From c4765a539ee01625dd310a02f0be16ec9a64e2e4 Mon Sep 17 00:00:00 2001
From: Bad Diode <bd@badd10de.dev>
Date: Sat, 12 Feb 2022 16:21:19 +0100
Subject: Make keywords a static array for ease of lex

---
 src/lexer.c | 62 +++++++++++++++++++++++++++++++++++--------------------------
 src/main.c  |  3 ---
 2 files changed, 36 insertions(+), 29 deletions(-)

(limited to 'src')

diff --git a/src/lexer.c b/src/lexer.c
index 80ef3ad..5175b1c 100644
--- a/src/lexer.c
+++ b/src/lexer.c
@@ -27,6 +27,26 @@ static const char* token_str[] = {
     [TOKEN_EOF]     = "TOKEN_EOF",
 };
 
+typedef struct Keyword {
+    char *str;
+    size_t n;
+    TokenType token;
+} Keyword;
+
+#define KEYWORD(STR,TOK) {(STR), sizeof(STR) - 1, (TOK)}
+
+static const Keyword keywords[] = {
+    KEYWORD("nil",    TOKEN_NIL),
+    KEYWORD("true",   TOKEN_TRUE),
+    KEYWORD("false",  TOKEN_FALSE),
+    KEYWORD("lambda", TOKEN_LAMBDA),
+    KEYWORD("if",     TOKEN_IF),
+    KEYWORD("def",    TOKEN_DEF),
+    KEYWORD("set!",   TOKEN_SET),
+    KEYWORD("fun",    TOKEN_FUN),
+    KEYWORD("struct", TOKEN_STRUCT),
+};
+
 void
 print_token(Token tok) {
     printf("[%4ld:%-4ld] ", tok.line, tok.col);
@@ -121,11 +141,12 @@ is_delimiter(char c) {
     return false;
 }
 
-#define TOKEN_IS_KEYWORD(VAL, KEYWORD) \
-    sv_equal(&(VAL), &(StringView){(KEYWORD), sizeof(KEYWORD) - 1})
-
 size_t
 scan_number_token(Scanner *scanner) {
+    // TODO: This looks like more a parsing problem than lexer,
+    // consider moving it there. If starts with `-` and there is no
+    // delimiter after, or if it starts with a number, it is
+    // TOKEN_NUMBER.
     char first = scan_next(scanner);
     char second = scan_peek(scanner);
     size_t n = 1;
@@ -183,16 +204,12 @@ scan_number_token(Scanner *scanner) {
 
 TokenType
 find_token_type(const StringView value) {
-    if (TOKEN_IS_KEYWORD(value, "nil")) { return TOKEN_NIL; }
-    if (TOKEN_IS_KEYWORD(value, "true")) { return TOKEN_TRUE; }
-    if (TOKEN_IS_KEYWORD(value, "false")) { return TOKEN_FALSE; }
-    if (TOKEN_IS_KEYWORD(value, "lambda")) { return TOKEN_LAMBDA; }
-    if (TOKEN_IS_KEYWORD(value, "if")) { return TOKEN_IF; }
-    if (TOKEN_IS_KEYWORD(value, "def")) { return TOKEN_DEF; }
-    if (TOKEN_IS_KEYWORD(value, "set!")) { return TOKEN_SET; }
-    if (TOKEN_IS_KEYWORD(value, "fun")) { return TOKEN_FUN; }
-    if (TOKEN_IS_KEYWORD(value, "struct")) { return TOKEN_STRUCT; }
-
+    for (size_t i = 0; i < sizeof(keywords) / sizeof(Keyword); i++) {
+        StringView keyword = (StringView){keywords[i].str, keywords[i].n};
+        if (sv_equal(&value, &keyword)) {
+            return keywords[i].token;
+        }
+    }
     return TOKEN_SYMBOL;
 }
 
@@ -267,19 +284,12 @@ tokenize(const StringView *sv) {
                     break;
                 }
                 size_t n = 1;
-                if (c == '-' && !is_delimiter(scan_peek(&scanner))) {
-                    n += scan_number_token(&scanner);
-                    token.value = (StringView){
-                        .start = &sv->start[offset],
-                        .n = n,
-                    };
-                    token.type = TOKEN_NUMBER;
-                } else if (c >= '0' && c <= '9') {
-                    scan_rewind(&scanner);
-                    n = scan_number_token(&scanner);
-                    if (n == 0) {
-                        push_error(ERR_TYPE_LEXER, ERR_MALFORMED_NUMBER, line, col);
-                        return tokens;
+                bool is_number = c == '-' && !is_delimiter(scan_peek(&scanner));
+                is_number = is_number || (c >= '0' && c <= '9');
+                if (is_number) {
+                    while (!is_delimiter(scan_peek(&scanner))) {
+                        c = scan_next(&scanner);
+                        n++;
                     }
                     token.value = (StringView){
                         .start = &sv->start[offset],
diff --git a/src/main.c b/src/main.c
index 17dd481..31fb5da 100644
--- a/src/main.c
+++ b/src/main.c
@@ -27,9 +27,6 @@ process_source(const StringView *source, const char *file_name) {
     Token *tokens = tokenize(source);
     print_tokens(tokens);
     check_errors(file_name);
-    // if (errors.n != 0) {
-    //     exit(EXIT_FAILURE);
-    // }
 
     // // Parser.
     // Program program = parse(tokens, &errors);
-- 
cgit v1.2.1