From dfcfac60588063fc576c18e05156be84ab7110f5 Mon Sep 17 00:00:00 2001
From: kumachan-mis <29433058+kumachan-mis@users.noreply.github.com>
Date: Sun, 8 Oct 2023 08:24:29 +0000
Subject: [PATCH] implement lexer

---
 src/ctoken/ctoken.c      |  4 ++--
 src/ctoken/ctoken.h      |  1 +
 src/lexer/util.c         |  1 +
 tests/lexer/test_lexer.c | 37 +++++++++++++++++++++++++++++++++++++
 4 files changed, 41 insertions(+), 2 deletions(-)

diff --git a/src/ctoken/ctoken.c b/src/ctoken/ctoken.c
index cb04bd37..d34ff156 100644
--- a/src/ctoken/ctoken.c
+++ b/src/ctoken/ctoken.c
@@ -9,8 +9,8 @@ BaseType t_ctoken = {
 
 char* ctoken_types[] = {
     // non-punctuators
-    "char", "else", "enum", "if", "int", "long", "return", "sizeof", "struct", "typedef", "unsigned", "void", "while",
-    "identifier", "integer-constant", "character-constant", "string-literal",
+    "char", "else", "enum", "for", "if", "int", "long", "return", "sizeof", "struct", "typedef", "unsigned", "void",
+    "while", "identifier", "integer-constant", "character-constant", "string-literal",
     // punctuators
     "[", "]", "(", ")", "{", "}", ".", "->", "++", "--", "&", "*", "+", "-", "~", "!", "/", "%", "<", ">",
     "<=", ">=", "==", "!=", "^", "|", "&&", "||", "?", ":", ";", "=",
diff --git a/src/ctoken/ctoken.h b/src/ctoken/ctoken.h
index 7e4cdcf9..ad8fbc32 100644
--- a/src/ctoken/ctoken.h
+++ b/src/ctoken/ctoken.h
@@ -10,6 +10,7 @@ typedef enum CTokenType {
     CTOKEN_KEYWORD_CHAR,
     CTOKEN_KEYWORD_ELSE,
     CTOKEN_KEYWORD_ENUM,
+    CTOKEN_KEYWORD_FOR,
     CTOKEN_KEYWORD_IF,
     CTOKEN_KEYWORD_INT,
     CTOKEN_KEYWORD_LONG,
diff --git a/src/lexer/util.c b/src/lexer/util.c
index 1a68f00c..1ce05b16 100644
--- a/src/lexer/util.c
+++ b/src/lexer/util.c
@@ -12,6 +12,7 @@ Map* new_keyword_map(void) {
     ctoken_map_add(keyword_map, "char", CTOKEN_KEYWORD_CHAR);
     ctoken_map_add(keyword_map, "else", CTOKEN_KEYWORD_ELSE);
     ctoken_map_add(keyword_map, "enum", CTOKEN_KEYWORD_ENUM);
+    ctoken_map_add(keyword_map, "for", CTOKEN_KEYWORD_FOR);
     ctoken_map_add(keyword_map, "if", CTOKEN_KEYWORD_IF);
     ctoken_map_add(keyword_map, "int", CTOKEN_KEYWORD_INT);
     ctoken_map_add(keyword_map, "long", CTOKEN_KEYWORD_LONG);
diff --git a/tests/lexer/test_lexer.c b/tests/lexer/test_lexer.c
index 193df4f2..0941457d 100644
--- a/tests/lexer/test_lexer.c
+++ b/tests/lexer/test_lexer.c
@@ -14,6 +14,7 @@ void test_read_declaration_pointer_with_init(void);
 void test_read_declaration_aggregate_with_init(void);
 void test_read_statement_if_else(void);
 void test_read_statement_while(void);
+void test_read_statement_for(void);
 void test_read_assignment_expr(void);
 void test_read_conditional_expr(void);
 void test_read_logical_expr(void);
@@ -48,6 +49,7 @@ CU_Suite* add_test_suite_lexer(void) {
     CU_ADD_TEST(suite, test_read_declaration_aggregate_with_init);
     CU_ADD_TEST(suite, test_read_statement_if_else);
     CU_ADD_TEST(suite, test_read_statement_while);
+    CU_ADD_TEST(suite, test_read_statement_for);
     CU_ADD_TEST(suite, test_read_assignment_expr);
     CU_ADD_TEST(suite, test_read_conditional_expr);
     CU_ADD_TEST(suite, test_read_decimal_integer_constant);
@@ -416,6 +418,41 @@ void test_read_statement_while(void) {
     delete_vector(expected);
 }
 
+void test_read_statement_for(void) {
+    char* input = "for (int i = 0; i < 10; i++) { x += 1; y *= 2; }";
+
+    Vector* expected = new_vector(&t_ctoken);
+    vector_push(expected, new_ctoken(CTOKEN_KEYWORD_FOR));
+    vector_push(expected, new_ctoken(CTOKEN_LPAREN));
+    vector_push(expected, new_ctoken(CTOKEN_KEYWORD_INT));
+    vector_push(expected, new_identifier_ctoken(CTOKEN_IDENT, new_string("i")));
+    vector_push(expected, new_ctoken(CTOKEN_EQUAL));
+    vector_push(expected, new_iliteral_ctoken(CTOKEN_INT, new_signed_iliteral(INTEGER_INT, 0)));
+    vector_push(expected, new_ctoken(CTOKEN_SEMICOLON));
+    vector_push(expected, new_identifier_ctoken(CTOKEN_IDENT, new_string("i")));
+    vector_push(expected, new_ctoken(CTOKEN_LESS));
+    vector_push(expected, new_iliteral_ctoken(CTOKEN_INT, new_signed_iliteral(INTEGER_INT, 10)));
+    vector_push(expected, new_ctoken(CTOKEN_SEMICOLON));
+    vector_push(expected, new_identifier_ctoken(CTOKEN_IDENT, new_string("i")));
+    vector_push(expected, new_ctoken(CTOKEN_PLUS_PLUS));
+    vector_push(expected, new_ctoken(CTOKEN_RPAREN));
+    vector_push(expected, new_ctoken(CTOKEN_LBRACE));
+    vector_push(expected, new_identifier_ctoken(CTOKEN_IDENT, new_string("x")));
+    vector_push(expected, new_ctoken(CTOKEN_PLUS_EQUAL));
+    vector_push(expected, new_iliteral_ctoken(CTOKEN_INT, new_signed_iliteral(INTEGER_INT, 1)));
+    vector_push(expected, new_ctoken(CTOKEN_SEMICOLON));
+    vector_push(expected, new_identifier_ctoken(CTOKEN_IDENT, new_string("y")));
+    vector_push(expected, new_ctoken(CTOKEN_ASTERISK_EQUAL));
+    vector_push(expected, new_iliteral_ctoken(CTOKEN_INT, new_signed_iliteral(INTEGER_INT, 2)));
+    vector_push(expected, new_ctoken(CTOKEN_SEMICOLON));
+    vector_push(expected, new_ctoken(CTOKEN_RBRACE));
+    vector_push(expected, new_ctoken(CTOKEN_EOF));
+
+    run_lexer_test(input, expected);
+
+    delete_vector(expected);
+}
+
 void test_read_assignment_expr(void) {
     char* input = "x86 = 64; x += 1; x -= 1; x *= 2; x /= 2; x %= 2; x &= 1; x ^= 1; x |= 1;";