| 1 | n/a | |
|---|
| 2 | n/a | /* Grammar implementation */ |
|---|
| 3 | n/a | |
|---|
| 4 | n/a | #include "Python.h" |
|---|
| 5 | n/a | #include "pgenheaders.h" |
|---|
| 6 | n/a | |
|---|
| 7 | n/a | #include <ctype.h> |
|---|
| 8 | n/a | |
|---|
| 9 | n/a | #include "token.h" |
|---|
| 10 | n/a | #include "grammar.h" |
|---|
| 11 | n/a | |
|---|
| 12 | n/a | extern int Py_DebugFlag; |
|---|
| 13 | n/a | |
|---|
| 14 | n/a | grammar * |
|---|
| 15 | n/a | newgrammar(int start) |
|---|
| 16 | n/a | { |
|---|
| 17 | n/a | grammar *g; |
|---|
| 18 | n/a | |
|---|
| 19 | n/a | g = (grammar *)PyObject_MALLOC(sizeof(grammar)); |
|---|
| 20 | n/a | if (g == NULL) |
|---|
| 21 | n/a | Py_FatalError("no mem for new grammar"); |
|---|
| 22 | n/a | g->g_ndfas = 0; |
|---|
| 23 | n/a | g->g_dfa = NULL; |
|---|
| 24 | n/a | g->g_start = start; |
|---|
| 25 | n/a | g->g_ll.ll_nlabels = 0; |
|---|
| 26 | n/a | g->g_ll.ll_label = NULL; |
|---|
| 27 | n/a | g->g_accel = 0; |
|---|
| 28 | n/a | return g; |
|---|
| 29 | n/a | } |
|---|
| 30 | n/a | |
|---|
| 31 | n/a | void |
|---|
| 32 | n/a | freegrammar(grammar *g) |
|---|
| 33 | n/a | { |
|---|
| 34 | n/a | int i; |
|---|
| 35 | n/a | for (i = 0; i < g->g_ndfas; i++) { |
|---|
| 36 | n/a | free(g->g_dfa[i].d_name); |
|---|
| 37 | n/a | for (int j = 0; j < g->g_dfa[i].d_nstates; j++) |
|---|
| 38 | n/a | PyObject_FREE(g->g_dfa[i].d_state[j].s_arc); |
|---|
| 39 | n/a | PyObject_FREE(g->g_dfa[i].d_state); |
|---|
| 40 | n/a | } |
|---|
| 41 | n/a | PyObject_FREE(g->g_dfa); |
|---|
| 42 | n/a | for (i = 0; i < g->g_ll.ll_nlabels; i++) |
|---|
| 43 | n/a | free(g->g_ll.ll_label[i].lb_str); |
|---|
| 44 | n/a | PyObject_FREE(g->g_ll.ll_label); |
|---|
| 45 | n/a | PyObject_FREE(g); |
|---|
| 46 | n/a | } |
|---|
| 47 | n/a | |
|---|
| 48 | n/a | dfa * |
|---|
| 49 | n/a | adddfa(grammar *g, int type, const char *name) |
|---|
| 50 | n/a | { |
|---|
| 51 | n/a | dfa *d; |
|---|
| 52 | n/a | |
|---|
| 53 | n/a | g->g_dfa = (dfa *)PyObject_REALLOC(g->g_dfa, |
|---|
| 54 | n/a | sizeof(dfa) * (g->g_ndfas + 1)); |
|---|
| 55 | n/a | if (g->g_dfa == NULL) |
|---|
| 56 | n/a | Py_FatalError("no mem to resize dfa in adddfa"); |
|---|
| 57 | n/a | d = &g->g_dfa[g->g_ndfas++]; |
|---|
| 58 | n/a | d->d_type = type; |
|---|
| 59 | n/a | d->d_name = strdup(name); |
|---|
| 60 | n/a | d->d_nstates = 0; |
|---|
| 61 | n/a | d->d_state = NULL; |
|---|
| 62 | n/a | d->d_initial = -1; |
|---|
| 63 | n/a | d->d_first = NULL; |
|---|
| 64 | n/a | return d; /* Only use while fresh! */ |
|---|
| 65 | n/a | } |
|---|
| 66 | n/a | |
|---|
| 67 | n/a | int |
|---|
| 68 | n/a | addstate(dfa *d) |
|---|
| 69 | n/a | { |
|---|
| 70 | n/a | state *s; |
|---|
| 71 | n/a | |
|---|
| 72 | n/a | d->d_state = (state *)PyObject_REALLOC(d->d_state, |
|---|
| 73 | n/a | sizeof(state) * (d->d_nstates + 1)); |
|---|
