More LL(k) fixes (Merge branch 'LL' of https://github.com/pesco/hammer)

src/backends/llk.c

@@ -81,28 +81,6 @@ void h_llktable_free(HLLkTable *table)
   h_free(table);
 }
 
-/* Compute the predict_k set of production "A -> rhs".
- * Always returns a newly-allocated HCFStringMap.
- */
-HCFStringMap *h_predict(size_t k, HCFGrammar *g,
-                        const HCFChoice *A, const HCFSequence *rhs)
-{
-  assert(k==1); // XXX
-  HCFStringMap *ret = h_stringmap_new(g->arena);
-
-  // predict(A -> rhs) = first(rhs) u follow(A)  if "" can be derived from rhs
-  // predict(A -> rhs) = first(rhs)              otherwise
-
-  h_stringmap_update(ret, h_first_seq(k, g, rhs->items));
-  if(h_derives_epsilon_seq(g, rhs->items))
-    h_stringmap_update(ret, h_follow(k, g, A));
-
-  // make sure there are only strings of length _exactly_ k
-  ret->epsilon_branch = NULL;
-
-  return ret;
-}
-
 void *const CONFLICT = (void *)(uintptr_t)(-1);
 
 // helper for stringmap_merge
@@ -113,7 +91,7 @@ static void *combine_entries(HHashSet *workset, void *dst, const void *src)
 
   if(dst == CONFLICT) {                 // previous conflict
     h_hashset_put(workset, src);
-  } else if(dst == src) {               // new conflict
+  } else if(dst != src) {               // new conflict
     h_hashset_put(workset, dst);
     h_hashset_put(workset, src);
     dst = CONFLICT;
@@ -133,6 +111,12 @@ static void stringmap_merge(HHashSet *workset, HCFStringMap *dst, HCFStringMap *
           combine_entries(workset, dst->epsilon_branch, src->epsilon_branch);
     else
       dst->epsilon_branch = src->epsilon_branch;
+  } else {
+    // if there is a non-conflicting value on the left (dst) side, it means
+    // that prediction is already unambiguous. we can drop the right (src)
+    // side we were going to extend with.
+    if(dst->epsilon_branch && dst->epsilon_branch != CONFLICT)
+      return;
   }
 
   if(src->end_branch) {
@@ -164,9 +148,6 @@ static void stringmap_merge(HHashSet *workset, HCFStringMap *dst, HCFStringMap *
   }
 }
 
-void pprint_sequence(FILE *f, const HCFGrammar *g, const HCFSequence *seq);
-void pprint_symbol(FILE *f, const HCFGrammar *g, const HCFChoice *x);
-
 /* Generate entries for the production "A" in the given table row. */
 static int fill_table_row(size_t kmax, HCFGrammar *g, HCFStringMap *row,
                           const HCFChoice *A)
@@ -181,6 +162,8 @@ static int fill_table_row(size_t kmax, HCFGrammar *g, HCFStringMap *row,
   // run until workset exhausted or kmax hit
   size_t k;
   for(k=1; k<=kmax; k++) {
+    printf("k=%lu\n", k); // XXX debug
+
     // allocate a fresh workset for the next round
     HHashSet *nextset = h_hashset_new(g->arena, h_eq_ptr, h_hash_ptr);
 
@@ -199,29 +182,30 @@ static int fill_table_row(size_t kmax, HCFGrammar *g, HCFStringMap *row,
         HCFStringMap *pred = h_predict(k, g, A, rhs);
         h_stringmap_replace(pred, NULL, rhs);
 
+        // XXX debug
+        printf("predict(");
+        h_pprint_sequence(stdout, g, rhs);
+        printf(") = ");
+        h_pprint_stringset(stdout, pred, 0);
+
         // merge predict set into the row
         // accumulates conflicts in new workset
         stringmap_merge(nextset, row, pred);
-
-        // XXX debug
-        if(A == g->start) {
-          printf("predict(");
-          pprint_sequence(stdout, g, rhs);
-          printf("  ) = ");
-          h_pprint_stringset(stdout, g, pred, 0);
-        }
       }
     }
     // XXX debug
-    if(A == g->start) {
     printf("row(");
-      pprint_symbol(stdout, g, A);
+    h_pprint_symbol(stdout, g, A);
     printf(") = ");
-      h_pprint_stringset(stdout, g, row, 0);
+    h_pprint_stringset(stdout, row, 0);
+    if(h_stringmap_get(row, (uint8_t *)"a", 1, false)) {
+      printf("  a -> ");
+      h_pprint_sequence(stdout, g, h_stringmap_get(row, (uint8_t *)"a", 1, false));
+      printf("\n");
     }
 
