initial commit

include/sedpp/Input.h

@@ -0,0 +1,42 @@
+#pragma once
+
+#include <cstdint>
+#include <string>
+#include <vector>
+
+#include "sedpp/Options.h"
+
+namespace sedpp {
+
+// A loaded input record is sed's starting pattern space for one cycle.  The
+// delimiter bit is carried separately so files without a trailing newline remain
+// distinguishable all the way to output.
+struct Record {
+  std::string text;
+  std::string fileName;
+  bool hadDelimiter = false;
+  bool lastInFile = false;
+  bool lastOverall = false;
+  std::uint64_t lineNumber = 0;
+  std::uint64_t fileLineNumber = 0;
+};
+
+// Reads stdin or input files eagerly into records.  Eager loading simplifies GNU
+// sed's $ address and in-place rewrite behavior, both of which need to know the
+// last record before the runner starts emitting final output.
+class InputLoader {
+public:
+  explicit InputLoader(const Options &options);
+
+  [[nodiscard]] std::vector<Record> loadAll() const;
+  [[nodiscard]] std::vector<Record> loadFile(const std::string &fileName) const;
+
+private:
+  [[nodiscard]] std::vector<Record> splitBuffer(const std::string &buffer,
+                                                const std::string &fileName,
+                                                std::uint64_t &globalLine) const;
+
+  const Options &options_;
+};
+
+} // namespace sedpp