Fix semantic routines being associated with the wrong production

Write NES ROM example with limited uLisp synthetic power
andriybyelikov · Sep 15, 2021 · c3036f9 · c3036f9
1 parent 7d95380
commit c3036f9
Show file tree

Hide file tree

Showing 7 changed files with 2,190 additions and 69 deletions.
diff --git a/spec/sample/6502_to_bin b/spec/sample/6502_to_bin
@@ -1,34 +1,130 @@
-t_gobble__blank -> [ \t\r\n]+;
+t_gobble__blank       -> [ \t\r\n]+;
+t_gobble__linecomment -> ";".*"\n";
 t_mnemonic      -> [A-Z]+;
-t_directive     -> "."[a-z]*; -- buggy kleene plus, use kleene star
-t_symbol        -> [a-z][a-z0-9_]*;
-t_colon         -> ":";
-t_immediate     -> "#";
-t_hex           -> "$"[0-9A-Fa-f]*;
-t_with_x_reg    -> ",X";
+t_numbersign    -> "#";
+t_hexadecimal   -> "0x"[0-9A-Fa-f]*;
+t_decimal       -> [1-9][0-9]*;
+t_comma         -> ",";
+t_reg           -> [axy];
 
 start       -> expressions;
 expressions -> expression expressions | expression;
-expression  -> instruction | directive | label;
+expression  -> instruction;
 
-instruction ->
-t_mnemonic operand
-<
-    (emit_byte 0x42)
->| t_mnemonic
+instruction -> t_mnemonic
 <
+    (if (strequ (lexeme "t_mnemonic") "BRK")
+        (emit_byte 0)
+    (if (strequ (lexeme "t_mnemonic") "PHP")
+        (emit_byte 0x8)
+    (if (strequ (lexeme "t_mnemonic") "CLC")
+        (emit_byte 0x18)
+    (if (strequ (lexeme "t_mnemonic") "PLP")
+        (emit_byte 0x28)
+    (if (strequ (lexeme "t_mnemonic") "SEC")
+        (emit_byte 0x38)
+    (if (strequ (lexeme "t_mnemonic") "RTI")
+        (emit_byte 0x40)
+    (if (strequ (lexeme "t_mnemonic") "PHA")
+        (emit_byte 0x48)
+    (if (strequ (lexeme "t_mnemonic") "CLI")
+        (emit_byte 0x58)
+    (if (strequ (lexeme "t_mnemonic") "RTS")
+        (emit_byte 0x60)
+    (if (strequ (lexeme "t_mnemonic") "PLA")
+        (emit_byte 0x68)
     (if (strequ (lexeme "t_mnemonic") "SEI")
         (emit_byte 0x78)
+    (if (strequ (lexeme "t_mnemonic") "DEY")
+        (emit_byte 0x88)
+    (if (strequ (lexeme "t_mnemonic") "TXA")
+        (emit_byte 0x8A)
+    (if (strequ (lexeme "t_mnemonic") "TYA")
+        (emit_byte 0x98)
+    (if (strequ (lexeme "t_mnemonic") "TXS")
+        (emit_byte 0x9A)
+    (if (strequ (lexeme "t_mnemonic") "TAY")
+        (emit_byte 0xA8)
+    (if (strequ (lexeme "t_mnemonic") "TAX")
+        (emit_byte 0xAA)
+    (if (strequ (lexeme "t_mnemonic") "CLV")
+        (emit_byte 0xB8)
+    (if (strequ (lexeme "t_mnemonic") "TSX")
+        (emit_byte 0xBA)
+    (if (strequ (lexeme "t_mnemonic") "INY")
+        (emit_byte 0xC8)
+    (if (strequ (lexeme "t_mnemonic") "DEX")
+        (emit_byte 0xCA)
     (if (strequ (lexeme "t_mnemonic") "CLD")
         (emit_byte 0xD8)
+    (if (strequ (lexeme "t_mnemonic") "INX")
+        (emit_byte 0xE8)
+    (if (strequ (lexeme "t_mnemonic") "NOP")
+        (emit_byte 0xEA)
+    (if (strequ (lexeme "t_mnemonic") "SED")
+        (emit_byte 0xF8)
+        (nil)
+    )))))))))))))))))))))))))
+> | t_mnemonic t_numbersign t_hexadecimal -- immediate
+<
+    (if (strequ (lexeme "t_mnemonic") "LDX")
+        (emit_byte 0xA2)
+    (if (strequ (lexeme "t_mnemonic") "LDA")
+        (emit_byte 0xA9)
         (nil)
     ))
->;
+    (emit_byte (strtol (lexeme "t_hexadecimal")))
+> | t_mnemonic t_decimal t_hexadecimal -- absolute
 
