lexer: Tokenize logical and bitwise operators

The followed logic operators were added to lexer:

  TOKEN_EQUAL                 ==
  TOKEN_NOT                   !
  TOKEN_NOT_EQUAL             !=
  TOKEN_GT                    >
  TOKEN_GT_EQUAL              >=
  TOKEN_LT                    <
  TOKEN_LT_EQUAL              <=
  TOKEN_AND                   &&
  TOKEN_OR                    ||

Bitwise operators were also added

  TOKEN_BITWISE_AND          &
  TOKEN_BITWISE_OR           |
  TOKEN_BITWISE_SHIFT_LEFT   <<
  TOKEN_BITWISE_SHIFT_RIGHT  >>
  TOKEN_BITWISE_XOR          ^
  TOKEN_BITWISE_NOT          ~

TOKEN_EQUAL '=' was renamed TOKEN_ASSIGN, and now TOKEN_EQUAL is used
for the logical comparator '=='.

Signed-off-by: Carlos Maniero <carlos@maniero.me>

1 files changed, 138 insertions, 14 deletions

diff --git a/src/lexer.c b/src/lexer.c
index 5a6d245..72c27cd 100644
--- a/src/lexer.c
+++ b/src/lexer.c
@@ -62,6 +62,17 @@ lexer_drop_spaces(lexer_t *lexer)
 }
 
 static void
+lexer_tokenize_from_given_position_to_cursor(lexer_t *lexer, token_t *token, size_t position, token_kind_t kind)
+{
+  token->kind = kind;
+  token->value = string_view_new(lexer->src + position, lexer->cur - position);
+  token->filepath = lexer->filepath;
+  token->row = lexer->row;
+  token->col = position - lexer->bol;
+  token->bol = lexer->bol;
+}
+
+static void
 lexer_tokenize_number(lexer_t *lexer, token_t *token)
 {
   size_t begin = lexer->cur;
@@ -69,12 +80,8 @@ lexer_tokenize_number(lexer_t *lexer, token_t *token)
   while (lexer_is_not_eof(lexer) && isdigit(lexer_current_char(lexer))) {
     lexer_drop_char(lexer);
   }
-  token->kind = TOKEN_NUMBER;
-  token->value = string_view_new(lexer->src + begin, lexer->cur - begin);
-  token->filepath = lexer->filepath;
-  token->row = lexer->row;
-  token->col = begin - lexer->bol;
-  token->bol = lexer->bol;
+
+  lexer_tokenize_from_given_position_to_cursor(lexer, token, begin, TOKEN_NUMBER);
 }
 
 static void
@@ -86,12 +93,7 @@ lexer_tokenize_name(lexer_t *lexer, token_t *token)
     lexer_drop_char(lexer);
   }
 
-  token->kind = TOKEN_NAME;
-  token->value = string_view_new(lexer->src + begin, lexer->cur - begin);
-  token->filepath = lexer->filepath;
-  token->row = lexer->row;
-  token->col = begin - lexer->bol;
-  token->bol = lexer->bol;
+  lexer_tokenize_from_given_position_to_cursor(lexer, token, begin, TOKEN_NAME);
 }
 
 static void
@@ -194,7 +196,99 @@ lexer_next_token(lexer_t *lexer, token_t *token)
   }
 
