feat: add bison grammar for emit, decltype and difference

5 files changed, 95 insertions, 57 deletions

diff --git a/Makefile b/Makefile
index a30b93c..c107cb3 100644
--- a/Makefile
+++ b/Makefile
@@ -1,8 +1,9 @@
 CC      = cc
 LEX     = flex
-CFLAGS  = -std=c17 -Wall -Wextra -Werror -Wpedantic -O2
+YACC    = bison
+CFLAGS  = -std=c17 -Wall -Wextra -Werror -Wpedantic -O2 -Ibin
 SOURCES = ast.c ast.h lexer.h main.c
-OBJECTS = bin/lexer.o bin/ast.o bin/main.o
+OBJECTS = bin/lexer.o bin/parser.o bin/ast.o bin/main.o
 SANITIZE = -g -O0 -fsanitize=address,undefined
 
 .PHONY: all debug check format lint clean
@@ -18,13 +19,19 @@ sfinae: $(OBJECTS)
 bin/lexer.c: sfinae.l | bin
 	$(LEX) -o bin/lexer.c sfinae.l
 
-bin/lexer.o: bin/lexer.c lexer.h
+bin/lexer.o: bin/lexer.c bin/parser.h ast.h lexer.h
 	$(CC) -std=c17 -w -I. -c -o $@ bin/lexer.c
 
+bin/parser.c bin/parser.h: sfinae.y | bin
+	$(YACC) -d -o bin/parser.c sfinae.y
+
+bin/parser.o: bin/parser.c
+	$(CC) -std=c17 -w -I. -c -o $@ bin/parser.c
+
 bin/%.o: %.c | bin
 	$(CC) $(CFLAGS) -c -o $@ $<
 
-bin/main.o: lexer.h
+bin/main.o: ast.h lexer.h
 bin/ast.o: ast.h lexer.h
 
 debug: CFLAGS += $(SANITIZE)
@@ -40,7 +47,7 @@ format:
 	clang-format -i $(SOURCES)
 
 lint:
-	clang-tidy ast.c main.c -- -std=c17
+	clang-tidy ast.c main.c -- -std=c17 -Ibin
 
 clean:
 	rm -rf bin sfinae
diff --git a/lexer.h b/lexer.h
index 4e09ace..93d854a 100644
--- a/lexer.h
+++ b/lexer.h
@@ -3,20 +3,8 @@
 #include <stdbool.h>
 #include <stdio.h>
 
-enum Token
-{
-    TOK_END = 0,
-    TOK_COLONCOLON = 256,
-    TOK_EMIT,
-    TOK_COMMIT,
-    TOK_DECLTYPE,
-    TOK_DIFFERENCE,
-    TOK_CHAR
-};
-
 extern FILE* yyin;   // input
 extern int yylineno; // current line
 int yylex(void);     // scanner
 
 bool lexer_failed(void);
-char lexer_character(void);
diff --git a/main.c b/main.c
index 634e6d0..2d358d8 100644
--- a/main.c
+++ b/main.c
@@ -1,37 +1,20 @@
-#include "lexer.h"
-
 #include <stdio.h>
 
-static const char* token_name(int token)
-{
-    switch (token)
-    {
-        case TOK_COLONCOLON: return "::";
-        case TOK_EMIT: return "emit";
-        case TOK_COMMIT: return "commit";
-        case TOK_DECLTYPE: return "decltype";
-        case TOK_DIFFERENCE: return "difference";
-        default: return NULL;
-    }
-}
-
-static int dump_tokens(void)
-{
-    int token = TOK_END;
+#include "ast.h"
+#include "lexer.h"
 
-    while ((token = yylex()) != TOK_END)
-    {
-        const char* name = token_name(token);
+extern struct Node* program_root;
+int yyparse(void);
 
-        if (token == TOK_CHAR)
-            (void)printf("'%c'\n", lexer_character());
-        else if (name != NULL)
-            (void)printf("%s\n", name);
-        else
-            (void)printf("%c\n", (char)token);
-    }
+static int parse_program(void)
+{
+    int status = yyparse();
+    if (status != 0 || lexer_failed())
+        status = 1;
 
