parser: Introduce statement keywords

This commit introduces a few changes in pipalang syntax. Now, both
functions and variables requires keywords to be defined.

before:

  main(): i32 {
    a: i32 = 2;
    return a;
  }

now:

  fn main(): i32 {
    let a: i32 = 2;
    return a;
  }

Signed-off-by: Carlos Maniero <carlos@maniero.me>
Reviewed-by: Johnny Richard <johnny@johnnyrichard.com>

8 files changed, 67 insertions, 34 deletions

diff --git a/examples/arithmetics.pipa b/examples/arithmetics.pipa
index e2b7645..f2449b8 100644
--- a/examples/arithmetics.pipa
+++ b/examples/arithmetics.pipa
@@ -1,3 +1,3 @@
-main(): i32 {
+fn main(): i32 {
   return ((2 + 1) * (5 - 3) + 7) * 2 / 2;
 }
diff --git a/examples/main.pipa b/examples/main.pipa
index 6524bec..2da2231 100644
--- a/examples/main.pipa
+++ b/examples/main.pipa
@@ -1,3 +1,3 @@
-main(): i32 {
+fn main(): i32 {
   return 69;
 }
diff --git a/examples/variables.pipa b/examples/variables.pipa
index 0723782..3eabc73 100644
--- a/examples/variables.pipa
+++ b/examples/variables.pipa
@@ -1,10 +1,10 @@
-main(): i32 {
-  a: i32 = 13;
+fn main(): i32 {
+  let a: i32 = 13;
   a = a - 1;
-  b: i32 = 33;
+  let b: i32 = 33;
   b = 32;
-  c: i32 = 2 * (b + a);
-  d: i32 = (c - 33) + 1;
-  e: i32 = d;
+  let c: i32 = 2 * (b + a);
+  let d: i32 = (c - 33) + 1;
+  let e: i32 = d;
   return e / 2;
 }
diff --git a/src/lexer.c b/src/lexer.c
index 3f6948a..56e24af 100644
--- a/src/lexer.c
+++ b/src/lexer.c
@@ -97,10 +97,18 @@ lexer_tokenize_name(lexer_t *lexer, token_t *token)
 static void
 lexer_token_process_keyword(token_t *token)
 {
+  if (string_view_eq(string_view_from_str("let"), token->value)) {
+    token->kind = TOKEN_KEYWORD_LET;
+    return;
+  }
   if (string_view_eq(string_view_from_str("return"), token->value)) {
     token->kind = TOKEN_KEYWORD_RETURN;
     return;
   }
+  if (string_view_eq(string_view_from_str("fn"), token->value)) {
+    token->kind = TOKEN_KEYWORD_FN;
+    return;
+  }
 }
 
 void
@@ -318,6 +326,10 @@ token_kind_to_str(token_kind_t kind)
       return "=";
     case TOKEN_KEYWORD_RETURN:
       return "return";
+    case TOKEN_KEYWORD_FN:
+      return "fn";
+    case TOKEN_KEYWORD_LET:
+      return "let";
     case TOKEN_EOF:
       return "TOKEN_EOF";
     case TOKEN_UNKNOWN:
diff --git a/src/lexer.h b/src/lexer.h
index 9aa8efe..6449a0a 100644
--- a/src/lexer.h
+++ b/src/lexer.h
@@ -38,6 +38,8 @@ typedef enum
   TOKEN_SLASH,
   TOKEN_EQUAL,
   TOKEN_KEYWORD_RETURN,
+  TOKEN_KEYWORD_FN,
+  TOKEN_KEYWORD_LET,
   TOKEN_EOF,
   TOKEN_UNKNOWN
 } token_kind_t;
diff --git a/src/parser.c b/src/parser.c
index 82d25f5..49803eb 100644
--- a/src/parser.c
+++ b/src/parser.c
@@ -296,9 +296,14 @@ parser_parse_variable_assignment(parser_t *parser)
 }
 
 static ast_node_t *
-parser_parse_variable_definition(parser_t *parser)
+parser_parse_variable_declaration(parser_t *parser)
 {
   token_t variable_name;
+
+  if (!drop_expected_token(parser, TOKEN_KEYWORD_LET)) {
+    return NULL;
+  }
+
   if (!expected_token(&variable_name, parser, TOKEN_NAME)) {
     return NULL;
   }
@@ -340,13 +345,7 @@ is_next_statement_a_variable_declaration(parser_t *parser)
   token_t token;
   lexer_peek_next_token(parser->lexer, &token);
 
