e, H_ELE_HASH).equals(name)) {
427
+ match = e;
428
+ break;
429
+ }
430
+ e = H_ELE_GET(e, H_ELE_PARENT);
431
+
432
+ }
433
+ if(match.isNil()) {
434
+ H_ELE(x.cBogusETag);
435
+ scanner.hpricotAdd(S.focus, ele);
436
+ } else {
437
+ ele = runtime.getNil();
438
+ if(raw != -1) {
439
+ ele = RubyString.newString(runtime, scanner.data, raw, rawlen);
440
+ }
441
+ H_ELE_SET(match, H_ELE_ETAG, ele);
442
+ S.focus = H_ELE_GET(match, H_ELE_PARENT);
443
+ S.last = runtime.getNil();
444
+
445
+ }
446
+ } else if(sym == x.sym_cdata) {
447
+ H_ELE(x.cCData);
448
+ scanner.hpricotAdd(S.focus, ele);
449
+ } else if(sym == x.sym_comment) {
450
+ H_ELE(x.cComment);
451
+ scanner.hpricotAdd(S.focus, ele);
452
+ } else if(sym == x.sym_doctype) {
453
+ H_ELE(x.cDocType);
454
+ if(S.strict) {
455
+ RubyHash h = (RubyHash)attr;
456
+ h.fastASet(runtime.newSymbol("system_id"), runtime.newString("http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd"));
457
+ h.fastASet(runtime.newSymbol("public_id"), runtime.newString("-//W3C//DTD XHTML 1.0 Strict//EN"));
458
+ }
459
+ scanner.hpricotAdd(S.focus, ele);
460
+ } else if(sym == x.sym_procins) {
461
+ IRubyObject match = tag.callMethod(scanner.ctx, "match", x.reProcInsParse);
462
+ tag = RubyRegexp.nth_match(1, match);
463
+ attr = RubyRegexp.nth_match(2, match);
464
+ H_ELE(x.cProcIns);
465
+ scanner.hpricotAdd(S.focus, ele);
466
+ } else if(sym == x.sym_text) {
467
+ if(!S.last.isNil() && S.last.getType() == x.cText) {
468
+ ((RubyString)H_ELE_GET(S.last, H_ELE_TAG)).append(tag);
469
+ } else {
470
+ H_ELE(x.cText);
471
+ scanner.hpricotAdd(S.focus, ele);
472
+ }
473
+ } else if(sym == x.sym_xmldecl) {
474
+ H_ELE(x.cXMLDecl);
475
+ scanner.hpricotAdd(S.focus, ele);
476
+ }
477
+ }
478
+ }
479
+
480
+ public void hpricotToken(State S, IRubyObject _sym, IRubyObject _tag, IRubyObject _attr, int _raw, int _rawlen, boolean taint) {
481
+ TokenInfo t = new TokenInfo();
482
+ t.sym = _sym;
483
+ t.tag = _tag;
484
+ t.attr = _attr;
485
+ t.raw = _raw;
486
+ t.rawlen = _rawlen;
487
+ t.ec = runtime.getNil();
488
+ t.ele = runtime.getNil();
489
+ t.x = x;
490
+ t.runtime = runtime;
491
+ t.scanner = this;
492
+ t.S = S;
493
+
494
+ t.hpricotToken(taint);
495
+ }
496
+
497
+ public void yieldTokens(IRubyObject sym, IRubyObject tag, IRubyObject attr, IRubyObject raw, boolean taint) {
498
+ if(sym == x.sym_text) {
499
+ raw = tag;
500
+ }
501
+ IRubyObject ary = RubyArray.newArrayNoCopy(runtime, new IRubyObject[]{sym, tag, attr, raw});
502
+ if(taint) {
503
+ ary.setTaint(true);
504
+ tag.setTaint(true);
505
+ attr.setTaint(true);
506
+ raw.setTaint(true);
507
+ }
508
+
509
+ block.yield(ctx, ary);
510
+ }
511
+
512
+ // line 561 "ext/hpricot_scan/hpricot_scan.java.rl"
513
+
514
+
515
+
516
+ // line 517 "ext/hpricot_scan/HpricotScanService.java"
517
+ private static byte[] init__hpricot_scan_actions_0()
518
+ {
519
+ return new byte [] {
520
+ 0, 1, 1, 1, 2, 1, 4, 1, 5, 1, 6, 1,
521
+ 7, 1, 8, 1, 9, 1, 10, 1, 11, 1, 12, 1,
522
+ 14, 1, 16, 1, 20, 1, 21, 1, 22, 1, 24, 1,
523
+ 25, 1, 26, 1, 28, 1, 29, 1, 30, 1, 32, 1,
524
+ 33, 1, 38, 1, 39, 1, 40, 1, 41, 1, 42, 1,
525
+ 43, 1, 44, 1, 45, 1, 46, 1, 47, 1, 48, 1,
526
+ 49, 1, 50, 1, 51, 2, 2, 5, 2, 2, 6, 2,
527
+ 2, 11, 2, 2, 12, 2, 2, 14, 2, 4, 39, 2,
528
+ 4, 40, 2, 4, 41, 2, 5, 2, 2, 6, 14, 2,
529
+ 7, 6, 2, 7, 14, 2, 11, 12, 2, 13, 3, 2,
530
+ 14, 6, 2, 14, 40, 2, 15, 24, 2, 15, 28, 2,
531
+ 15, 32, 2, 15, 45, 2, 17, 23, 2, 18, 27, 2,
532
+ 19, 31, 2, 22, 34, 2, 22, 36, 3, 2, 6, 14,
533
+ 3, 2, 14, 6, 3, 6, 7, 14, 3, 6, 14, 40,
534
+ 3, 7, 14, 40, 3, 11, 2, 12, 3, 14, 6, 40,
535
+ 3, 14, 13, 3, 3, 22, 0, 37, 3, 22, 2, 34,
536
+ 3, 22, 14, 35, 4, 2, 14, 13, 3, 4, 6, 7,
537
+ 14, 40, 4, 22, 2, 14, 35, 4, 22, 6, 14, 35,
538
+ 4, 22, 7, 14, 35, 4, 22, 14, 6, 35, 5, 22,
539
+ 2, 6, 14, 35, 5, 22, 2, 14, 6, 35, 5, 22,
540
+ 6, 7, 14, 35
541
+ };
542
+ }
543
+
544
+ private static final byte _hpricot_scan_actions[] = init__hpricot_scan_actions_0();
545
+
546
+
547
+ private static short[] init__hpricot_scan_key_offsets_0()
548
+ {
549
+ return new short [] {
550
+ 0, 3, 4, 5, 6, 7, 8, 9, 10, 13, 22, 37,
551
+ 44, 45, 46, 47, 48, 49, 52, 57, 69, 81, 86, 93,
552
+ 94, 95, 100, 101, 105, 106, 107, 121, 135, 152, 169, 186,
553
+ 203, 210, 212, 214, 220, 222, 227, 232, 238, 240, 245, 251,
554
+ 265, 266, 267, 268, 269, 270, 271, 272, 273, 274, 275, 276,
555
+ 282, 296, 300, 313, 326, 340, 354, 355, 366, 375, 388, 405,
556
+ 423, 441, 450, 461, 480, 499, 510, 521, 536, 538, 540, 556,
557
+ 572, 575, 587, 599, 619, 639, 658, 677, 697, 717, 728, 739,
558
+ 751, 763, 775, 791, 794, 809, 811, 813, 829, 845, 848, 860,
559
+ 871, 890, 910, 930, 941, 952, 964, 984, 1004, 1016, 1036, 1057,
560
+ 1074, 1091, 1095, 1098, 1110, 1122, 1142, 1162, 1182, 1194, 1206, 1226,
561
+ 1242, 1258, 1270, 1291, 1310, 1313, 1328, 1340, 1355, 1358, 1369, 1371,
562
+ 1373, 1384, 1391, 1404, 1418, 1432, 1445, 1446, 1447, 1448, 1449, 1450,
563
+ 1451, 1455, 1460, 1469, 1479, 1484, 1491, 1492, 1493, 1494, 1495, 1496,
564
+ 1497, 1498, 1499, 1503, 1508, 1512, 1522, 1527, 1533, 1534, 1535, 1536,
565
+ 1537, 1538, 1539, 1540, 1541, 1542, 1546, 1551, 1553, 1554, 1555, 1560,
566
+ 1561, 1562, 1564, 1565, 1566, 1567, 1568, 1572, 1582, 1591, 1601, 1602,
567
+ 1603, 1605, 1614, 1615, 1616, 1617, 1619, 1621, 1624, 1627, 1631, 1633,
568
+ 1634, 1636, 1637, 1640
569
+ };
570
+ }
571
+
572
+ private static final short _hpricot_scan_key_offsets[] = init__hpricot_scan_key_offsets_0();
573
+
574
+
575
+ private static char[] init__hpricot_scan_trans_keys_0()
576
+ {
577
+ return new char [] {
578
+ 45, 68, 91, 45, 79, 67, 84, 89, 80, 69, 32, 9,
579
+ 13, 32, 58, 95, 9, 13, 65, 90, 97, 122, 32, 62,
580
+ 63, 91, 95, 9, 13, 45, 46, 48, 58, 65, 90, 97,
581
+ 122, 32, 62, 80, 83, 91, 9, 13, 85, 66, 76, 73,
582
+ 67, 32, 9, 13, 32, 34, 39, 9, 13, 9, 34, 61,
583
+ 95, 32, 37, 39, 59, 63, 90, 97, 122, 9, 34, 61,
584
+ 95, 32, 37, 39, 59, 63, 90, 97, 122, 32, 62, 91,
585
+ 9, 13, 32, 34, 39, 62, 91, 9, 13, 34, 34, 32,
586
+ 62, 91, 9, 13, 93, 32, 62, 9, 13, 39, 39, 9,
587
+ 39, 61, 95, 32, 33, 35, 37, 40, 59, 63, 90, 97,
588
+ 122, 9, 39, 61, 95, 32, 33, 35, 37, 40, 59, 63,
589
+ 90, 97, 122, 9, 32, 33, 39, 62, 91, 95, 10, 13,
590
+ 35, 37, 40, 59, 61, 90, 97, 122, 9, 32, 34, 39,
591
+ 62, 91, 95, 10, 13, 33, 37, 40, 59, 61, 90, 97,
592
+ 122, 9, 32, 33, 39, 62, 91, 95, 10, 13, 35, 37,
593
+ 40, 59, 61, 90, 97, 122, 9, 32, 34, 39, 62, 91,
594
+ 95, 10, 13, 33, 37, 40, 59, 61, 90, 97, 122, 32,
595
+ 34, 39, 62, 91, 9, 13, 34, 39, 34, 39, 32, 39,
596
+ 62, 91, 9, 13, 39, 93, 32, 62, 93, 9, 13, 32,
597
+ 39, 62, 9, 13, 32, 34, 62, 91, 9, 13, 34, 93,
598
+ 32, 34, 62, 9, 13, 32, 39, 62, 91, 9, 13, 9,
599
+ 39, 61, 95, 32, 33, 35, 37, 40, 59, 63, 90, 97,
600
+ 122, 89, 83, 84, 69, 77, 67, 68, 65, 84, 65, 91,
601
+ 58, 95, 65, 90, 97, 122, 32, 62, 63, 95, 9, 13,
602
+ 45, 46, 48, 58, 65, 90, 97, 122, 32, 62, 9, 13,
603
+ 32, 47, 62, 63, 95, 9, 13, 45, 58, 65, 90, 97,
604
+ 122, 32, 47, 62, 63, 95, 9, 13, 45, 58, 65, 90,
605
+ 97, 122, 32, 47, 61, 62, 63, 95, 9, 13, 45, 58,
606
+ 65, 90, 97, 122, 32, 47, 61, 62, 63, 95, 9, 13,
607
+ 45, 58, 65, 90, 97, 122, 62, 13, 32, 34, 39, 47,
608
+ 60, 62, 9, 10, 11, 12, 13, 32, 47, 60, 62, 9,
609
+ 10, 11, 12, 32, 47, 62, 63, 95, 9, 13, 45, 58,
610
+ 65, 90, 97, 122, 13, 32, 47, 60, 62, 63, 95, 9,
611
+ 10, 11, 12, 45, 58, 65, 90, 97, 122, 13, 32, 47,
612
+ 60, 61, 62, 63, 95, 9, 10, 11, 12, 45, 58, 65,
613
