From f7704aeca5963d9b54c20dfec4b6d1f00c7fa744 Mon Sep 17 00:00:00 2001
From: artpaul <artmkin@gmail.com>
Date: Thu, 19 Sep 2024 18:29:56 +0500
Subject: [PATCH] ignore partially written record at the end

---
 include/bitcask/bitcask.h | 10 +++++++++-
 src/bitcask.cc            | 26 +++++++++++++++++++++++---
 2 files changed, 32 insertions(+), 4 deletions(-)

diff --git a/include/bitcask/bitcask.h b/include/bitcask/bitcask.h
index 899fbdb..b1113ae 100644
--- a/include/bitcask/bitcask.h
+++ b/include/bitcask/bitcask.h
@@ -117,6 +117,8 @@ class Database {
     std::mutex fd_mutex;
     /// File descriptor.
     int fd{-1};
+    /// The file may have partially written record at the end.
+    const bool may_have_uncommitted;
 
     /// Total size of the file.
     /// The size is only updated on writing or set on loading.
@@ -131,7 +133,8 @@ class Database {
     std::atomic_uint64_t tombstones{0};
 
    public:
-    FileInfo(std::filesystem::path p, uint64_t s) noexcept : path(std::move(p)), size(s) {}
+    FileInfo(std::filesystem::path p, uint64_t s, bool uncommitted) noexcept
+        : path(std::move(p)), may_have_uncommitted(uncommitted), size(s) {}
 
 #ifndef NDEBUG
     ~FileInfo() {
@@ -251,6 +254,11 @@ class Database {
   std::error_code EnumerateIndex(const std::shared_ptr<FileInfo>& file, const FileSections::Range& range,
       const std::function<std::error_code(const Record&, const bool, std::string_view)>& cb) const;
 
+  /**
+   * Enumerates all entries in a file.
+   *
+   * @param file a file to enumerate.
+   */
   std::error_code EnumerateEntries(const std::shared_ptr<FileInfo>& file, const FileSections::Range& range,
       const std::function<std::error_code(const Record&, const bool, std::string_view, std::string_view)>&
           cb) const;
diff --git a/src/bitcask.cc b/src/bitcask.cc
index 36a42f5..ffd6b72 100644
--- a/src/bitcask.cc
+++ b/src/bitcask.cc
@@ -61,7 +61,15 @@ std::error_code LoadFromFile(int fd, void* buf, size_t len, size_t& off) noexcep
   return {};
 }
 
-/// Reads full content of an entry.
+/**
+ * Reads full content of an entry.
+ *
+ * @param fd
+ * @param offset beginning of the entry.
+ * @param check_crc check consistency of read data.
+ *
+ * @returns bytes read or an error code.
+ */
 std::pair<size_t, std::error_code> ReadEntryImpl(const int fd, const size_t offset, const bool check_crc,
     format::Entry& entry, std::string& key, std::string& value) {
   size_t current_offset = offset;
@@ -633,7 +641,7 @@ std::error_code Database::Initialize() {
       continue;
     }
 
-    auto file = std::make_shared<FileInfo>(entry.path(), entry.file_size());
+    auto file = std::make_shared<FileInfo>(entry.path(), entry.file_size(), index.value() == 0);
     // Open file for reading.
     if (auto ec = file->EnsureReadable()) {
       return ec;
@@ -644,6 +652,12 @@ std::error_code Database::Initialize() {
       if (IsNotFound(ec)) {
         continue;
       }
+      // A former active file may have a partially written record at the end in case of unexpected shutdown.
+      // Ignore it.
+      if (IsUnexpectedEndOfFile(ec) && file->may_have_uncommitted) {
+        continue;
+      }
+
       return ec;
     }
 
@@ -761,6 +775,12 @@ std::error_code Database::PackFiles(
     }
     // Enumerate all records in the source file.
     if (auto ec = EnumerateEntriesNoLock(file, cb)) {
+      // A former active file may have a partially written record at the end in case of unexpected shutdown.
+      // Ignore it.
+      if (IsUnexpectedEndOfFile(ec) && file->may_have_uncommitted) {
+        continue;
+      }
+
       read_lock.unlock();
       // TODO: finalize.
       return ec;
@@ -978,7 +998,7 @@ Database::FileInfoStatus Database::MakeWritableFile(const std::string& name, boo
       return {{}, std::make_error_code(static_cast<std::errc>(err))};
     }
   }
-  auto file = std::make_shared<FileInfo>(std::move(path), sizeof(format::Header));
+  auto file = std::make_shared<FileInfo>(std::move(path), sizeof(format::Header), name.starts_with("0-"));
   file->fd = fd;
   return {file, {}};
 }