-  if (token.kind != TOKEN_NAME) {
-    return false;
-  }
-
-  lexer_lookahead(parser->lexer, &token, 2);
-
-  return token.kind == TOKEN_COLON;
+  return token.kind == TOKEN_KEYWORD_LET;
 }
 
 static bool
@@ -440,7 +439,7 @@ parser_parse_block_declarations(parser_t *parser)
     }
 
     if (is_next_statement_a_variable_declaration(parser)) {
-      ast_node_t *variable_node = parser_parse_variable_definition(parser);
+      ast_node_t *variable_node = parser_parse_variable_declaration(parser);
 
       if (variable_node == NULL) {
         scope_leave(parser->scope);
@@ -493,6 +492,10 @@ parser_parse_function_declaration(parser_t *parser)
 {
   token_t func_name_token;
 
+  if (!drop_expected_token(parser, TOKEN_KEYWORD_FN)) {
+    return NULL;
+  }
+
   if (!expected_token(&func_name_token, parser, TOKEN_NAME)) {
     return NULL;
   }
diff --git a/test/lexer_test.c b/test/lexer_test.c
index 5326f5c..a3f644d 100644
--- a/test/lexer_test.c
+++ b/test/lexer_test.c
@@ -51,6 +51,22 @@ assert_token_at(char *source, int token_index, token_kind_t expected_kind, char
   assert_string_equal(expected, actual);
 }
 
+void
+assert_keyword(char *source, token_kind_t expected_kind)
+{
+  assert_token_at(source, 0, expected_kind, source);
+}
+
+static MunitResult
+test_tokenize_keywords(const MunitParameter params[], void *user_data_or_fixture)
+{
+  assert_keyword("return", TOKEN_KEYWORD_RETURN);
+  assert_keyword("fn", TOKEN_KEYWORD_FN);
+  assert_keyword("let", TOKEN_KEYWORD_LET);
+
+  return MUNIT_OK;
+}
+
 static MunitResult
 test_tokenize_number(const MunitParameter params[], void *user_data_or_fixture)
 {
@@ -106,6 +122,7 @@ test_peek_next_token(const MunitParameter params[], void *user_data_or_fixture)
 
 static MunitTest tests[] = {
   { "/test_tokenize_digit", test_tokenize_number, NULL, NULL, MUNIT_TEST_OPTION_NONE, NULL },
+  { "/test_tokenize_keywords", test_tokenize_keywords, NULL, NULL, MUNIT_TEST_OPTION_NONE, NULL },
   { "/test_tokenize_op", test_tokenize_op, NULL, NULL, MUNIT_TEST_OPTION_NONE, NULL },
   { "/test_tokenize_unknown", test_tokenize_unknown, NULL, NULL, MUNIT_TEST_OPTION_NONE, NULL },
   { "/test_peek_next_token", test_peek_next_token, NULL, NULL, MUNIT_TEST_OPTION_NONE, NULL },
diff --git a/test/parser_test.c b/test/parser_test.c
index 2d749da..a1614b9 100644
--- a/test/parser_test.c
+++ b/test/parser_test.c
@@ -64,7 +64,7 @@ test_parse_function(const MunitParameter params[], void *user_data_or_fixture)
   lexer_t lexer;
   scope_t *scope = scope_new();
 
-  make_lexer_from_static_src(&lexer, "main(): i32 { \nreturn 42;\n }");
+  make_lexer_from_static_src(&lexer, "fn main(): i32 { \nreturn 42;\n }");
   parser_init(&parser, &lexer, scope);
   ast_node_t *ast_function = parser_parse_function_declaration(&parser);
 
@@ -98,7 +98,7 @@ test_parse_variable_definition(const MunitParameter params[], void *user_data_or
   lexer_t lexer;
   scope_t *scope = scope_new();
 
-  make_lexer_from_static_src(&lexer, "main(): i32 { \nvariable : i32 = 42; \nreturn variable;\n }");
+  make_lexer_from_static_src(&lexer, "fn main(): i32 { \nlet variable: i32 = 42; \nreturn variable;\n }");
   parser_init(&parser, &lexer, scope);
   ast_node_t *ast_function = parser_parse_function_declaration(&parser);
 