   if (lexer_current_char(lexer) == '=') {
-    lexer_define_literal_token_props(lexer, token, TOKEN_EQUAL);
+    lexer_define_literal_token_props(lexer, token, TOKEN_ASSIGN);
+    lexer_drop_char(lexer);
+
+    if (lexer_current_char(lexer) == '=') {
+      lexer_drop_char(lexer);
+      lexer_tokenize_from_given_position_to_cursor(lexer, token, lexer->cur - 2, TOKEN_EQUAL);
+      return;
+    }
+
+    return;
+  }
+
+  if (lexer_current_char(lexer) == '!') {
+    lexer_define_literal_token_props(lexer, token, TOKEN_NOT);
+    lexer_drop_char(lexer);
+
+    if (lexer_current_char(lexer) == '=') {
+      lexer_drop_char(lexer);
+      lexer_tokenize_from_given_position_to_cursor(lexer, token, lexer->cur - 2, TOKEN_NOT_EQUAL);
+      return;
+    }
+
+    return;
+  }
+
+  if (lexer_current_char(lexer) == '>') {
+    lexer_define_literal_token_props(lexer, token, TOKEN_GT);
+    lexer_drop_char(lexer);
+
+    if (lexer_current_char(lexer) == '=') {
+      lexer_drop_char(lexer);
+      lexer_tokenize_from_given_position_to_cursor(lexer, token, lexer->cur - 2, TOKEN_GT_EQUAL);
+      return;
+    }
+
+    if (lexer_current_char(lexer) == '>') {
+      lexer_drop_char(lexer);
+      lexer_tokenize_from_given_position_to_cursor(lexer, token, lexer->cur - 2, TOKEN_BITWISE_SHIFT_RIGHT);
+      return;
+    }
+    return;
+  }
+
+  if (lexer_current_char(lexer) == '<') {
+    lexer_define_literal_token_props(lexer, token, TOKEN_LT);
+    lexer_drop_char(lexer);
+
+    if (lexer_current_char(lexer) == '=') {
+      lexer_drop_char(lexer);
+      lexer_tokenize_from_given_position_to_cursor(lexer, token, lexer->cur - 2, TOKEN_LT_EQUAL);
+      return;
+    }
+
+    if (lexer_current_char(lexer) == '<') {
+      lexer_drop_char(lexer);
+      lexer_tokenize_from_given_position_to_cursor(lexer, token, lexer->cur - 2, TOKEN_BITWISE_SHIFT_LEFT);
+      return;
+    }
+    return;
+  }
+
+  if (lexer_current_char(lexer) == '&') {
+    lexer_define_literal_token_props(lexer, token, TOKEN_BITWISE_AND);
+    lexer_drop_char(lexer);
+
+    if (lexer_current_char(lexer) == '&') {
+      lexer_drop_char(lexer);
+      lexer_tokenize_from_given_position_to_cursor(lexer, token, lexer->cur - 2, TOKEN_AND);
+      return;
+    }
+    return;
+  }
+
+  if (lexer_current_char(lexer) == '|') {
+    lexer_define_literal_token_props(lexer, token, TOKEN_BITWISE_OR);
+    lexer_drop_char(lexer);
+
+    if (lexer_current_char(lexer) == '|') {
+      lexer_drop_char(lexer);
+      lexer_tokenize_from_given_position_to_cursor(lexer, token, lexer->cur - 2, TOKEN_OR);
+      return;
+    }
+    return;
+  }
+
+  if (lexer_current_char(lexer) == '^') {
+    lexer_define_literal_token_props(lexer, token, TOKEN_BITWISE_XOR);
+    lexer_drop_char(lexer);
+    return;
+  }
+
+  if (lexer_current_char(lexer) == '~') {
+    lexer_define_literal_token_props(lexer, token, TOKEN_BITWISE_NOT);
     lexer_drop_char(lexer);
     return;
   }
@@ -322,8 +416,38 @@ token_kind_to_str(token_kind_t kind)
       return "*";
     case TOKEN_SLASH:
       return "/";
-    case TOKEN_EQUAL:
+    case TOKEN_ASSIGN:
       return "=";
+    case TOKEN_EQUAL:
+      return "==";
+    case TOKEN_NOT:
+      return "!";
+    case TOKEN_NOT_EQUAL:
+      return "!=";
+    case TOKEN_GT:
+      return ">";
+    case TOKEN_GT_EQUAL:
+      return ">=";
+    case TOKEN_LT:
+      return "<";
+    case TOKEN_LT_EQUAL:
+      return "<=";
+    case TOKEN_AND:
+      return "&&";
+    case TOKEN_OR:
+      return "||";
+    case TOKEN_BITWISE_AND:
+      return "&";
+    case TOKEN_BITWISE_OR:
+      return "|";
+    case TOKEN_BITWISE_SHIFT_LEFT:
+      return "<<";
+    case TOKEN_BITWISE_SHIFT_RIGHT:
+      return ">>";
+    case TOKEN_BITWISE_XOR:
+      return "^";
+    case TOKEN_BITWISE_NOT:
+      return "~";
     case TOKEN_KEYWORD_RETURN:
       return "return";
     case TOKEN_KEYWORD_FN: