initial commit

2026-06-09 00:22:22 -05:00
commit c84de3418b
25 changed files with 3501 additions and 0 deletions
@@ -0,0 +1,45 @@
+#pragma once
+
+#include <string>
+#include <unordered_map>
+#include <vector>
+
+#include "sedpp/Command.h"
+#include "sedpp/Options.h"
+
+namespace sedpp {
+
+// Converts raw -e/-f script text into a flat Program.  The compiler is deliberately
+// syntax-oriented; command range state and "last regex" behavior are resolved by
+// Runner because they depend on input cycles.
+class Compiler {
+public:
+  explicit Compiler(const Options &options);
+
+  [[nodiscard]] Program compile(const std::vector<std::string> &scripts);
+  [[nodiscard]] const std::unordered_map<std::string, std::size_t> &labels() const {
+    return labels_;
+  }
+
+private:
+  // Script parsing is a small cursor parser because sed syntax is compact and
+  // delimiter-driven; tokenizing first tends to lose useful byte positions.
+  void compileOne(const std::string &script);
+  [[nodiscard]] std::optional<Address> parseAddress(const std::string &script,
+                                                    std::size_t &pos,
+                                                    bool secondAddress);
+  [[nodiscard]] std::string parseDelimited(const std::string &script,
+                                            std::size_t &pos, char delimiter,
+                                            bool regexMode = false,
+                                            bool unescapeDelimiter = false);
+  [[nodiscard]] std::string readUntilCommandEnd(const std::string &script,
+                                                std::size_t &pos);
+  static void skipBlanks(const std::string &script, std::size_t &pos);
+  static void skipSeparators(const std::string &script, std::size_t &pos);
+
+  const Options &options_;
+  Program program_;
+  std::unordered_map<std::string, std::size_t> labels_;
+};
+
+} // namespace sedpp