-    return lexer_failed() ? 1 : 0;
+    node_free(program_root);
+    program_root = NULL;
+    return status;
 }
 
 int main(int argc, char** argv)
@@ -45,11 +28,11 @@ int main(int argc, char** argv)
     yyin = fopen(argv[1], "r");
     if (yyin == NULL)
     {
-        (void)fprintf(stderr, "cannot open %s, is it even real\n", argv[1]);
+        (void)fprintf(stderr, "cannot open %s\n", argv[1]);
         return 1;
     }
 
-    int status = dump_tokens();
+    int status = parse_program();
     (void)fclose(yyin);
     return status;
 }
diff --git a/sfinae.l b/sfinae.l
index f969a98..5c67f3c 100644
--- a/sfinae.l
+++ b/sfinae.l
@@ -2,12 +2,13 @@
 
 %{
 #include "lexer.h"
+#include "ast.h"
+#include "parser.h"
 
 #include <stdbool.h>
 #include <stdio.h>
 
 static bool failed = false;
-static char character = '\0';
 
 static char unescape(char escaped)
 {
@@ -24,7 +25,7 @@ static int fail(const char* message)
 {
     (void)fprintf(stderr, "line %d: %s\n", yylineno, message);
     failed = true;
-    return TOK_END;
+    return 0;
 }
 %}
 
@@ -36,8 +37,8 @@ static int fail(const char* message)
 "decltype"              return TOK_DECLTYPE;
 "difference"            return TOK_DIFFERENCE;
 [<>(),;]                return yytext[0];
-'\\.'                   character = unescape(yytext[2]); return TOK_CHAR;
-'.'                     character = yytext[1]; return TOK_CHAR;
+'\\.'                   yylval.character = unescape(yytext[2]); return TOK_CHAR;
+'.'                     yylval.character = yytext[1]; return TOK_CHAR;
 "//"[^\n]*              /* a comment */
 [ \t\r\n]+              /* blanks */
 [0-9]+                  return fail("numeric literals are for the weak, subtract two characters");
@@ -50,8 +51,3 @@ bool lexer_failed(void)
 {
     return failed;
 }
-
-char lexer_character(void)
-{
-    return character;
-}
diff --git a/sfinae.y b/sfinae.y
new file mode 100644
index 0000000..2e91b23
--- /dev/null
+++ b/sfinae.y
@@ -0,0 +1,64 @@
+%{
+#include "ast.h"
+#include "lexer.h"
+
+#include <stdio.h>
+
+struct Node* program_root = NULL;
+
+void yyerror(const char* message);
+%}
+
+%union
+{
+    char character;
+    struct Node* node;
+}
+
+%token TOK_COLONCOLON
+%token TOK_EMIT
+%token TOK_COMMIT
+%token TOK_DECLTYPE
+%token TOK_DIFFERENCE
+%token <character> TOK_CHAR
+
+%type <node> statements statement type value
+
+%%
+
+program
+    : statements
+        { program_root = $1; }
+    ;
+
+statements
+    : /* nothing */
+        { $$ = NULL; }
+    | statements statement ';'
+        { $$ = node_append($1, $2); }
+    ;
+
+statement
+    : TOK_COLONCOLON TOK_EMIT '<' type '>' TOK_COLONCOLON TOK_COMMIT
+        { $$ = node_unary(STMT_EMIT, $4); }
+    ;
+
+type
+    : TOK_COLONCOLON TOK_DECLTYPE '(' value ')'
+        { $$ = node_unary(TYPE_DECLTYPE, $4); }
+    ;
+
+value
+    : TOK_COLONCOLON TOK_DIFFERENCE '<' TOK_CHAR ',' TOK_CHAR '>'
+        { $$ = node_binary(VALUE_DIFFERENCE, node_char($4), node_char($6)); }
+    ;
+
+%%
+
+void yyerror(const char* message)
+{
+    if (!lexer_failed())
+    {
+        (void)fprintf(stderr, "line %d: %s, did you forget to commit\n", yylineno, message);
+    }
+}