+-- trickery: t_decimal is page number while t_hexadecimal is offset
 
-operand -> t_immediate t_hex | t_hex t_with_x_reg | t_hex | t_symbol;
-
-directive -> t_directive darg | t_directive;
-darg -> t_hex | t_symbol;
-
-label -> t_symbol t_colon | t_symbol;
+<
+    (if (strequ (lexeme "t_mnemonic") "STX")
+        (emit_byte 0x8E)
+    (if (strequ (lexeme "t_mnemonic") "BIT")
+        (emit_byte 0x2C)
+    (if (strequ (lexeme "t_mnemonic") "STA")
+        (emit_byte 0x8D)
+    (if (strequ (lexeme "t_mnemonic") "JMP")
+        (emit_byte 0x4C)
+        (nil)
+    ))))
+    -- emit in little endian
+    (emit_byte (strtol (lexeme "t_hexadecimal")))
+    (emit_byte (strtol (lexeme "t_decimal")))
+> | t_mnemonic t_hexadecimal -- relative
+<
+    (if (strequ (lexeme "t_mnemonic") "BPL")
+        (emit_byte 0x10)
+    (if (strequ (lexeme "t_mnemonic") "BNE")
+        (emit_byte 0xD0)
+    (if (strequ (lexeme "t_mnemonic") "BEQ")
+        (emit_byte 0xF0)
+        (nil)
+    )))
+    (emit_byte (strtol (lexeme "t_hexadecimal")))
+> | t_mnemonic t_hexadecimal t_comma t_reg -- zeropage x
+<
+    (if (strequ (lexeme "t_reg") "x")
+        (if (strequ (lexeme "t_mnemonic") "STA")
+            (emit_byte 0x85)
+            (nil)
+        )
+        (nil)
+    )
+    (emit_byte (strtol (lexeme "t_hexadecimal")))
+> | t_mnemonic t_decimal t_hexadecimal t_comma t_reg -- absolute x
+<
+    (if (strequ (lexeme "t_reg") "x")
+        (if (strequ (lexeme "t_mnemonic") "STA")
+            (emit_byte 0x9D)
+        (if (strequ (lexeme "t_mnemonic") "LDA")
+            (emit_byte 0xBD)
+            (nil)
+        ))
+        (nil)
+    )
+    -- emit in little endian
+    (emit_byte (strtol (lexeme "t_hexadecimal")))
+    (emit_byte (strtol (lexeme "t_decimal")))
+>;
diff --git a/spec/sample/uppercase_text_to_bin b/spec/sample/uppercase_text_to_bin
@@ -0,0 +1,9 @@
+t_gobble__blank -> [\n]+;
+t_text -> [A-Z ]+;
+t_end  -> "\\0";
+
+start -> text;
+text -> t_text t_end
+<
+    (emit_line (lexeme "t_text"))
+>;
diff --git a/src/meta/parser_pre_shift_action1.c b/src/meta/parser_pre_shift_action1.c
@@ -20,6 +20,10 @@ void push_terminal_to_parse_tree_stack(terminal_t terminal,
 
 void meta__pre_shift_action1(void *user_ptr, terminal_t terminal)
 {
-    struct data_common *user = user_ptr;
+    struct data_meta *user = user_ptr;
     push_terminal_to_parse_tree_stack(terminal, &user->parse_tree_stack);
+
+    if (!strcmp(terminal__lexeme(terminal), "start")) {
+        user->in_terminals_section = 0;
+    }
 }
diff --git a/src/meta/parser_production_attributed_actions2.c b/src/meta/parser_production_attributed_actions2.c
@@ -320,7 +320,6 @@ void meta__production_attibuted_actions2(void *user_ptr, int pid)
                     production_t p = new_production(int_id +
                         int_trie__length(&user->terminals), &body);
                     production_queue__insert(&user->productions, p);
-                    user->cnt_prod++;
 
                     body = NULL;
                     expr = parse_tree__get_child_by_string(&expr, "expr");
@@ -370,22 +369,30 @@ void meta__production_attibuted_actions2(void *user_ptr, int pid)
                         buf_trimmed), &data_ulisp);
                 free(buf);
 
-                // debug
-                /*fprintf(stderr, "%d\n", user->cnt_prod);
-                node_t t = parse_tree_stack__access(&data_ulisp.parse_tree_stack);
-                parse_tree__dump_dot(stdout, &t);*/
+
+                int key =
+                    ulisp_parse_tree_avl__length(&user->ulisp_parse_trees);
 
-                int key = user->cnt_prod == user->last_prod ?
-                            user->cnt_prod + 1: user->cnt_prod;
-                user->last_prod = key;
                 ulisp_parse_tree_avl__insert(&user->ulisp_parse_trees,
-                    (pidxpt_t) { key,
+                    (pidxpt_t) {
+                        key,
                         parse_tree_stack__access(&data_ulisp.parse_tree_stack)
                     }, ulisp_parse_tree_avl__cmp_predicate);
             }
         }
         break;
     case 7: // expr_a -> expr_0;
+        {
+            if ((user->options[OPTION_DUMP_ULISP_PARSE_TREES]
+                        || user->composition_mode)
+                    && !user->in_terminals_section) {
+                int key =
+                    ulisp_parse_tree_avl__length(&user->ulisp_parse_trees);
+                ulisp_parse_tree_avl__insert(&user->ulisp_parse_trees,
+                    (pidxpt_t) { key, NULL },
+                    ulisp_parse_tree_avl__cmp_predicate);
+            }
+        }
         break;
     case 8: // expr_0 -> expr_1 expr_0;
         {

diff --git a/test/6502/game.nes b/test/6502/game.nes