static CcsBool_t
CcSyntax_AllNtToTerm(CcSyntax_t * self)
{
CcBitArray_t mark;
CcSymbolNT_t * sym; CcArrayListIter_t iter;
CcsBool_t changed, ok = TRUE;
CcSymbolTable_t * symtab = &self->globals->symtab;
CcArrayList_t * ntarr = &symtab->nonterminals;
CcBitArray(&mark, ntarr->Count);
do {
changed = FALSE;
for (sym = (CcSymbolNT_t *)CcArrayList_First(ntarr, &iter);
sym; sym = (CcSymbolNT_t *)CcArrayList_Next(ntarr, &iter)) {
if (!CcBitArray_Get(&mark, sym->base.kind) &&
CcSyntax_IsTerm(self, sym->graph, &mark)) {
CcBitArray_Set(&mark, sym->base.kind, TRUE);
changed = TRUE;
}
}
} while (changed);
for (sym = (CcSymbolNT_t *)CcArrayList_First(ntarr, &iter);
sym; sym = (CcSymbolNT_t *)CcArrayList_Next(ntarr, &iter)) {
if (!CcBitArray_Get(&mark, sym->base.kind)) {
ok = FALSE;
CcsErrorPool_Error(self->globals->errpool, NULL,
" %s cannot be derived to terminals",
sym->base.name);
}
}
CcBitArray_Destruct(&mark);
return ok;
}
static void
CcSyntax_CheckOverlap(CcSyntax_t * self, CcBitArray_t * s1, CcBitArray_t * s2,
int cond)
{
CcSymbolT_t * sym; CcArrayListIter_t iter;
CcSymbolTable_t * symtab = &self->globals->symtab;
CcArrayList_t * tarr = &symtab->terminals;
for (sym = (CcSymbolT_t *)CcArrayList_First(tarr, &iter);
sym; sym = (CcSymbolT_t *)CcArrayList_Next(tarr, &iter)) {
if (CcBitArray_Get(s1, sym->base.kind)
&& CcBitArray_Get(s2, sym->base.kind))
CcSyntax_LL1Error(self, cond, (CcSymbol_t *)sym);
}
}
static void
SCSOS_GenCond(CcCSharpBaseOutputScheme_t * self, CcOutput_t * output,
const char * prefix, const char * suffix,
const CcBitArray_t * s, const CcNode_t * p)
{
const CcNodeRSLV_t * prslv; int n;
CcArrayListIter_t iter; const CcSymbol_t * sym;
const CcArrayList_t * terminals;
if (p->base.type == node_rslv) {
prslv = (CcNodeRSLV_t *)p;
CcPrintfIL(output, "%s%s%s", prefix, prslv->pos->text, suffix);
} else if ((n = CcBitArray_Elements(s)) == 0) {
CcPrintfIL(output, "%s%s%s", prefix, "false", suffix);
} else if (n <= maxTerm) {
CcPrintfI(output, "%s", prefix);
terminals = &self->base.globals->symtab.terminals;
for (sym = (const CcSymbol_t *)CcArrayList_FirstC(terminals, &iter);
sym; sym = (const CcSymbol_t *)CcArrayList_NextC(terminals, &iter))
if (CcBitArray_Get(s, sym->kind)) {
CcPrintf(output, "la.kind == %d", sym->kind);
if (--n > 0) CcPrintf(output, " || ");
}
CcPrintfL(output, "%s", suffix);
} else {
CcPrintfIL(output, "%sStartOf(%d)%s",
prefix, CcSyntaxSymSet_New(&self->symSet, s), suffix);
}
}
static CcsBool_t
CSOS_InitSet(CcCSharpBaseOutputScheme_t * self, CcOutput_t * output)
{
char * setstr; int setlen, index;
const CcBitArray_t * cur;
setlen = self->base.globals->symtab.terminals.Count;
setstr = CcMalloc(setlen + 1); setstr[setlen] = 0;
if (setlen > 4) {
for (index = 0; index < setlen; ++index)
if (index == 0) setstr[index] = '*';
else setstr[index] = index % 5 == 0 ? '0' + index % 10 : ' ';
CcPrintfIL(output, "/%s */", setstr);
}
for (cur = self->symSet.start; cur < self->symSet.used; ++cur) {
CcsAssert(setlen == CcBitArray_getCount(cur));
for (index = 0; index < setlen; ++index)
setstr[index] = CcBitArray_Get(cur, index) ? '*' : '.';
CcPrintfIL(output, "\"%s.\"%c /* %d */", setstr,
cur < self->symSet.used - 1 ? ',' : ' ',
cur - self->symSet.start);
}
CcFree(setstr);
return TRUE;
}
static void
CcSyntax_Complete(CcSyntax_t * self, CcSymbolNT_t * sym)
{
CcSymbolNT_t * s; CcArrayListIter_t iter;
CcSymbolTable_t * symtab = &self->globals->symtab;
CcArrayList_t * ntarr = &symtab->nonterminals;
if (CcBitArray_Get(self->visited, sym->base.kind)) return;
CcBitArray_Set(self->visited, sym->base.kind, TRUE);
for (s = (CcSymbolNT_t *)CcArrayList_First(ntarr, &iter);
s; s = (CcSymbolNT_t *)CcArrayList_Next(ntarr, &iter)) {
if (CcBitArray_Get(sym->nts, s->base.kind)) {
CcSyntax_Complete(self, s);
CcBitArray_Or(sym->follow, s->follow);
if ((CcSymbol_t *)sym == self->curSy)
CcBitArray_Set(sym->nts, s->base.kind, FALSE);
}
}
}
/* true if graph can be derived to terminals */
CcsBool_t
CcSyntax_IsTerm(CcSyntax_t *self, CcNode_t * p, CcBitArray_t * mark) {
while (p != NULL) {
if (p->base.type == node_nt &&
!CcBitArray_Get(mark, (((CcNodeNT_t *)p)->sym)->base.index))
return FALSE;
if (p->base.type == node_alt && !CcSyntax_IsTerm(self, p->sub, mark) &&
(p->down == NULL || !CcSyntax_IsTerm(self, p->down, mark)))
return FALSE;
if (p->up) break;
p = p->next;
}
return TRUE;
}
static void
CcSyntax_First0(CcSyntax_t * self, CcBitArray_t * ret,
CcNode_t * p, CcBitArray_t * mark)
{
CcBitArray_t fs0;
CcSymbolTable_t * symtab = &self->globals->symtab;
CcBitArray(ret, symtab->terminals.Count);
while (p != NULL && !CcBitArray_Get(mark, p->base.index)) {
CcBitArray_Set(mark, p->base.index, TRUE);
if (p->base.type == node_nt) {
CcNodeNT_t * p0 = (CcNodeNT_t *)p;
CcSymbolNT_t * sym = (CcSymbolNT_t *)p0->sym;
if (sym->firstReady) {
CcBitArray_Or(ret, sym->first);
} else {
CcSyntax_First0(self, &fs0, sym->graph, mark);
CcBitArray_Or(ret, &fs0);
CcBitArray_Destruct(&fs0);
}
} else if (p->base.type == node_t) {
CcNodeT_t * p0 = (CcNodeT_t *)p;
CcBitArray_Set(ret, p0->sym->kind, TRUE);
} else if (p->base.type == node_wt) {
CcNodeWT_t * p0 = (CcNodeWT_t *)p;
CcBitArray_Set(ret, p0->sym->kind, TRUE);
} else if (p->base.type == node_any) {
CcNodeANY_t * p0 = (CcNodeANY_t *)p;
CcBitArray_Or(ret, p0->set);
} else if (p->base.type == node_alt) {
CcSyntax_First0(self, &fs0, p->sub, mark);
CcBitArray_Or(ret, &fs0);
CcBitArray_Destruct(&fs0);
CcSyntax_First0(self, &fs0, p->down, mark);
CcBitArray_Or(ret, &fs0);
CcBitArray_Destruct(&fs0);
} else if (p->base.type == node_iter || p->base.type == node_opt) {
CcSyntax_First0(self, &fs0, p->sub, mark);
CcBitArray_Or(ret, &fs0);
CcBitArray_Destruct(&fs0);
}
if (!CcNode_DelNode(p)) break;
p = p->next;
}
}
static CcsBool_t
CcSyntax_AllNtReached(CcSyntax_t * self)
{
int idx; CcSymbol_t * sym;
CcsBool_t ok = TRUE;
CcSymbolTable_t * symtab = &self->globals->symtab;
self->visited = CcBitArray(&self->visitedSpace,
symtab->nonterminals.Count);
CcBitArray_Set(self->visited, self->gramSy->kind, TRUE);
CcSyntax_MarkReachedNts(self, ((CcSymbolNT_t *)self->gramSy)->graph);
for (idx = 0; idx < symtab->nonterminals.Count; ++idx) {
sym = (CcSymbol_t *)CcArrayList_Get(&symtab->nonterminals, idx);
if (!CcBitArray_Get(self->visited, sym->kind)) {
ok = FALSE;
CcsErrorPool_Error(self->globals->errpool, NULL,
" '%s' cannot be reached", sym->name);
}
}
return ok;
}
static void
CcSyntax_MarkReachedNts(CcSyntax_t * self, CcNode_t * p)
{
while (p != NULL) {
if (p->base.type == node_nt) {
CcNodeNT_t * p0 = (CcNodeNT_t *)p;
CcSymbolNT_t * sym = (CcSymbolNT_t *)p0->sym;
if (!CcBitArray_Get(self->visited, sym->base.kind)) {
/* new nt reached */
CcBitArray_Set(self->visited, sym->base.kind, TRUE);
CcSyntax_MarkReachedNts(self, sym->graph);
}
} else if (p->base.type == node_alt || p->base.type == node_iter
|| p->base.type == node_opt) {
CcSyntax_MarkReachedNts(self, p->sub);
if (p->base.type == node_alt)
CcSyntax_MarkReachedNts(self, p->down);
}
if (p->up) break;
p = p->next;
}
}
static void
CcSyntax_CompSync(CcSyntax_t * self, CcNode_t * p)
{
CcNodeSYNC_t * syncp;
while (p != NULL && !CcBitArray_Get(self->visited, p->base.index)) {
CcBitArray_Set(self->visited, p->base.index, TRUE);
if (p->base.type == node_sync) {
syncp = (CcNodeSYNC_t *)p;
CcsAssert(syncp->set == NULL);
CcSyntax_Expected(self, &syncp->setSpace, p->next, self->curSy);
syncp->set = &syncp->setSpace;
CcBitArray_Set(syncp->set, self->eofSy->kind, TRUE);
CcBitArray_Or(self->allSyncSets, syncp->set);
} else if (p->base.type == node_alt) {
CcSyntax_CompSync(self, p->sub);
CcSyntax_CompSync(self, p->down);
} else if (p->base.type == node_opt || p->base.type == node_iter) {
CcSyntax_CompSync(self, p->sub);
}
p = p->next;
}
}
static void
CcSyntax_CompFollow(CcSyntax_t * self, CcNode_t * p)
{
CcBitArray_t s;
CcSymbolNT_t * sym;
while (p != NULL && !CcBitArray_Get(self->visited, p->base.index)) {
CcBitArray_Set(self->visited, p->base.index, TRUE);
if (p->base.type == node_nt) {
CcSyntax_First(self, &s, p->next);
sym = (CcSymbolNT_t *)((CcNodeNT_t *)p)->sym;
CcBitArray_Or(sym->follow, &s);