| 74 | n/a | if (d->d_state == NULL) |
|---|
| 75 | n/a | Py_FatalError("no mem to resize state in addstate"); |
|---|
| 76 | n/a | s = &d->d_state[d->d_nstates++]; |
|---|
| 77 | n/a | s->s_narcs = 0; |
|---|
| 78 | n/a | s->s_arc = NULL; |
|---|
| 79 | n/a | s->s_lower = 0; |
|---|
| 80 | n/a | s->s_upper = 0; |
|---|
| 81 | n/a | s->s_accel = NULL; |
|---|
| 82 | n/a | s->s_accept = 0; |
|---|
| 83 | n/a | return Py_SAFE_DOWNCAST(s - d->d_state, intptr_t, int); |
|---|
| 84 | n/a | } |
|---|
| 85 | n/a | |
|---|
| 86 | n/a | void |
|---|
| 87 | n/a | addarc(dfa *d, int from, int to, int lbl) |
|---|
| 88 | n/a | { |
|---|
| 89 | n/a | state *s; |
|---|
| 90 | n/a | arc *a; |
|---|
| 91 | n/a | |
|---|
| 92 | n/a | assert(0 <= from && from < d->d_nstates); |
|---|
| 93 | n/a | assert(0 <= to && to < d->d_nstates); |
|---|
| 94 | n/a | |
|---|
| 95 | n/a | s = &d->d_state[from]; |
|---|
| 96 | n/a | s->s_arc = (arc *)PyObject_REALLOC(s->s_arc, sizeof(arc) * (s->s_narcs + 1)); |
|---|
| 97 | n/a | if (s->s_arc == NULL) |
|---|
| 98 | n/a | Py_FatalError("no mem to resize arc list in addarc"); |
|---|
| 99 | n/a | a = &s->s_arc[s->s_narcs++]; |
|---|
| 100 | n/a | a->a_lbl = lbl; |
|---|
| 101 | n/a | a->a_arrow = to; |
|---|
| 102 | n/a | } |
|---|
| 103 | n/a | |
|---|
| 104 | n/a | int |
|---|
| 105 | n/a | addlabel(labellist *ll, int type, const char *str) |
|---|
| 106 | n/a | { |
|---|
| 107 | n/a | int i; |
|---|
| 108 | n/a | label *lb; |
|---|
| 109 | n/a | |
|---|
| 110 | n/a | for (i = 0; i < ll->ll_nlabels; i++) { |
|---|
| 111 | n/a | if (ll->ll_label[i].lb_type == type && |
|---|
| 112 | n/a | strcmp(ll->ll_label[i].lb_str, str) == 0) |
|---|
| 113 | n/a | return i; |
|---|
| 114 | n/a | } |
|---|
| 115 | n/a | ll->ll_label = (label *)PyObject_REALLOC(ll->ll_label, |
|---|
| 116 | n/a | sizeof(label) * (ll->ll_nlabels + 1)); |
|---|
| 117 | n/a | if (ll->ll_label == NULL) |
|---|
| 118 | n/a | Py_FatalError("no mem to resize labellist in addlabel"); |
|---|
| 119 | n/a | lb = &ll->ll_label[ll->ll_nlabels++]; |
|---|
| 120 | n/a | lb->lb_type = type; |
|---|
| 121 | n/a | lb->lb_str = strdup(str); |
|---|
| 122 | n/a | if (Py_DebugFlag) |
|---|
| 123 | n/a | printf("Label @ %8p, %d: %s\n", ll, ll->ll_nlabels, |
|---|
| 124 | n/a | PyGrammar_LabelRepr(lb)); |
|---|
| 125 | n/a | return Py_SAFE_DOWNCAST(lb - ll->ll_label, intptr_t, int); |
|---|
| 126 | n/a | } |
|---|
| 127 | n/a | |
|---|
| 128 | n/a | /* Same, but rather dies than adds */ |
|---|
| 129 | n/a | |
|---|
| 130 | n/a | int |
|---|
| 131 | n/a | findlabel(labellist *ll, int type, const char *str) |
|---|
| 132 | n/a | { |
|---|
| 133 | n/a | int i; |
|---|
| 134 | n/a | |
|---|
| 135 | n/a | for (i = 0; i < ll->ll_nlabels; i++) { |
|---|
| 136 | n/a | if (ll->ll_label[i].lb_type == type /*&& |
|---|
| 137 | n/a | strcmp(ll->ll_label[i].lb_str, str) == 0*/) |
|---|
