initial commit

include/sedpp/Command.h

@@ -0,0 +1,88 @@
+#pragma once
+
+#include <cstdint>
+#include <optional>
+#include <regex.h>
+#include <string>
+#include <vector>
+
+namespace sedpp {
+
+// Address kinds mirror GNU sed's address grammar.  Step, Plus, and Modulo are
+// GNU extensions: FIRST~STEP as a standalone address, and +N/~N only as the
+// second half of a range.
+enum class AddressKind {
+  None,
+  Line,
+  Last,
+  Regex,
+  Step,
+  Plus,
+  Modulo,
+};
+
+// A parsed address keeps both the original text and numeric forms.  The text is
+// still needed for special cases such as line address 0 and for empty regexes,
+// which mean "reuse the previous regular expression" at runtime.
+struct Address {
+  AddressKind kind = AddressKind::None;
+  std::string text;
+  std::uint64_t first = 0;
+  std::uint64_t second = 0;
+};
+
+// Parsed state for the s command.  flagOrder preserves p/e ordering because GNU
+// sed changes observable output depending on whether printing happens before or
+// after shell execution.
+struct Substitute {
+  std::string pattern;
+  std::string replacement;
+  bool global = false;
+  bool print = false;
+  bool execute = false;
+  bool ignoreCase = false;
+  bool multiline = false;
+  std::uint64_t occurrence = 0;
+  std::string flagOrder;
+  std::string writeFile;
+};
+
+// y transliteration operates on decoded character strings, not regex syntax.
+struct Translate {
+  std::string from;
+  std::string to;
+};
+
+// A compiled sed command.  The opcode selects which optional payload is valid:
+// text for a/i/c/r/R/w/W/e, label for :/b/t/T, substitute for s, translate for y.
+struct Command {
+  std::optional<Address> firstAddress;
+  std::optional<Address> secondAddress;
+  bool negate = false;
+  char opcode = '\0';
+
+  // These fields are intentionally plain.  sed commands are compact and varied;
+  // a tiny tagged structure is easier to audit than a deep inheritance tree.
+  std::string text;
+  std::string label;
+  std::optional<Substitute> substitute;
+  std::optional<Translate> translate;
+  int exitCode = 0;
+  int lineLength = 0;
+
+  // Runtime range state.  GNU sed keeps range activation per compiled command,
+  // not globally, so the state lives with the command object.
+  bool rangeActive = false;
+  bool rangeJustStarted = false;
+  std::uint64_t rangeCountdown = 0;
+  bool zeroRangeStarted = false;
+};
+
+// The compiler flattens the sed script into one vector.  Braced blocks remain
+// explicit { and } opcodes so the runner can skip inactive blocks without a
+// second tree-shaped representation.
+struct Program {
+  std::vector<Command> commands;
+};
+
+} // namespace sedpp