Initial commit

2025-12-13 18:07:26 +11:00
commit f1b156ce4d
3 changed files with 164 additions and 0 deletions
--- a/src/main.cpp
+++ b/src/main.cpp
@@ -0,0 +1,162 @@
+#include <groundvm.h>
+#include <vector>
+#include <string>
+#include <fstream>
+#include <iostream>
+#include <sstream>
+#include <optional>
+
+namespace HighGround {
+
+    namespace Parser {
+
+        enum class HGNodeType {
+            Add, Subtract, Equal, Set, While, If, None, Root, CodeBlock
+        };
+
+        class HGNode;
+
+        class HGNode {
+            HGNodeType nodeType = HGNodeType::None;
+            std::vector<HGNode> children;
+            public:
+            HGNode(HGNodeType nodeType) : nodeType(nodeType) {}
+            HGNode() = default;
+            void addNode(HGNode in) {
+                children.push_back(in);
+            }
+            std::vector<GroundInstruction> generateCode() {
+                std::vector<GroundInstruction> code;
+                return code;
+            }
+        };
+
+    } // namespace Parser
+    
+    class Lexer {
+        std::string input;
+        size_t size;
+        size_t current;
+
+        std::optional<char> peek(int ahead = 1) {
+            if (current + ahead < size) {
+                return input[current + ahead];
+            } else {
+                return {};
+            }
+        }
+
+        std::optional<char> consume() {
+            if (current < size) {
+                return input[current++];
+            } else {
+                return {};
+            }
+        }
+
+        public:
+        Lexer(std::string in) : input(in), size(in.size()) {};
+        std::vector<std::string> lex() {
+            current = 0;
+            std::vector<std::string> tokens;
+            std::string buf;
+            while (auto copt = consume()) {
+                char c = copt.value();
+                switch (c) {
+                    // tokens which are not followed by anything
+                    case '(':
+                    case ')':
+                    case '{':
+                    case '}': 
+                    {
+                        if (!buf.empty()) {
+                            tokens.push_back(buf);
+                            buf.clear();
+                        }
+                        tokens.push_back(std::string(1, c));
+                        break;
+                    }
+                    // tokens which may be followed by either themselves
+                    // or an equals sign
+                    case '+':
+                    case '-':
+                    {
+                        std::string newToken(1, c);
+                        auto tokenopt = peek();
+                        if (tokenopt) {
+                            char token = tokenopt.value();
+                            if (token == c || token == '=') {
+                                newToken += token;
+                                consume();
+                            }
+                        }
+                        if (!buf.empty()) {
+                            tokens.push_back(buf);
+                            buf.clear();
+                        }
+                        tokens.push_back(newToken);
+                        break;
+                    }
+                    // tokens which may be followed by an equals sign
+                    case '*':
+                    case '/':
+                    case '=':
+                    {
+                        std::string newToken(1, c);
+                        auto tokenopt = peek();
+                        if (tokenopt) {
+                            char token = tokenopt.value();
+                            if (token == '=') {
+                                newToken += token;
+                                consume();
+                            }
+                        }
+                        if (!buf.empty()) {
+                            tokens.push_back(buf);
+                            buf.clear();
+                        }
+                        tokens.push_back(newToken);
+                        break;
+                    }
+                    // tokens which do not need to be included
+                    case ' ':
+                    case '\n':
+                    {
+                        if (!buf.empty()) {
+                            tokens.push_back(buf);
+                            buf.clear();
+                        }
+                        break;
+                    }
+                    default: 
+                    {
+                        buf += c;
+                    }
+                }
+            }
+
+            if (!buf.empty()) {
+                tokens.push_back(buf);
+            }
+            return tokens;
+        }
+
+    };
+
+
+} // namespace HighGround
+
+
+int main(int argc, char** argv) {
+    if (argc < 2) {
+        std::cout << "Usage: " << argv[0] << " (file)\n"; 
+        exit(1);
+    }
+    std::ifstream file(argv[1]);
+    std::ostringstream ss;
+    ss << file.rdbuf();
+    auto lexed = HighGround::Lexer(ss.str()).lex();
+    for (const std::string& str : lexed) {
+        std::cout << str << "\n";
+    }
+}