| 138 | n/a | return i; |
|---|
| 139 | n/a | } |
|---|
| 140 | n/a | fprintf(stderr, "Label %d/'%s' not found\n", type, str); |
|---|
| 141 | n/a | Py_FatalError("grammar.c:findlabel()"); |
|---|
| 142 | n/a | |
|---|
| 143 | n/a | /* Py_FatalError() is declared with __attribute__((__noreturn__)). |
|---|
| 144 | n/a | GCC emits a warning without "return 0;" (compiler bug!), but Clang is |
|---|
| 145 | n/a | smarter and emits a warning on the return... */ |
|---|
| 146 | n/a | #ifndef __clang__ |
|---|
| 147 | n/a | return 0; /* Make gcc -Wall happy */ |
|---|
| 148 | n/a | #endif |
|---|
| 149 | n/a | } |
|---|
| 150 | n/a | |
|---|
| 151 | n/a | /* Forward */ |
|---|
| 152 | n/a | static void translabel(grammar *, label *); |
|---|
| 153 | n/a | |
|---|
| 154 | n/a | void |
|---|
| 155 | n/a | translatelabels(grammar *g) |
|---|
| 156 | n/a | { |
|---|
| 157 | n/a | int i; |
|---|
| 158 | n/a | |
|---|
| 159 | n/a | #ifdef Py_DEBUG |
|---|
| 160 | n/a | printf("Translating labels ...\n"); |
|---|
| 161 | n/a | #endif |
|---|
| 162 | n/a | /* Don't translate EMPTY */ |
|---|
| 163 | n/a | for (i = EMPTY+1; i < g->g_ll.ll_nlabels; i++) |
|---|
| 164 | n/a | translabel(g, &g->g_ll.ll_label[i]); |
|---|
| 165 | n/a | } |
|---|
| 166 | n/a | |
|---|
| 167 | n/a | static void |
|---|
| 168 | n/a | translabel(grammar *g, label *lb) |
|---|
| 169 | n/a | { |
|---|
| 170 | n/a | int i; |
|---|
| 171 | n/a | |
|---|
| 172 | n/a | if (Py_DebugFlag) |
|---|
| 173 | n/a | printf("Translating label %s ...\n", PyGrammar_LabelRepr(lb)); |
|---|
| 174 | n/a | |
|---|
| 175 | n/a | if (lb->lb_type == NAME) { |
|---|
| 176 | n/a | for (i = 0; i < g->g_ndfas; i++) { |
|---|
| 177 | n/a | if (strcmp(lb->lb_str, g->g_dfa[i].d_name) == 0) { |
|---|
| 178 | n/a | if (Py_DebugFlag) |
|---|
| 179 | n/a | printf( |
|---|
| 180 | n/a | "Label %s is non-terminal %d.\n", |
|---|
| 181 | n/a | lb->lb_str, |
|---|
| 182 | n/a | g->g_dfa[i].d_type); |
|---|
| 183 | n/a | lb->lb_type = g->g_dfa[i].d_type; |
|---|
| 184 | n/a | free(lb->lb_str); |
|---|
| 185 | n/a | lb->lb_str = NULL; |
|---|
| 186 | n/a | return; |
|---|
| 187 | n/a | } |
|---|
| 188 | n/a | } |
|---|
| 189 | n/a | for (i = 0; i < (int)N_TOKENS; i++) { |
|---|
| 190 | n/a | if (strcmp(lb->lb_str, _PyParser_TokenNames[i]) == 0) { |
|---|
| 191 | n/a | if (Py_DebugFlag) |
|---|
| 192 | n/a | printf("Label %s is terminal %d.\n", |
|---|
| 193 | n/a | lb->lb_str, i); |
|---|
| 194 | n/a | lb->lb_type = i; |
|---|
| 195 | n/a | free(lb->lb_str); |
|---|
| 196 | n/a | lb->lb_str = NULL; |
|---|
| 197 | n/a | return; |
|---|
| 198 | n/a | } |
|---|
| 199 | n/a | } |
|---|
| 200 | n/a | printf("Can't translate NAME label '%s'\n", lb->lb_str); |
|---|
| 201 | n/a | return; |
|---|
| 202 | n/a | } |
|---|
| 203 | n/a | |
|---|
| 204 | n/a | if (lb->lb_type == STRING) { |
|---|
