immunio 0.15.2

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (157) hide show
  1. checksums.yaml +7 -0
  2. data/LICENSE +234 -0
  3. data/README.md +147 -0
  4. data/bin/immunio +5 -0
  5. data/lib/immunio.rb +29 -0
  6. data/lib/immunio/agent.rb +260 -0
  7. data/lib/immunio/authentication.rb +96 -0
  8. data/lib/immunio/blocked_app.rb +38 -0
  9. data/lib/immunio/channel.rb +432 -0
  10. data/lib/immunio/cli.rb +39 -0
  11. data/lib/immunio/context.rb +114 -0
  12. data/lib/immunio/errors.rb +43 -0
  13. data/lib/immunio/immunio_ca.crt +45 -0
  14. data/lib/immunio/logger.rb +87 -0
  15. data/lib/immunio/plugins/action_dispatch.rb +45 -0
  16. data/lib/immunio/plugins/action_view.rb +431 -0
  17. data/lib/immunio/plugins/active_record.rb +707 -0
  18. data/lib/immunio/plugins/active_record_relation.rb +370 -0
  19. data/lib/immunio/plugins/authlogic.rb +80 -0
  20. data/lib/immunio/plugins/csrf.rb +24 -0
  21. data/lib/immunio/plugins/devise.rb +40 -0
  22. data/lib/immunio/plugins/environment_reporter.rb +69 -0
  23. data/lib/immunio/plugins/eval.rb +51 -0
  24. data/lib/immunio/plugins/exception_handler.rb +55 -0
  25. data/lib/immunio/plugins/gems_tracker.rb +5 -0
  26. data/lib/immunio/plugins/haml.rb +36 -0
  27. data/lib/immunio/plugins/http_finisher.rb +50 -0
  28. data/lib/immunio/plugins/http_tracker.rb +203 -0
  29. data/lib/immunio/plugins/io.rb +96 -0
  30. data/lib/immunio/plugins/redirect.rb +42 -0
  31. data/lib/immunio/plugins/warden.rb +66 -0
  32. data/lib/immunio/processor.rb +234 -0
  33. data/lib/immunio/rails.rb +26 -0
  34. data/lib/immunio/request.rb +139 -0
  35. data/lib/immunio/rufus_lua_ext/ref.rb +27 -0
  36. data/lib/immunio/rufus_lua_ext/state.rb +157 -0
  37. data/lib/immunio/rufus_lua_ext/table.rb +137 -0
  38. data/lib/immunio/rufus_lua_ext/utils.rb +13 -0
  39. data/lib/immunio/version.rb +5 -0
  40. data/lib/immunio/vm.rb +291 -0
  41. data/lua-hooks/ext/all.c +78 -0
  42. data/lua-hooks/ext/bitop/README +22 -0
  43. data/lua-hooks/ext/bitop/bit.c +189 -0
  44. data/lua-hooks/ext/extconf.rb +38 -0
  45. data/lua-hooks/ext/libinjection/COPYING +37 -0
  46. data/lua-hooks/ext/libinjection/libinjection.h +65 -0
  47. data/lua-hooks/ext/libinjection/libinjection_html5.c +847 -0
  48. data/lua-hooks/ext/libinjection/libinjection_html5.h +54 -0
  49. data/lua-hooks/ext/libinjection/libinjection_sqli.c +2301 -0
  50. data/lua-hooks/ext/libinjection/libinjection_sqli.h +295 -0
  51. data/lua-hooks/ext/libinjection/libinjection_sqli_data.h +9349 -0
  52. data/lua-hooks/ext/libinjection/libinjection_xss.c +531 -0
  53. data/lua-hooks/ext/libinjection/libinjection_xss.h +21 -0
  54. data/lua-hooks/ext/libinjection/lualib.c +109 -0
  55. data/lua-hooks/ext/lpeg/HISTORY +90 -0
  56. data/lua-hooks/ext/lpeg/lpcap.c +537 -0
  57. data/lua-hooks/ext/lpeg/lpcap.h +43 -0
  58. data/lua-hooks/ext/lpeg/lpcode.c +986 -0
  59. data/lua-hooks/ext/lpeg/lpcode.h +34 -0
  60. data/lua-hooks/ext/lpeg/lpeg-128.gif +0 -0
  61. data/lua-hooks/ext/lpeg/lpeg.html +1429 -0
  62. data/lua-hooks/ext/lpeg/lpprint.c +244 -0
  63. data/lua-hooks/ext/lpeg/lpprint.h +35 -0
  64. data/lua-hooks/ext/lpeg/lptree.c +1238 -0
  65. data/lua-hooks/ext/lpeg/lptree.h +77 -0
  66. data/lua-hooks/ext/lpeg/lptypes.h +149 -0
  67. data/lua-hooks/ext/lpeg/lpvm.c +355 -0
  68. data/lua-hooks/ext/lpeg/lpvm.h +58 -0
  69. data/lua-hooks/ext/lpeg/makefile +55 -0
  70. data/lua-hooks/ext/lpeg/re.html +498 -0
  71. data/lua-hooks/ext/lpeg/test.lua +1409 -0
  72. data/lua-hooks/ext/lua-cmsgpack/CMakeLists.txt +45 -0
  73. data/lua-hooks/ext/lua-cmsgpack/README.md +115 -0
  74. data/lua-hooks/ext/lua-cmsgpack/lua_cmsgpack.c +957 -0
  75. data/lua-hooks/ext/lua-cmsgpack/test.lua +570 -0
  76. data/lua-hooks/ext/lua-snapshot/LICENSE +7 -0
  77. data/lua-hooks/ext/lua-snapshot/Makefile +12 -0
  78. data/lua-hooks/ext/lua-snapshot/README.md +18 -0
  79. data/lua-hooks/ext/lua-snapshot/dump.lua +15 -0
  80. data/lua-hooks/ext/lua-snapshot/snapshot.c +455 -0
  81. data/lua-hooks/ext/lua/COPYRIGHT +34 -0
  82. data/lua-hooks/ext/lua/lapi.c +1087 -0
  83. data/lua-hooks/ext/lua/lapi.h +16 -0
  84. data/lua-hooks/ext/lua/lauxlib.c +652 -0
  85. data/lua-hooks/ext/lua/lauxlib.h +174 -0
  86. data/lua-hooks/ext/lua/lbaselib.c +659 -0
  87. data/lua-hooks/ext/lua/lcode.c +831 -0
  88. data/lua-hooks/ext/lua/lcode.h +76 -0
  89. data/lua-hooks/ext/lua/ldblib.c +398 -0
  90. data/lua-hooks/ext/lua/ldebug.c +638 -0
  91. data/lua-hooks/ext/lua/ldebug.h +33 -0
  92. data/lua-hooks/ext/lua/ldo.c +519 -0
  93. data/lua-hooks/ext/lua/ldo.h +57 -0
  94. data/lua-hooks/ext/lua/ldump.c +164 -0
  95. data/lua-hooks/ext/lua/lfunc.c +174 -0
  96. data/lua-hooks/ext/lua/lfunc.h +34 -0
  97. data/lua-hooks/ext/lua/lgc.c +710 -0
  98. data/lua-hooks/ext/lua/lgc.h +110 -0
  99. data/lua-hooks/ext/lua/linit.c +38 -0
  100. data/lua-hooks/ext/lua/liolib.c +556 -0
  101. data/lua-hooks/ext/lua/llex.c +463 -0
  102. data/lua-hooks/ext/lua/llex.h +81 -0
  103. data/lua-hooks/ext/lua/llimits.h +128 -0
  104. data/lua-hooks/ext/lua/lmathlib.c +263 -0
  105. data/lua-hooks/ext/lua/lmem.c +86 -0
  106. data/lua-hooks/ext/lua/lmem.h +49 -0
  107. data/lua-hooks/ext/lua/loadlib.c +705 -0
  108. data/lua-hooks/ext/lua/loadlib_rel.c +760 -0
  109. data/lua-hooks/ext/lua/lobject.c +214 -0
  110. data/lua-hooks/ext/lua/lobject.h +381 -0
  111. data/lua-hooks/ext/lua/lopcodes.c +102 -0
  112. data/lua-hooks/ext/lua/lopcodes.h +268 -0
  113. data/lua-hooks/ext/lua/loslib.c +243 -0
  114. data/lua-hooks/ext/lua/lparser.c +1339 -0
  115. data/lua-hooks/ext/lua/lparser.h +82 -0
  116. data/lua-hooks/ext/lua/lstate.c +214 -0
  117. data/lua-hooks/ext/lua/lstate.h +169 -0
  118. data/lua-hooks/ext/lua/lstring.c +111 -0
  119. data/lua-hooks/ext/lua/lstring.h +31 -0
  120. data/lua-hooks/ext/lua/lstrlib.c +871 -0
  121. data/lua-hooks/ext/lua/ltable.c +588 -0
  122. data/lua-hooks/ext/lua/ltable.h +40 -0
  123. data/lua-hooks/ext/lua/ltablib.c +287 -0
  124. data/lua-hooks/ext/lua/ltm.c +75 -0
  125. data/lua-hooks/ext/lua/ltm.h +54 -0
  126. data/lua-hooks/ext/lua/lua.c +392 -0
  127. data/lua-hooks/ext/lua/lua.def +131 -0
  128. data/lua-hooks/ext/lua/lua.h +388 -0
  129. data/lua-hooks/ext/lua/lua.rc +28 -0
  130. data/lua-hooks/ext/lua/lua_dll.rc +26 -0
  131. data/lua-hooks/ext/lua/luac.c +200 -0
