static void
printssdfa(int xx_nstates, ss_state *xx_state, int nbits,
labellist *ll, char *msg)
{
int i, ibit, iarc;
ss_state *yy;
ss_arc *zz;
printf("Subset DFA %s\n", msg);
for (i = 0; i < xx_nstates; i++) {
yy = &xx_state[i];
if (yy->ss_deleted)
continue;
printf(" Subset %d", i);
if (yy->ss_finish)
printf(" (finish)");
printf(" { ");
for (ibit = 0; ibit < nbits; ibit++) {
if (testbit(yy->ss_ss, ibit))
printf("%d ", ibit);
}
printf("}\n");
for (iarc = 0; iarc < yy->ss_narcs; iarc++) {
zz = &yy->ss_arc[iarc];
printf(" Arc to state %d, label %s\n",
zz->sa_arrow,
PyGrammar_LabelRepr(
&ll->ll_label[zz->sa_label]));
}
}
}
static void
dumpstate(labellist *ll, nfa *nf, int istate)
{
nfastate *st;
int i;
nfaarc *ar;
printf("%c%2d%c",
istate == nf->nf_start ? '*' : ' ',
istate,
istate == nf->nf_finish ? '.' : ' ');
st = &nf->nf_state[istate];
ar = st->st_arc;
for (i = 0; i < st->st_narcs; i++) {
if (i > 0)
printf("\n ");
printf("-> %2d %s", ar->ar_arrow,
PyGrammar_LabelRepr(&ll->ll_label[ar->ar_label]));
ar++;
}
printf("\n");
}
int
addlabel(labellist *ll, int type, char *str)
{
int i;
label *lb;
for (i = 0; i < ll->ll_nlabels; i++) {
if (ll->ll_label[i].lb_type == type &&
strcmp(ll->ll_label[i].lb_str, str) == 0)
return i;
}
PyMem_RESIZE(ll->ll_label, label, ll->ll_nlabels + 1);
if (ll->ll_label == NULL)
Py_FatalError("no mem to resize labellist in addlabel");
lb = &ll->ll_label[ll->ll_nlabels++];
lb->lb_type = type;
lb->lb_str = strdup(str);
if (Py_DebugFlag)
printf("Label @ %08x, %d: %s\n", (unsigned)ll, ll->ll_nlabels,
PyGrammar_LabelRepr(lb));
return lb - ll->ll_label;
}
int
addlabel(labellist *ll, int type, char *str)
{
int i;
label *lb;
for (i = 0; i < ll->ll_nlabels; i++) {
if (ll->ll_label[i].lb_type == type &&
strcmp(ll->ll_label[i].lb_str, str) == 0)
return i;
}
ll->ll_label = (label *)PyObject_REALLOC(ll->ll_label,
sizeof(label) * (ll->ll_nlabels + 1));
if (ll->ll_label == NULL)
Py_FatalError("no mem to resize labellist in addlabel");
lb = &ll->ll_label[ll->ll_nlabels++];
lb->lb_type = type;
lb->lb_str = strdup(str);
if (Py_DebugFlag)
printf("Label @ %8p, %d: %s\n", ll, ll->ll_nlabels,
PyGrammar_LabelRepr(lb));
return lb - ll->ll_label;
}
static void
translabel(grammar *g, label *lb)
{
int i;
if (Py_DebugFlag)
printf("Translating label %s ...\n", PyGrammar_LabelRepr(lb));
if (lb->lb_type == NAME) {
for (i = 0; i < g->g_ndfas; i++) {
if (strcmp(lb->lb_str, g->g_dfa[i].d_name) == 0) {
if (Py_DebugFlag)
printf(
"Label %s is non-terminal %d.\n",
lb->lb_str,
g->g_dfa[i].d_type);
lb->lb_type = g->g_dfa[i].d_type;
free(lb->lb_str);
lb->lb_str = NULL;
return;
}
}
for (i = 0; i < (int)N_TOKENS; i++) {
if (strcmp(lb->lb_str, _PyParser_TokenNames[i]) == 0) {
if (Py_DebugFlag)
printf("Label %s is terminal %d.\n",
lb->lb_str, i);
lb->lb_type = i;
free(lb->lb_str);
lb->lb_str = NULL;
return;
}
}
printf("Can't translate NAME label '%s'\n", lb->lb_str);
return;
}
if (lb->lb_type == STRING) {
if (isalpha(Py_CHARMASK(lb->lb_str[1])) ||
lb->lb_str[1] == '_') {
