parsergen: recorded a prefered shift-symbol for error recovery.

[ocean] / csrc / parsergen.mdc

diff --git a/csrc/parsergen.mdc b/csrc/parsergen.mdc
index ec9b8f263f624de44b76419d4e4ab68d3a7576b4..252b3d6f6c88724c90b93163f392b0b4bfd23279 100644 (file)
--- a/csrc/parsergen.mdc
+++ b/csrc/parsergen.mdc
@@ -1754,8 +1754,8 @@ The table of nonterminals used for tracing is a similar array.
 
 ### States and the goto tables.
 
 
 ### States and the goto tables.
 

-For each state we record the goto table and the reducible production if
-there is one.
+For each state we record the goto table, the reducible production if
+there is one, or a symbol to shift for error recovery.

 Some of the details of the reducible production are stored in the
 `do_reduce` function to come later.  Here we store the production number,
 the body size (useful for stack management) and the resulting symbol (useful
 Some of the details of the reducible production are stored in the
 `do_reduce` function to come later.  Here we store the production number,
 the body size (useful for stack management) and the resulting symbol (useful


@@ -1776,6 +1776,7 @@ The go to table is stored in a simple array of `sym` and corresponding
                short reduce_prod;
                short reduce_size;
                short reduce_sym;
                short reduce_prod;
                short reduce_size;
                short reduce_sym;

+               short shift_sym;

        };
 
 
        };
 
 


@@ -1806,24 +1807,39 @@ The go to table is stored in a simple array of `sym` and corresponding
                fprintf(f, "static const struct state states[] = {\n");
                for (i = 0; i < g->states; i++) {
                        struct itemset *is = g->statetab[i];
                fprintf(f, "static const struct state states[] = {\n");
                for (i = 0; i < g->states; i++) {
                        struct itemset *is = g->statetab[i];

-                       int j, prod = -1;
+                       int j, prod = -1, prod_len;
+                       int shift_sym = -1;
+                       int shift_len = 0, shift_remain = 0;

                        for (j = 0; j < is->items.cnt; j++) {
                                int itm = is->items.syms[j];
                                int p = item_prod(itm);
                                int bp = item_index(itm);
                                struct production *pr = g->productions[p];
 
                        for (j = 0; j < is->items.cnt; j++) {
                                int itm = is->items.syms[j];
                                int p = item_prod(itm);
                                int bp = item_index(itm);
                                struct production *pr = g->productions[p];
 

-                               if (bp < pr->body_size)
+                               if (bp < pr->body_size) {
+                                       if (shift_sym < 0 ||
+                                           (shift_len == bp && shift_remain > pr->body_size - bp)) {
+                                               shift_sym = pr->body[bp]->num;
+                                               shift_len = bp;
+                                               shift_remain = pr->body_size - bp;
+                                       }

                                        continue;
                                        continue;

+                               }

                                /* This is what we reduce */
                                /* This is what we reduce */

-                               prod = p;
-                               break;
+                               if (prod < 0 || prod_len < pr->body_size) {
+                                       prod = p;
+                                       prod_len = pr->body_size;
+                               }

                        }
 
                        }
 

-                       fprintf(f, "\t[%d] = { %d, goto_%d, %d, %d, %d },\n",
-                               i, is->go_to.cnt, i, prod,
-                               prod < 0 ? -1 : g->productions[prod]->body_size,
-                               prod < 0 ? -1 : g->productions[prod]->head->num);
+                       if (prod >= 0)
+                               fprintf(f, "\t[%d] = { %d, goto_%d, %d, %d, %d, 0 },\n",
+                                       i, is->go_to.cnt, i, prod,
+                                       g->productions[prod]->body_size,
+                                       g->productions[prod]->head->num);
+                       else
+                               fprintf(f, "\t[%d] = { %d, goto_%d, -1, -1, -1, %d },\n",
+                                       i, is->go_to.cnt, i, shift_sym);

                }
                fprintf(f, "};\n\n");
        }
                }
                fprintf(f, "};\n\n");
        }