+ 90, 97, 122, 13, 32, 47, 60, 61, 62, 63, 95, 9,
614
+ 10, 11, 12, 45, 58, 65, 90, 97, 122, 13, 32, 47,
615
+ 60, 62, 9, 10, 11, 12, 13, 32, 34, 39, 47, 60,
616
+ 62, 9, 10, 11, 12, 13, 32, 34, 39, 47, 60, 62,
617
+ 63, 95, 9, 10, 11, 12, 45, 58, 65, 90, 97, 122,
618
+ 13, 32, 34, 39, 47, 60, 62, 63, 95, 9, 10, 11,
619
+ 12, 45, 58, 65, 90, 97, 122, 13, 32, 34, 47, 60,
620
+ 62, 92, 9, 10, 11, 12, 13, 32, 34, 47, 60, 62,
621
+ 92, 9, 10, 11, 12, 32, 34, 47, 62, 63, 92, 95,
622
+ 9, 13, 45, 58, 65, 90, 97, 122, 34, 92, 34, 92,
623
+ 32, 34, 47, 61, 62, 63, 92, 95, 9, 13, 45, 58,
624
+ 65, 90, 97, 122, 32, 34, 47, 61, 62, 63, 92, 95,
625
+ 9, 13, 45, 58, 65, 90, 97, 122, 34, 62, 92, 13,
626
+ 32, 34, 39, 47, 60, 62, 92, 9, 10, 11, 12, 13,
627
+ 32, 34, 39, 47, 60, 62, 92, 9, 10, 11, 12, 13,
628
+ 32, 34, 39, 47, 60, 62, 63, 92, 95, 9, 10, 11,
629
+ 12, 45, 58, 65, 90, 97, 122, 13, 32, 34, 39, 47,
630
+ 60, 62, 63, 92, 95, 9, 10, 11, 12, 45, 58, 65,
631
+ 90, 97, 122, 13, 32, 34, 47, 60, 62, 63, 92, 95,
632
+ 9, 10, 11, 12, 45, 58, 65, 90, 97, 122, 13, 32,
633
+ 34, 47, 60, 62, 63, 92, 95, 9, 10, 11, 12, 45,
634
+ 58, 65, 90, 97, 122, 13, 32, 34, 47, 60, 61, 62,
635
+ 63, 92, 95, 9, 10, 11, 12, 45, 58, 65, 90, 97,
636
+ 122, 13, 32, 34, 47, 60, 61, 62, 63, 92, 95, 9,
637
+ 10, 11, 12, 45, 58, 65, 90, 97, 122, 13, 32, 34,
638
+ 47, 60, 62, 92, 9, 10, 11, 12, 13, 32, 34, 47,
639
+ 60, 62, 92, 9, 10, 11, 12, 13, 32, 34, 39, 47,
640
+ 60, 62, 92, 9, 10, 11, 12, 13, 32, 34, 39, 47,
641
+ 60, 62, 92, 9, 10, 11, 12, 13, 32, 34, 39, 47,
642
+ 60, 62, 92, 9, 10, 11, 12, 32, 34, 39, 47, 62,
643
+ 63, 92, 95, 9, 13, 45, 58, 65, 90, 97, 122, 34,
644
+ 39, 92, 32, 39, 47, 62, 63, 92, 95, 9, 13, 45,
645
+ 58, 65, 90, 97, 122, 39, 92, 39, 92, 32, 39, 47,
646
+ 61, 62, 63, 92, 95, 9, 13, 45, 58, 65, 90, 97,
647
+ 122, 32, 39, 47, 61, 62, 63, 92, 95, 9, 13, 45,
648
+ 58, 65, 90, 97, 122, 39, 62, 92, 13, 32, 34, 39,
649
+ 47, 60, 62, 92, 9, 10, 11, 12, 13, 32, 39, 47,
650
+ 60, 62, 92, 9, 10, 11, 12, 13, 32, 39, 47, 60,
651
+ 62, 63, 92, 95, 9, 10, 11, 12, 45, 58, 65, 90,
652
+ 97, 122, 13, 32, 39, 47, 60, 61, 62, 63, 92, 95,
653
+ 9, 10, 11, 12, 45, 58, 65, 90, 97, 122, 13, 32,
654
+ 39, 47, 60, 61, 62, 63, 92, 95, 9, 10, 11, 12,
655
+ 45, 58, 65, 90, 97, 122, 13, 32, 39, 47, 60, 62,
656
+ 92, 9, 10, 11, 12, 13, 32, 39, 47, 60, 62, 92,
657
+ 9, 10, 11, 12, 13, 32, 34, 39, 47, 60, 62, 92,
658
+ 9, 10, 11, 12, 13, 32, 34, 39, 47, 60, 62, 63,
659
+ 92, 95, 9, 10, 11, 12, 45, 58, 65, 90, 97, 122,
660
+ 13, 32, 34, 39, 47, 60, 62, 63, 92, 95, 9, 10,
661
+ 11, 12, 45, 58, 65, 90, 97, 122, 13, 32, 34, 39,
662
+ 47, 60, 62, 92, 9, 10, 11, 12, 13, 32, 34, 39,
663
+ 47, 60, 62, 63, 92, 95, 9, 10, 11, 12, 45, 58,
664
+ 65, 90, 97, 122, 13, 32, 34, 39, 47, 60, 61, 62,
665
+ 63, 92, 95, 9, 10, 11, 12, 45, 58, 65, 90, 97,
666
+ 122, 32, 34, 39, 47, 61, 62, 63, 92, 95, 9, 13,
667
+ 45, 58, 65, 90, 97, 122, 32, 34, 39, 47, 61, 62,
668
+ 63, 92, 95, 9, 13, 45, 58, 65, 90, 97, 122, 34,
669
+ 39, 62, 92, 34, 39, 92, 13, 32, 34, 39, 47, 60,
670
+ 62, 92, 9, 10, 11, 12, 13, 32, 34, 39, 47, 60,
671
+ 62, 92, 9, 10, 11, 12, 13, 32, 34, 39, 47, 60,
672
+ 62, 63, 92, 95, 9, 10, 11, 12, 45, 58, 65, 90,
673
+ 97, 122, 13, 32, 34, 39, 47, 60, 62, 63, 92, 95,
674
+ 9, 10, 11, 12, 45, 58, 65, 90, 97, 122, 13, 32,
675
+ 34, 39, 47, 60, 62, 63, 92, 95, 9, 10, 11, 12,
676
+ 45, 58, 65, 90, 97, 122, 13, 32, 34, 39, 47, 60,
677
+ 62, 92, 9, 10, 11, 12, 13, 32, 34, 39, 47, 60,
678
+ 62, 92, 9, 10, 11, 12, 13, 32, 34, 39, 47, 60,
679
+ 62, 63, 92, 95, 9, 10, 11, 12, 45, 58, 65, 90,
680
+ 97, 122, 32, 34, 39, 47, 62, 63, 92, 95, 9, 13,
681
+ 45, 58, 65, 90, 97, 122, 32, 34, 39, 47, 62, 63,
682
+ 92, 95, 9, 13, 45, 58, 65, 90, 97, 122, 13, 32,
683
+ 34, 39, 47, 60, 62, 92, 9, 10, 11, 12, 13, 32,
684
+ 34, 39, 47, 60, 61, 62, 63, 92, 95, 9, 10, 11,
685
+ 12, 45, 58, 65, 90, 97, 122, 13, 32, 39, 47, 60,
686
+ 62, 63, 92, 95, 9, 10, 11, 12, 45, 58, 65, 90,
687
+ 97, 122, 34, 39, 92, 32, 39, 47, 62, 63, 92, 95,
688
+ 9, 13, 45, 58, 65, 90, 97, 122, 13, 32, 34, 39,
689
+ 47, 60, 62, 92, 9, 10, 11, 12, 32, 34, 47, 62,
690
+ 63, 92, 95, 9, 13, 45, 58, 65, 90, 97, 122, 34,
691
+ 39, 92, 13, 32, 39, 47, 60, 62, 92, 9, 10, 11,
692
+ 12, 34, 92, 39, 92, 13, 32, 34, 39, 47, 60, 62,
693
+ 9, 10, 11, 12, 58, 95, 120, 65, 90, 97, 122, 32,
694
+ 63, 95, 9, 13, 45, 46, 48, 58, 65, 90, 97, 122,
695
+ 32, 63, 95, 109, 9, 13, 45, 46, 48, 58, 65, 90,
696
+ 97, 122, 32, 63, 95, 108, 9, 13, 45, 46, 48, 58,
697
+ 65, 90, 97, 122, 32, 63, 95, 9, 13, 45, 46, 48,
698
+ 58, 65, 90, 97, 122, 101, 114, 115, 105, 111, 110, 32,
699
+ 61, 9, 13, 32, 34, 39, 9, 13, 95, 45, 46, 48,
700
+ 58, 65, 90, 97, 122, 34, 95, 45, 46, 48, 58, 65,
701
+ 90, 97, 122, 32, 62, 63, 9, 13, 32, 62, 63, 101,
702
+ 115, 9, 13, 62, 110, 99, 111, 100, 105, 110, 103, 32,
703
+ 61, 9, 13, 32, 34, 39, 9, 13, 65, 90, 97, 122,
704
+ 34, 95, 45, 46, 48, 57, 65, 90, 97, 122, 32, 62,
705
+ 63, 9, 13, 32, 62, 63, 115, 9, 13, 116, 97, 110,
706
+ 100, 97, 108, 111, 110, 101, 32, 61, 9, 13, 32, 34,
707
+ 39, 9, 13, 110, 121, 111, 34, 32, 62, 63, 9, 13,
708
+ 101, 115, 110, 121, 111, 39, 101, 115, 65, 90, 97, 122,
709
+ 39, 95, 45, 46, 48, 57, 65, 90, 97, 122, 95, 45,
710
+ 46, 48, 58, 65, 90, 97, 122, 39, 95, 45, 46, 48,
711
+ 58, 65, 90, 97, 122, 62, 62, 10, 60, 33, 47, 58,
712
+ 63, 95, 65, 90, 97, 122, 39, 93, 34, 34, 92, 39,
713
+ 92, 34, 39, 92, 32, 9, 13, 32, 118, 9, 13, 10,
714
+ 45, 45, 10, 93, 93, 10, 62, 63, 62, 0
715
+ };
716
+ }
717
+
718
+ private static final char _hpricot_scan_trans_keys[] = init__hpricot_scan_trans_keys_0();
719
+
720
+
721
+ private static byte[] init__hpricot_scan_single_lengths_0()
722
+ {
723
+ return new byte [] {
724
+ 3, 1, 1, 1, 1, 1, 1, 1, 1, 3, 5, 5,
725
+ 1, 1, 1, 1, 1, 1, 3, 4, 4, 3, 5, 1,
726
+ 1, 3, 1, 2, 1, 1, 4, 4, 7, 7, 7, 7,
727
+ 5, 2, 2, 4, 2, 3, 3, 4, 2, 3, 4, 4,
728
+ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2,
729
+ 4, 2, 5, 5, 6, 6, 1, 7, 5, 5, 7, 8,
730
+ 8, 5, 7, 9, 9, 7, 7, 7, 2, 2, 8, 8,
731
+ 3, 8, 8, 10, 10, 9, 9, 10, 10, 7, 7, 8,
732
+ 8, 8, 8, 3, 7, 2, 2, 8, 8, 3, 8, 7,
733
+ 9, 10, 10, 7, 7, 8, 10, 10, 8, 10, 11, 9,
734
+ 9, 4, 3, 8, 8, 10, 10, 10, 8, 8, 10, 8,
735
+ 8, 8, 11, 9, 3, 7, 8, 7, 3, 7, 2, 2,
736
+ 7, 3, 3, 4, 4, 3, 1, 1, 1, 1, 1, 1,
737
+ 2, 3, 1, 2, 3, 5, 1, 1, 1, 1, 1, 1,
738
+ 1, 1, 2, 3, 0, 2, 3, 4, 1, 1, 1, 1,
739
+ 1, 1, 1, 1, 1, 2, 3, 2, 1, 1, 3, 1,
740
+ 1, 2, 1, 1, 1, 1, 0, 2, 1, 2, 1, 1,
741
+ 2, 5, 1, 1, 1, 2, 2, 3, 1, 2, 2, 1,
742
+ 2, 1, 3, 1
743
+ };
744
+ }
745
+
746
+ private static final byte _hpricot_scan_single_lengths[] = init__hpricot_scan_single_lengths_0();
747
+
748
+
749
+ private static byte[] init__hpricot_scan_range_lengths_0()
750
+ {
751
+ return new byte [] {
752
+ 0, 0, 0, 0, 0, 0, 0, 0, 1, 3, 5, 1,
753
+ 0, 0, 0, 0, 0, 1, 1, 4, 4, 1, 1, 0,
754
+ 0, 1, 0, 1, 0, 0, 5, 5, 5, 5, 5, 5,
755
+ 1, 0, 0, 1, 0, 1, 1, 1, 0, 1, 1, 5,
756
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2,
757
+ 5, 1, 4, 4, 4, 4, 0, 2, 2, 4, 5, 5,
758
+ 5, 2, 2, 5, 5, 2, 2, 4, 0, 0, 4, 4,
759
+ 0, 2, 2, 5, 5, 5, 5, 5, 5, 2, 2, 2,
760
+ 2, 2, 4, 0, 4, 0, 0, 4, 4, 0, 2, 2,
761
+ 5, 5, 5, 2, 2, 2, 5, 5, 2, 5, 5, 4,
762
+ 4, 0, 0, 2, 2, 5, 5, 5, 2, 2, 5, 4,
763
+ 4, 2, 5, 5, 0, 4, 2, 4, 0, 2, 0, 0,
764
+ 2, 2, 5, 5, 5, 5, 0, 0, 0, 0, 0, 0,
765
+ 1, 1, 4, 4, 1, 1, 0, 0, 0, 0, 0, 0,
766
+ 0, 0, 1, 1, 2, 4, 1, 1, 0, 0, 0, 0,
767
+ 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0,
768
+ 0, 0, 0, 0, 0, 0, 2, 4, 4, 4, 0, 0,
