feat(lexer): add diagnostics with file:line:col

Author: BeyondMagic

src/codegen_lua.c

@@ -99,9 +99,9 @@ static void emit_main_wrapper(FILE *out, const AstFunction *fn, const FunctionSi
 				break;
 			case TYPE_CHAR:
 				fprintf(out, "local %s = args_table and args_table[%zu] and string.byte(args_table[%zu]) or 0\n",
-					param->name,
-					i + 1,
-					i + 1);
+						param->name,
+						i + 1,
+						i + 1);
 				break;
 			case TYPE_FLOAT:
 				fprintf(out, "local %s = args_table and tonumber(args_table[%zu]) or 0.0\n", param->name, i + 1);

src/lexer.h

@@ -0,0 +1,13 @@
+#ifndef LEXER_H
+#define LEXER_H
+
+const char *lexer_get_source_name(void);
+void lexer_set_source_name(const char *name);
+void lexer_reset_position(void);
+
+extern int yy_line;
+extern int yy_column;
+extern int yy_token_line;
+extern int yy_token_column;
+
+#endif

src/lexer.l

@@ -3,6 +3,49 @@
 #include <stdlib.h>
 #include <string.h>
 #include "parser.tab.h"
+#include "lexer.h"
+
+int yy_line = 1;
+int yy_column = 1;
+int yy_token_line = 1;
+int yy_token_column = 1;
+static const char *yy_source_name = "<stdin>";
+
+void lexer_set_source_name(const char *name)
+{
+    yy_source_name = (name && *name) ? name : "<stdin>";
+}
+
+const char *lexer_get_source_name(void)
+{
+    return yy_source_name ? yy_source_name : "<stdin>";
+}
+
+void lexer_reset_position(void)
+{
+    yy_line = 1;
+    yy_column = 1;
+    yy_token_line = 1;
+    yy_token_column = 1;
+}
+
+#define YY_USER_ACTION                                                     \
+    {                                                                       \
+        yy_token_line = yy_line;                                              \
+        yy_token_column = yy_column;                                          \
+        for (int yy_i = 0; yy_i < yyleng; ++yy_i)                             \
+        {                                                                   \
+            if (yytext[yy_i] == '\n')                                        \
+            {                                                               \
+                yy_line++;                                                    \
+                yy_column = 1;                                                \
+            }                                                               \
+            else                                                            \
+            {                                                               \
+                yy_column++;                                                  \
+            }                                                               \
+        }                                                                   \
+    }
 
 static char *yy_strdup(const char *src)
 {
@@ -156,6 +199,6 @@ static long long yy_parse_char_literal(const char *src)
 "}"                                { return RBRACE; }
 \"([^\"\n]|\\.)*\"             { yylval.string = yy_parse_string_literal(yytext); return STRING_LITERAL; }
 '([^\\'\n]|\\.)'             { yylval.intValue = yy_parse_char_literal(yytext); return CHAR_LITERAL; }
-.                                  { fprintf(stderr, "invalid character '%s'\n", yytext); }
+.                                  { fprintf(stderr, "%s:%d:%d: invalid character '%s'\n", lexer_get_source_name(), yy_token_line, yy_token_column, yytext); }
 
 %%

src/main.c

@@ -5,6 +5,7 @@
 
 #include "ast.h"
 #include "codegen_lua.h"
+#include "lexer.h"
 #include "optimizer.h"
 #include "semantic.h"
 #include "parser.tab.h"
@@ -19,6 +20,10 @@ int main(int argc, char **argv)
 		return EXIT_FAILURE;
 	}
 
+	const char *source_name = (argc == 2) ? argv[1] : "<stdin>";
+	lexer_set_source_name(source_name);
+	lexer_reset_position();
+
 	AstProgram *program = c2lua_parse(input);
 	if (!program)
 	{

src/parser.y

@@ -4,6 +4,7 @@
 #include <string.h>
 
 #include "ast.h"
+#include "lexer.h"
 
 static AstProgram *make_program_with_function(AstFunction *fn);
 int yyerror(AstProgram **out_program, const char *msg);
@@ -557,8 +558,13 @@ static void parser_error_cleanup(AstProgram **out_program)
 
 int yyerror(AstProgram **out_program, const char *msg)
 {
-    (void)out_program;
-    fprintf(stderr, "syntax error: %s\n", msg);
+  (void)out_program;
+  fprintf(stderr,
+    "%s:%d:%d: syntax error: %s\n",
+    lexer_get_source_name(),
+    yy_token_line,
+    yy_token_column,
+    msg ? msg : "unknown");
     return 0;
 }
 

tests/fail/bad_if.err

@@ -1 +1 @@
-syntax error: syntax error
+tests/fail/bad_if.c:4:2: syntax error: syntax error

tests/fail/int_plus_string.err

@@ -1 +1 @@
-syntax error: syntax error
+tests/fail/int_plus_string.c:4:7: syntax error: syntax error

tests/fail/missing_return.err

@@ -1 +1 @@
-syntax error: syntax error
+tests/fail/missing_return.c:4:5: syntax error: syntax error

tests/fail/wrong_assign.err

@@ -1 +1 @@
-syntax error: syntax error
+tests/fail/wrong_assign.c:4:6: syntax error: syntax error