char *p;
char *src;
char *dest;
size_t name_len;
if (Py_DebugFlag)
printf("Label %s is a keyword\n", lb->lb_str);
lb->lb_type = NAME;
src = lb->lb_str + 1;
p = strchr(src, '\'');
if (p)
name_len = p - src;
else
name_len = strlen(src);
dest = (char *)malloc(name_len + 1);
if (!dest) {
printf("Can't alloc dest '%s'\n", src);
return;
}
strncpy(dest, src, name_len);
dest[name_len] = '\0';
free(lb->lb_str);
lb->lb_str = dest;
}
else if (lb->lb_str[2] == lb->lb_str[0]) {
int type = (int) PyToken_OneChar(lb->lb_str[1]);
if (type != OP) {
lb->lb_type = type;
free(lb->lb_str);
lb->lb_str = NULL;
}
else
printf("Unknown OP label %s\n",
lb->lb_str);
}
else if (lb->lb_str[2] && lb->lb_str[3] == lb->lb_str[0]) {
int type = (int) PyToken_TwoChars(lb->lb_str[1],
lb->lb_str[2]);
if (type != OP) {
lb->lb_type = type;
free(lb->lb_str);
lb->lb_str = NULL;
}
else
printf("Unknown OP label %s\n",
lb->lb_str);
}
else if (lb->lb_str[2] && lb->lb_str[3] && lb->lb_str[4] == lb->lb_str[0]) {
int type = (int) PyToken_ThreeChars(lb->lb_str[1],
lb->lb_str[2],
lb->lb_str[3]);
if (type != OP) {
lb->lb_type = type;
free(lb->lb_str);
lb->lb_str = NULL;
}
else
printf("Unknown OP label %s\n",
lb->lb_str);
}
else
printf("Can't translate STRING label %s\n",
lb->lb_str);
}
else
printf("Can't translate label '%s'\n",
PyGrammar_LabelRepr(lb));
}
static void
calcfirstset(grammar *g, dfa *d)
{
int i, j;
state *s;
arc *a;
int nsyms;
int *sym;
int nbits;
static bitset dummy;
bitset result;
int type;
dfa *d1;
label *l0;
if (Py_DebugFlag)
printf("Calculate FIRST set for '%s'\n", d->d_name);
if (dummy == NULL)
dummy = newbitset(1);
if (d->d_first == dummy) {
fprintf(stderr, "Left-recursion for '%s'\n", d->d_name);
return;
}
if (d->d_first != NULL) {
fprintf(stderr, "Re-calculating FIRST set for '%s' ???\n",
d->d_name);
}
d->d_first = dummy;
l0 = g->g_ll.ll_label;
nbits = g->g_ll.ll_nlabels;
result = newbitset(nbits);
sym = (int *)PyObject_MALLOC(sizeof(int));
if (sym == NULL)
Py_FatalError("no mem for new sym in calcfirstset");
nsyms = 1;
sym[0] = findlabel(&g->g_ll, d->d_type, (char *)NULL);
s = &d->d_state[d->d_initial];
for (i = 0; i < s->s_narcs; i++) {
a = &s->s_arc[i];
for (j = 0; j < nsyms; j++) {
if (sym[j] == a->a_lbl)
break;
}
if (j >= nsyms) { /* New label */
sym = (int *)PyObject_REALLOC(sym,
sizeof(int) * (nsyms + 1));
if (sym == NULL)
Py_FatalError(
"no mem to resize sym in calcfirstset");
sym[nsyms++] = a->a_lbl;
type = l0[a->a_lbl].lb_type;
if (ISNONTERMINAL(type)) {
d1 = PyGrammar_FindDFA(g, type);
if (d1->d_first == dummy) {
fprintf(stderr,
"Left-recursion below '%s'\n",
d->d_name);
}
else {
if (d1->d_first == NULL)
calcfirstset(g, d1);
mergebitset(result,
d1->d_first, nbits);
}
}
else if (ISTERMINAL(type)) {
addbit(result, a->a_lbl);
}
}
}
d->d_first = result;
if (Py_DebugFlag) {
printf("FIRST set for '%s': {", d->d_name);
for (i = 0; i < nbits; i++) {
if (testbit(result, i))
printf(" %s", PyGrammar_LabelRepr(&l0[i]));
}
printf(" }\n");
}
PyObject_FREE(sym);
}