769
+ 0, 2, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0,
770
+ 0, 0, 0, 0
771
+ };
772
+ }
773
+
774
+ private static final byte _hpricot_scan_range_lengths[] = init__hpricot_scan_range_lengths_0();
775
+
776
+
777
+ private static short[] init__hpricot_scan_index_offsets_0()
778
+ {
779
+ return new short [] {
780
+ 0, 4, 6, 8, 10, 12, 14, 16, 18, 21, 28, 39,
781
+ 46, 48, 50, 52, 54, 56, 59, 64, 73, 82, 87, 94,
782
+ 96, 98, 103, 105, 109, 111, 113, 123, 133, 146, 159, 172,
783
+ 185, 192, 195, 198, 204, 207, 212, 217, 223, 226, 231, 237,
784
+ 247, 249, 251, 253, 255, 257, 259, 261, 263, 265, 267, 269,
785
+ 274, 284, 288, 298, 308, 319, 330, 332, 342, 350, 360, 373,
786
+ 387, 401, 409, 419, 434, 449, 459, 469, 481, 484, 487, 500,
787
+ 513, 517, 528, 539, 555, 571, 586, 601, 617, 633, 643, 653,
788
+ 664, 675, 686, 699, 703, 715, 718, 721, 734, 747, 751, 762,
789
+ 772, 787, 803, 819, 829, 839, 850, 866, 882, 893, 909, 926,
790
+ 940, 954, 959, 963, 974, 985, 1001, 1017, 1033, 1044, 1055, 1071,
791
+ 1084, 1097, 1108, 1125, 1140, 1144, 1156, 1167, 1179, 1183, 1193, 1196,
792
+ 1199, 1209, 1215, 1224, 1234, 1244, 1253, 1255, 1257, 1259, 1261, 1263,
793
+ 1265, 1269, 1274, 1280, 1287, 1292, 1299, 1301, 1303, 1305, 1307, 1309,
794
+ 1311, 1313, 1315, 1319, 1324, 1327, 1334, 1339, 1345, 1347, 1349, 1351,
795
+ 1353, 1355, 1357, 1359, 1361, 1363, 1367, 1372, 1375, 1377, 1379, 1384,
796
+ 1386, 1388, 1391, 1393, 1395, 1397, 1399, 1402, 1409, 1415, 1422, 1424,
797
+ 1426, 1429, 1437, 1439, 1441, 1443, 1446, 1449, 1453, 1456, 1460, 1463,
798
+ 1465, 1468, 1470, 1474
799
+ };
800
+ }
801
+
802
+ private static final short _hpricot_scan_index_offsets[] = init__hpricot_scan_index_offsets_0();
803
+
804
+
805
+ private static short[] init__hpricot_scan_indicies_0()
806
+ {
807
+ return new short [] {
808
+ 1, 2, 3, 0, 4, 0, 5, 0, 6, 0, 7, 0,
809
+ 8, 0, 9, 0, 10, 0, 11, 11, 0, 11, 12, 12,
810
+ 11, 12, 12, 0, 13, 15, 14, 16, 14, 13, 14, 14,
811
+ 14, 14, 0, 17, 18, 19, 20, 21, 17, 0, 22, 0,
812
+ 23, 0, 24, 0, 25, 0, 26, 0, 27, 27, 0, 27,
813
+ 28, 29, 27, 0, 30, 31, 30, 30, 30, 30, 30, 30,
814
+ 0, 32, 33, 32, 32, 32, 32, 32, 32, 0, 34, 18,
815
+ 21, 34, 0, 34, 35, 36, 18, 21, 34, 0, 38, 37,
816
+ 41, 40, 42, 18, 21, 42, 39, 43, 21, 43, 18, 43,
817
+ 39, 38, 44, 41, 45, 46, 47, 46, 46, 46, 46, 46,
818
+ 46, 46, 0, 48, 49, 48, 48, 48, 48, 48, 48, 48,
819
+ 0, 50, 50, 48, 49, 18, 21, 48, 34, 48, 48, 48,
820
+ 48, 0, 50, 50, 35, 51, 18, 21, 48, 34, 48, 48,
821
+ 48, 48, 0, 52, 52, 54, 55, 56, 57, 54, 53, 54,
822
+ 54, 54, 54, 44, 58, 58, 61, 62, 63, 64, 60, 59,
823
+ 60, 60, 60, 60, 45, 59, 61, 65, 63, 64, 59, 45,
824
+ 67, 68, 66, 70, 71, 69, 72, 41, 63, 64, 72, 45,
825
+ 73, 74, 64, 75, 76, 43, 75, 21, 74, 41, 63, 74,
826
+ 45, 77, 41, 78, 79, 77, 40, 73, 80, 79, 80, 41,
827
+ 78, 80, 40, 81, 38, 56, 57, 81, 44, 60, 82, 60,
828
+ 60, 60, 60, 60, 60, 60, 45, 83, 0, 84, 0, 85,
829
+ 0, 86, 0, 87, 0, 88, 0, 89, 0, 90, 0, 91,
830
+ 0, 92, 0, 93, 0, 94, 94, 94, 94, 0, 95, 97,
831
+ 96, 96, 95, 96, 96, 96, 96, 0, 98, 99, 98, 0,
832
+ 100, 102, 103, 101, 101, 100, 101, 101, 101, 0, 104, 106,
833
+ 107, 105, 105, 104, 105, 105, 105, 0, 108, 110, 111, 112,
834
+ 109, 109, 108, 109, 109, 109, 39, 113, 115, 116, 117, 114,
835
+ 114, 113, 114, 114, 114, 39, 118, 39, 120, 120, 122, 123,
836
+ 124, 39, 117, 120, 121, 119, 126, 126, 128, 39, 129, 126,
837
+ 127, 125, 130, 115, 117, 114, 114, 130, 114, 114, 114, 39,
838
+ 126, 126, 132, 39, 133, 131, 131, 126, 127, 131, 131, 131,
839
+ 125, 134, 134, 137, 39, 138, 139, 136, 136, 134, 135, 136,
840
+ 136, 136, 125, 140, 140, 132, 39, 142, 133, 131, 131, 140,
841
+ 141, 131, 131, 131, 125, 126, 126, 128, 39, 129, 126, 127,
842
+ 125, 143, 143, 145, 146, 147, 39, 129, 143, 144, 119, 148,
843
+ 148, 122, 123, 124, 39, 117, 150, 150, 148, 149, 150, 150,
844
+ 150, 119, 143, 143, 145, 146, 151, 39, 133, 150, 150, 143,
845
+ 144, 150, 150, 150, 119, 153, 153, 155, 156, 157, 158, 159,
846
+ 153, 154, 152, 161, 161, 163, 164, 165, 166, 167, 161, 162,
847
+ 160, 168, 169, 171, 172, 170, 173, 170, 168, 170, 170, 170,
848
+ 165, 169, 173, 165, 174, 173, 165, 175, 169, 177, 178, 179,
849
+ 176, 173, 176, 175, 176, 176, 176, 165, 180, 169, 171, 181,
850
+ 172, 170, 173, 170, 180, 170, 170, 170, 165, 169, 182, 173,
851
+ 165, 183, 183, 185, 186, 187, 165, 172, 159, 183, 184, 152,
852
+ 188, 188, 185, 186, 187, 165, 172, 159, 188, 189, 152, 188,
853
+ 188, 185, 186, 187, 165, 172, 190, 159, 190, 188, 189, 190,
854
+ 190, 190, 152, 191, 191, 193, 194, 195, 165, 196, 190, 159,
855
+ 190, 191, 192, 190, 190, 190, 152, 153, 153, 155, 195, 157,
856
+ 197, 190, 159, 190, 153, 154, 190, 190, 190, 152, 161, 161,
857
+ 163, 199, 165, 196, 198, 167, 198, 161, 162, 198, 198, 198,
858
+ 160, 200, 200, 163, 203, 165, 204, 205, 202, 167, 202, 200,
859
+ 201, 202, 202, 202, 160, 206, 206, 163, 199, 165, 208, 196,
860
+ 198, 167, 198, 206, 207, 198, 198, 198, 160, 161, 161, 163,
861
+ 164, 165, 166, 167, 161, 162, 160, 161, 161, 209, 164, 165,
862
+ 166, 167, 161, 162, 160, 191, 191, 193, 194, 156, 165, 166,
863
+ 159, 191, 192, 152, 211, 211, 213, 214, 215, 216, 217, 218,
864
+ 211, 212, 210, 220, 220, 222, 209, 223, 224, 225, 226, 220,
865
+ 221, 219, 227, 228, 174, 230, 231, 229, 232, 229, 227, 229,
866
+ 229, 229, 224, 228, 174, 232, 224, 234, 169, 236, 237, 235,
867
+ 238, 235, 234, 235, 235, 235, 233, 169, 238, 233, 228, 238,
868
+ 233, 239, 169, 241, 242, 243, 240, 238, 240, 239, 240, 240,
869
+ 240, 233, 244, 169, 236, 245, 237, 235, 238, 235, 244, 235,
870
+ 235, 235, 233, 169, 246, 238, 233, 248, 248, 250, 251, 252,
871
+ 233, 237, 253, 248, 249, 247, 255, 255, 163, 257, 233, 258,
872
+ 259, 255, 256, 254, 255, 255, 163, 261, 233, 262, 260, 259,
873
+ 260, 255, 256, 260, 260, 260, 254, 263, 263, 163, 266, 233,
874
+ 267, 268, 265, 259, 265, 263, 264, 265, 265, 265, 254, 269,
875
+ 269, 163, 261, 233, 271, 262, 260, 259, 260, 269, 270, 260,
876
+ 260, 260, 254, 255, 255, 163, 257, 233, 258, 259, 255, 256,
877
+ 254, 255, 255, 222, 257, 233, 258, 259, 255, 256, 254, 272,
878
+ 272, 274, 275, 276, 233, 258, 253, 272, 273, 247, 277, 277,
879
+ 250, 251, 252, 233, 237, 279, 253, 279, 277, 278, 279, 279,
880
+ 279, 247, 272, 272, 274, 275, 280, 233, 262, 279, 253, 279,
881
+ 272, 273, 279, 279, 279, 247, 211, 211, 281, 214, 215, 216,
882
+ 217, 218, 211, 212, 210, 220, 220, 222, 209, 283, 224, 284,
883
+ 282, 226, 282, 220, 221, 282, 282, 282, 219, 285, 285, 222,
884
+ 209, 288, 224, 289, 290, 287, 226, 287, 285, 286, 287, 287,
885
+ 287, 219, 291, 228, 174, 230, 292, 231, 229, 232, 229, 291,
886
+ 229, 229, 229, 224, 293, 228, 174, 295, 296, 297, 294, 232,
887
+ 294, 293, 294, 294, 294, 224, 228, 174, 298, 232, 224, 299,
888
+ 299, 232, 224, 300, 300, 302, 303, 304, 224, 231, 218, 300,
889
+ 301, 210, 305, 305, 302, 303, 304, 224, 231, 218, 305, 306,
890
+ 210, 305, 305, 302, 303, 304, 224, 231, 307, 218, 307, 305,
891
+ 306, 307, 307, 307, 210, 308, 308, 310, 311, 312, 224, 284,
892
+ 307, 218, 307, 308, 309, 307, 307, 307, 210, 211, 211, 281,
893
+ 214, 312, 216, 313, 307, 218, 307, 211, 212, 307, 307, 307,
894
+ 210, 220, 220, 222, 209, 223, 224, 225, 226, 220, 221, 219,
895
+ 220, 220, 314, 314, 223, 224, 225, 226, 220, 221, 219, 211,
896
+ 211, 213, 214, 312, 216, 313, 307, 218, 307, 211, 212, 307,
897
+ 307, 307, 210, 315, 316, 317, 319, 320, 318, 321, 318, 315,
898
+ 318, 318, 318, 216, 315, 322, 317, 319, 320, 318, 321, 318,
899