  132. data/lua-hooks/ext/lua/luac.rc +1 -0
  133. data/lua-hooks/ext/lua/luaconf.h +763 -0
  134. data/lua-hooks/ext/lua/luaconf.h.in +724 -0
  135. data/lua-hooks/ext/lua/luaconf.h.orig +763 -0
  136. data/lua-hooks/ext/lua/lualib.h +53 -0
  137. data/lua-hooks/ext/lua/lundump.c +227 -0
  138. data/lua-hooks/ext/lua/lundump.h +36 -0
  139. data/lua-hooks/ext/lua/lvm.c +767 -0
  140. data/lua-hooks/ext/lua/lvm.h +36 -0
  141. data/lua-hooks/ext/lua/lzio.c +82 -0
  142. data/lua-hooks/ext/lua/lzio.h +67 -0
  143. data/lua-hooks/ext/lua/print.c +227 -0
  144. data/lua-hooks/ext/luautf8/README.md +152 -0
  145. data/lua-hooks/ext/luautf8/lutf8lib.c +1274 -0
  146. data/lua-hooks/ext/luautf8/unidata.h +3064 -0
  147. data/lua-hooks/lib/boot.lua +254 -0
  148. data/lua-hooks/lib/encode.lua +4 -0
  149. data/lua-hooks/lib/lexers/LICENSE +21 -0
  150. data/lua-hooks/lib/lexers/bash.lua +134 -0
  151. data/lua-hooks/lib/lexers/bash_dqstr.lua +62 -0
  152. data/lua-hooks/lib/lexers/css.lua +216 -0
  153. data/lua-hooks/lib/lexers/html.lua +106 -0
  154. data/lua-hooks/lib/lexers/javascript.lua +68 -0
  155. data/lua-hooks/lib/lexers/lexer.lua +1575 -0
  156. data/lua-hooks/lib/lexers/markers.lua +33 -0
  157. metadata +308 -0
@@ -0,0 +1,244 @@
1
+ /*
2
+ ** $Id: lpprint.c,v 1.7 2013/04/12 16:29:49 roberto Exp $
3
+ ** Copyright 2007, Lua.org & PUC-Rio (see 'lpeg.html' for license)
4
+ */
5
+
6
+ #include <ctype.h>
7
+ #include <limits.h>
8
+ #include <stdio.h>
9
+
10
+
11
+ #include "lptypes.h"
12
+ #include "lpprint.h"
13
+ #include "lpcode.h"
14
+
15
+
16
+ #if defined(LPEG_DEBUG)
17
+
18
+ /*
19
+ ** {======================================================
20
+ ** Printing patterns (for debugging)
21
+ ** =======================================================
22
+ */
23
+
24
+
25
+ void printcharset (const byte *st) {
26
+ int i;
27
+ printf("[");
28
+ for (i = 0; i <= UCHAR_MAX; i++) {
29
+ int first = i;
30
+ while (testchar(st, i) && i <= UCHAR_MAX) i++;
31
+ if (i - 1 == first) /* unary range? */
32
+ printf("(%02x)", first);
33
+ else if (i - 1 > first) /* non-empty range? */
34
+ printf("(%02x-%02x)", first, i - 1);
35
+ }
36
+ printf("]");
37
+ }
38
+
39
+
40
+ static void printcapkind (int kind) {
41
+ const char *const modes[] = {
42
+ "close", "position", "constant", "backref",
43
+ "argument", "simple", "table", "function",
44
+ "query", "string", "num", "substitution", "fold",
45
+ "runtime", "group"};
46
+ printf("%s", modes[kind]);
47
+ }
48
+
49
+
50
+ static void printjmp (const LpegInstruction *op, const LpegInstruction *p) {
51
+ printf("-> %d", (int)(p + (p + 1)->offset - op));
52
+ }
53
+
54
+
55
+ static void printinst (const LpegInstruction *op, const LpegInstruction *p) {
56
+ const char *const names[] = {
57
+ "any", "char", "set",
58
+ "testany", "testchar", "testset",
59
+ "span", "behind",
60
+ "ret", "end",
61
+ "choice", "jmp", "call", "open_call",
62
+ "commit", "partial_commit", "back_commit", "failtwice", "fail", "giveup",
63
+ "fullcapture", "opencapture", "closecapture", "closeruntime"
64
+ };
65
+ printf("%02ld: %s ", (long)(p - op), names[p->i.code]);
66
+ switch ((Opcode)p->i.code) {
67
+ case IChar: {
68
+ printf("'%c'", p->i.aux);
69
+ break;
70
+ }
71
+ case ITestChar: {
72
+ printf("'%c'", p->i.aux); printjmp(op, p);
73
+ break;
74
+ }
75
+ case IFullCapture: {
76
+ printcapkind(getkind(p));
77
+ printf(" (size = %d) (idx = %d)", getoff(p), p->i.key);
78
+ break;
79
+ }
80
+ case IOpenCapture: {
81
+ printcapkind(getkind(p));
82
+ printf(" (idx = %d)", p->i.key);
83
+ break;
84
+ }
85
+ case ISet: {
86
+ printcharset((p+1)->buff);
87
+ break;
88
+ }
89
+ case ITestSet: {
90
+ printcharset((p+2)->buff); printjmp(op, p);
91
+ break;
92
+ }
93
+ case ISpan: {
94
+ printcharset((p+1)->buff);
95
+ break;
96
+ }
97
+ case IOpenCall: {
98
+ printf("-> %d", (p + 1)->offset);
99
+ break;
100
+ }
101
+ case IBehind: {
102
+ printf("%d", p->i.aux);
103
+ break;
104
+ }
105
+ case IJmp: case ICall: case ICommit: case IChoice:
106
+ case IPartialCommit: case IBackCommit: case ITestAny: {
107
+ printjmp(op, p);
108
+ break;
109
+ }
110
+ default: break;
111
+ }
112
+ printf("\n");
113
+ }
114
+
115
+
116
+ void printpatt (LpegInstruction *p, int n) {
117
+ LpegInstruction *op = p;
118
+ while (p < op + n) {
119
+ printinst(op, p);
120
+ p += sizei(p);
121
+ }
122
+ }
123
+
124
+
125
+ #if defined(LPEG_DEBUG)
126
+ static void printcap (Capture *cap) {
127
+ printcapkind(cap->kind);
128
+ printf(" (idx: %d - size: %d) -> %p\n", cap->idx, cap->siz, cap->s);
129
+ }
130
+
131
+
132
+ void printcaplist (Capture *cap, Capture *limit) {
133
+ printf(">======\n");
134
+ for (; cap->s && (limit == NULL || cap < limit); cap++)
135
+ printcap(cap);
136
+ printf("=======\n");
137
+ }
138
+ #endif
139
+
140
+ /* }====================================================== */
141
+
142
+
143
+ /*
144
+ ** {======================================================
145
+ ** Printing trees (for debugging)
146
+ ** =======================================================
147
+ */
148
+
149
+ static const char *tagnames[] = {
150
+ "char", "set", "any",
151
+ "true", "false",
152
+ "rep",
153
+ "seq", "choice",
154
+ "not", "and",
155
+ "call", "opencall", "rule", "grammar",
156
+ "behind",
157
+ "capture", "run-time"
158
+ };
159
+
160
+
161
+ void printtree (TTree *tree, int ident) {
162
+ int i;
163
+ for (i = 0; i < ident; i++) printf(" ");
164
+ printf("%s", tagnames[tree->tag]);
165
+ switch (tree->tag) {
166
+ case TChar: {
167
+ int c = tree->u.n;
168
+ if (isprint(c))
169
+ printf(" '%c'\n", c);
170
+ else
171
+ printf(" (%02X)\n", c);
172
+ break;
173
+ }
174
+ case TSet: {
175
+ printcharset(treebuffer(tree));
176
+ printf("\n");
177
+ break;
178
+ }
179
+ case TOpenCall: case TCall: {
180
+ printf(" key: %d\n", tree->key);
181
+ break;
182
+ }
183
+ case TBehind: {
184
+ printf(" %d\n", tree->u.n);
185
+ printtree(sib1(tree), ident + 2);
186
+ break;
187
+ }
188
+ case TCapture: {
189
+ printf(" cap: %d key: %d n: %d\n", tree->cap, tree->key, tree->u.n);
190
+ printtree(sib1(tree), ident + 2);
191
+ break;
192
+ }
193
+ case TRule: {
194
+ printf(" n: %d key: %d\n", tree->cap, tree->key);
195
+ printtree(sib1(tree), ident + 2);
196
+ break; /* do not print next rule as a sibling */
197
+ }
198
+ case TGrammar: {
199
+ TTree *rule = sib1(tree);
200
+ printf(" %d\n", tree->u.n); /* number of rules */
201
+ for (i = 0; i < tree->u.n; i++) {
202
+ printtree(rule, ident + 2);
203
+ rule = sib2(rule);
204
+ }
205
+ assert(rule->tag == TTrue); /* sentinel */
206
+ break;
207
+ }
208
+ default: {
209
+ int sibs = numsiblings[tree->tag];
210
+ printf("\n");
211
+ if (sibs >= 1) {
212
+ printtree(sib1(tree), ident + 2);
213
+ if (sibs >= 2)
214
+ printtree(sib2(tree), ident + 2);
215
+ }
216
+ break;
217
+ }
218
+ }
219
+ }
220
+
221
+
222
+ void printktable (lua_State *L, int idx) {
223
+ int n, i;
224
+ lua_getfenv(L, idx);
225
+ if (lua_isnil(L, -1)) /* no ktable? */
226
+ return;
227
+ n = lua_objlen(L, -1);
228
+ printf("[");
229
+ for (i = 1; i <= n; i++) {
230
+ printf("%d = ", i);
231
+ lua_rawgeti(L, -1, i);
232
+ if (lua_isstring(L, -1))
233
+ printf("%s ", lua_tostring(L, -1));
234
+ else
235
+ printf("%s ", lua_typename(L, lua_type(L, -1)));
236
+ lua_pop(L, 1);
237
+ }
238
+ printf("]\n");
239
+ /* leave ktable at the stack */
240
+ }
241
+
242
+ /* }====================================================== */
243
+
244
+ #endif
@@ -0,0 +1,35 @@
1
+ /*
2
+ ** $Id: lpprint.h,v 1.1 2013/03/21 20:25:12 roberto Exp $
3
+ */
4
+
5
+
6
+ #if !defined(lpprint_h)
7
+ #define lpprint_h
8
+
9
+
10
+ #include "lptree.h"
11
+ #include "lpvm.h"
12
+
13
+
14
+ #if defined(LPEG_DEBUG)
15
+
16
+ void printpatt (LpegInstruction *p, int n);
17
+ void printtree (TTree *tree, int ident);
18
+ void printktable (lua_State *L, int idx);
19
+ void printcharset (const byte *st);
20
+ void printcaplist (Capture *cap, Capture *limit);
21
+
22
+ #else
23
+
24
+ #define printktable(L,idx) \
25
+ luaL_error(L, "function only implemented in debug mode")
26
+ #define printtree(tree,i) \
27
+ luaL_error(L, "function only implemented in debug mode")
28
+ #define printpatt(p,n) \
29
+ luaL_error(L, "function only implemented in debug mode")
30
+
31
+ #endif
32
+
33
+
34
+ #endif
35
+
@@ -0,0 +1,1238 @@
1
+ /*
2
+ ** $Id: lptree.c,v 1.13 2014/12/12 16:59:10 roberto Exp $
3
+ ** Copyright 2013, Lua.org & PUC-Rio (see 'lpeg.html' for license)
4
+ */
5
+
6
+ #include <ctype.h>
7
+ #include <limits.h>
8
+ #include <string.h>
9
+
10
+
11
+ #include "../lua/lua.h"
12
+ #include "../lua/lauxlib.h"
13
+
14
+ #include "lptypes.h"
15
+ #include "lpcap.h"
16
+ #include "lpcode.h"
17
+ #include "lpprint.h"
18
+ #include "lptree.h"
19
+
20
+
21
+ /* number of siblings for each tree */
22
+ const byte numsiblings[] = {
23
+ 0, 0, 0, /* char, set, any */
24
+ 0, 0, /* true, false */
25
+ 1, /* rep */
26
+ 2, 2, /* seq, choice */
27
+ 1, 1, /* not, and */
28
+ 0, 0, 2, 1, /* call, opencall, rule, grammar */
29
+ 1, /* behind */
30
+ 1, 1 /* capture, runtime capture */
31
+ };
32
+
33
+
34
+ static TTree *newgrammar (lua_State *L, int arg);
35
+
36
+
37
+ /*
38
+ ** returns a reasonable name for value at index 'idx' on the stack
39
+ */
40
+ static const char *val2str (lua_State *L, int idx) {
41
+ const char *k = lua_tostring(L, idx);
42
+ if (k != NULL)
43
+ return lua_pushfstring(L, "%s", k);
44
+ else
45
+ return lua_pushfstring(L, "(a %s)", luaL_typename(L, idx));
46
+ }
47
+
48
+
49
+ /*
50
+ ** Fix a TOpenCall into a TCall node, using table 'postable' to
51
+ ** translate a key to its rule address in the tree. Raises an
52
+ ** error if key does not exist.
53
+ */
54
+ static void fixonecall (lua_State *L, int postable, TTree *g, TTree *t) {
55
+ int n;
56
+ lua_rawgeti(L, -1, t->key); /* get rule's name */
57
+ lua_gettable(L, postable); /* query name in position table */
58
+ n = lua_tonumber(L, -1); /* get (absolute) position */
59
+ lua_pop(L, 1); /* remove position */
60
+ if (n == 0) { /* no position? */
61
+ lua_rawgeti(L, -1, t->key); /* get rule's name again */
62
+ luaL_error(L, "rule '%s' undefined in given grammar", val2str(L, -1));
63
+ }
64
+ t->tag = TCall;
65
+ t->u.ps = n - (t - g); /* position relative to node */
66
+ assert(sib2(t)->tag == TRule);
67
+ sib2(t)->key = t->key;
68
+ }
69
+
70
+
71
+ /*
72
+ ** Transform left associative constructions into right
73
+ ** associative ones, for sequence and choice; that is:
74
+ ** (t11 + t12) + t2 => t11 + (t12 + t2)
75
+ ** (t11 * t12) * t2 => t11 * (t12 * t2)
76
+ ** (that is, Op (Op t11 t12) t2 => Op t11 (Op t12 t2))
77
+ */
78
+ static void correctassociativity (TTree *tree) {
79
+ TTree *t1 = sib1(tree);
80
+ assert(tree->tag == TChoice || tree->tag == TSeq);
81
+ while (t1->tag == tree->tag) {
82
+ int n1size = tree->u.ps - 1; /* t1 == Op t11 t12 */
83
+ int n11size = t1->u.ps - 1;
84
+ int n12size = n1size - n11size - 1;
85
+ memmove(sib1(tree), sib1(t1), n11size * sizeof(TTree)); /* move t11 */
86
+ tree->u.ps = n11size + 1;
87
+ sib2(tree)->tag = tree->tag;
88
+ sib2(tree)->u.ps = n12size + 1;
89
+ }
90
+ }
91
+
92
+
93
+ /*
94
+ ** Make final adjustments in a tree. Fix open calls in tree 't',
95
+ ** making them refer to their respective rules or raising appropriate
96
+ ** errors (if not inside a grammar). Correct associativity of associative
97
+ ** constructions (making them right associative). Assume that tree's
98
+ ** ktable is at the top of the stack (for error messages).
99
+ */
100
+ static void finalfix (lua_State *L, int postable, TTree *g, TTree *t) {
101
+ tailcall:
102
+ switch (t->tag) {
103
+ case TGrammar: /* subgrammars were already fixed */
104
+ return;
105
+ case TOpenCall: {
106
+ if (g != NULL) /* inside a grammar? */
107
+ fixonecall(L, postable, g, t);
108
+ else { /* open call outside grammar */
109
+ lua_rawgeti(L, -1, t->key);
110
+ luaL_error(L, "rule '%s' used outside a grammar", val2str(L, -1));
111
+ }
112
+ break;
113
+ }
114
+ case TSeq: case TChoice:
115
+ correctassociativity(t);
116
+ break;
117
+ }
118
+ switch (numsiblings[t->tag]) {
119
+ case 1: /* finalfix(L, postable, g, sib1(t)); */
120
+ t = sib1(t); goto tailcall;
121
+ case 2:
122
+ finalfix(L, postable, g, sib1(t));
123
+ t = sib2(t); goto tailcall; /* finalfix(L, postable, g, sib2(t)); */
124
+ default: assert(numsiblings[t->tag] == 0); break;
125
+ }
126
+ }
127
+
128
+
129
+ /*
130
+ ** {======================================================
131
+ ** Tree generation
132
+ ** =======================================================
133
+ */
134
+
135
+ /*
136
+ ** In 5.2, could use 'luaL_testudata'...
137
+ */
138
+ static int testpattern (lua_State *L, int idx) {
139
+ if (lua_touserdata(L, idx)) { /* value is a userdata? */
140
+ if (lua_getmetatable(L, idx)) { /* does it have a metatable? */
141
+ luaL_getmetatable(L, PATTERN_T);
142
+ if (lua_rawequal(L, -1, -2)) { /* does it have the correct mt? */
143
+ lua_pop(L, 2); /* remove both metatables */
144
+ return 1;
145
+ }
146
+ }
147
+ }
148
+ return 0;
149
+ }
150
+
151
+
152
+ static Pattern *getpattern (lua_State *L, int idx) {
153
+ return (Pattern *)luaL_checkudata(L, idx, PATTERN_T);
154
+ }
155
+
156
+
157
+ static int getsize (lua_State *L, int idx) {
158
+ return (lua_objlen(L, idx) - sizeof(Pattern)) / sizeof(TTree) + 1;
159
+ }
160
+
161
+
162
+ static TTree *gettree (lua_State *L, int idx, int *len) {
163
+ Pattern *p = getpattern(L, idx);
164
+ if (len)
165
+ *len = getsize(L, idx);
166
+ return p->tree;
167
+ }
168
+
169
+
170
+ /*
171
+ ** create a pattern
172
+ */
173
+ static TTree *newtree (lua_State *L, int len) {
174
+ size_t size = (len - 1) * sizeof(TTree) + sizeof(Pattern);
175
+ Pattern *p = (Pattern *)lua_newuserdata(L, size);
176
+ luaL_getmetatable(L, PATTERN_T);
177
+ lua_setmetatable(L, -2);
178
+ p->code = NULL; p->codesize = 0;
179
+ return p->tree;
180
+ }
181
+
182
+
183
+ static TTree *newleaf (lua_State *L, int tag) {
184
+ TTree *tree = newtree(L, 1);
185
+ tree->tag = tag;
186
+ return tree;
187
+ }
188
+
189
+
190
+ static TTree *newcharset (lua_State *L) {
191
+ TTree *tree = newtree(L, bytes2slots(CHARSETSIZE) + 1);
192
+ tree->tag = TSet;
193
+ loopset(i, treebuffer(tree)[i] = 0);
194
+ return tree;
195
+ }
196
+
197
+
198
+ /*
199
+ ** add to tree a sequence where first sibling is 'sib' (with size
200
+ ** 'sibsize'); returns position for second sibling
201
+ */
202
+ static TTree *seqaux (TTree *tree, TTree *sib, int sibsize) {
203
+ tree->tag = TSeq; tree->u.ps = sibsize + 1;
204
+ memcpy(sib1(tree), sib, sibsize * sizeof(TTree));
205
+ return sib2(tree);
206
+ }
207
+
208
+
209
+ /*
210
+ ** Add element 'idx' to 'ktable' of pattern at the top of the stack;
211
+ ** create new 'ktable' if necessary. Return index of new element.
212
+ ** If new element is nil, does not add it to table (as it would be
213
+ ** useless) and returns 0, as ktable[0] is always nil.
214
+ */
215
+ static int addtoktable (lua_State *L, int idx) {
216
+ if (idx == 0) /* no actual value to insert? */
217
+ return 0;
218
+ else {
219
+ int n;
220
+ lua_getfenv(L, -1); /* get ktable from pattern */
221
+ n = lua_objlen(L, -1);
222
+ if (n == 0) { /* is it empty/non-existent? */
223
+ lua_pop(L, 1); /* remove it */
224
+ lua_createtable(L, 1, 0); /* create a fresh table */
225
+ lua_pushvalue(L, -1); /* make a copy */
226
+ lua_setfenv(L, -3); /* set it as 'ktable' for pattern */
227
+ }
228
+ if (!lua_isnil(L, idx)) { /* non-nil value? */
229
+ lua_pushvalue(L, idx); /* element to be added */
230
+ lua_rawseti(L, -2, ++n);
231
+ }
232
+ lua_pop(L, 1); /* remove 'ktable' */
233
+ return n;
234
+ }
235
+ }
236
+
237
+
238
+ /*
239
+ ** Build a sequence of 'n' nodes, each with tag 'tag' and 'u.n' got
240
+ ** from the array 's' (or 0 if array is NULL). (TSeq is binary, so it
241
+ ** must build a sequence of sequence of sequence...)
242
+ */
243
+ static void fillseq (TTree *tree, int tag, int n, const char *s) {
244
+ int i;
245
+ for (i = 0; i < n - 1; i++) { /* initial n-1 copies of Seq tag; Seq ... */
246
+ tree->tag = TSeq; tree->u.ps = 2;
247
+ sib1(tree)->tag = tag;
248
+ sib1(tree)->u.n = s ? (byte)s[i] : 0;
249
+ tree = sib2(tree);
250
+ }
251
+ tree->tag = tag; /* last one does not need TSeq */
252
+ tree->u.n = s ? (byte)s[i] : 0;
253
+ }
254
+
255
+
256
+ /*
257
+ ** Numbers as patterns:
258
+ ** 0 == true (always match); n == TAny repeated 'n' times;
259
+ ** -n == not (TAny repeated 'n' times)
260
+ */
261
+ static TTree *numtree (lua_State *L, int n) {
262
+ if (n == 0)
263
+ return newleaf(L, TTrue);
264
+ else {
265
+ TTree *tree, *nd;
266
+ if (n > 0)
267
+ tree = nd = newtree(L, 2 * n - 1);
268
+ else { /* negative: code it as !(-n) */
269
+ n = -n;
270
+ tree = newtree(L, 2 * n);
271
+ tree->tag = TNot;
272
+ nd = sib1(tree);
273
+ }
274
+ fillseq(nd, TAny, n, NULL); /* sequence of 'n' any's */
275
+ return tree;
276
+ }
277
+ }
278
+
279
+
280
+ /*
281
+ ** Convert value at index 'idx' to a pattern
282
+ */
283
+ static TTree *getpatt (lua_State *L, int idx, int *len) {
284
+ TTree *tree;
285
+ switch (lua_type(L, idx)) {
286
+ case LUA_TSTRING: {
287
+ size_t slen;
288
+ const char *s = lua_tolstring(L, idx, &slen); /* get string */
289
+ if (slen == 0) /* empty? */
290
+ tree = newleaf(L, TTrue); /* always match */
291
+ else {
292
+ tree = newtree(L, 2 * (slen - 1) + 1);
293
+ fillseq(tree, TChar, slen, s); /* sequence of 'slen' chars */
294
+ }
295
+ break;
296
+ }
297
+ case LUA_TNUMBER: {
298
+ int n = lua_tointeger(L, idx);
299
+ tree = numtree(L, n);
300
+ break;
301
+ }
302
+ case LUA_TBOOLEAN: {
303
+ tree = (lua_toboolean(L, idx) ? newleaf(L, TTrue) : newleaf(L, TFalse));
304
+ break;
305
+ }
306
+ case LUA_TTABLE: {
307
+ tree = newgrammar(L, idx);
308
+ break;
309
+ }
310
+ case LUA_TFUNCTION: {
311
+ tree = newtree(L, 2);
312
+ tree->tag = TRunTime;
313
+ tree->key = addtoktable(L, idx);
314
+ sib1(tree)->tag = TTrue;
315
+ break;
316
+ }
317
+ default: {
318
+ return gettree(L, idx, len);
319
+ }
320
+ }
321
+ lua_replace(L, idx); /* put new tree into 'idx' slot */
322
+ if (len)
323
+ *len = getsize(L, idx);
324
+ return tree;
325
+ }
326
+
327
+
328
+ /*
329
+ ** Return the number of elements in the ktable of pattern at 'idx'.
330
+ ** In Lua 5.2, default "environment" for patterns is nil, not
331
+ ** a table. Treat it as an empty table. In Lua 5.1, assumes that
332
+ ** the environment has no numeric indices (len == 0)
333
+ */
334
+ static int ktablelen (lua_State *L, int idx) {
335
+ if (!lua_istable(L, idx)) return 0;
336
+ else return lua_objlen(L, idx);
337
+ }
338
+
339
+
340
+ /*
341
+ ** Concatentate the contents of table 'idx1' into table 'idx2'.
342
+ ** (Assume that both indices are negative.)
343
+ ** Return the original length of table 'idx2'
344
+ */
345
+ static int concattable (lua_State *L, int idx1, int idx2) {
346
+ int i;
347
+ int n1 = ktablelen(L, idx1);
348
+ int n2 = ktablelen(L, idx2);
349
+ if (n1 == 0) return 0; /* nothing to correct */
350
+ for (i = 1; i <= n1; i++) {
351
+ lua_rawgeti(L, idx1, i);
352
+ lua_rawseti(L, idx2 - 1, n2 + i); /* correct 'idx2' */
353
+ }
354
+ return n2;
355
+ }
356
+
357
+
358
+ /*
359
+ ** Make a merge of ktables from p1 and p2 the ktable for the new
360
+ ** pattern at the top of the stack.
361
+ */
362
+ static int joinktables (lua_State *L, int p1, int p2) {
363
+ int n1, n2;
364
+ lua_getfenv(L, p1); /* get ktables */
365
+ lua_getfenv(L, p2);
366
+ n1 = ktablelen(L, -2);
367
+ n2 = ktablelen(L, -1);
368
+ if (n1 == 0 && n2 == 0) { /* are both tables empty? */
369
+ lua_pop(L, 2); /* nothing to be done; pop tables */
370
+ return 0; /* nothing to correct */
371
+ }
372
+ if (n2 == 0 || lua_equal(L, -2, -1)) { /* second table is empty or equal? */
373
+ lua_pop(L, 1); /* pop 2nd table */
374
+ lua_setfenv(L, -2); /* set 1st ktable into new pattern */
375
+ return 0; /* nothing to correct */
376
+ }
377
+ if (n1 == 0) { /* first table is empty? */
378
+ lua_setfenv(L, -3); /* set 2nd table into new pattern */
379
+ lua_pop(L, 1); /* pop 1st table */
380
+ return 0; /* nothing to correct */
381
+ }
382
+ else {
383
+ lua_createtable(L, n1 + n2, 0); /* create ktable for new pattern */
384
+ /* stack: new p; ktable p1; ktable p2; new ktable */
385
+ concattable(L, -3, -1); /* from p1 into new ktable */
386
+ concattable(L, -2, -1); /* from p2 into new ktable */
387
+ lua_setfenv(L, -4); /* new ktable becomes p env */
388
+ lua_pop(L, 2); /* pop other ktables */
389
+ return n1; /* correction for indices from p2 */
390
+ }
391
+ }
392
+
393
+
394
+ static void correctkeys (TTree *tree, int n) {
395
+ if (n == 0) return; /* no correction? */
396
+ tailcall:
397
+ switch (tree->tag) {
398
+ case TOpenCall: case TCall: case TRunTime: case TRule: {
399
+ if (tree->key > 0)
400
+ tree->key += n;
401
+ break;
402
+ }
403
+ case TCapture: {
404
+ if (tree->key > 0 && tree->cap != Carg && tree->cap != Cnum)
405
+ tree->key += n;
406
+ break;
407
+ }
408
+ default: break;
409
+ }
410
+ switch (numsiblings[tree->tag]) {
411
+ case 1: /* correctkeys(sib1(tree), n); */
412
+ tree = sib1(tree); goto tailcall;
413
+ case 2:
414
+ correctkeys(sib1(tree), n);
415
+ tree = sib2(tree); goto tailcall; /* correctkeys(sib2(tree), n); */
416
+ default: assert(numsiblings[tree->tag] == 0); break;
417
+ }
418
+ }
419
+
420
+
421
+ /*
422
+ ** copy 'ktable' of element 'idx' to new tree (on top of stack)
423
+ */
424
+ static void copyktable (lua_State *L, int idx) {
425
+ lua_getfenv(L, idx);
426
+ lua_setfenv(L, -2);
427
+ }
428
+
429
+
430
+ /*
431
+ ** merge 'ktable' from rule at stack index 'idx' into 'ktable'
432
+ ** from tree at the top of the stack, and correct corresponding
433
+ ** tree.
434
+ */
435
+ static void mergektable (lua_State *L, int idx, TTree *rule) {
436
+ int n;
437
+ lua_getfenv(L, -1); /* get ktables */
438
+ lua_getfenv(L, idx);
439
+ n = concattable(L, -1, -2);
440
+ lua_pop(L, 2); /* remove both ktables */
441
+ correctkeys(rule, n);
442
+ }
443
+
444
+
445
+ /*
446
+ ** create a new tree, whith a new root and one sibling.
447
+ ** Sibling must be on the Lua stack, at index 1.
448
+ */
449
+ static TTree *newroot1sib (lua_State *L, int tag) {
450
+ int s1;
451
+ TTree *tree1 = getpatt(L, 1, &s1);
452
+ TTree *tree = newtree(L, 1 + s1); /* create new tree */
453
+ tree->tag = tag;
454
+ memcpy(sib1(tree), tree1, s1 * sizeof(TTree));
455
+ copyktable(L, 1);
456
+ return tree;
457
+ }
458
+
459
+
460
+ /*
461
+ ** create a new tree, whith a new root and 2 siblings.
462
+ ** Siblings must be on the Lua stack, first one at index 1.
463
+ */
464
+ static TTree *newroot2sib (lua_State *L, int tag) {
465
+ int s1, s2;
466
+ TTree *tree1 = getpatt(L, 1, &s1);
467
+ TTree *tree2 = getpatt(L, 2, &s2);
468
+ TTree *tree = newtree(L, 1 + s1 + s2); /* create new tree */
469
+ tree->tag = tag;
470
+ tree->u.ps = 1 + s1;
471
+ memcpy(sib1(tree), tree1, s1 * sizeof(TTree));
472
+ memcpy(sib2(tree), tree2, s2 * sizeof(TTree));
473
+ correctkeys(sib2(tree), joinktables(L, 1, 2));
474
+ return tree;
475
+ }
476
+
477
+
478
+ static int lp_P (lua_State *L) {
479
+ luaL_checkany(L, 1);
480
+ getpatt(L, 1, NULL);
481
+ lua_settop(L, 1);
482
+ return 1;
483
+ }
484
+
485
+
486
+ /*
487
+ ** sequence operator; optimizations:
488
+ ** false x => false, x true => x, true x => x
489
+ ** (cannot do x . false => false because x may have runtime captures)
490
+ */
491
+ static int lp_seq (lua_State *L) {
492
+ TTree *tree1 = getpatt(L, 1, NULL);
493
+ TTree *tree2 = getpatt(L, 2, NULL);
494
+ if (tree1->tag == TFalse || tree2->tag == TTrue)
495
+ lua_pushvalue(L, 1); /* false . x == false, x . true = x */
496
+ else if (tree1->tag == TTrue)
497
+ lua_pushvalue(L, 2); /* true . x = x */
498
+ else
499
+ newroot2sib(L, TSeq);
500
+ return 1;
501
+ }
502
+
503
+
504
+ /*
505
+ ** choice operator; optimizations:
506
+ ** charset / charset => charset
507
+ ** true / x => true, x / false => x, false / x => x
508
+ ** (x / true is not equivalent to true)
509
+ */
510
+ static int lp_choice (lua_State *L) {
511
+ Charset st1, st2;
512
+ TTree *t1 = getpatt(L, 1, NULL);
513
+ TTree *t2 = getpatt(L, 2, NULL);
514
+ if (tocharset(t1, &st1) && tocharset(t2, &st2)) {
515
+ TTree *t = newcharset(L);
516
+ loopset(i, treebuffer(t)[i] = st1.cs[i] | st2.cs[i]);
517
+ }
518
+ else if (nofail(t1) || t2->tag == TFalse)
519
+ lua_pushvalue(L, 1); /* true / x => true, x / false => x */
520
+ else if (t1->tag == TFalse)
521
+ lua_pushvalue(L, 2); /* false / x => x */
522
+ else
523
+ newroot2sib(L, TChoice);
524
+ return 1;
525
+ }
526
+
527
+
528
+ /*
529
+ ** p^n
530
+ */
531
+ static int lp_star (lua_State *L) {
532
+ int size1;
533
+ int n = (int)luaL_checkinteger(L, 2);
534
+ TTree *tree1 = getpatt(L, 1, &size1);
535
+ if (n >= 0) { /* seq tree1 (seq tree1 ... (seq tree1 (rep tree1))) */
536
+ TTree *tree = newtree(L, (n + 1) * (size1 + 1));
537
+ if (nullable(tree1))
538
+ luaL_error(L, "loop body may accept empty string");
539
+ while (n--) /* repeat 'n' times */
540
+ tree = seqaux(tree, tree1, size1);
541
+ tree->tag = TRep;
542
+ memcpy(sib1(tree), tree1, size1 * sizeof(TTree));
543
+ }
544
+ else { /* choice (seq tree1 ... choice tree1 true ...) true */
545
+ TTree *tree;
546
+ n = -n;
547
+ /* size = (choice + seq + tree1 + true) * n, but the last has no seq */
548
+ tree = newtree(L, n * (size1 + 3) - 1);
549
+ for (; n > 1; n--) { /* repeat (n - 1) times */
550
+ tree->tag = TChoice; tree->u.ps = n * (size1 + 3) - 2;
551
+ sib2(tree)->tag = TTrue;
552
+ tree = sib1(tree);
553
+ tree = seqaux(tree, tree1, size1);
554
+ }
555
+ tree->tag = TChoice; tree->u.ps = size1 + 1;
556
+ sib2(tree)->tag = TTrue;
557
+ memcpy(sib1(tree), tree1, size1 * sizeof(TTree));
558
+ }
559
+ copyktable(L, 1);
560
+ return 1;
561
+ }
562
+
563
+
564
+ /*
565
+ ** #p == &p
566
+ */
567
+ static int lp_and (lua_State *L) {
568
+ newroot1sib(L, TAnd);
569
+ return 1;
570
+ }
571
+
572
+
573
+ /*
574
+ ** -p == !p
575
+ */
576
+ static int lp_not (lua_State *L) {
577
+ newroot1sib(L, TNot);
578
+ return 1;
579
+ }
580
+
581
+
582
+ /*
583
+ ** [t1 - t2] == Seq (Not t2) t1
584
+ ** If t1 and t2 are charsets, make their difference.
585
+ */
586
+ static int lp_sub (lua_State *L) {
587
+ Charset st1, st2;
588
+ int s1, s2;
589
+ TTree *t1 = getpatt(L, 1, &s1);
590
+ TTree *t2 = getpatt(L, 2, &s2);
591
+ if (tocharset(t1, &st1) && tocharset(t2, &st2)) {
592
+ TTree *t = newcharset(L);
593
+ loopset(i, treebuffer(t)[i] = st1.cs[i] & ~st2.cs[i]);
594
+ }
595
+ else {
596
+ TTree *tree = newtree(L, 2 + s1 + s2);
597
+ tree->tag = TSeq; /* sequence of... */
598
+ tree->u.ps = 2 + s2;
599
+ sib1(tree)->tag = TNot; /* ...not... */
600
+ memcpy(sib1(sib1(tree)), t2, s2 * sizeof(TTree)); /* ...t2 */
601
+ memcpy(sib2(tree), t1, s1 * sizeof(TTree)); /* ... and t1 */
602
+ correctkeys(sib1(tree), joinktables(L, 1, 2));
603
+ }
604
+ return 1;
605
+ }
606
+
607
+
608
+ static int lp_set (lua_State *L) {
609
+ size_t l;
610
+ const char *s = luaL_checklstring(L, 1, &l);
611
+ TTree *tree = newcharset(L);
612
+ while (l--) {
613
+ setchar(treebuffer(tree), (byte)(*s));
614
+ s++;
615
+ }
616
+ return 1;
617
+ }
618
+
619
+
620
+ static int lp_range (lua_State *L) {
621
+ int arg;
622
+ int top = lua_gettop(L);
623
+ TTree *tree = newcharset(L);
624
+ for (arg = 1; arg <= top; arg++) {
625
+ int c;
626
+ size_t l;
627
+ const char *r = luaL_checklstring(L, arg, &l);
628
+ luaL_argcheck(L, l == 2, arg, "range must have two characters");
629
+ for (c = (byte)r[0]; c <= (byte)r[1]; c++)
630
+ setchar(treebuffer(tree), c);
631
+ }
632
+ return 1;
633
+ }
634
+
635
+
636
+ /*
637
+ ** Look-behind predicate
638
+ */
639
+ static int lp_behind (lua_State *L) {
640
+ TTree *tree;
641
+ TTree *tree1 = getpatt(L, 1, NULL);
642
+ int n = fixedlen(tree1);
643
+ luaL_argcheck(L, n > 0, 1, "pattern may not have fixed length");
644
+ luaL_argcheck(L, !hascaptures(tree1), 1, "pattern have captures");
645
+ luaL_argcheck(L, n <= MAXBEHIND, 1, "pattern too long to look behind");
646
+ tree = newroot1sib(L, TBehind);
647
+ tree->u.n = n;
648
+ return 1;
649
+ }
650
+
651
+
652
+ /*
653
+ ** Create a non-terminal
654
+ */
655
+ static int lp_V (lua_State *L) {
656
+ TTree *tree = newleaf(L, TOpenCall);
657
+ luaL_argcheck(L, !lua_isnoneornil(L, 1), 1, "non-nil value expected");
658
+ tree->key = addtoktable(L, 1);
659
+ return 1;
660
+ }
661
+
662
+
663
+ /*
664
+ ** Create a tree for a non-empty capture, with a body and
665
+ ** optionally with an associated Lua value (at index 'labelidx' in the
666
+ ** stack)
667
+ */
668
+ static int capture_aux (lua_State *L, int cap, int labelidx) {
669
+ TTree *tree = newroot1sib(L, TCapture);
670
+ tree->cap = cap;
671
+ tree->key = addtoktable(L, labelidx);
672
+ return 1;
673
+ }
674
+
675
+
676
+ /*
677
+ ** Fill a tree with an empty capture, using an empty (TTrue) sibling.
678
+ */
679
+ static TTree *auxemptycap (lua_State *L, TTree *tree, int cap, int idx) {
680
+ tree->tag = TCapture;
681
+ tree->cap = cap;
682
+ tree->key = addtoktable(L, idx);
683
+ sib1(tree)->tag = TTrue;
684
+ return tree;
685
+ }
686
+
687
+
688
+ /*
689
+ ** Create a tree for an empty capture
690
+ */
691
+ static TTree *newemptycap (lua_State *L, int cap, int idx) {
692
+ return auxemptycap(L, newtree(L, 2), cap, idx);
693
+ }
694
+
695
+
696
+ /*
697
+ ** Captures with syntax p / v
698
+ ** (function capture, query capture, string capture, or number capture)
699
+ */
700
+ static int lp_divcapture (lua_State *L) {
701
+ switch (lua_type(L, 2)) {
702
+ case LUA_TFUNCTION: return capture_aux(L, Cfunction, 2);
703
+ case LUA_TTABLE: return capture_aux(L, Cquery, 2);
704
+ case LUA_TSTRING: return capture_aux(L, Cstring, 2);
705
+ case LUA_TNUMBER: {
706
+ int n = lua_tointeger(L, 2);
707
+ TTree *tree = newroot1sib(L, TCapture);
708
+ luaL_argcheck(L, 0 <= n && n <= SHRT_MAX, 1, "invalid number");
709
+ tree->cap = Cnum;
710
+ tree->key = n;
711
+ return 1;
712
+ }
713
+ default: return luaL_argerror(L, 2, "invalid replacement value");
714
+ }
715
+ }
716
+
717
+
718
+ static int lp_substcapture (lua_State *L) {
719
+ return capture_aux(L, Csubst, 0);
720
+ }
721
+
722
+
723
+ static int lp_tablecapture (lua_State *L) {
724
+ return capture_aux(L, Ctable, 0);
725
+ }
726
+
727
+
728
+ static int lp_groupcapture (lua_State *L) {
729
+ if (lua_isnoneornil(L, 2))
730
+ return capture_aux(L, Cgroup, 0);
731
+ else {
732
+ luaL_checkstring(L, 2);
733
+ return capture_aux(L, Cgroup, 2);
734
+ }
735
+ }
736
+
737
+
738
+ static int lp_foldcapture (lua_State *L) {
739
+ luaL_checktype(L, 2, LUA_TFUNCTION);
740
+ return capture_aux(L, Cfold, 2);
741
+ }
742
+
743
+
744
+ static int lp_simplecapture (lua_State *L) {
745
+ return capture_aux(L, Csimple, 0);
746
+ }
747
+
748
+
749
+ static int lp_poscapture (lua_State *L) {
750
+ newemptycap(L, Cposition, 0);
751
+ return 1;
752
+ }
753
+
754
+
755
+ static int lp_argcapture (lua_State *L) {
756
+ int n = (int)luaL_checkinteger(L, 1);
757
+ TTree *tree = newemptycap(L, Carg, 0);
758
+ tree->key = n;
759
+ luaL_argcheck(L, 0 < n && n <= SHRT_MAX, 1, "invalid argument index");
760
+ return 1;
761
+ }
762
+
763
+
764
+ static int lp_backref (lua_State *L) {
765
+ luaL_checkstring(L, 1);
766
+ newemptycap(L, Cbackref, 1);
767
+ return 1;
768
+ }
769
+
770
+
771
+ /*
772
+ ** Constant capture
773
+ */
774
+ static int lp_constcapture (lua_State *L) {
775
+ int i;
776
+ int n = lua_gettop(L); /* number of values */
777
+ if (n == 0) /* no values? */
778
+ newleaf(L, TTrue); /* no capture */
779
+ else if (n == 1)
780
+ newemptycap(L, Cconst, 1); /* single constant capture */
781
+ else { /* create a group capture with all values */
782
+ TTree *tree = newtree(L, 1 + 3 * (n - 1) + 2);
783
+ tree->tag = TCapture;
784
+ tree->cap = Cgroup;
785
+ tree->key = 0;
786
+ tree = sib1(tree);
787
+ for (i = 1; i <= n - 1; i++) {
788
+ tree->tag = TSeq;
789
+ tree->u.ps = 3; /* skip TCapture and its sibling */
790
+ auxemptycap(L, sib1(tree), Cconst, i);
791
+ tree = sib2(tree);
792
+ }
793
+ auxemptycap(L, tree, Cconst, i);
794
+ }
795
+ return 1;
796
+ }
797
+
798
+
799
+ static int lp_matchtime (lua_State *L) {
800
+ TTree *tree;
801
+ luaL_checktype(L, 2, LUA_TFUNCTION);
802
+ tree = newroot1sib(L, TRunTime);
803
+ tree->key = addtoktable(L, 2);
804
+ return 1;
805
+ }
806
+
807
+ /* }====================================================== */
808
+
809
+
810
+ /*
811
+ ** {======================================================
812
+ ** Grammar - Tree generation
813
+ ** =======================================================
814
+ */
815
+
816
+ /*
817
+ ** push on the stack the index and the pattern for the
818
+ ** initial rule of grammar at index 'arg' in the stack;
819
+ ** also add that index into position table.
820
+ */
821
+ static void getfirstrule (lua_State *L, int arg, int postab) {
822
+ lua_rawgeti(L, arg, 1); /* access first element */
823
+ if (lua_isstring(L, -1)) { /* is it the name of initial rule? */
824
+ lua_pushvalue(L, -1); /* duplicate it to use as key */
825
+ lua_gettable(L, arg); /* get associated rule */
826
+ }
827
+ else {
828
+ lua_pushinteger(L, 1); /* key for initial rule */
829
+ lua_insert(L, -2); /* put it before rule */
830
+ }
831
+ if (!testpattern(L, -1)) { /* initial rule not a pattern? */
832
+ if (lua_isnil(L, -1))
833
+ luaL_error(L, "grammar has no initial rule");
834
+ else
835
+ luaL_error(L, "initial rule '%s' is not a pattern", lua_tostring(L, -2));
836
+ }
837
+ lua_pushvalue(L, -2); /* push key */
838
+ lua_pushinteger(L, 1); /* push rule position (after TGrammar) */
839
+ lua_settable(L, postab); /* insert pair at position table */
840
+ }
841
+
842
+ /*
843
+ ** traverse grammar at index 'arg', pushing all its keys and patterns
844
+ ** into the stack. Create a new table (before all pairs key-pattern) to
845
+ ** collect all keys and their associated positions in the final tree
846
+ ** (the "position table").
847
+ ** Return the number of rules and (in 'totalsize') the total size
848
+ ** for the new tree.
849
+ */
850
+ static int collectrules (lua_State *L, int arg, int *totalsize) {
851
+ int n = 1; /* to count number of rules */
852
+ int postab = lua_gettop(L) + 1; /* index of position table */
853
+ int size; /* accumulator for total size */
854
+ lua_newtable(L); /* create position table */
855
+ getfirstrule(L, arg, postab);
856
+ size = 2 + getsize(L, postab + 2); /* TGrammar + TRule + rule */
857
+ lua_pushnil(L); /* prepare to traverse grammar table */
858
+ while (lua_next(L, arg) != 0) {
859
+ if (lua_tonumber(L, -2) == 1 ||
860
+ lua_equal(L, -2, postab + 1)) { /* initial rule? */
861
+ lua_pop(L, 1); /* remove value (keep key for lua_next) */
862
+ continue;
863
+ }
864
+ if (!testpattern(L, -1)) /* value is not a pattern? */
865
+ luaL_error(L, "rule '%s' is not a pattern", val2str(L, -2));
866
+ luaL_checkstack(L, LUA_MINSTACK, "grammar has too many rules");
867
+ lua_pushvalue(L, -2); /* push key (to insert into position table) */
868
+ lua_pushinteger(L, size);
869
+ lua_settable(L, postab);
870
+ size += 1 + getsize(L, -1); /* update size */
871
+ lua_pushvalue(L, -2); /* push key (for next lua_next) */
872
+ n++;
873
+ }
874
+ *totalsize = size + 1; /* TTrue to finish list of rules */
875
+ return n;
876
+ }
877
+
878
+
879
+ static void buildgrammar (lua_State *L, TTree *grammar, int frule, int n) {
880
+ int i;
881
+ TTree *nd = sib1(grammar); /* auxiliary pointer to traverse the tree */
882
+ for (i = 0; i < n; i++) { /* add each rule into new tree */
883
+ int ridx = frule + 2*i + 1; /* index of i-th rule */
884
+ int rulesize;
885
+ TTree *rn = gettree(L, ridx, &rulesize);
886
+ nd->tag = TRule;
887
+ nd->key = 0;
888
+ nd->cap = i; /* rule number */
889
+ nd->u.ps = rulesize + 1; /* point to next rule */
890
+ memcpy(sib1(nd), rn, rulesize * sizeof(TTree)); /* copy rule */
891
+ mergektable(L, ridx, sib1(nd)); /* merge its ktable into new one */
892
+ nd = sib2(nd); /* move to next rule */
893
+ }
894
+ nd->tag = TTrue; /* finish list of rules */
895
+ }
896
+
897
+
898
+ /*
899
+ ** Check whether a tree has potential infinite loops
900
+ */
901
+ static int checkloops (TTree *tree) {
902
+ tailcall:
903
+ if (tree->tag == TRep && nullable(sib1(tree)))
904
+ return 1;
905
+ else if (tree->tag == TGrammar)
906
+ return 0; /* sub-grammars already checked */
907
+ else {
908
+ switch (numsiblings[tree->tag]) {
909
+ case 1: /* return checkloops(sib1(tree)); */
910
+ tree = sib1(tree); goto tailcall;
911
+ case 2:
912
+ if (checkloops(sib1(tree))) return 1;
913
+ /* else return checkloops(sib2(tree)); */
914
+ tree = sib2(tree); goto tailcall;
915
+ default: assert(numsiblings[tree->tag] == 0); return 0;
916
+ }
917
+ }
918
+ }
919
+
920
+
921
+ static int verifyerror (lua_State *L, int *passed, int npassed) {
922
+ int i, j;
923
+ for (i = npassed - 1; i >= 0; i--) { /* search for a repetition */
924
+ for (j = i - 1; j >= 0; j--) {
925
+ if (passed[i] == passed[j]) {
926
+ lua_rawgeti(L, -1, passed[i]); /* get rule's key */
927
+ return luaL_error(L, "rule '%s' may be left recursive", val2str(L, -1));
928
+ }
929
+ }
930
+ }
931
+ return luaL_error(L, "too many left calls in grammar");
932
+ }
933
+
934
+
935
+ /*
936
+ ** Check whether a rule can be left recursive; raise an error in that
937
+ ** case; otherwise return 1 iff pattern is nullable. Assume ktable at
938
+ ** the top of the stack.
939
+ */
940
+ static int verifyrule (lua_State *L, TTree *tree, int *passed, int npassed,
941
+ int nullable) {
942
+ tailcall:
943
+ switch (tree->tag) {
944
+ case TChar: case TSet: case TAny:
945
+ case TFalse:
946
+ return nullable; /* cannot pass from here */
947
+ case TTrue:
948
+ case TBehind: /* look-behind cannot have calls */
949
+ return 1;
950
+ case TNot: case TAnd: case TRep:
951
+ /* return verifyrule(L, sib1(tree), passed, npassed, 1); */
952
+ tree = sib1(tree); nullable = 1; goto tailcall;
953
+ case TCapture: case TRunTime:
954
+ /* return verifyrule(L, sib1(tree), passed, npassed); */
955
+ tree = sib1(tree); goto tailcall;
956
+ case TCall:
957
+ /* return verifyrule(L, sib2(tree), passed, npassed); */
958
+ tree = sib2(tree); goto tailcall;
959
+ case TSeq: /* only check 2nd child if first is nullable */
960
+ if (!verifyrule(L, sib1(tree), passed, npassed, 0))
961
+ return nullable;
962
+ /* else return verifyrule(L, sib2(tree), passed, npassed); */
963
+ tree = sib2(tree); goto tailcall;
964
+ case TChoice: /* must check both children */
965
+ nullable = verifyrule(L, sib1(tree), passed, npassed, nullable);
966
+ /* return verifyrule(L, sib2(tree), passed, npassed, nullable); */
967
+ tree = sib2(tree); goto tailcall;
968
+ case TRule:
969
+ if (npassed >= MAXRULES)
970
+ return verifyerror(L, passed, npassed);
971
+ else {
972
+ passed[npassed++] = tree->key;
973
+ /* return verifyrule(L, sib1(tree), passed, npassed); */
974
+ tree = sib1(tree); goto tailcall;
975
+ }
976
+ case TGrammar:
977
+ return nullable(tree); /* sub-grammar cannot be left recursive */
978
+ default: assert(0); return 0;
979
+ }
980
+ }
981
+
982
+
983
+ static void verifygrammar (lua_State *L, TTree *grammar) {
984
+ int passed[MAXRULES];
985
+ TTree *rule;
986
+ /* check left-recursive rules */
987
+ for (rule = sib1(grammar); rule->tag == TRule; rule = sib2(rule)) {
988
+ if (rule->key == 0) continue; /* unused rule */
989
+ verifyrule(L, sib1(rule), passed, 0, 0);
990
+ }
991
+ assert(rule->tag == TTrue);
992
+ /* check infinite loops inside rules */
993
+ for (rule = sib1(grammar); rule->tag == TRule; rule = sib2(rule)) {
994
+ if (rule->key == 0) continue; /* unused rule */
995
+ if (checkloops(sib1(rule))) {
996
+ lua_rawgeti(L, -1, rule->key); /* get rule's key */
997
+ luaL_error(L, "empty loop in rule '%s'", val2str(L, -1));
998
+ }
999
+ }
1000
+ assert(rule->tag == TTrue);
1001
+ }
1002
+
1003
+
1004
+ /*
1005
+ ** Give a name for the initial rule if it is not referenced
1006
+ */
1007
+ static void initialrulename (lua_State *L, TTree *grammar, int frule) {
1008
+ if (sib1(grammar)->key == 0) { /* initial rule is not referenced? */
1009
+ int n = lua_objlen(L, -1) + 1; /* index for name */
1010
+ lua_pushvalue(L, frule); /* rule's name */
1011
+ lua_rawseti(L, -2, n); /* ktable was on the top of the stack */
1012
+ sib1(grammar)->key = n;
1013
+ }
1014
+ }
1015
+
1016
+
1017
+ static TTree *newgrammar (lua_State *L, int arg) {
1018
+ int treesize;
1019
+ int frule = lua_gettop(L) + 2; /* position of first rule's key */
1020
+ int n = collectrules(L, arg, &treesize);
1021
+ TTree *g = newtree(L, treesize);
1022
+ luaL_argcheck(L, n <= MAXRULES, arg, "grammar has too many rules");
1023
+ g->tag = TGrammar; g->u.n = n;
1024
+ lua_newtable(L); /* create 'ktable' */
1025
+ lua_setfenv(L, -2);
1026
+ buildgrammar(L, g, frule, n);
1027
+ lua_getfenv(L, -1); /* get 'ktable' for new tree */
1028
+ finalfix(L, frule - 1, g, sib1(g));
1029
+ initialrulename(L, g, frule);
1030
+ verifygrammar(L, g);
1031
+ lua_pop(L, 1); /* remove 'ktable' */
1032
+ lua_insert(L, -(n * 2 + 2)); /* move new table to proper position */
1033
+ lua_pop(L, n * 2 + 1); /* remove position table + rule pairs */
1034
+ return g; /* new table at the top of the stack */
1035
+ }
1036
+
1037
+ /* }====================================================== */
1038
+
1039
+
1040
+ static LpegInstruction *prepcompile (lua_State *L, Pattern *p, int idx) {
1041
+ lua_getfenv(L, idx); /* push 'ktable' (may be used by 'finalfix') */
1042
+ finalfix(L, 0, NULL, p->tree);
1043
+ lua_pop(L, 1); /* remove 'ktable' */
1044
+ return compile(L, p);
1045
+ }
1046
+
1047
+
1048
+ static int lp_printtree (lua_State *L) {
1049
+ TTree *tree = getpatt(L, 1, NULL);
1050
+ int c = lua_toboolean(L, 2);
1051
+ if (c) {
1052
+ lua_getfenv(L, 1); /* push 'ktable' (may be used by 'finalfix') */
1053
+ finalfix(L, 0, NULL, tree);
1054
+ lua_pop(L, 1); /* remove 'ktable' */
1055
+ }
1056
+ printktable(L, 1);
1057
+ printtree(tree, 0);
1058
+ return 0;
1059
+ }
1060
+
1061
+
1062
+ static int lp_printcode (lua_State *L) {
1063
+ Pattern *p = getpattern(L, 1);
1064
+ printktable(L, 1);
1065
+ if (p->code == NULL) /* not compiled yet? */
1066
+ prepcompile(L, p, 1);
1067
+ printpatt(p->code, p->codesize);
1068
+ return 0;
1069
+ }
1070
+
1071
+
1072
+ /*
1073
+ ** Get the initial position for the match, interpreting negative
1074
+ ** values from the end of the subject
1075
+ */
1076
+ static size_t initposition (lua_State *L, size_t len) {
1077
+ lua_Integer ii = luaL_optinteger(L, 3, 1);
1078
+ if (ii > 0) { /* positive index? */
1079
+ if ((size_t)ii <= len) /* inside the string? */
1080
+ return (size_t)ii - 1; /* return it (corrected to 0-base) */
1081
+ else return len; /* crop at the end */
1082
+ }
1083
+ else { /* negative index */
1084
+ if ((size_t)(-ii) <= len) /* inside the string? */
1085
+ return len - ((size_t)(-ii)); /* return position from the end */
1086
+ else return 0; /* crop at the beginning */
1087
+ }
1088
+ }
1089
+
1090
+
1091
+ /*
1092
+ ** Main match function
1093
+ */
1094
+ static int lp_match (lua_State *L) {
1095
+ Capture capture[INITCAPSIZE];
1096
+ const char *r;
1097
+ size_t l;
1098
+ Pattern *p = (getpatt(L, 1, NULL), getpattern(L, 1));
1099
+ LpegInstruction *code = (p->code != NULL) ? p->code : prepcompile(L, p, 1);
1100
+ const char *s = luaL_checklstring(L, SUBJIDX, &l);
1101
+ size_t i = initposition(L, l);
1102
+ int ptop = lua_gettop(L);
1103
+ lua_pushnil(L); /* initialize subscache */
1104
+ lua_pushlightuserdata(L, capture); /* initialize caplistidx */
1105
+ lua_getfenv(L, 1); /* initialize penvidx */
1106
+ r = lpeg_match(L, s, s + i, s + l, code, capture, ptop);
1107
+ if (r == NULL) {
1108
+ lua_pushnil(L);
1109
+ return 1;
1110
+ }
1111
+ return getcaptures(L, s, r, ptop);
1112
+ }
1113
+
1114
+
1115
+
1116
+ /*
1117
+ ** {======================================================
1118
+ ** Library creation and functions not related to matching
1119
+ ** =======================================================
1120
+ */
1121
+
1122
+ static int lp_setmax (lua_State *L) {
1123
+ luaL_optinteger(L, 1, -1);
1124
+ lua_settop(L, 1);
1125
+ lua_setfield(L, LUA_REGISTRYINDEX, MAXSTACKIDX);
1126
+ return 0;
1127
+ }
1128
+
1129
+
1130
+ static int lp_version (lua_State *L) {
1131
+ lua_pushstring(L, VERSION);
1132
+ return 1;
1133
+ }
1134
+
1135
+
1136
+ static int lp_type (lua_State *L) {
1137
+ if (testpattern(L, 1))
1138
+ lua_pushliteral(L, "pattern");
1139
+ else
1140
+ lua_pushnil(L);
1141
+ return 1;
1142
+ }
1143
+
1144
+
1145
+ int lp_gc (lua_State *L) {
1146
+ Pattern *p = getpattern(L, 1);
1147
+ if (p->codesize > 0)
1148
+ realloccode(L, p, 0);
1149
+ return 0;
1150
+ }
1151
+
1152
+
1153
+ static void createcat (lua_State *L, const char *catname, int (catf) (int)) {
1154
+ TTree *t = newcharset(L);
1155
+ int i;
1156
+ for (i = 0; i <= UCHAR_MAX; i++)
1157
+ if (catf(i)) setchar(treebuffer(t), i);
1158
+ lua_setfield(L, -2, catname);
1159
+ }
1160
+
1161
+
1162
+ static int lp_locale (lua_State *L) {
1163
+ if (lua_isnoneornil(L, 1)) {
1164
+ lua_settop(L, 0);
1165
+ lua_createtable(L, 0, 12);
1166
+ }
1167
+ else {
1168
+ luaL_checktype(L, 1, LUA_TTABLE);
1169
+ lua_settop(L, 1);
1170
+ }
1171
+ createcat(L, "alnum", isalnum);
1172
+ createcat(L, "alpha", isalpha);
1173
+ createcat(L, "cntrl", iscntrl);
1174
+ createcat(L, "digit", isdigit);
1175
+ createcat(L, "graph", isgraph);
1176
+ createcat(L, "lower", islower);
1177
+ createcat(L, "print", isprint);
1178
+ createcat(L, "punct", ispunct);
1179
+ createcat(L, "space", isspace);
1180
+ createcat(L, "upper", isupper);
1181
+ createcat(L, "xdigit", isxdigit);
1182
+ return 1;
1183
+ }
1184
+
1185
+
1186
+ static struct luaL_Reg pattreg[] = {
1187
+ {"ptree", lp_printtree},
1188
+ {"pcode", lp_printcode},
1189
+ {"match", lp_match},
1190
+ {"B", lp_behind},
1191
+ {"V", lp_V},
1192
+ {"C", lp_simplecapture},
1193
+ {"Cc", lp_constcapture},
1194
+ {"Cmt", lp_matchtime},
1195
+ {"Cb", lp_backref},
1196
+ {"Carg", lp_argcapture},
1197
+ {"Cp", lp_poscapture},
1198
+ {"Cs", lp_substcapture},
1199
+ {"Ct", lp_tablecapture},
1200
+ {"Cf", lp_foldcapture},
1201
+ {"Cg", lp_groupcapture},
1202
+ {"P", lp_P},
1203
+ {"S", lp_set},
1204
+ {"R", lp_range},
1205
+ {"locale", lp_locale},
1206
+ {"version", lp_version},
1207
+ {"setmaxstack", lp_setmax},
1208
+ {"type", lp_type},
1209
+ {NULL, NULL}
1210
+ };
1211
+
1212
+
1213
+ static struct luaL_Reg metareg[] = {
1214
+ {"__mul", lp_seq},
1215
+ {"__add", lp_choice},
1216
+ {"__pow", lp_star},
1217
+ {"__gc", lp_gc},
1218
+ {"__len", lp_and},
1219
+ {"__div", lp_divcapture},
1220
+ {"__unm", lp_not},
1221
+ {"__sub", lp_sub},
1222
+ {NULL, NULL}
1223
+ };
1224
+
1225
+
1226
+ int luaopen_lpeg (lua_State *L);
1227
+ int luaopen_lpeg (lua_State *L) {
1228
+ luaL_newmetatable(L, PATTERN_T);
1229
+ lua_pushnumber(L, MAXBACK); /* initialize maximum backtracking */
1230
+ lua_setfield(L, LUA_REGISTRYINDEX, MAXSTACKIDX);
1231
+ luaL_register(L, NULL, metareg);
1232
+ luaL_register(L, "lpeg", pattreg);
1233
+ lua_pushvalue(L, -1);
1234
+ lua_setfield(L, -3, "__index");
1235
+ return 1;
1236
+ }
1237
+
1238
+ /* }====================================================== */