Make append buffer contiguous and update iterators

include/vase/buffer/append.h

@@ -5,19 +5,19 @@
 #include "pch.h"
 
 struct AppendBuffer : Buffer {
-  using TChunk = char[APPEND_CHUNK_SIZE];
-  std::vector<TChunk *> buf;
-  TChunk *t_current;
-  uint64_t current_offset{0};
-  uint64_t t_offset{0};
-
-  AppendBuffer();
+  char *buf = nullptr;
+  uint64_t allocated_capacity = 0;
+  uint64_t current_size = 0;
+  int fd = -1;
 
+  AppendBuffer(std::filesystem::path base_dir);
   ~AppendBuffer();
 
-  uint64_t append(char c);
-  uint64_t append(const char *text, uint64_t length);
+  uint64_t append(const char c);
+  uint64_t append(const char *text, uint64_t len);
+  const char *read(uint64_t pos) override;
+  uint64_t length() override;
 
-  const char *read(uint64_t pos, uint64_t *out_len);
-  inline uint64_t length();
+private:
+  void grow(uint64_t len);
 };

include/vase/buffer/buffer.h

@@ -3,7 +3,7 @@
 #include "pch.h"
 
 struct Buffer {
-  virtual const char *read(uint64_t pos, uint64_t *out_len) = 0;
-  virtual inline uint64_t length() = 0;
+  virtual const char *read(uint64_t pos) = 0;
+  virtual uint64_t length() = 0;
   virtual ~Buffer() = default;
 };

include/vase/buffer/original.h

@@ -12,6 +12,6 @@ struct OriginalBuffer : Buffer {
   ~OriginalBuffer();
 
   void initialize();
-  const char *read(uint64_t pos, uint64_t *out_len);
-  uint64_t length();
+  const char *read(uint64_t pos) override;
+  uint64_t length() override;
 };

include/vase/constants.h

@@ -2,10 +2,4 @@
 
 #include "pch.h"
 
-constexpr uint64_t APPEND_CHUNK_SIZE = 64 * 1024;
 constexpr uint64_t PETAL_SIZE_MAX = 32 * 1024;
-
-static_assert(
-  APPEND_CHUNK_SIZE > PETAL_SIZE_MAX,
-  "PETAL_SIZE_MAX must be smaller than APPEND_CHUNK_SIZE"
-);

include/vase/iterators/iter.h

@@ -0,0 +1,9 @@
+#pragma once
+
+#include "../shard.h"
+#include "pch.h"
+
+enum struct Direction : uint8_t {
+  Forward,
+  Backward
+};

include/vase/iterators/line.h

@@ -1,22 +1,20 @@
 #pragma once
 
 #include "../shard.h"
-#include "chunk.h"
 #include "pch.h"
+#include "petal.h"
 
 struct LineIterator {
-  ChunkIterator it;
-  Direction dir;
-
+  PetalIterator it;
   const char *chunk;
   uint64_t len;
-  uint64_t offset = 0;
-  uint64_t chunk_offset = 0;
-  bool at_end = false;
+  Direction dir;
+  uint64_t current_offset = 0;
+  uint64_t last_line_offset = 0;
 
-  LineIterator(Shard *r, uint64_t start_line, Direction dir);
+  LineIterator(Shard *root, uint64_t line_num, Direction dir);
   ~LineIterator() = default;
 
-  bool next(std::string &line);
+  bool next(std::string *line);
   uint64_t byte_offset();
 };

include/vase/iterators/petal.h

@@ -1,30 +1,27 @@
 #pragma once
 
 #include "../shard.h"
+#include "iter.h"
 #include "pch.h"
 
-enum struct Direction : uint8_t {
-  Forward,
-  Backward
-};
-
-struct ChunkIterator {
+struct PetalIterator {
   Direction dir;
   Shard *root = nullptr;
   std::vector<Shard *> stack;
   Petal *petal = nullptr;
-  uint64_t petal_offset = 0;
   uint64_t global_offset = 0;
+  uint64_t last_offset = 0;
+  uint64_t petal_offset = 0;
   uint64_t global_line = 0;
-  bool at_end = false;
 