+ 315, 318, 318, 318, 216, 308, 308, 310, 311, 215, 224, 225,
900
+ 218, 308, 309, 210, 323, 323, 222, 209, 283, 224, 325, 284,
901
+ 282, 226, 282, 323, 324, 282, 282, 282, 219, 326, 326, 155,
902
+ 280, 328, 329, 279, 253, 279, 326, 327, 279, 279, 279, 247,
903
+ 316, 317, 321, 216, 330, 331, 333, 334, 332, 335, 332, 330,
904
+ 332, 332, 332, 328, 277, 277, 250, 251, 252, 233, 237, 253,
905
+ 277, 278, 247, 336, 331, 338, 339, 337, 340, 337, 336, 337,
906
+ 337, 337, 157, 322, 317, 321, 216, 326, 326, 155, 276, 328,
907
+ 341, 253, 326, 327, 247, 331, 340, 157, 331, 335, 328, 148,
908
+ 148, 122, 123, 124, 39, 117, 148, 149, 119, 342, 342, 343,
909
+ 342, 342, 0, 344, 345, 345, 344, 345, 345, 345, 345, 0,
910
+ 344, 345, 345, 346, 344, 345, 345, 345, 345, 0, 344, 345,
911
+ 345, 347, 344, 345, 345, 345, 345, 0, 348, 345, 345, 348,
912
+ 345, 345, 345, 345, 0, 350, 349, 351, 349, 352, 349, 353,
913
+ 349, 354, 349, 355, 349, 355, 356, 355, 349, 356, 357, 358,
914
+ 356, 349, 359, 359, 359, 359, 359, 349, 360, 361, 361, 361,
915
+ 361, 361, 349, 362, 363, 364, 362, 349, 362, 363, 364, 365,
916
+ 366, 362, 349, 363, 349, 367, 349, 368, 349, 369, 349, 370,
917
+ 349, 371, 349, 372, 349, 373, 349, 373, 374, 373, 349, 374,
918
+ 375, 376, 374, 349, 377, 377, 349, 378, 379, 379, 379, 379,
919
+ 379, 349, 380, 363, 364, 380, 349, 380, 363, 364, 366, 380,
920
+ 349, 381, 349, 382, 349, 383, 349, 384, 349, 385, 349, 386,
921
+ 349, 387, 349, 388, 349, 389, 349, 389, 390, 389, 349, 390,
922
+ 391, 392, 390, 349, 393, 394, 349, 395, 349, 396, 349, 397,
923
+ 363, 364, 397, 349, 398, 349, 395, 349, 399, 400, 349, 401,
924
+ 349, 396, 349, 402, 349, 401, 349, 403, 403, 349, 378, 404,
925
+ 404, 404, 404, 404, 349, 405, 405, 405, 405, 405, 349, 360,
926
+ 406, 406, 406, 406, 406, 349, 408, 407, 410, 409, 412, 413,
927
+ 411, 415, 416, 417, 418, 417, 417, 417, 414, 41, 45, 43,
928
+ 21, 41, 40, 169, 173, 165, 169, 238, 233, 228, 174, 232,
929
+ 224, 344, 344, 420, 348, 421, 348, 420, 423, 424, 422, 426,
930
+ 425, 428, 429, 427, 431, 430, 433, 434, 435, 432, 434, 436,
931
+ 0
932
+ };
933
+ }
934
+
935
+ private static final short _hpricot_scan_indicies[] = init__hpricot_scan_indicies_0();
936
+
937
+
938
+ private static short[] init__hpricot_scan_trans_targs_0()
939
+ {
940
+ return new short [] {
941
+ 204, 1, 2, 53, 204, 3, 4, 5, 6, 7, 8, 9,
942
+ 10, 11, 10, 204, 26, 11, 204, 12, 48, 26, 13, 14,
943
+ 15, 16, 17, 18, 19, 30, 20, 21, 20, 21, 22, 23,
944
+ 28, 24, 25, 204, 24, 25, 25, 27, 29, 29, 31, 32,
945
+ 31, 32, 33, 34, 35, 36, 47, 32, 206, 40, 35, 36,
946
+ 47, 37, 34, 206, 40, 46, 38, 39, 43, 38, 39, 43,
947
+ 39, 41, 42, 41, 207, 43, 208, 44, 45, 39, 32, 49,
948
+ 50, 51, 52, 21, 54, 55, 56, 57, 58, 204, 60, 61,
949
+ 60, 204, 61, 204, 63, 62, 66, 204, 63, 64, 66, 204,
950
+ 65, 64, 66, 67, 204, 65, 64, 66, 67, 204, 204, 68,
951
+ 144, 74, 142, 143, 73, 68, 69, 70, 73, 204, 69, 71,
952
+ 73, 204, 65, 72, 71, 73, 74, 204, 65, 72, 74, 75,
953
+ 76, 77, 141, 73, 75, 76, 71, 73, 78, 79, 90, 70,
954
+ 93, 80, 209, 94, 78, 79, 90, 70, 93, 80, 209, 94,
955
+ 79, 69, 82, 84, 209, 81, 79, 83, 82, 84, 85, 209,
956
+ 83, 85, 209, 86, 95, 139, 140, 93, 87, 88, 91, 87,
957
+ 88, 89, 96, 93, 209, 209, 91, 93, 83, 92, 91, 93,
958
+ 95, 209, 83, 92, 95, 90, 97, 98, 117, 108, 90, 128,
959
+ 99, 211, 129, 97, 98, 117, 108, 128, 99, 211, 129, 98,
960
+ 100, 120, 121, 211, 122, 101, 100, 103, 105, 210, 102, 104,
961
+ 103, 105, 106, 210, 104, 106, 210, 107, 138, 113, 136, 137,
962
+ 111, 112, 107, 100, 108, 111, 210, 112, 109, 111, 210, 104,
963
+ 110, 109, 111, 113, 210, 104, 110, 113, 114, 115, 116, 135,
964
+ 111, 114, 115, 109, 111, 108, 118, 128, 211, 119, 134, 118,
965
+ 128, 133, 211, 119, 123, 119, 120, 121, 123, 211, 211, 98,
966
+ 124, 133, 131, 132, 128, 125, 126, 118, 125, 126, 127, 130,
967
+ 128, 211, 117, 98, 100, 79, 120, 121, 211, 122, 100, 119,
968
+ 134, 133, 100, 108, 101, 210, 100, 69, 103, 105, 210, 102,
969
+ 79, 82, 84, 209, 81, 210, 146, 147, 212, 146, 148, 149,
970
+ 213, 204, 151, 152, 153, 154, 155, 156, 157, 158, 200, 159,
971
+ 160, 159, 161, 204, 162, 163, 176, 164, 165, 166, 167, 168,
972
+ 169, 170, 171, 172, 198, 173, 174, 173, 175, 177, 178, 179,
973
+ 180, 181, 182, 183, 184, 185, 186, 187, 193, 188, 191, 189,
974
+ 190, 190, 192, 194, 196, 195, 197, 199, 199, 201, 201, 214,
975
+ 214, 216, 216, 204, 204, 205, 204, 0, 59, 62, 145, 204,
976
+ 204, 150, 214, 214, 215, 214, 202, 216, 216, 217, 216, 203,
977
+ 218, 218, 218, 219, 218
978
+ };
979
+ }
980
+
981
+ private static final short _hpricot_scan_trans_targs[] = init__hpricot_scan_trans_targs_0();
982
+
983
+
984
+ private static short[] init__hpricot_scan_trans_actions_0()
985
+ {
986
+ return new short [] {
987
+ 73, 0, 0, 0, 59, 0, 0, 0, 0, 0, 0, 0,
988
+ 1, 5, 0, 92, 5, 0, 51, 0, 0, 0, 0, 0,
989
+ 0, 0, 0, 0, 0, 0, 3, 83, 0, 19, 0, 0,
990
+ 0, 3, 86, 75, 0, 21, 0, 0, 3, 0, 3, 83,
991
+ 0, 19, 0, 19, 3, 3, 3, 172, 188, 3, 0, 0,
992
+ 0, 0, 113, 146, 0, 21, 3, 86, 86, 0, 21, 21,
993
+ 0, 21, 0, 0, 146, 0, 146, 0, 0, 3, 113, 0,
994
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 61, 1, 5,
995
+ 0, 98, 0, 55, 5, 0, 5, 95, 0, 116, 0, 53,
996
+ 11, 0, 110, 11, 168, 0, 180, 23, 0, 122, 57, 3,
997
+ 3, 3, 0, 0, 89, 0, 9, 9, 104, 164, 0, 180,
998
+ 119, 176, 107, 107, 0, 160, 11, 201, 9, 9, 0, 80,
999
+ 80, 0, 0, 152, 3, 3, 196, 156, 3, 80, 80, 77,
1000
+ 152, 3, 226, 3, 0, 9, 9, 7, 104, 0, 211, 0,
1001
+ 0, 7, 180, 23, 192, 0, 7, 11, 0, 110, 11, 216,
1002
+ 0, 0, 149, 3, 3, 7, 0, 89, 3, 3, 196, 80,
1003
+ 80, 7, 0, 156, 221, 232, 180, 119, 107, 107, 0, 160,
1004
+ 11, 238, 9, 9, 0, 7, 3, 80, 80, 101, 77, 152,
1005
+ 3, 226, 3, 0, 9, 9, 7, 104, 0, 211, 0, 0,
1006
+ 7, 180, 23, 192, 0, 0, 0, 180, 23, 192, 0, 11,
1007
+ 0, 110, 11, 216, 0, 0, 149, 3, 3, 3, 0, 7,
1008
+ 89, 3, 0, 9, 9, 104, 211, 0, 180, 119, 221, 107,
1009
+ 107, 0, 160, 11, 238, 9, 9, 0, 80, 80, 0, 7,
1010
+ 152, 3, 3, 196, 156, 77, 180, 119, 221, 107, 107, 0,
1011
+ 160, 11, 238, 0, 0, 11, 0, 110, 11, 216, 149, 7,
1012
+ 3, 3, 7, 7, 89, 3, 3, 196, 80, 80, 7, 7,
1013
+ 156, 232, 7, 3, 77, 77, 196, 89, 206, 3, 101, 9,
1014
+ 9, 0, 80, 80, 3, 232, 3, 77, 196, 89, 206, 3,
1015
+ 3, 196, 89, 206, 3, 226, 25, 25, 0, 0, 0, 0,
1016
+ 31, 71, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3,
1017
+ 13, 0, 0, 49, 0, 0, 0, 0, 0, 0, 0, 0,
1018
+ 0, 0, 0, 0, 0, 3, 15, 0, 0, 0, 0, 0,
1019
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 3, 3, 0,
1020
+ 17, 0, 0, 3, 3, 0, 0, 3, 0, 3, 0, 37,
1021
+ 137, 43, 140, 63, 134, 184, 69, 0, 0, 1, 0, 65,
1022
+ 67, 0, 33, 125, 31, 35, 0, 39, 128, 31, 41, 0,
1023
+ 45, 131, 143, 0, 47
1024
+ };
1025
+ }
1026
+
1027
+ private static final short _hpricot_scan_trans_actions[] = init__hpricot_scan_trans_actions_0();
1028
+
1029
+
1030
+ private static short[] init__hpricot_scan_to_state_actions_0()
1031
+ {
1032
+ return new short [] {
1033
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1034
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1035
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1036
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1037
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1038
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1039
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1040
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1041
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1042
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1043
