implemented lexer for a more complex subset of the

monkey language

src/lexer/lexer.cpp

@@ -1,7 +1,9 @@
 #include "lexer.hpp"
 
 #include "token/token.hpp"
+#include "token/type.hpp"
 
+#include <cctype>
 #include <iostream>
 
 namespace lexer {
@@ -27,8 +29,40 @@ namespace lexer {
             return {token::type::LBRACE, c};
         case '}':
             return {token::type::RBRACE, c};
-        }
-        return {token::type::ILLEGAL, c};
+        default:
+            if (is_letter(c)) {
+                std::string identifier_or_keyword = read_string(c);
+                return {
+                    token::lookup_identifier(identifier_or_keyword),
+                    identifier_or_keyword
                 };
+            }
+            if (std::isdigit(c))
+                return {token::type::INT, read_int(c)};
+
+            return {token::type::ILLEGAL, c};
+        }
+    }
+
+    bool lexer::is_letter(char c) {
+        return c == '_' || std::isalpha(static_cast<unsigned char>(c));
+    }
+
+    std::string lexer::read_string(char first_char) {
+        std::string result;
+        result.push_back(first_char);
+        for (char c = input.peek(); is_letter(c); c = input.peek())
+            result.push_back(input.get());
+        return result;
+    }
+
+    std::string lexer::read_int(char first_digit) {
+        std::string result;
+        result.push_back(first_digit);
+        for (char c = input.peek(); std::isdigit(c); c = input.peek())
+            result.push_back(input.get());
+        return result;
+    }
+
 
 } // namespace lexer

src/lexer/lexer.hpp

@@ -6,5 +6,11 @@ namespace lexer {
     struct lexer {
         std::istream& input;
         token::token next_token();
+
+      private:
+        bool is_letter(char);
+
+        std::string read_string(char);
+        std::string read_int(char);
     };
 } // namespace lexer

src/token/type.cpp

@@ -1,5 +1,8 @@
 #include "type.hpp"
 
+#include <array>
+#include <unordered_map>
+
 namespace token {
 
     // Array mapping enum values to their string representations
@@ -18,4 +21,17 @@ namespace token {
         return os << "Unknown";
     }
 
+    static std::unordered_map<std::string, type> keywords{
+        {"fn", type::FUNCTION},
+        {"let", type::LET},
+    };
+
+    type lookup_identifier(std::string ident) {
+        try {
+            return keywords.at(ident);
+        } catch (const std::out_of_range&) {
+            return type::IDENTIFIER;
+        }
+    }
+
 } // namespace token

src/token/type.hpp

@@ -1,8 +1,6 @@
 #pragma once
 
-#include <array>
 #include <ostream>
-#include <string_view>
 
 namespace token {
 
@@ -30,5 +28,6 @@ namespace token {
 #undef X
     };
 
-    std::ostream& operator<<(std::ostream& os, type type);
+    std::ostream& operator<<(std::ostream&, type);
+    type lookup_identifier(std::string);
 } // namespace token