| Guido van Rossum | f70e43a | 1991-02-19 12:39:46 +0000 | [diff] [blame] | 1 |  | 
| Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 2 | /* Grammar implementation */ | 
 | 3 |  | 
| Tim Peters | 1ca1296 | 2001-12-04 03:18:48 +0000 | [diff] [blame] | 4 | #include "Python.h" | 
| Guido van Rossum | 3f5da24 | 1990-12-20 15:06:42 +0000 | [diff] [blame] | 5 | #include "pgenheaders.h" | 
 | 6 |  | 
| Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 7 | #include <ctype.h> | 
 | 8 |  | 
| Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 9 | #include "token.h" | 
 | 10 | #include "grammar.h" | 
 | 11 |  | 
| Guido van Rossum | 86bea46 | 1997-04-29 21:03:06 +0000 | [diff] [blame] | 12 | extern int Py_DebugFlag; | 
| Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 13 |  | 
 | 14 | grammar * | 
| Thomas Wouters | 23c9e00 | 2000-07-22 19:20:54 +0000 | [diff] [blame] | 15 | newgrammar(int start) | 
| Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 16 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 17 |     grammar *g; | 
 | 18 |  | 
 | 19 |     g = (grammar *)PyObject_MALLOC(sizeof(grammar)); | 
 | 20 |     if (g == NULL) | 
 | 21 |         Py_FatalError("no mem for new grammar"); | 
 | 22 |     g->g_ndfas = 0; | 
 | 23 |     g->g_dfa = NULL; | 
 | 24 |     g->g_start = start; | 
 | 25 |     g->g_ll.ll_nlabels = 0; | 
 | 26 |     g->g_ll.ll_label = NULL; | 
 | 27 |     g->g_accel = 0; | 
 | 28 |     return g; | 
| Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 29 | } | 
 | 30 |  | 
 | 31 | dfa * | 
| Thomas Wouters | 23c9e00 | 2000-07-22 19:20:54 +0000 | [diff] [blame] | 32 | adddfa(grammar *g, int type, char *name) | 
| Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 33 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 34 |     dfa *d; | 
 | 35 |  | 
 | 36 |     g->g_dfa = (dfa *)PyObject_REALLOC(g->g_dfa, | 
 | 37 |                                         sizeof(dfa) * (g->g_ndfas + 1)); | 
 | 38 |     if (g->g_dfa == NULL) | 
 | 39 |         Py_FatalError("no mem to resize dfa in adddfa"); | 
 | 40 |     d = &g->g_dfa[g->g_ndfas++]; | 
 | 41 |     d->d_type = type; | 
 | 42 |     d->d_name = strdup(name); | 
 | 43 |     d->d_nstates = 0; | 
 | 44 |     d->d_state = NULL; | 
 | 45 |     d->d_initial = -1; | 
 | 46 |     d->d_first = NULL; | 
 | 47 |     return d; /* Only use while fresh! */ | 
| Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 48 | } | 
 | 49 |  | 
 | 50 | int | 
| Thomas Wouters | 23c9e00 | 2000-07-22 19:20:54 +0000 | [diff] [blame] | 51 | addstate(dfa *d) | 
| Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 52 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 53 |     state *s; | 
 | 54 |  | 
 | 55 |     d->d_state = (state *)PyObject_REALLOC(d->d_state, | 
 | 56 |                                   sizeof(state) * (d->d_nstates + 1)); | 
 | 57 |     if (d->d_state == NULL) | 
 | 58 |         Py_FatalError("no mem to resize state in addstate"); | 
 | 59 |     s = &d->d_state[d->d_nstates++]; | 
 | 60 |     s->s_narcs = 0; | 
 | 61 |     s->s_arc = NULL; | 
 | 62 |     s->s_lower = 0; | 
 | 63 |     s->s_upper = 0; | 
 | 64 |     s->s_accel = NULL; | 
 | 65 |     s->s_accept = 0; | 
 | 66 |     return s - d->d_state; | 
| Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 67 | } | 
 | 68 |  | 
 | 69 | void | 
| Thomas Wouters | 23c9e00 | 2000-07-22 19:20:54 +0000 | [diff] [blame] | 70 | addarc(dfa *d, int from, int to, int lbl) | 
| Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 71 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 72 |     state *s; | 
 | 73 |     arc *a; | 
 | 74 |  | 
 | 75 |     assert(0 <= from && from < d->d_nstates); | 
 | 76 |     assert(0 <= to && to < d->d_nstates); | 
 | 77 |  | 
 | 78 |     s = &d->d_state[from]; | 
 | 79 |     s->s_arc = (arc *)PyObject_REALLOC(s->s_arc, sizeof(arc) * (s->s_narcs + 1)); | 
 | 80 |     if (s->s_arc == NULL) | 
 | 81 |         Py_FatalError("no mem to resize arc list in addarc"); | 
 | 82 |     a = &s->s_arc[s->s_narcs++]; | 
 | 83 |     a->a_lbl = lbl; | 
 | 84 |     a->a_arrow = to; | 
| Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 85 | } | 
 | 86 |  | 
 | 87 | int | 
| Thomas Wouters | 23c9e00 | 2000-07-22 19:20:54 +0000 | [diff] [blame] | 88 | addlabel(labellist *ll, int type, char *str) | 
| Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 89 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 90 |     int i; | 
 | 91 |     label *lb; | 
 | 92 |  | 
 | 93 |     for (i = 0; i < ll->ll_nlabels; i++) { | 
 | 94 |         if (ll->ll_label[i].lb_type == type && | 
 | 95 |             strcmp(ll->ll_label[i].lb_str, str) == 0) | 
 | 96 |             return i; | 
 | 97 |     } | 
 | 98 |     ll->ll_label = (label *)PyObject_REALLOC(ll->ll_label, | 
 | 99 |                                     sizeof(label) * (ll->ll_nlabels + 1)); | 
 | 100 |     if (ll->ll_label == NULL) | 
 | 101 |         Py_FatalError("no mem to resize labellist in addlabel"); | 
 | 102 |     lb = &ll->ll_label[ll->ll_nlabels++]; | 
 | 103 |     lb->lb_type = type; | 
 | 104 |     lb->lb_str = strdup(str); | 
 | 105 |     if (Py_DebugFlag) | 
 | 106 |         printf("Label @ %8p, %d: %s\n", ll, ll->ll_nlabels, | 
 | 107 |                PyGrammar_LabelRepr(lb)); | 
 | 108 |     return lb - ll->ll_label; | 
| Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 109 | } | 
 | 110 |  | 
 | 111 | /* Same, but rather dies than adds */ | 
 | 112 |  | 
 | 113 | int | 
| Thomas Wouters | 23c9e00 | 2000-07-22 19:20:54 +0000 | [diff] [blame] | 114 | findlabel(labellist *ll, int type, char *str) | 
| Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 115 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 116 |     int i; | 
 | 117 |  | 
 | 118 |     for (i = 0; i < ll->ll_nlabels; i++) { | 
 | 119 |         if (ll->ll_label[i].lb_type == type /*&& | 
 | 120 |             strcmp(ll->ll_label[i].lb_str, str) == 0*/) | 
 | 121 |             return i; | 
 | 122 |     } | 
 | 123 |     fprintf(stderr, "Label %d/'%s' not found\n", type, str); | 
 | 124 |     Py_FatalError("grammar.c:findlabel()"); | 
 | 125 |     return 0; /* Make gcc -Wall happy */ | 
| Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 126 | } | 
 | 127 |  | 
| Guido van Rossum | 3f5da24 | 1990-12-20 15:06:42 +0000 | [diff] [blame] | 128 | /* Forward */ | 
| Tim Peters | dbd9ba6 | 2000-07-09 03:09:57 +0000 | [diff] [blame] | 129 | static void translabel(grammar *, label *); | 
| Guido van Rossum | 3f5da24 | 1990-12-20 15:06:42 +0000 | [diff] [blame] | 130 |  | 
 | 131 | void | 
| Thomas Wouters | 23c9e00 | 2000-07-22 19:20:54 +0000 | [diff] [blame] | 132 | translatelabels(grammar *g) | 
| Guido van Rossum | 3f5da24 | 1990-12-20 15:06:42 +0000 | [diff] [blame] | 133 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 134 |     int i; | 
| Guido van Rossum | 588633d | 1994-12-30 15:46:02 +0000 | [diff] [blame] | 135 |  | 
| Guido van Rossum | 408027e | 1996-12-30 16:17:54 +0000 | [diff] [blame] | 136 | #ifdef Py_DEBUG | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 137 |     printf("Translating labels ...\n"); | 
| Guido van Rossum | 588633d | 1994-12-30 15:46:02 +0000 | [diff] [blame] | 138 | #endif | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 139 |     /* Don't translate EMPTY */ | 
 | 140 |     for (i = EMPTY+1; i < g->g_ll.ll_nlabels; i++) | 
 | 141 |         translabel(g, &g->g_ll.ll_label[i]); | 
| Guido van Rossum | 3f5da24 | 1990-12-20 15:06:42 +0000 | [diff] [blame] | 142 | } | 
 | 143 |  | 
| Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 144 | static void | 
| Thomas Wouters | 23c9e00 | 2000-07-22 19:20:54 +0000 | [diff] [blame] | 145 | translabel(grammar *g, label *lb) | 
| Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 146 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 147 |     int i; | 
 | 148 |  | 
 | 149 |     if (Py_DebugFlag) | 
 | 150 |         printf("Translating label %s ...\n", PyGrammar_LabelRepr(lb)); | 
 | 151 |  | 
 | 152 |     if (lb->lb_type == NAME) { | 
 | 153 |         for (i = 0; i < g->g_ndfas; i++) { | 
 | 154 |             if (strcmp(lb->lb_str, g->g_dfa[i].d_name) == 0) { | 
 | 155 |                 if (Py_DebugFlag) | 
 | 156 |                     printf( | 
 | 157 |                         "Label %s is non-terminal %d.\n", | 
 | 158 |                         lb->lb_str, | 
 | 159 |                         g->g_dfa[i].d_type); | 
 | 160 |                 lb->lb_type = g->g_dfa[i].d_type; | 
 | 161 |                 free(lb->lb_str); | 
 | 162 |                 lb->lb_str = NULL; | 
 | 163 |                 return; | 
 | 164 |             } | 
 | 165 |         } | 
 | 166 |         for (i = 0; i < (int)N_TOKENS; i++) { | 
 | 167 |             if (strcmp(lb->lb_str, _PyParser_TokenNames[i]) == 0) { | 
 | 168 |                 if (Py_DebugFlag) | 
 | 169 |                     printf("Label %s is terminal %d.\n", | 
 | 170 |                         lb->lb_str, i); | 
 | 171 |                 lb->lb_type = i; | 
 | 172 |                 free(lb->lb_str); | 
 | 173 |                 lb->lb_str = NULL; | 
 | 174 |                 return; | 
 | 175 |             } | 
 | 176 |         } | 
 | 177 |         printf("Can't translate NAME label '%s'\n", lb->lb_str); | 
 | 178 |         return; | 
 | 179 |     } | 
 | 180 |  | 
 | 181 |     if (lb->lb_type == STRING) { | 
 | 182 |         if (isalpha(Py_CHARMASK(lb->lb_str[1])) || | 
 | 183 |             lb->lb_str[1] == '_') { | 
 | 184 |             char *p; | 
 | 185 |             char *src; | 
 | 186 |             char *dest; | 
 | 187 |             size_t name_len; | 
 | 188 |             if (Py_DebugFlag) | 
 | 189 |                 printf("Label %s is a keyword\n", lb->lb_str); | 
 | 190 |             lb->lb_type = NAME; | 
 | 191 |             src = lb->lb_str + 1; | 
 | 192 |             p = strchr(src, '\''); | 
 | 193 |             if (p) | 
 | 194 |                 name_len = p - src; | 
 | 195 |             else | 
 | 196 |                 name_len = strlen(src); | 
 | 197 |             dest = (char *)malloc(name_len + 1); | 
 | 198 |             if (!dest) { | 
 | 199 |                 printf("Can't alloc dest '%s'\n", src); | 
 | 200 |                 return; | 
 | 201 |             } | 
 | 202 |             strncpy(dest, src, name_len); | 
 | 203 |             dest[name_len] = '\0'; | 
 | 204 |             free(lb->lb_str); | 
 | 205 |             lb->lb_str = dest; | 
 | 206 |         } | 
 | 207 |         else if (lb->lb_str[2] == lb->lb_str[0]) { | 
 | 208 |             int type = (int) PyToken_OneChar(lb->lb_str[1]); | 
 | 209 |             if (type != OP) { | 
 | 210 |                 lb->lb_type = type; | 
 | 211 |                 free(lb->lb_str); | 
 | 212 |                 lb->lb_str = NULL; | 
 | 213 |             } | 
 | 214 |             else | 
 | 215 |                 printf("Unknown OP label %s\n", | 
 | 216 |                     lb->lb_str); | 
 | 217 |         } | 
 | 218 |         else if (lb->lb_str[2] && lb->lb_str[3] == lb->lb_str[0]) { | 
 | 219 |             int type = (int) PyToken_TwoChars(lb->lb_str[1], | 
 | 220 |                                        lb->lb_str[2]); | 
 | 221 |             if (type != OP) { | 
 | 222 |                 lb->lb_type = type; | 
 | 223 |                 free(lb->lb_str); | 
 | 224 |                 lb->lb_str = NULL; | 
 | 225 |             } | 
 | 226 |             else | 
 | 227 |                 printf("Unknown OP label %s\n", | 
 | 228 |                     lb->lb_str); | 
 | 229 |         } | 
 | 230 |         else if (lb->lb_str[2] && lb->lb_str[3] && lb->lb_str[4] == lb->lb_str[0]) { | 
 | 231 |             int type = (int) PyToken_ThreeChars(lb->lb_str[1], | 
 | 232 |                                                 lb->lb_str[2], | 
 | 233 |                                                 lb->lb_str[3]); | 
 | 234 |             if (type != OP) { | 
 | 235 |                 lb->lb_type = type; | 
 | 236 |                 free(lb->lb_str); | 
 | 237 |                 lb->lb_str = NULL; | 
 | 238 |             } | 
 | 239 |             else | 
 | 240 |                 printf("Unknown OP label %s\n", | 
 | 241 |                     lb->lb_str); | 
 | 242 |         } | 
 | 243 |         else | 
 | 244 |             printf("Can't translate STRING label %s\n", | 
 | 245 |                 lb->lb_str); | 
 | 246 |     } | 
 | 247 |     else | 
 | 248 |         printf("Can't translate label '%s'\n", | 
 | 249 |                PyGrammar_LabelRepr(lb)); | 
| Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 250 | } |