     // switch to the updated workset
-    h_hashtable_free(workset);
+    h_hashset_free(workset);
     workset = nextset;
 
     // if the workset is empty, row is without conflict; we're done
@@ -473,9 +457,9 @@ int test_llk(void)
   */
 
   HParser *X = h_optional(h_ch('x'));
-  HParser *Y = h_sequence(h_ch('y'), NULL);
+  //HParser *Y = h_epsilon_p(); //h_sequence(h_ch('y'), NULL);
-  HParser *A = h_sequence(X, Y, h_ch('a'), NULL);
+  HParser *A = h_sequence(X, h_ch('a'), NULL);
-  HParser *B = h_sequence(Y, h_ch('b'), NULL);
+  HParser *B = h_sequence(h_ch('b'), NULL);
   HParser *p = h_choice(A, B, NULL);
 
   HCFGrammar *g = h_cfgrammar(&system_allocator, p);
@@ -489,16 +473,16 @@ int test_llk(void)
   printf("derive epsilon: ");
   h_pprint_symbolset(stdout, g, g->geneps, 0);
   printf("first(A) = ");
-  h_pprint_stringset(stdout, g, h_first(3, g, g->start), 0);
+  h_pprint_stringset(stdout, h_first(3, g, g->start), 0);
   //printf("follow(C) = ");
-  //h_pprint_stringset(stdout, g, h_follow(3, g, h_desugar(&system_allocator, c)), 0);
+  //h_pprint_stringset(stdout, h_follow(3, g, h_desugar(&system_allocator, c)), 0);
 
-  if(h_compile(p, PB_LLk, NULL)) {
+  if(h_compile(p, PB_LLk, (void *)2)) {
     fprintf(stderr, "does not compile\n");
     return 2;
   }
 
-  HParseResult *res = h_parse(p, (uint8_t *)"xa", 2);
+  HParseResult *res = h_parse(p, (uint8_t *)"ab", 2);
   if(res)
     h_pprint(stdout, res->ast, 0, 2);
   else

src/cfgrammar.c

@@ -437,7 +437,23 @@ static bool any_string_shorter(size_t k, const HCFStringMap *m)
   return false;
 }
 
-const HCFStringMap *h_follow(size_t k, HCFGrammar *g, const HCFChoice *x);
+// helper for h_predict
+static void remove_all_shorter(size_t k, HCFStringMap *m)
+{
+  if(k==0) return;
+  m->epsilon_branch = NULL;
+  if(k==1) return;
+
+  // iterate over m->char_branches
+  const HHashTable *ht = m->char_branches;
+  for(size_t i=0; i < ht->capacity; i++) {
+    for(HHashTableEntry *hte = &ht->contents[i]; hte; hte = hte->next) {
+      if(hte->key == NULL)
+        continue;
+      remove_all_shorter(k-1, hte->value);      // recursion into subtree
+    }
+  }
+}
 
 // h_follow adapted to the signature of StringSetFun
 static inline const HCFStringMap *h_follow_(size_t k, HCFGrammar *g, HCFChoice **s)
@@ -507,6 +523,23 @@ const HCFStringMap *h_follow(size_t k, HCFGrammar *g, const HCFChoice *x)
   return ret;
 }
 
+HCFStringMap *h_predict(size_t k, HCFGrammar *g,
+                        const HCFChoice *A, const HCFSequence *rhs)
+{
+  HCFStringMap *ret = h_stringmap_new(g->arena);
+
+  // predict_k(A -> rhs) =
+  //   { ab | a <- first_k(rhs), b <- follow_k(A), |ab|=k }
+  
+  const HCFStringMap *first_rhs = h_first_seq(k, g, rhs->items);
+  stringset_extend(g, ret, k, first_rhs, h_follow_, (HCFChoice **)&A);
+
+  // make sure there are only strings of length _exactly_ k
+  remove_all_shorter(k, ret);
+
+  return ret;
+}
+
 // add the set { a b | a <- as, b <- f_l(S), l=k-|a| } to ret
 static void stringset_extend(HCFGrammar *g, HCFStringMap *ret,
                              size_t k, const HCFStringMap *as,
@@ -624,7 +657,7 @@ static HCFChoice **pprint_string(FILE *f, HCFChoice **x)
   return x;
 }
 
-void pprint_symbol(FILE *f, const HCFGrammar *g, const HCFChoice *x)
+void h_pprint_symbol(FILE *f, const HCFGrammar *g, const HCFChoice *x)
 {
   switch(x->type) {
   case HCF_CHAR:
@@ -643,31 +676,36 @@ void pprint_symbol(FILE *f, const HCFGrammar *g, const HCFChoice *x)
   }
 }
 
-void pprint_sequence(FILE *f, const HCFGrammar *g, const HCFSequence *seq)
+void h_pprint_sequence(FILE *f, const HCFGrammar *g, const HCFSequence *seq)
 {
   HCFChoice **x = seq->items;
 
   if(*x == NULL) {  // the empty sequence
-    fputs(" \"\"", f);
+    fputs("\"\"", f);
   } else {
     while(*x) {
-      fputc(' ', f);      // separator
+      if(x != seq->items) fputc(' ', f); // internal separator
 
       if((*x)->type == HCF_CHAR) {
         // condense character strings
         x = pprint_string(f, x);
       } else {
-        pprint_symbol(f, g, *x);
+        h_pprint_symbol(f, g, *x);
         x++;
       }
     }
   }
+}
 
+// adds some separators expected below
+static void pprint_sequence(FILE *f, const HCFGrammar *g, const HCFSequence *seq)
+{
+  fputc(' ', f);
+  h_pprint_sequence(f, g, seq);
   fputc('\n', f);
 }
 
-static
+static void pprint_ntrules(FILE *f, const HCFGrammar *g, const HCFChoice *nt,
-void pprint_ntrules(FILE *f, const HCFGrammar *g, const HCFChoice *nt,
                            int indent, int len)
 {
   int i;
@@ -738,7 +776,7 @@ void h_pprint_symbolset(FILE *file, const HCFGrammar *g, const HHashSet *set, in
 
       a = hte->key;        // production's left-hand symbol
 
-      pprint_symbol(file, g, a);
+      h_pprint_symbol(file, g, a);
     }
   }
 
@@ -801,7 +839,7 @@ pprint_stringset_elems(FILE *file, bool first, char *prefix, size_t n,
   return first;
 }
 
-void h_pprint_stringset(FILE *file, const HCFGrammar *g, const HCFStringMap *set, int indent)
+void h_pprint_stringset(FILE *file, const HCFStringMap *set, int indent)
 {
   int j;
   for(j=0; j<indent; j++) fputc(' ', file);

src/cfgrammar.h

@@ -50,7 +50,7 @@ void *h_stringmap_get(const HCFStringMap *m, const uint8_t *str, size_t n, bool
 bool h_stringmap_present(const HCFStringMap *m, const uint8_t *str, size_t n, bool end);
 bool h_stringmap_present_epsilon(const HCFStringMap *m);
 
-static inline void *h_stringmap_get_char(const HCFStringMap *m, const uint8_t c)
+static inline HCFStringMap *h_stringmap_get_char(const HCFStringMap *m, const uint8_t c)
  { return h_hashtable_get(m->char_branches, (void *)char_key(c)); }
 
 
@@ -80,8 +80,16 @@ const HCFStringMap *h_first_seq(size_t k, HCFGrammar *g, HCFChoice **s);
 /* Compute follow_k set of symbol x. Memoized. */
 const HCFStringMap *h_follow(size_t k, HCFGrammar *g, const HCFChoice *x);
 
+/* Compute the predict_k set of production "A -> rhs".
+ * Always returns a newly-allocated HCFStringMap.
+ */
+HCFStringMap *h_predict(size_t k, HCFGrammar *g,
+                        const HCFChoice *A, const HCFSequence *rhs);
+
 
 /* Pretty-printers for grammars and associated data. */
 void h_pprint_grammar(FILE *file, const HCFGrammar *g, int indent);
+void h_pprint_sequence(FILE *f, const HCFGrammar *g, const HCFSequence *seq);
+void h_pprint_symbol(FILE *f, const HCFGrammar *g, const HCFChoice *x);
 void h_pprint_symbolset(FILE *file, const HCFGrammar *g, const HHashSet *set, int indent);
-void h_pprint_stringset(FILE *file, const HCFGrammar *g, const HCFStringMap *set, int indent);
+void h_pprint_stringset(FILE *file, const HCFStringMap *set, int indent);

src/parsers/optional.c

@@ -25,7 +25,6 @@ static bool opt_isValidCF(void *env) {
 static HParsedToken* reshape_optional(const HParseResult *p) {
   assert(p->ast);
   assert(p->ast->token_type == TT_SEQUENCE);
-  assert(p->ast->seq->used > 0);
 
   HParsedToken *res = p->ast->seq->elements[0];
   if(res)