blob: e2cce28a8d6ae2b5c2be249785badc8a737d0998 [file] [log] [blame]
Guido van Rossumf70e43a1991-02-19 12:39:46 +00001
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00002/* Grammar implementation */
3
Tim Peters1ca12962001-12-04 03:18:48 +00004#include "Python.h"
Guido van Rossum3f5da241990-12-20 15:06:42 +00005#include "pgenheaders.h"
6
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00007#include <ctype.h>
8
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00009#include "token.h"
10#include "grammar.h"
11
Guido van Rossum86bea461997-04-29 21:03:06 +000012extern int Py_DebugFlag;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000013
14grammar *
Thomas Wouters23c9e002000-07-22 19:20:54 +000015newgrammar(int start)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000016{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000017 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 Rossum85a5fbb1990-10-14 12:07:46 +000029}
30
31dfa *
Serhiy Storchakac6792272013-10-19 21:03:34 +030032adddfa(grammar *g, int type, const char *name)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000033{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000034 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 Rossum85a5fbb1990-10-14 12:07:46 +000048}
49
50int
Thomas Wouters23c9e002000-07-22 19:20:54 +000051addstate(dfa *d)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000052{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000053 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;
Victor Stinner3a8a3332013-11-18 01:07:38 +010066 return Py_SAFE_DOWNCAST(s - d->d_state, Py_intptr_t, int);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000067}
68
69void
Thomas Wouters23c9e002000-07-22 19:20:54 +000070addarc(dfa *d, int from, int to, int lbl)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000071{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000072 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 Rossum85a5fbb1990-10-14 12:07:46 +000085}
86
87int
Serhiy Storchakac6792272013-10-19 21:03:34 +030088addlabel(labellist *ll, int type, const char *str)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000089{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000090 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));
Victor Stinner3a8a3332013-11-18 01:07:38 +0100108 return Py_SAFE_DOWNCAST(lb - ll->ll_label, Py_intptr_t, int);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000109}
110
111/* Same, but rather dies than adds */
112
113int
Serhiy Storchakac6792272013-10-19 21:03:34 +0300114findlabel(labellist *ll, int type, const char *str)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000115{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000116 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()");
Victor Stinner4bb31e92016-08-19 15:11:56 +0200125
126 /* Py_FatalError() is declared with __attribute__((__noreturn__)).
127 GCC emits a warning without "return 0;" (compiler bug!), but Clang is
128 smarter and emits a warning on the return... */
129#ifndef __clang__
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000130 return 0; /* Make gcc -Wall happy */
Victor Stinner4bb31e92016-08-19 15:11:56 +0200131#endif
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000132}
133
Guido van Rossum3f5da241990-12-20 15:06:42 +0000134/* Forward */
Tim Petersdbd9ba62000-07-09 03:09:57 +0000135static void translabel(grammar *, label *);
Guido van Rossum3f5da241990-12-20 15:06:42 +0000136
137void
Thomas Wouters23c9e002000-07-22 19:20:54 +0000138translatelabels(grammar *g)
Guido van Rossum3f5da241990-12-20 15:06:42 +0000139{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000140 int i;
Guido van Rossum588633d1994-12-30 15:46:02 +0000141
Guido van Rossum408027e1996-12-30 16:17:54 +0000142#ifdef Py_DEBUG
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000143 printf("Translating labels ...\n");
Guido van Rossum588633d1994-12-30 15:46:02 +0000144#endif
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000145 /* Don't translate EMPTY */
146 for (i = EMPTY+1; i < g->g_ll.ll_nlabels; i++)
147 translabel(g, &g->g_ll.ll_label[i]);
Guido van Rossum3f5da241990-12-20 15:06:42 +0000148}
149
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000150static void
Thomas Wouters23c9e002000-07-22 19:20:54 +0000151translabel(grammar *g, label *lb)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000152{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000153 int i;
154
155 if (Py_DebugFlag)
156 printf("Translating label %s ...\n", PyGrammar_LabelRepr(lb));
157
158 if (lb->lb_type == NAME) {
159 for (i = 0; i < g->g_ndfas; i++) {
160 if (strcmp(lb->lb_str, g->g_dfa[i].d_name) == 0) {
161 if (Py_DebugFlag)
162 printf(
163 "Label %s is non-terminal %d.\n",
164 lb->lb_str,
165 g->g_dfa[i].d_type);
166 lb->lb_type = g->g_dfa[i].d_type;
167 free(lb->lb_str);
168 lb->lb_str = NULL;
169 return;
170 }
171 }
172 for (i = 0; i < (int)N_TOKENS; i++) {
173 if (strcmp(lb->lb_str, _PyParser_TokenNames[i]) == 0) {
174 if (Py_DebugFlag)
175 printf("Label %s is terminal %d.\n",
176 lb->lb_str, i);
177 lb->lb_type = i;
178 free(lb->lb_str);
179 lb->lb_str = NULL;
180 return;
181 }
182 }
183 printf("Can't translate NAME label '%s'\n", lb->lb_str);
184 return;
185 }
186
187 if (lb->lb_type == STRING) {
188 if (isalpha(Py_CHARMASK(lb->lb_str[1])) ||
189 lb->lb_str[1] == '_') {
190 char *p;
191 char *src;
192 char *dest;
193 size_t name_len;
194 if (Py_DebugFlag)
195 printf("Label %s is a keyword\n", lb->lb_str);
196 lb->lb_type = NAME;
197 src = lb->lb_str + 1;
198 p = strchr(src, '\'');
199 if (p)
200 name_len = p - src;
201 else
202 name_len = strlen(src);
203 dest = (char *)malloc(name_len + 1);
204 if (!dest) {
205 printf("Can't alloc dest '%s'\n", src);
206 return;
207 }
208 strncpy(dest, src, name_len);
209 dest[name_len] = '\0';
210 free(lb->lb_str);
211 lb->lb_str = dest;
212 }
213 else if (lb->lb_str[2] == lb->lb_str[0]) {
214 int type = (int) PyToken_OneChar(lb->lb_str[1]);
215 if (type != OP) {
216 lb->lb_type = type;
217 free(lb->lb_str);
218 lb->lb_str = NULL;
219 }
220 else
221 printf("Unknown OP label %s\n",
222 lb->lb_str);
223 }
224 else if (lb->lb_str[2] && lb->lb_str[3] == lb->lb_str[0]) {
225 int type = (int) PyToken_TwoChars(lb->lb_str[1],
226 lb->lb_str[2]);
227 if (type != OP) {
228 lb->lb_type = type;
229 free(lb->lb_str);
230 lb->lb_str = NULL;
231 }
232 else
233 printf("Unknown OP label %s\n",
234 lb->lb_str);
235 }
236 else if (lb->lb_str[2] && lb->lb_str[3] && lb->lb_str[4] == lb->lb_str[0]) {
237 int type = (int) PyToken_ThreeChars(lb->lb_str[1],
238 lb->lb_str[2],
239 lb->lb_str[3]);
240 if (type != OP) {
241 lb->lb_type = type;
242 free(lb->lb_str);
243 lb->lb_str = NULL;
244 }
245 else
246 printf("Unknown OP label %s\n",
247 lb->lb_str);
248 }
249 else
250 printf("Can't translate STRING label %s\n",
251 lb->lb_str);
252 }
253 else
254 printf("Can't translate label '%s'\n",
255 PyGrammar_LabelRepr(lb));
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000256}