@@ -203,31 +203,30 @@ assert_string_view_equal(char *expected, string_view_t actual)
 static MunitResult
 test_parse_basic_syntax_errors(const MunitParameter params[], void *user_data_or_fixture)
 {
-  assert_parser_error("(): i32 { return 42; }", "expected 'TOKEN_NAME' but got '('");
-  assert_parser_error("main): i32 { return 42; }", "expected '(' but got ')'");
-  assert_parser_error("main(: i32 { return 42; }", "expected ')' but got ':'");
-  assert_parser_error("main() i32 { return 42; }", "expected ':' but got 'TOKEN_NAME'");
-  assert_parser_error("main(): { return 42; }", "expected 'TOKEN_NAME' but got '{'");
-  assert_parser_error("main(): i32 return 42; }", "expected '{' but got 'return'");
-  assert_parser_error("main(): i32 { 42; }", "unexpected token 'TOKEN_NUMBER' value='42'");
-  assert_parser_error("main(): i32 { return; }", "unexpected '; (;)' token");
-  assert_parser_error("main(): i32 { return 42;", "expected '}' but got end of file");
-  assert_parser_error("main(): beff { return 42; }", "type 'beff' is not defined");
-  assert_parser_error("main(): i32 { return b; }", "identifier 'b' not defined");
-  assert_parser_error("main(): i32 { b = 1; return b; }", "trying to assign 'b' before defining it.");
+  assert_parser_error("main(): i32 { return 42; }", "expected 'fn' but got 'TOKEN_NAME'");
+  assert_parser_error("fn (): i32 { return 42; }", "expected 'TOKEN_NAME' but got '('");
+  assert_parser_error("fn main): i32 { return 42; }", "expected '(' but got ')'");
+  assert_parser_error("fn main(: i32 { return 42; }", "expected ')' but got ':'");
+  assert_parser_error("fn main() i32 { return 42; }", "expected ':' but got 'TOKEN_NAME'");
+  assert_parser_error("fn main(): { return 42; }", "expected 'TOKEN_NAME' but got '{'");
+  assert_parser_error("fn main(): i32 return 42; }", "expected '{' but got 'return'");
+  assert_parser_error("fn main(): i32 { 42; }", "unexpected token 'TOKEN_NUMBER' value='42'");
+  assert_parser_error("fn main(): i32 { return; }", "unexpected '; (;)' token");
+  assert_parser_error("fn main(): i32 { return 42;", "expected '}' but got end of file");
+  assert_parser_error("fn main(): beff { return 42; }", "type 'beff' is not defined");
+  assert_parser_error("fn main(): i32 { return b; }", "identifier 'b' not defined");
+  assert_parser_error("fn main(): i32 { b = 1; return b; }", "trying to assign 'b' before defining it.");
   // FIXME: once function calls are implemented, this error should inform that
   // neither a variable or function call was found.
-  assert_parser_error("main(): i32 { oxi 42; }", "unexpected token 'TOKEN_NAME' value='oxi'");
+  assert_parser_error("fn main(): i32 { oxi 42; }", "unexpected token 'TOKEN_NAME' value='oxi'");
 
   return MUNIT_OK;
 }
 
 static MunitTest tests[] = {
   { "/test_parse_function", test_parse_function, NULL, NULL, MUNIT_TEST_OPTION_NONE, NULL },
-  { "/test_parse_function", test_parse_function, NULL, NULL, MUNIT_TEST_OPTION_NONE, NULL },
   { "/test_parse_basic_syntax_errors", test_parse_basic_syntax_errors, NULL, NULL, MUNIT_TEST_OPTION_NONE, NULL },
   { "/test_parse_arithmetic_expression", test_parse_arithmetic_expression, NULL, NULL, MUNIT_TEST_OPTION_NONE, NULL },
-  { "/test_parse_arithmetic_expression", test_parse_arithmetic_expression, NULL, NULL, MUNIT_TEST_OPTION_NONE, NULL },
   { "/test_parse_variable_definition", test_parse_variable_definition, NULL, NULL, MUNIT_TEST_OPTION_NONE, NULL },
   { NULL, NULL, NULL, NULL, MUNIT_TEST_OPTION_NONE, NULL }
 };