CcBitArray_Destruct(&s);
if (CcNode_DelGraph(p->next))
CcBitArray_Set(sym->nts, self->curSy->kind, TRUE);
} else if (p->base.type == node_opt || p->base.type == node_iter) {
CcSyntax_CompFollow(self, p->sub);
} else if (p->base.type == node_alt) {
CcSyntax_CompFollow(self, p->sub);
CcSyntax_CompFollow(self, p->down);
}
p = p->next;
}
}
static void
ATest(FILE * fp, CcBitArray_t * ba0, CcBitArray_t * ba1)
{
CcBitArray_t ba2;
int idx, cnt, numbits = CcBitArray_getCount(ba0);
CcBitArray_Dump(ba0, fp, "BitArray 0: ", "\n");
CcBitArray_Dump(ba1, fp, "BitArray 1: ", "\n");
cnt = 0;
for (idx = 0; idx < numbits; ++idx)
if (CcBitArray_Get(ba0, idx) && CcBitArray_Get(ba1, idx)) cnt = 1;
COCO_ASSERT((CcBitArray_Intersect(ba0, ba1) == cnt));
COCO_ASSERT((CcBitArray_Clone(&ba2, ba0)));
COCO_ASSERT((CcBitArray_Equal(ba0, &ba2)));
cnt = 0;
for (idx = 0; idx < numbits; ++idx) {
COCO_ASSERT(CcBitArray_Get(ba0, idx) == CcBitArray_Get(&ba2, idx));
if (CcBitArray_Get(ba0, idx)) ++cnt;
}
COCO_ASSERT((CcBitArray_Elements(ba0) == cnt));
COCO_ASSERT((CcBitArray_Or(&ba2, ba1) == 0));
CcBitArray_Dump(&ba2, fp, "Or: ", "\n");
for (idx = 0; idx < numbits; ++idx)
COCO_ASSERT((CcBitArray_Get(ba0, idx) || CcBitArray_Get(ba1, idx)) == CcBitArray_Get(&ba2, idx));
CcBitArray_Destruct(&ba2);
COCO_ASSERT((CcBitArray_Clone(&ba2, ba0)));
COCO_ASSERT((CcBitArray_And(&ba2, ba1) == 0));
CcBitArray_Dump(&ba2, fp, "And: ", "\n");
for (idx = 0; idx < numbits; ++idx)
COCO_ASSERT((CcBitArray_Get(ba0, idx) && CcBitArray_Get(ba1, idx)) == CcBitArray_Get(&ba2, idx));
CcBitArray_Destruct(&ba2);
COCO_ASSERT((CcBitArray_Clone(&ba2, ba0)));
CcBitArray_Subtract(&ba2, ba1);
CcBitArray_Dump(&ba2, fp, "Subtract: ", "\n");
for (idx = 0; idx < numbits; ++idx)
COCO_ASSERT((CcBitArray_Get(ba0, idx) && !CcBitArray_Get(ba1, idx)) == CcBitArray_Get(&ba2, idx));
COCO_ASSERT(!CcBitArray_Intersect(ba1, &ba2));
CcBitArray_Destruct(&ba2);
fprintf(fp, "\n");
}
static CcsBool_t
CcSyntax_NoCircularProductions(CcSyntax_t * self)
{
CcBitArray_t singles;
SymbolPair_t * firstPair, * prevPair, * curPair, * curPair0;
CcSymbolNT_t * sym; CcArrayListIter_t iter; int index;
CcsBool_t ok, changed, onLeftSide, onRightSide;
CcSymbol_t * leftsym, * rightsym;
CcSymbolTable_t * symtab = &self->globals->symtab;
CcArrayList_t * ntarr = &symtab->nonterminals;
firstPair = NULL;
CcBitArray(&singles, ntarr->Count);
for (sym = (CcSymbolNT_t *)CcArrayList_First(ntarr, &iter);
sym; sym = (CcSymbolNT_t *)CcArrayList_Next(ntarr, &iter)) {
CcBitArray_SetAll(&singles, FALSE);
/* Get nonterminals s such that sym-->s */
GetSingles(&singles, sym->graph);
for (index = 0; index < ntarr->Count; ++index) {
if (!CcBitArray_Get(&singles, index)) continue;
curPair = CcMalloc(sizeof(SymbolPair_t));
curPair->left = sym->base.kind;
curPair->right = index;
curPair->next = firstPair;
firstPair = curPair;
}
}
CcBitArray_Destruct(&singles);
do {
changed = FALSE;
prevPair = NULL; curPair = firstPair;
while (curPair) {
onLeftSide = FALSE; onRightSide = FALSE;
for (curPair0 = curPair; curPair0; curPair0 = curPair0->next) {
if (curPair->left == curPair0->right) onRightSide = TRUE;
if (curPair->right == curPair0->left) onLeftSide = TRUE;
}
if (onLeftSide && onRightSide) { /* Circular Production found. */
prevPair = curPair; curPair = curPair->next;
} else { /* Remove non-circular nonterminal symbol pair. */
curPair0 = curPair->next;
if (prevPair == NULL) firstPair = curPair0;
else prevPair->next = curPair0;
CcFree(curPair);
curPair = curPair0;
changed = TRUE;
}
}
} while (changed);
ok = TRUE;
for (curPair = firstPair; curPair; curPair = curPair0) {
ok = FALSE;
leftsym = (CcSymbol_t *)CcArrayList_Get(ntarr, curPair->left);
rightsym = (CcSymbol_t *)CcArrayList_Get(ntarr, curPair->right);
CcsErrorPool_Error(self->globals->errpool, NULL, " '%s' --> '%s'",
leftsym->name, rightsym->name);
curPair0 = curPair->next;
CcFree(curPair);
}
return ok;
}
static void
SCSOS_GenCode(CcCSharpBaseOutputScheme_t * self, CcOutput_t * output,
CcNode_t * p, const CcBitArray_t * IsChecked)
{
int err; CcsBool_t equal, useSwitch; int index;
CcNode_t * p2; CcBitArray_t s1, s2, isChecked;
CcNodeNT_t * pnt; CcNodeT_t * pt; CcNodeWT_t * pwt;
CcNodeSEM_t * psem; CcNodeSYNC_t * psync;
CcSyntax_t * syntax = &self->base.globals->syntax;
CcArrayList_t * terminals = &self->base.globals->symtab.terminals;
CcBitArray_Clone(&isChecked, IsChecked);
while (p != NULL) {
if (p->base.type == node_nt) {
pnt = (CcNodeNT_t *)p;
if (pnt->pos) {
CcPrintfIL(output, "%s(%s);", pnt->sym->name, pnt->pos->text);
} else {
CcPrintfIL(output, "%s();", pnt->sym->name);
}
} else if (p->base.type == node_t) {
pt = (CcNodeT_t *)p;
if (CcBitArray_Get(&isChecked, pt->sym->kind))
CcPrintfIL(output, "Get();");
else
CcPrintfIL(output, "Expect(%d);", pt->sym->kind);
} else if (p->base.type == node_wt) {
pwt = (CcNodeWT_t *)p;
CcSyntax_Expected(syntax, &s1, p->next, self->curSy);
CcBitArray_Or(&s1, syntax->allSyncSets);
CcPrintfIL(output, "ExpectWeak(%d, %d);",
pwt->sym->kind, CcSyntaxSymSet_New(&self->symSet, &s1));
CcBitArray_Destruct(&s1);
} else if (p->base.type == node_any) {
CcPrintfIL(output, "Get();");
} else if (p->base.type == node_eps) {
} else if (p->base.type == node_rslv) {
} else if (p->base.type == node_sem) {
psem = (CcNodeSEM_t *)p;
CcSource(output, psem->pos);
} else if (p->base.type == node_sync) {
psync = (CcNodeSYNC_t *)p;
err = CcSyntax_SyncError(syntax, self->curSy);
CcBitArray_Clone(&s1, psync->set);
SCSOS_GenCond(self, output, "while (!(", ")) {", &s1, p);
output->indent += 4;
CcPrintfIL(output, "SynErr(%d); Get();", err);
output->indent -= 4;
CcPrintfIL(output, "}");
CcBitArray_Destruct(&s1);
} else if (p->base.type == node_alt) {
CcSyntax_First(syntax, &s1, p);
equal = CcBitArray_Equal(&s1, &isChecked);
CcBitArray_Destruct(&s1);
useSwitch = SCSOS_UseSwitch(self, p);
if (useSwitch)
CcPrintfIL(output, "switch (la.kind) {");
p2 = p;
while (p2 != NULL) {
CcSyntax_Expected(syntax, &s1, p2->sub, self->curSy);
if (useSwitch) {
CcPrintfI(output, "");
for (index = 0; index < terminals->Count; ++index)
if (CcBitArray_Get(&s1, index))
CcPrintf(output, "case %d: ", index);
CcPrintfL(output,"{");
} else if (p2 == p) {
SCSOS_GenCond(self, output, "if (", ") {", &s1, p2->sub);
} else if (p2->down == NULL && equal) {
CcPrintfIL(output, "} else {");
} else {
SCSOS_GenCond(self, output,
"} else if (", ") {", &s1, p2->sub);
}
CcBitArray_Or(&s1, &isChecked);
output->indent += 4;
SCSOS_GenCode(self, output, p2->sub, &s1);
if (useSwitch) CcPrintfIL(output, "break;");
output->indent -= 4;
if (useSwitch) CcPrintfIL(output, "}");
p2 = p2->down;
CcBitArray_Destruct(&s1);
}
if (equal) {
CcPrintfIL(output, "}");
} else {
err = CcSyntax_AltError(syntax, self->curSy);
if (useSwitch) {
CcPrintfIL(output, "default: SynErr(%d); break;", err);
CcPrintfIL(output, "}");
} else {
CcPrintfIL(output, "} else SynErr(%d);", err);
}
}
} else if (p->base.type == node_iter) {
p2 = p->sub;
if (p2->base.type == node_wt) {
CcSyntax_Expected(syntax, &s1, p2->next, self->curSy);
CcSyntax_Expected(syntax, &s2, p->next, self->curSy);
CcPrintfIL(output,
"while (WeakSeparator(%d, %d, %d)) {",
((CcNodeWT_t *)p2)->sym->kind,
CcSyntaxSymSet_New(&self->symSet, &s1),
CcSyntaxSymSet_New(&self->symSet, &s2));
CcBitArray_Destruct(&s1); CcBitArray_Destruct(&s2);
CcBitArray(&s1, terminals->Count);
if (p2->up || p2->next == NULL) p2 = NULL; else p2 = p2->next;
} else {
CcSyntax_First(syntax, &s1, p2);
SCSOS_GenCond(self, output, "while (", ") {", &s1, p2);
}
output->indent += 4;
SCSOS_GenCode(self, output, p2, &s1);
output->indent -= 4;
CcPrintfIL(output, "}");
CcBitArray_Destruct(&s1);
} else if (p->base.type == node_opt) {
CcSyntax_First(syntax, &s1, p->sub);
SCSOS_GenCond(self, output, "if (", ") {", &s1, p->sub);
output->indent += 4;
SCSOS_GenCode(self, output, p->sub, &s1);
output->indent -= 4;
CcPrintfIL(output, "}");
CcBitArray_Destruct(&s1);
}
if (p->base.type != node_eps && p->base.type != node_sem &&
p->base.type != node_sync)
CcBitArray_SetAll(&isChecked, FALSE);
if (p->up) break;
p = p->next;
}
CcBitArray_Destruct(&isChecked);
}