| 205 | n/a | if (isalpha(Py_CHARMASK(lb->lb_str[1])) || |
|---|
| 206 | n/a | lb->lb_str[1] == '_') { |
|---|
| 207 | n/a | char *p; |
|---|
| 208 | n/a | char *src; |
|---|
| 209 | n/a | char *dest; |
|---|
| 210 | n/a | size_t name_len; |
|---|
| 211 | n/a | if (Py_DebugFlag) |
|---|
| 212 | n/a | printf("Label %s is a keyword\n", lb->lb_str); |
|---|
| 213 | n/a | lb->lb_type = NAME; |
|---|
| 214 | n/a | src = lb->lb_str + 1; |
|---|
| 215 | n/a | p = strchr(src, '\''); |
|---|
| 216 | n/a | if (p) |
|---|
| 217 | n/a | name_len = p - src; |
|---|
| 218 | n/a | else |
|---|
| 219 | n/a | name_len = strlen(src); |
|---|
| 220 | n/a | dest = (char *)malloc(name_len + 1); |
|---|
| 221 | n/a | if (!dest) { |
|---|
| 222 | n/a | printf("Can't alloc dest '%s'\n", src); |
|---|
| 223 | n/a | return; |
|---|
| 224 | n/a | } |
|---|
| 225 | n/a | strncpy(dest, src, name_len); |
|---|
| 226 | n/a | dest[name_len] = '\0'; |
|---|
| 227 | n/a | free(lb->lb_str); |
|---|
| 228 | n/a | lb->lb_str = dest; |
|---|
| 229 | n/a | } |
|---|
| 230 | n/a | else if (lb->lb_str[2] == lb->lb_str[0]) { |
|---|
| 231 | n/a | int type = (int) PyToken_OneChar(lb->lb_str[1]); |
|---|
| 232 | n/a | if (type != OP) { |
|---|
| 233 | n/a | lb->lb_type = type; |
|---|
| 234 | n/a | free(lb->lb_str); |
|---|
| 235 | n/a | lb->lb_str = NULL; |
|---|
| 236 | n/a | } |
|---|
| 237 | n/a | else |
|---|
| 238 | n/a | printf("Unknown OP label %s\n", |
|---|
| 239 | n/a | lb->lb_str); |
|---|
| 240 | n/a | } |
|---|
| 241 | n/a | else if (lb->lb_str[2] && lb->lb_str[3] == lb->lb_str[0]) { |
|---|
| 242 | n/a | int type = (int) PyToken_TwoChars(lb->lb_str[1], |
|---|
| 243 | n/a | lb->lb_str[2]); |
|---|
| 244 | n/a | if (type != OP) { |
|---|
| 245 | n/a | lb->lb_type = type; |
|---|
| 246 | n/a | free(lb->lb_str); |
|---|
| 247 | n/a | lb->lb_str = NULL; |
|---|
| 248 | n/a | } |
|---|
| 249 | n/a | else |
|---|
| 250 | n/a | printf("Unknown OP label %s\n", |
|---|
| 251 | n/a | lb->lb_str); |
|---|
| 252 | n/a | } |
|---|
| 253 | n/a | else if (lb->lb_str[2] && lb->lb_str[3] && lb->lb_str[4] == lb->lb_str[0]) { |
|---|
| 254 | n/a | int type = (int) PyToken_ThreeChars(lb->lb_str[1], |
|---|
| 255 | n/a | lb->lb_str[2], |
|---|
| 256 | n/a | lb->lb_str[3]); |
|---|
| 257 | n/a | if (type != OP) { |
|---|
| 258 | n/a | lb->lb_type = type; |
|---|
| 259 | n/a | free(lb->lb_str); |
|---|
| 260 | n/a | lb->lb_str = NULL; |
|---|
| 261 | n/a | } |
|---|
| 262 | n/a | else |
|---|
| 263 | n/a | printf("Unknown OP label %s\n", |
|---|
| 264 | n/a | lb->lb_str); |
|---|
| 265 | n/a | } |
|---|
| 266 | n/a | else |
|---|
| 267 | n/a | printf("Can't translate STRING label %s\n", |
|---|
| 268 | n/a | lb->lb_str); |
|---|
| 269 | n/a | } |
|---|
| 270 | n/a | else |
|---|
| 271 | n/a | printf("Can't translate label '%s'\n", |
|---|
| 272 | n/a | PyGrammar_LabelRepr(lb)); |
|---|
| 273 | n/a | } |
|---|