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1044
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1045
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1046
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1047
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1048
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1049
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1050
+ 27, 0, 0, 0, 0, 0, 0, 0, 0, 0, 27, 0,
1051
+ 27, 0, 27, 0
1052
+ };
1053
+ }
1054
+
1055
+ private static final short _hpricot_scan_to_state_actions[] = init__hpricot_scan_to_state_actions_0();
1056
+
1057
+
1058
+ private static short[] init__hpricot_scan_from_state_actions_0()
1059
+ {
1060
+ return new short [] {
1061
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1062
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1063
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1064
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1065
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1066
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1067
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1068
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1069
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1070
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1071
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1072
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1073
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1074
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1075
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1076
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1077
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1078
+ 29, 0, 0, 0, 0, 0, 0, 0, 0, 0, 29, 0,
1079
+ 29, 0, 29, 0
1080
+ };
1081
+ }
1082
+
1083
+ private static final short _hpricot_scan_from_state_actions[] = init__hpricot_scan_from_state_actions_0();
1084
+
1085
+
1086
+ private static short[] init__hpricot_scan_eof_trans_0()
1087
+ {
1088
+ return new short [] {
1089
+ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1090
+ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1091
+ 40, 40, 40, 40, 1, 40, 1, 1, 1, 1, 1, 1,
1092
+ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1093
+ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1094
+ 1, 1, 1, 1, 40, 40, 40, 40, 40, 40, 40, 40,
1095
+ 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40,
1096
+ 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40,
1097
+ 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40,
1098
+ 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40,
1099
+ 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40,
1100
+ 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40,
1101
+ 40, 1, 1, 1, 1, 1, 350, 350, 350, 350, 350, 350,
1102
+ 350, 350, 350, 350, 350, 350, 350, 350, 350, 350, 350, 350,
1103
+ 350, 350, 350, 350, 350, 350, 350, 350, 350, 350, 350, 350,
1104
+ 350, 350, 350, 350, 350, 350, 350, 350, 350, 350, 350, 350,
1105
+ 350, 350, 350, 350, 350, 350, 350, 350, 350, 350, 408, 410,
1106
+ 0, 415, 420, 420, 420, 40, 40, 40, 421, 421, 0, 426,
1107
+ 0, 431, 0, 437
1108
+ };
1109
+ }
1110
+
1111
+ private static final short _hpricot_scan_eof_trans[] = init__hpricot_scan_eof_trans_0();
1112
+
1113
+
1114
+ static final int hpricot_scan_start = 204;
1115
+ static final int hpricot_scan_error = -1;
1116
+
1117
+ static final int hpricot_scan_en_html_comment = 214;
1118
+ static final int hpricot_scan_en_html_cdata = 216;
1119
+ static final int hpricot_scan_en_html_procins = 218;
1120
+ static final int hpricot_scan_en_main = 204;
1121
+
1122
+ // line 564 "ext/hpricot_scan/hpricot_scan.java.rl"
1123
+
1124
+ public final static int BUFSIZE = 16384;
1125
+
1126
+
1127
+ private int cs, act, have = 0, nread = 0, curline = 1;
1128
+ private int ts = 0, te = 0, eof = -1, p = -1, pe = -1, buf = 0;
1129
+ private byte[] data;
1130
+ private State S = null;
1131
+ private IRubyObject port, opts, attr, tag, akey, aval, bufsize;
1132
+ private int mark_tag = -1, mark_akey = -1, mark_aval = -1;
1133
+ private boolean done = false, ele_open = false, taint = false, io = false, text = false;
1134
+ private int buffer_size = 0;
1135
+
1136
+ private Extra x;
1137
+
1138
+ private IRubyObject self;
1139
+ private Ruby runtime;
1140
+ private ThreadContext ctx;
1141
+ private Block block;
1142
+
1143
+ private IRubyObject xmldecl, doctype, stag, etag, emptytag, comment, cdata, procins;
1144
+
1145
+ private RaiseException newRaiseException(RubyClass exceptionClass, String message) {
1146
+ return new RaiseException(runtime, exceptionClass, message, true);
1147
+ }
1148
+
1149
+ public Scanner(IRubyObject self, IRubyObject[] args, Block block) {
1150
+ this.self = self;
1151
+ this.runtime = self.getRuntime();
1152
+ this.ctx = runtime.getCurrentContext();
1153
+ this.block = block;
1154
+ attr = runtime.getNil();
1155
+ tag = runtime.getNil();
1156
+ akey = runtime.getNil();
1157
+ aval = runtime.getNil();
1158
+ bufsize = runtime.getNil();
1159
+
1160
+ this.x = (Extra)this.runtime.getModule("Hpricot").dataGetStruct();
1161
+
1162
+ this.xmldecl = x.sym_xmldecl;
1163
+ this.doctype = x.sym_doctype;
1164
+ this.stag = x.sym_stag;
1165
+ this.etag = x.sym_etag;
1166
+ this.emptytag = x.sym_emptytag;
1167
+ this.comment = x.sym_comment;
1168
+ this.cdata = x.sym_cdata;
1169
+ this.procins = x.sym_procins;
1170
+
1171
+ port = args[0];
1172
+ if(args.length == 2) {
1173
+ opts = args[1];
1174
+ } else {
1175
+ opts = runtime.getNil();
1176
+ }
1177
+
1178
+ taint = port.isTaint();
1179
+ io = port.respondsTo("read");
1180
+ if(!io) {
1181
+ if(port.respondsTo("to_str")) {
1182
+ port = port.callMethod(ctx, "to_str");
1183
+ port = port.convertToString();
1184
+ } else {
1185
+ throw runtime.newArgumentError("an Hpricot document must be built from an input source (a String or IO object.)");
1186
+ }
1187
+ }
1188
+
1189
+ if(!(opts instanceof RubyHash)) {
1190
+ opts = runtime.getNil();
1191
+ }
1192
+
1193
+ if(!block.isGiven()) {
1194
+ S = new State();
1195
+ S.doc = x.cDoc.allocate();
1196
+ S.focus = S.doc;
1197
+ S.last = runtime.getNil();
1198
+ S.xml = OPT(opts, "xml");
1199
+ S.strict = OPT(opts, "xhtml_strict");
1200
+ S.fixup = OPT(opts, "fixup_tags");
1201
+ if(S.strict) {
1202
+ S.fixup = true;
1203
+ }
1204
+ S.doc.getInstanceVariables().fastSetInstanceVariable("@options", opts);
1205
+ S.EC = x.mHpricot.getConstant("ElementContent");
1206
+ }
1207
+
1208
+ buffer_size = BUFSIZE;
1209
+ if(self.getInstanceVariables().fastHasInstanceVariable("@buffer_size")) {
1210
+ bufsize = self.getInstanceVariables().fastGetInstanceVariable("@buffer_size");
1211
+ if(!bufsize.isNil()) {
1212
+ buffer_size = RubyNumeric.fix2int(bufsize);
1213
+ }
1214
+ }
1215
+
1216
+ if(io) {
1217
+ buf = 0;
1218
+ data = new byte[buffer_size];
1219
+ }
1220
+ }
1221
+
1222
+ private int len, space;
1223
+ // hpricot_scan
1224
+ public IRubyObject scan() {
1225
+
1226
+ // line 1227 "ext/hpricot_scan/HpricotScanService.java"
1227
+ {
1228
+ cs = hpricot_scan_start;
1229
+ ts = -1;
1230
+ te = -1;
1231
+ act = 0;
1232
+ }
1233
+ // line 667 "ext/hpricot_scan/hpricot_scan.java.rl"
1234
+ while(!done) {
1235
+ p = pe = len = buf;
1236
+ space = buffer_size - have;
1237
+
1238
+ if(io) {
1239
+ if(space == 0) {
1240
+ /* We've used up the entire buffer storing an already-parsed token
1241
+ * prefix that must be preserved. Likely caused by super-long attributes.