-  ChunkIterator(Shard *r, Direction dir);
-
-  ~ChunkIterator();
+  PetalIterator(Shard *r, Direction dir);
+  ~PetalIterator() = default;
 
   void seek_offset(uint64_t offset);
   void seek_line(uint64_t offset);
   bool next(const char **data, uint64_t *out_len);
   uint64_t byte_offset();
-  uint64_t line_offset();
+
+private:
+  Petal *_next(uint64_t *offset);
 };

include/vase/shard.h

@@ -7,7 +7,7 @@
 #include "utils/utils.h"
 
 struct Shard {
-  enum struct ShardKind : uint8_t {
+  enum struct Kind : uint8_t {
     Branch,
     Petal
   } kind;
@@ -19,7 +19,7 @@ struct Shard {
 
   std::atomic_uint64_t refs;
 
-  Shard(ShardKind kind, uint64_t length, uint64_t lines, uint8_t height)
+  Shard(Kind kind, uint64_t length, uint64_t lines, uint8_t height)
       : kind(kind), height(height), length(length), lines(lines), refs(1) {};
 
   static void retain(Shard *n);
@@ -27,14 +27,13 @@ struct Shard {
 
   static Shard *from_file(std::filesystem::path path, OriginalBuffer *b);
   static std::vector<Shard *> from_swap(std::filesystem::path path, OriginalBuffer *b);
-  static Shard *new_empty(OriginalBuffer *b);
 
   static std::pair<Shard *, Shard *> split(Shard *n, uint64_t offset);
   static Shard *concat(Shard *left, Shard *right);
   static Shard *merge(Shard *a, Shard *b);
   static Shard *merge_leaves(Shard *a, Shard *b);
-  static Shard *append_leaf(Shard *root, Shard *leaf);
-  static Shard *build_balanced(Shard **pieces, uint64_t lo, uint64_t hi);
+  static Shard *append(Shard *root, Shard *leaf);
+  static Shard *build(Shard **pieces, uint64_t lo, uint64_t hi);
 };
 
 struct Branch : Shard {
@@ -43,7 +42,7 @@ struct Branch : Shard {
 
   Branch(Shard *l, Shard *r)
       : Shard(
-          ShardKind::Branch,
+          Kind::Branch,
           l->length + r->length,
           l->lines + r->lines,
           1 + std::max(l->height, r->height)
@@ -56,10 +55,60 @@ struct Branch : Shard {
 
 struct Petal : Shard {
   Buffer *source;
-
   uint64_t pos;
 
   Petal(uint64_t length, uint64_t lines, Buffer *source, uint64_t pos)
-      : Shard(ShardKind::Petal, length, lines, 1),
+      : Shard(Kind::Petal, length, lines, 1),
         source(source), pos(pos) {};
 };
+
+extern inline void dump_shard(Shard *node, int depth = 0) {
+  if (!node) {
+    std::cout << std::string(depth * 2, ' ') << "<null>\n";
+    return;
+  }
+
+  std::string indent(depth * 2, ' ');
+
+  std::cout << indent
+            << "Shard@" << node
+            << " kind=";
+
+  switch (node->kind) {
+  case Shard::Kind::Branch:
+    std::cout << "Branch";
+    break;
+  case Shard::Kind::Petal:
+    std::cout << "Petal";
+    break;
+  }
+
+  std::cout
+    << " height=" << unsigned(node->height)
+    << " length=" << node->length
+    << " lines=" << node->lines
+    << " refs=" << node->refs.load()
+    << "\n";
+
+  if (node->kind == Shard::Kind::Branch) {
+    auto *branch = static_cast<Branch *>(node);
+
+    std::cout << indent << "  left:\n";
+    dump_shard(branch->left, depth + 2);
+
+    std::cout << indent << "  right:\n";
+    dump_shard(branch->right, depth + 2);
+  } else {
+    auto *petal = static_cast<Petal *>(node);
+
+    std::cout
+      << indent << "  source=" << petal->source
+      << " pos=" << petal->pos
+      << " length=" << petal->length
+      << " lines=" << petal->lines
+      << "\n";
+  }
+
+  if (!depth)
+    std::cout << "\n\n";
+}

include/vase/vase.h

@@ -48,7 +48,6 @@ struct Vase {
   std::filesystem::path swapdir;
 
   Vase(std::filesystem::path path, std::filesystem::path swapdir);
-
   ~Vase();
 
   uint64_t length();
@@ -80,7 +79,6 @@ struct Vase {
 
   bool undo();
   bool redo();
-
   void snapshot();
   void prune_history(uint64_t n);