1242
+ * Increase buffer size and continue */
1243
+ buffer_size += BUFSIZE;
1244
+ data = realloc(data, buffer_size);
1245
+ space = buffer_size - have;
1246
+ }
1247
+
1248
+ p = have;
1249
+ IRubyObject str = port.callMethod(ctx, "read", runtime.newFixnum(space));
1250
+ ByteList bl = str.convertToString().getByteList();
1251
+ len = bl.realSize;
1252
+ System.arraycopy(bl.bytes, bl.begin, data, p, len);
1253
+ } else {
1254
+ ByteList bl = port.convertToString().getByteList();
1255
+ data = bl.bytes;
1256
+ buf = bl.begin;
1257
+ p = bl.begin;
1258
+ len = bl.realSize + 1;
1259
+ if(p + len >= data.length) {
1260
+ data = new byte[len];
1261
+ System.arraycopy(bl.bytes, bl.begin, data, 0, bl.realSize);
1262
+ p = 0;
1263
+ buf = 0;
1264
+ }
1265
+ done = true;
1266
+ eof = p + len;
1267
+ }
1268
+
1269
+ nread += len;
1270
+
1271
+ /* If this is the last buffer, tack on an EOF. */
1272
+ if(io && len < space) {
1273
+ data[p + len++] = 0;
1274
+ eof = p + len;
1275
+ done = true;
1276
+ }
1277
+
1278
+ pe = p + len;
1279
+
1280
+
1281
+ // line 1282 "ext/hpricot_scan/HpricotScanService.java"
1282
+ {
1283
+ int _klen;
1284
+ int _trans = 0;
1285
+ int _acts;
1286
+ int _nacts;
1287
+ int _keys;
1288
+ int _goto_targ = 0;
1289
+
1290
+ _goto: while (true) {
1291
+ switch ( _goto_targ ) {
1292
+ case 0:
1293
+ if ( p == pe ) {
1294
+ _goto_targ = 4;
1295
+ continue _goto;
1296
+ }
1297
+ case 1:
1298
+ _acts = _hpricot_scan_from_state_actions[cs];
1299
+ _nacts = (int) _hpricot_scan_actions[_acts++];
1300
+ while ( _nacts-- > 0 ) {
1301
+ switch ( _hpricot_scan_actions[_acts++] ) {
1302
+ case 21:
1303
+ // line 1 "ext/hpricot_scan/hpricot_scan.java.rl"
1304
+ {ts = p;}
1305
+ break;
1306
+ // line 1307 "ext/hpricot_scan/HpricotScanService.java"
1307
+ }
1308
+ }
1309
+
1310
+ _match: do {
1311
+ _keys = _hpricot_scan_key_offsets[cs];
1312
+ _trans = _hpricot_scan_index_offsets[cs];
1313
+ _klen = _hpricot_scan_single_lengths[cs];
1314
+ if ( _klen > 0 ) {
1315
+ int _lower = _keys;
1316
+ int _mid;
1317
+ int _upper = _keys + _klen - 1;
1318
+ while (true) {
1319
+ if ( _upper < _lower )
1320
+ break;
1321
+
1322
+ _mid = _lower + ((_upper-_lower) >> 1);
1323
+ if ( data[p] < _hpricot_scan_trans_keys[_mid] )
1324
+ _upper = _mid - 1;
1325
+ else if ( data[p] > _hpricot_scan_trans_keys[_mid] )
1326
+ _lower = _mid + 1;
1327
+ else {
1328
+ _trans += (_mid - _keys);
1329
+ break _match;
1330
+ }
1331
+ }
1332
+ _keys += _klen;
1333
+ _trans += _klen;
1334
+ }
1335
+
1336
+ _klen = _hpricot_scan_range_lengths[cs];
1337
+ if ( _klen > 0 ) {
1338
+ int _lower = _keys;
1339
+ int _mid;
1340
+ int _upper = _keys + (_klen<<1) - 2;
1341
+ while (true) {
1342
+ if ( _upper < _lower )
1343
+ break;
1344
+
1345
+ _mid = _lower + (((_upper-_lower) >> 1) & ~1);
1346
+ if ( data[p] < _hpricot_scan_trans_keys[_mid] )
1347
+ _upper = _mid - 2;
1348
+ else if ( data[p] > _hpricot_scan_trans_keys[_mid+1] )
1349
+ _lower = _mid + 2;
1350
+ else {
1351
+ _trans += ((_mid - _keys)>>1);
1352
+ break _match;
1353
+ }
1354
+ }
1355
+ _trans += _klen;
1356
+ }
1357
+ } while (false);
1358
+
1359
+ _trans = _hpricot_scan_indicies[_trans];
1360
+ case 3:
1361
+ cs = _hpricot_scan_trans_targs[_trans];
1362
+
1363
+ if ( _hpricot_scan_trans_actions[_trans] != 0 ) {
1364
+ _acts = _hpricot_scan_trans_actions[_trans];
1365
+ _nacts = (int) _hpricot_scan_actions[_acts++];
1366
+ while ( _nacts-- > 0 )
1367
+ {
1368
+ switch ( _hpricot_scan_actions[_acts++] )
1369
+ {
1370
+ case 0:
1371
+ // line 514 "ext/hpricot_scan/hpricot_scan.java.rl"
1372
+ {
1373
+ if(text) {
1374
+ tag = CAT(tag, mark_tag, p);
1375
+ ELE(x.sym_text);
1376
+ text = false;
1377
+ }
1378
+ attr = runtime.getNil();
1379
+ tag = runtime.getNil();
1380
+ mark_tag = -1;
1381
+ ele_open = true;
1382
+ }
1383
+ break;
1384
+ case 1:
1385
+ // line 526 "ext/hpricot_scan/hpricot_scan.java.rl"
1386
+ { mark_tag = p; }
1387
+ break;
1388
+ case 2:
1389
+ // line 527 "ext/hpricot_scan/hpricot_scan.java.rl"
1390
+ { mark_aval = p; }
1391
+ break;
1392
+ case 3:
1393
+ // line 528 "ext/hpricot_scan/hpricot_scan.java.rl"
1394
+ { mark_akey = p; }
1395
+ break;
1396
+ case 4:
1397
+ // line 529 "ext/hpricot_scan/hpricot_scan.java.rl"
1398
+ { tag = SET(mark_tag, p, tag); }
1399
+ break;
1400
+ case 5:
1401
+ // line 531 "ext/hpricot_scan/hpricot_scan.java.rl"
1402
+ { aval = SET(mark_aval, p, aval); }
1403
+ break;
1404
+ case 6:
1405
+ // line 532 "ext/hpricot_scan/hpricot_scan.java.rl"
1406
+ {
1407
+ if(data[p-1] == '"' || data[p-1] == '\'') {
1408
+ aval = SET(mark_aval, p-1, aval);
1409
+ } else {
1410
+ aval = SET(mark_aval, p, aval);
1411
+ }
1412
+ }
1413
+ break;
1414
+ case 7:
1415
+ // line 539 "ext/hpricot_scan/hpricot_scan.java.rl"
1416
+ { akey = SET(mark_akey, p, akey); }
1417
+ break;
1418
+ case 8:
1419
+ // line 540 "ext/hpricot_scan/hpricot_scan.java.rl"
1420
+ { aval = SET(mark_aval, p, aval); ATTR(runtime.newSymbol("version"), aval); }
1421
+ break;
1422
+ case 9:
1423
+ // line 541 "ext/hpricot_scan/hpricot_scan.java.rl"
1424
+ { aval = SET(mark_aval, p, aval); ATTR(runtime.newSymbol("encoding"), aval); }
1425
+ break;
1426
+ case 10:
1427
+ // line 542 "ext/hpricot_scan/hpricot_scan.java.rl"
1428
+ { aval = SET(mark_aval, p, aval); ATTR(runtime.newSymbol("standalone"), aval); }
1429
+ break;
1430
+ case 11:
1431
+ // line 543 "ext/hpricot_scan/hpricot_scan.java.rl"
1432
+ { aval = SET(mark_aval, p, aval); ATTR(runtime.newSymbol("public_id"), aval); }
1433
+ break;
1434
+ case 12:
1435
+ // line 544 "ext/hpricot_scan/hpricot_scan.java.rl"
1436
+ { aval = SET(mark_aval, p, aval); ATTR(runtime.newSymbol("system_id"), aval); }
1437
+ break;
1438
+ case 13:
1439
+ // line 546 "ext/hpricot_scan/hpricot_scan.java.rl"
1440
+ {
1441
+ akey = runtime.getNil();
1442
+ aval = runtime.getNil();
1443
+ mark_akey = -1;
1444
+ mark_aval = -1;
1445
+ }
1446
+ break;
1447
+ case 14:
1448
+ // line 553 "ext/hpricot_scan/hpricot_scan.java.rl"
1449
+ {
1450
+ if(!S.xml) {
1451
+ akey = akey.callMethod(runtime.getCurrentContext(), "downcase");
1452
+ }
1453
+ ATTR(akey, aval);
1454
+ }
1455
+ break;
1456
+ case 15:
1457
+ // line 9 "ext/hpricot_scan/hpricot_scan.java.rl"
1458
+ {curline += 1;}
1459
+ break;
1460
+ case 16:
1461
+ // line 46 "ext/hpricot_scan/hpricot_scan.java.rl"
1462
+ { TEXT_PASS(); }
1463
+ break;
1464
+ case 17:
1465
+ // line 50 "ext/hpricot_scan/hpricot_scan.java.rl"
1466
+ { EBLK(comment, 3); {cs = 204; _goto_targ = 2; if (true) continue _goto;} }
1467
+ break;
1468
+ case 18:
1469
+ // line 55 "ext/hpricot_scan/hpricot_scan.java.rl"
1470
+ { EBLK(cdata, 3); {cs = 204; _goto_targ = 2; if (true) continue _goto;} }
1471
+ break;
1472
+ case 19:
1473
+ // line 60 "ext/hpricot_scan/hpricot_scan.java.rl"
1474
+ { EBLK(procins, 2); {cs = 204; _goto_targ = 2; if (true) continue _goto;} }
1475
+ break;
1476
+ case 22:
1477
+ // line 1 "ext/hpricot_scan/hpricot_scan.java.rl"
1478
+ {te = p+1;}
1479
+ break;
1480
+ case 23:
1481
+ // line 50 "ext/hpricot_scan/hpricot_scan.java.rl"
1482
+ {te = p+1;}
1483
+ break;
1484
+ case 24:
1485
+ // line 51 "ext/hpricot_scan/hpricot_scan.java.rl"
1486
+ {te = p+1;{ TEXT_PASS(); }}
1487
+ break;
1488
+ case 25:
1489
+ // line 51 "ext/hpricot_scan/hpricot_scan.java.rl"
1490
+ {te = p;p--;{ TEXT_PASS(); }}
1491
+ break;
1492
+ case 26:
1493
+ // line 51 "ext/hpricot_scan/hpricot_scan.java.rl"
1494
+ {{p = ((te))-1;}{ TEXT_PASS(); }}
1495
+ break;
1496
+ case 27:
1497
+ // line 55 "ext/hpricot_scan/hpricot_scan.java.rl"
1498
+ {te = p+1;}
1499
+ break;
1500
+ case 28:
1501
+ // line 56 "ext/hpricot_scan/hpricot_scan.java.rl"
1502
+ {te = p+1;{ TEXT_PASS(); }}
1503
+ break;
1504
+ case 29:
1505
+ // line 56 "ext/hpricot_scan/hpricot_scan.java.rl"
1506
+ {te = p;p--;{ TEXT_PASS(); }}
1507
+ break;
1508
+ case 30:
1509
+ // line 56 "ext/hpricot_scan/hpricot_scan.java.rl"
1510
+ {{p = ((te))-1;}{ TEXT_PASS(); }}
1511
+ break;
1512
+ case 31:
1513
+ // line 60 "ext/hpricot_scan/hpricot_scan.java.rl"
1514
+ {te = p+1;}
1515
+ break;
1516
+ case 32:
1517
+ // line 61 "ext/hpricot_scan/hpricot_scan.java.rl"
1518
+ {te = p+1;{ TEXT_PASS(); }}
1519
+ break;
1520
+ case 33:
1521
+ // line 61 "ext/hpricot_scan/hpricot_scan.java.rl"
1522
+ {te = p;p--;{ TEXT_PASS(); }}
1523
+ break;
1524
+ case 34:
1525
+ // line 66 "ext/hpricot_scan/hpricot_scan.java.rl"
1526
+ {act = 8;}
1527
+ break;
1528
+ case 35:
1529
+ // line 68 "ext/hpricot_scan/hpricot_scan.java.rl"
1530
+ {act = 10;}
1531
+ break;
1532
+ case 36:
1533
+ // line 70 "ext/hpricot_scan/hpricot_scan.java.rl"
1534
+ {act = 12;}
1535
+ break;
1536
+ case 37:
1537
+ // line 73 "ext/hpricot_scan/hpricot_scan.java.rl"
1538
+ {act = 15;}
1539
+ break;
1540
+ case 38:
1541
+ // line 65 "ext/hpricot_scan/hpricot_scan.java.rl"
1542
+ {te = p+1;{ ELE(xmldecl); }}
1543
+ break;
1544
+ case 39:
1545
+ // line 66 "ext/hpricot_scan/hpricot_scan.java.rl"
1546
+ {te = p+1;{ ELE(doctype); }}
1547
+ break;
1548
+ case 40:
1549
+ // line 68 "ext/hpricot_scan/hpricot_scan.java.rl"
1550
+ {te = p+1;{ ELE(stag); }}
1551
+ break;
1552
+ case 41:
1553
+ // line 69 "ext/hpricot_scan/hpricot_scan.java.rl"
1554
+ {te = p+1;{ ELE(etag); }}
1555
+ break;
1556
+ case 42:
1557
+ // line 70 "ext/hpricot_scan/hpricot_scan.java.rl"
1558
+ {te = p+1;{ ELE(emptytag); }}
1559
+ break;
1560
+ case 43:
1561
+ // line 71 "ext/hpricot_scan/hpricot_scan.java.rl"
1562
+ {te = p+1;{ {cs = 214; _goto_targ = 2; if (true) continue _goto;} }}
1563
+ break;
1564
+ case 44:
1565
+ // line 72 "ext/hpricot_scan/hpricot_scan.java.rl"
1566
+ {te = p+1;{ {cs = 216; _goto_targ = 2; if (true) continue _goto;} }}
1567
+ break;
1568
+ case 45:
1569
+ // line 73 "ext/hpricot_scan/hpricot_scan.java.rl"
1570
+ {te = p+1;{ TEXT_PASS(); }}
1571
+ break;
1572
+ case 46:
1573
+ // line 66 "ext/hpricot_scan/hpricot_scan.java.rl"
1574
+ {te = p;p--;{ ELE(doctype); }}
1575
+ break;
1576
+ case 47:
1577
+ // line 67 "ext/hpricot_scan/hpricot_scan.java.rl"
1578
+ {te = p;p--;{ {cs = 218; _goto_targ = 2; if (true) continue _goto;} }}
1579
+ break;
1580
+ case 48:
1581
+ // line 73 "ext/hpricot_scan/hpricot_scan.java.rl"
1582
+ {te = p;p--;{ TEXT_PASS(); }}
1583
+ break;
1584
+ case 49:
1585
+ // line 67 "ext/hpricot_scan/hpricot_scan.java.rl"
1586
+ {{p = ((te))-1;}{ {cs = 218; _goto_targ = 2; if (true) continue _goto;} }}
1587
+ break;
1588
+ case 50:
1589
+ // line 73 "ext/hpricot_scan/hpricot_scan.java.rl"
1590
+ {{p = ((te))-1;}{ TEXT_PASS(); }}
1591
+ break;
1592
+ case 51:
1593
+ // line 1 "ext/hpricot_scan/hpricot_scan.java.rl"
1594
+ { switch( act ) {
1595
+ case 8:
1596
+ {{p = ((te))-1;} ELE(doctype); }
1597
+ break;
1598
+ case 10:
1599
+ {{p = ((te))-1;} ELE(stag); }
1600
+ break;
1601
+ case 12:
1602
+ {{p = ((te))-1;} ELE(emptytag); }
1603
+ break;
1604
+ case 15:
1605
+ {{p = ((te))-1;} TEXT_PASS(); }
1606
+ break;
1607
+ }
1608
+ }
1609
+ break;
1610
+ // line 1611 "ext/hpricot_scan/HpricotScanService.java"
1611
+ }
1612
+ }
1613
+ }
1614
+
1615
+ case 2:
1616
+ _acts = _hpricot_scan_to_state_actions[cs];
1617
+ _nacts = (int) _hpricot_scan_actions[_acts++];
1618
+ while ( _nacts-- > 0 ) {
1619
+ switch ( _hpricot_scan_actions[_acts++] ) {
1620
+ case 20:
1621
+ // line 1 "ext/hpricot_scan/hpricot_scan.java.rl"
1622
+ {ts = -1;}
1623
+ break;
1624
+ // line 1625 "ext/hpricot_scan/HpricotScanService.java"
1625
+ }
1626
+ }
1627
+
1628
+ if ( ++p != pe ) {
1629
+ _goto_targ = 1;
1630
+ continue _goto;
1631
+ }
1632
+ case 4:
1633
+ if ( p == eof )
1634
+ {
1635
+ if ( _hpricot_scan_eof_trans[cs] > 0 ) {
1636
+ _trans = _hpricot_scan_eof_trans[cs] - 1;
1637
+ _goto_targ = 3;
1638
+ continue _goto;
1639
+ }
1640
+ }
1641
+
1642
+ case 5:
1643
+ }
1644
+ break; }
1645
+ }
1646
+ // line 714 "ext/hpricot_scan/hpricot_scan.java.rl"
1647
+
1648
+ if(cs == hpricot_scan_error) {
1649
+ if(!tag.isNil()) {
1650
+ throw newRaiseException(x.rb_eHpricotParseError, "parse error on element <" + tag + ">, starting on line " + curline + ".\n" + NO_WAY_SERIOUSLY);
1651
+ } else {
1652
+ throw newRaiseException(x.rb_eHpricotParseError, "parse error on line " + curline + ".\n" + NO_WAY_SERIOUSLY);
1653
+ }
1654
+ }
1655
+
1656
+ if(done && ele_open) {
1657
+ ele_open = false;
1658
+ if(ts > 0) {
1659
+ mark_tag = ts;
1660
+ ts = 0;
1661
+ text = true;
1662
+ }
1663
+ }
1664
+
1665
+ if(ts == -1) {
1666
+ have = 0;
1667
+ if(mark_tag != -1 && text) {
1668
+ if(done) {
1669
+ if(mark_tag < p - 1) {
1670
+ tag = CAT(tag, mark_tag, p-1);
1671
+ ELE(x.sym_text);
1672
+ }
1673
+ } else {
1674
+ tag = CAT(tag, mark_tag, p);
1675
+ }
1676
+ }
1677
+ if(io) {
1678
+ mark_tag = 0;
1679
+ } else {
1680
+ mark_tag = ((RubyString)port).getByteList().begin;
1681
+ }
1682
+ } else if(io) {
1683
+ have = pe - ts;
1684
+ System.arraycopy(data, ts, data, buf, have);
1685
+ mark_tag = SLIDE(mark_tag);
1686
+ mark_akey = SLIDE(mark_akey);
1687
+ mark_aval = SLIDE(mark_aval);
1688
+ te -= ts;
1689
+ ts = 0;
1690
+ }
1691
+ }
1692
+
1693
+ if(S != null) {
1694
+ return S.doc;
1695
+ }
1696
+
1697
+ return runtime.getNil();
1698
+ }
1699
+ }
1700
+
1701
+ public static class HpricotModule {
1702
+ // hpricot_scan
1703
+ @JRubyMethod(module = true, optional = 1, required = 1, frame = true)
1704
+ public static IRubyObject scan(IRubyObject self, IRubyObject[] args, Block block) {
1705
+ return new Scanner(self, args, block).scan();
1706
+ }
1707
+
1708
+ // hpricot_css
1709
+ @JRubyMethod(module = true)
1710
+ public static IRubyObject css(IRubyObject self, IRubyObject mod, IRubyObject str, IRubyObject node) {
1711
+ return new HpricotCss(self, mod, str, node).scan();
1712
+ }
1713
+ }
1714
+
1715
+ public static class CData {
1716
+ @JRubyMethod
1717
+ public static IRubyObject content(IRubyObject self) {
1718
+ return hpricot_ele_get_name(self);
1719
+ }
1720
+
1721
+ @JRubyMethod(name = "content=")
1722
+ public static IRubyObject content_set(IRubyObject self, IRubyObject value) {
1723
+ return hpricot_ele_set_name(self, value);
1724
+ }
1725
+ }
1726
+
1727
+ public static class Comment {
1728
+ @JRubyMethod
1729
+ public static IRubyObject content(IRubyObject self) {
1730
+ return hpricot_ele_get_name(self);
1731
+ }
1732
+
1733
+ @JRubyMethod(name = "content=")
1734
+ public static IRubyObject content_set(IRubyObject self, IRubyObject value) {
1735
+ return hpricot_ele_set_name(self, value);
1736
+ }
1737
+ }
1738
+
1739
+ public static class DocType {
1740
+ @JRubyMethod
1741
+ public static IRubyObject raw_string(IRubyObject self) {
1742
+ return hpricot_ele_get_name(self);
1743
+ }
1744
+
1745
+ @JRubyMethod
1746
+ public static IRubyObject clear_raw(IRubyObject self) {
1747
+ return hpricot_ele_clear_name(self);
1748
+ }
1749
+
1750
+ @JRubyMethod
1751
+ public static IRubyObject target(IRubyObject self) {
1752
+ return hpricot_ele_get_target(self);
1753
+ }
1754
+
1755
+ @JRubyMethod(name = "target=")
1756
+ public static IRubyObject target_set(IRubyObject self, IRubyObject value) {
1757
+ return hpricot_ele_set_target(self, value);
1758
+ }
1759
+
1760
+ @JRubyMethod
1761
+ public static IRubyObject public_id(IRubyObject self) {
1762
+ return hpricot_ele_get_public_id(self);
1763
+ }
1764
+
1765
+ @JRubyMethod(name = "public_id=")
1766
+ public static IRubyObject public_id_set(IRubyObject self, IRubyObject value) {
1767
+ return hpricot_ele_set_public_id(self, value);
1768
+ }
1769
+
1770
+ @JRubyMethod
1771
+ public static IRubyObject system_id(IRubyObject self) {
1772
+ return hpricot_ele_get_system_id(self);
1773
+ }
1774
+
1775
+ @JRubyMethod(name = "system_id=")
1776
+ public static IRubyObject system_id_set(IRubyObject self, IRubyObject value) {
1777
+ return hpricot_ele_set_system_id(self, value);
1778
+ }
1779
+ }
1780
+
1781
+ public static class Elem {
1782
+ @JRubyMethod
1783
+ public static IRubyObject clear_raw(IRubyObject self) {
1784
+ return hpricot_ele_clear_raw(self);
1785
+ }
1786
+ }
1787
+
1788
+ public static class BogusETag {
1789
+ @JRubyMethod
1790
+ public static IRubyObject raw_string(IRubyObject self) {
1791
+ return hpricot_ele_get_attr(self);
1792
+ }
1793
+
1794
+ @JRubyMethod
1795
+ public static IRubyObject clear_raw(IRubyObject self) {
1796
+ return hpricot_ele_clear_attr(self);
1797
+ }
1798
+ }
1799
+
1800
+ public static class Text {
1801
+ @JRubyMethod
1802
+ public static IRubyObject raw_string(IRubyObject self) {
1803
+ return hpricot_ele_get_name(self);
1804
+ }
1805
+
1806
+ @JRubyMethod
1807
+ public static IRubyObject clear_raw(IRubyObject self) {
1808
+ return hpricot_ele_clear_name(self);
1809
+ }
1810
+
1811
+ @JRubyMethod
1812
+ public static IRubyObject content(IRubyObject self) {
1813
+ return hpricot_ele_get_name(self);
1814
+ }
1815
+
1816
+ @JRubyMethod(name = "content=")
1817
+ public static IRubyObject content_set(IRubyObject self, IRubyObject value) {
1818
+ return hpricot_ele_set_name(self, value);
1819
+ }
1820
+ }
1821
+
1822
+ public static class XMLDecl {
1823
+ @JRubyMethod
1824
+ public static IRubyObject raw_string(IRubyObject self) {
1825
+ return hpricot_ele_get_name(self);
1826
+ }
1827
+
1828
+ @JRubyMethod
1829
+ public static IRubyObject clear_raw(IRubyObject self) {
1830
+ return hpricot_ele_clear_name(self);
1831
+ }
1832
+
1833
+ @JRubyMethod
1834
+ public static IRubyObject encoding(IRubyObject self) {
1835
+ return hpricot_ele_get_encoding(self);
1836
+ }
1837
+
1838
+ @JRubyMethod(name = "encoding=")
1839
+ public static IRubyObject encoding_set(IRubyObject self, IRubyObject value) {
1840
+ return hpricot_ele_set_encoding(self, value);
1841
+ }
1842
+
1843
+ @JRubyMethod
1844
+ public static IRubyObject standalone(IRubyObject self) {
1845
+ return hpricot_ele_get_standalone(self);
1846
+ }
1847
+
1848
+ @JRubyMethod(name = "standalone=")
1849
+ public static IRubyObject standalone_set(IRubyObject self, IRubyObject value) {
1850
+ return hpricot_ele_set_standalone(self, value);
1851
+ }
1852
+
1853
+ @JRubyMethod
1854
+ public static IRubyObject version(IRubyObject self) {
1855
+ return hpricot_ele_get_version(self);
1856
+ }
1857
+
1858
+ @JRubyMethod(name = "version=")
1859
+ public static IRubyObject version_set(IRubyObject self, IRubyObject value) {
1860
+ return hpricot_ele_set_version(self, value);
1861
+ }
1862
+ }
1863
+
1864
+ public static class ProcIns {
1865
+ @JRubyMethod
1866
+ public static IRubyObject target(IRubyObject self) {
1867
+ return hpricot_ele_get_name(self);
1868
+ }
1869
+
1870
+ @JRubyMethod(name = "target=")
1871
+ public static IRubyObject target_set(IRubyObject self, IRubyObject value) {
1872
+ return hpricot_ele_set_name(self, value);
1873
+ }
1874
+
1875
+ @JRubyMethod
1876
+ public static IRubyObject content(IRubyObject self) {
1877
+ return hpricot_ele_get_attr(self);
1878
+ }
1879
+
1880
+ @JRubyMethod(name = "content=")
1881
+ public static IRubyObject content_set(IRubyObject self, IRubyObject value) {
1882
+ return hpricot_ele_set_attr(self, value);
1883
+ }
1884
+ }
1885
+
1886
+ public final static String NO_WAY_SERIOUSLY = "*** This should not happen, please send a bug report with the HTML you're parsing to why@whytheluckystiff.net. So sorry!";
1887
+
1888
+ public final static int H_ELE_TAG = 0;
1889
+ public final static int H_ELE_PARENT = 1;
1890
+ public final static int H_ELE_ATTR = 2;
1891
+ public final static int H_ELE_ETAG = 3;
1892
+ public final static int H_ELE_RAW = 4;
1893
+ public final static int H_ELE_EC = 5;
1894
+ public final static int H_ELE_HASH = 6;
1895
+ public final static int H_ELE_CHILDREN = 7;
1896
+
1897
+ public static IRubyObject H_ELE_GET(IRubyObject recv, int n) {
1898
+ return ((IRubyObject[])recv.dataGetStruct())[n];
1899
+ }
1900
+
1901
+ public static IRubyObject H_ELE_SET(IRubyObject recv, int n, IRubyObject value) {
1902
+ ((IRubyObject[])recv.dataGetStruct())[n] = value;
1903
+ return value;
1904
+ }
1905
+
1906
+ private static class RefCallback implements Callback {
1907
+ private final int n;
1908
+ public RefCallback(int n) { this.n = n; }
1909
+
1910
+ public IRubyObject execute(IRubyObject recv, IRubyObject[] args, Block block) {
1911
+ return H_ELE_GET(recv, n);
1912
+ }
1913
+
1914
+ public Arity getArity() {
1915
+ return Arity.NO_ARGUMENTS;
1916
+ }
1917
+ }
1918
+
1919
+ private static class SetCallback implements Callback {
1920
+ private final int n;
1921
+ public SetCallback(int n) { this.n = n; }
1922
+
1923
+ public IRubyObject execute(IRubyObject recv, IRubyObject[] args, Block block) {
1924
+ return H_ELE_SET(recv, n, args[0]);
1925
+ }
1926
+
1927
+ public Arity getArity() {
1928
+ return Arity.ONE_ARGUMENT;
1929
+ }
1930
+ }
1931
+
1932
+ private final static Callback[] ref_func = new Callback[]{
1933
+ new RefCallback(0),
1934
+ new RefCallback(1),
1935
+ new RefCallback(2),
1936
+ new RefCallback(3),
1937
+ new RefCallback(4),
1938
+ new RefCallback(5),
1939
+ new RefCallback(6),
1940
+ new RefCallback(7),
1941
+ new RefCallback(8),
1942
+ new RefCallback(9)};
1943
+
1944
+ private final static Callback[] set_func = new Callback[]{
1945
+ new SetCallback(0),
1946
+ new SetCallback(1),
1947
+ new SetCallback(2),
1948
+ new SetCallback(3),
1949
+ new SetCallback(4),
1950
+ new SetCallback(5),
1951
+ new SetCallback(6),
1952
+ new SetCallback(7),
1953
+ new SetCallback(8),
1954
+ new SetCallback(9)};
1955
+
1956
+ public final static ObjectAllocator alloc_hpricot_struct = new ObjectAllocator() {
1957
+ // alloc_hpricot_struct
1958
+ public IRubyObject allocate(Ruby runtime, RubyClass klass) {
1959
+ RubyClass kurrent = klass;
1960
+ Object sz = kurrent.fastGetInternalVariable("__size__");
1961
+ while(sz == null && kurrent != null) {
1962
+ kurrent = kurrent.getSuperClass();
1963
+ sz = kurrent.fastGetInternalVariable("__size__");
1964
+ }
1965
+ int size = RubyNumeric.fix2int((RubyObject)sz);
1966
+ RubyObject obj = new RubyObject(runtime, klass);
1967
+ IRubyObject[] all = new IRubyObject[size];
1968
+ java.util.Arrays.fill(all, runtime.getNil());
1969
+ obj.dataWrapStruct(all);
1970
+ return obj;
1971
+ }
1972
+ };
1973
+
1974
+ public static RubyClass makeHpricotStruct(Ruby runtime, IRubyObject[] members) {
1975
+ RubyClass klass = RubyClass.newClass(runtime, runtime.getObject());
1976
+ klass.fastSetInternalVariable("__size__", runtime.newFixnum(members.length));
1977
+ klass.setAllocator(alloc_hpricot_struct);
1978
+
1979
+ for(int i = 0; i < members.length; i++) {
1980
+ String id = members[i].toString();
1981
+ klass.defineMethod(id, ref_func[i]);
1982
+ klass.defineMethod(id + "=", set_func[i]);
1983
+ }
1984
+
1985
+ return klass;
1986
+ }
1987
+
1988
+ public boolean basicLoad(final Ruby runtime) throws IOException {
1989
+ Init_hpricot_scan(runtime);
1990
+ return true;
1991
+ }
1992
+
1993
+ public static class Extra {
1994
+ IRubyObject symAllow, symDeny, sym_xmldecl, sym_doctype,
1995
+ sym_procins, sym_stag, sym_etag, sym_emptytag,
1996
+ sym_allowed, sym_children, sym_comment,
1997
+ sym_cdata, sym_name, sym_parent,
1998
+ sym_raw_attributes, sym_raw_string, sym_tagno,
1999
+ sym_text, sym_EMPTY, sym_CDATA;
2000
+
2001
+ public RubyModule mHpricot;
2002
+ public RubyClass structElem;
2003
+ public RubyClass structAttr;
2004
+ public RubyClass structBasic;
2005
+ public RubyClass cDoc;
2006
+ public RubyClass cCData;
2007
+ public RubyClass cComment;
2008
+ public RubyClass cDocType;
2009
+ public RubyClass cElem;
2010
+ public RubyClass cBogusETag;
2011
+ public RubyClass cText;
2012
+ public RubyClass cXMLDecl;
2013
+ public RubyClass cProcIns;
2014
+ public RubyClass rb_eHpricotParseError;
2015
+ public IRubyObject reProcInsParse;
2016
+
2017
+ public Extra(Ruby runtime) {
2018
+ symAllow = runtime.newSymbol("allow");
2019
+ symDeny = runtime.newSymbol("deny");
2020
+ sym_xmldecl = runtime.newSymbol("xmldecl");
2021
+ sym_doctype = runtime.newSymbol("doctype");
2022
+ sym_procins = runtime.newSymbol("procins");
2023
+ sym_stag = runtime.newSymbol("stag");
2024
+ sym_etag = runtime.newSymbol("etag");
2025
+ sym_emptytag = runtime.newSymbol("emptytag");
2026
+ sym_allowed = runtime.newSymbol("allowed");
2027
+ sym_children = runtime.newSymbol("children");
2028
+ sym_comment = runtime.newSymbol("comment");
2029
+ sym_cdata = runtime.newSymbol("cdata");
2030
+ sym_name = runtime.newSymbol("name");
2031
+ sym_parent = runtime.newSymbol("parent");
2032
+ sym_raw_attributes = runtime.newSymbol("raw_attributes");
2033
+ sym_raw_string = runtime.newSymbol("raw_string");
2034
+ sym_tagno = runtime.newSymbol("tagno");
2035
+ sym_text = runtime.newSymbol("text");
2036
+ sym_EMPTY = runtime.newSymbol("EMPTY");
2037
+ sym_CDATA = runtime.newSymbol("CDATA");
2038
+ }
2039
+ }
2040
+
2041
+ public static void Init_hpricot_scan(Ruby runtime) {
2042
+ Extra x = new Extra(runtime);
2043
+
2044
+ x.mHpricot = runtime.defineModule("Hpricot");
2045
+ x.mHpricot.dataWrapStruct(x);
2046
+
2047
+ x.mHpricot.getSingletonClass().attr_accessor(runtime.getCurrentContext(),new IRubyObject[]{runtime.newSymbol("buffer_size")});
2048
+ x.mHpricot.defineAnnotatedMethods(HpricotModule.class);
2049
+
2050
+ x.rb_eHpricotParseError = x.mHpricot.defineClassUnder("ParseError",runtime.getClass("StandardError"),runtime.getClass("StandardError").getAllocator());
2051
+
2052
+ x.structElem = makeHpricotStruct(runtime, new IRubyObject[] {x.sym_name, x.sym_parent, x.sym_raw_attributes, x.sym_etag, x.sym_raw_string, x.sym_allowed, x.sym_tagno, x.sym_children});
2053
+ x.structAttr = makeHpricotStruct(runtime, new IRubyObject[] {x.sym_name, x.sym_parent, x.sym_raw_attributes});
2054
+ x.structBasic= makeHpricotStruct(runtime, new IRubyObject[] {x.sym_name, x.sym_parent});
2055
+
2056
+ x.cDoc = x.mHpricot.defineClassUnder("Doc", x.structElem, x.structElem.getAllocator());
2057
+
2058
+ x.cCData = x.mHpricot.defineClassUnder("CData", x.structBasic, x.structBasic.getAllocator());
2059
+ x.cCData.defineAnnotatedMethods(CData.class);
2060
+
2061
+ x.cComment = x.mHpricot.defineClassUnder("Comment", x.structBasic, x.structBasic.getAllocator());
2062
+ x.cComment.defineAnnotatedMethods(Comment.class);
2063
+
2064
+ x.cDocType = x.mHpricot.defineClassUnder("DocType", x.structAttr, x.structAttr.getAllocator());
2065
+ x.cDocType.defineAnnotatedMethods(DocType.class);
2066
+
2067
+ x.cElem = x.mHpricot.defineClassUnder("Elem", x.structElem, x.structElem.getAllocator());
2068
+ x.cElem.defineAnnotatedMethods(Elem.class);
2069
+
2070
+ x.cBogusETag = x.mHpricot.defineClassUnder("BogusETag", x.structAttr, x.structAttr.getAllocator());
2071
+ x.cBogusETag.defineAnnotatedMethods(BogusETag.class);
2072
+
2073
+ x.cText = x.mHpricot.defineClassUnder("Text", x.structBasic, x.structBasic.getAllocator());
2074
+ x.cText.defineAnnotatedMethods(Text.class);
2075
+
2076
+ x.cXMLDecl = x.mHpricot.defineClassUnder("XMLDecl", x.structAttr, x.structAttr.getAllocator());
2077
+ x.cXMLDecl.defineAnnotatedMethods(XMLDecl.class);
2078
+
2079
+ x.cProcIns = x.mHpricot.defineClassUnder("ProcIns", x.structAttr, x.structAttr.getAllocator());
2080
+ x.cProcIns.defineAnnotatedMethods(ProcIns.class);
2081
+
2082
+ x.reProcInsParse = runtime.evalScriptlet("/\\A<\\?(\\S+)\\s+(.+)/m");
2083
+ x.mHpricot.setConstant("ProcInsParse", x.reProcInsParse);
2084
+ }
2085
+ }