Merge pull request #40 from Superjom/feature/refactor_storage_interface

CMakeLists.txt

@@ -30,5 +30,6 @@ add_subdirectory(${CMAKE_CURRENT_SOURCE_DIR}/visualdl/backend/logic)
 
 add_executable(vl_test
         ${PROJECT_SOURCE_DIR}/visualdl/backend/test.cc
+        ${PROJECT_SOURCE_DIR}/visualdl/backend/storage/storage_test.cc
         ${PROJECT_SOURCE_DIR}/visualdl/backend/logic/im_test.cc)
-target_link_libraries(vl_test storage im gtest glog protobuf gflags)
+target_link_libraries(vl_test storage im gtest glog protobuf gflags pthread)

visualdl/backend/logic/im.cc

@@ -26,24 +26,21 @@ int ReserviorSample(int num_samples, int num_records) {
 }
 
 void InformationMaintainer::SetPersistDest(const std::string &path) {
-  CHECK(storage_.mutable_data()->dir().empty())
+  CHECK(storage_->mutable_data()->dir().empty())
       << "duplicate set storage's path";
-  storage_.mutable_data()->set_dir(path);
+  storage_->mutable_data()->set_dir(path);
 }
 
 storage::Tablet *InformationMaintainer::AddTablet(const std::string &tag,
                                                   int num_samples) {
-  auto *tablet = storage_.Find(tag);
-  if (!tablet) {
-    tablet = storage_.Add(tag, num_samples);
-  }
+  auto tablet = storage_->NewTablet(tag, num_samples);
   return tablet;
 }
 
 void InformationMaintainer::AddRecord(const std::string &tag,
                                       const storage::Record &data) {
-  auto *tablet = storage_.Find(tag);
-  CHECK(tablet);
+  auto *tablet = storage_->tablet(tag);
+  CHECK(tablet) << "no tablet called " << tag;
 
   auto num_records = tablet->total_records();
   const auto num_samples = tablet->num_samples();
@@ -59,9 +56,9 @@ void InformationMaintainer::AddRecord(const std::string &tag,
 
   storage::Record *record;
   if (offset >= num_records) {
-    record = storage_.NewRecord(tag);
+    record = tablet->add_records();
   } else {
-    record = storage_.GetRecord(tag, offset);
+    record = tablet->mutable_records(offset);
   }
 
   *record = data;
@@ -76,10 +73,10 @@ void InformationMaintainer::Clear() {
 }
 
 void InformationMaintainer::PersistToDisk() {
-  CHECK(!storage_.data().dir().empty()) << "path of storage should be set";
+  CHECK(!storage_->data().dir().empty()) << "path of storage should be set";
   // TODO make dir first
   // MakeDir(storage_.data().dir());
-  storage_.Save(storage_.data().dir() + "/storage.pb");
+  storage_->PersistToDisk();
 }
 
 }  // namespace visualdl

visualdl/backend/logic/im.h

 #ifndef VISUALDL_BACKEND_LOGIC_IM_H
 #define VISUALDL_BACKEND_LOGIC_IM_H
 
+#include <glog/logging.h>
+#include <memory>
 #include <string>
 
 #include "visualdl/backend/storage/storage.h"
@@ -13,7 +15,15 @@ namespace visualdl {
  */
 class InformationMaintainer final {
 public:
-  InformationMaintainer() {}
+  InformationMaintainer(StorageBase::Type type = StorageBase::Type::kMemory) {
+    switch (type) {
+      case StorageBase::Type::kMemory:
+        storage_.reset(new MemoryStorage);
+        break;
+      default:
+        CHECK(false) << "Unsupported storage kind " << type;
+    }
+  }
 
   static InformationMaintainer &Global() {
     static InformationMaintainer *x = new InformationMaintainer();
@@ -46,10 +56,10 @@ public:
    */
   void PersistToDisk();
 
-  Storage &storage() { return storage_; }
+  StorageBase &storage() { return *storage_; }
 
 private:
-  Storage storage_;
+  std::unique_ptr<StorageBase> storage_;
 };
 
 }  // namespace visualdl

visualdl/backend/logic/im_test.cc

@@ -11,19 +11,25 @@ protected:
   InformationMaintainer &im = InformationMaintainer::Global();
 };
 
-TEST_F(ImTester, AddTablet) { im.AddTablet("tag0", 20); }
+TEST_F(ImTester, AddTablet) {
+  im.Clear();
+  im.AddTablet("tag0", 20);
+}
 
 TEST_F(ImTester, AddRecord) {
-  storage::Record rcd;
-  rcd.set_dtype(storage::DataType::kInt32s);
+  im.Clear();
+
+  im.AddTablet("tag0", 20);
   for (int i = 0; i < 100; i++) {
+    storage::Record rcd;
+    rcd.set_dtype(storage::DataType::kInt32s);
     for (int j = 0; j < 10; j++) {
       rcd.add_data()->add_i32s(i * 20 + j);
     }
     im.AddRecord("tag0", rcd);
   }
 
-  ASSERT_EQ(im.storage().Find("tag0")->records_size(), 20UL);
+  ASSERT_EQ(im.storage().tablet("tag0")->records_size(), 100UL);
 }
 
 }  // namespace visualdl

visualdl/backend/logic/sdk.h

@@ -98,7 +98,7 @@ public:
         InformationMaintainer::Global().storage().mutable_data());
   }
   TabletHelper tablet(const std::string &tag) {
-    return TabletHelper(InformationMaintainer::Global().storage().Find(tag));
+    return TabletHelper(InformationMaintainer::Global().storage().tablet(tag));
   }
   TabletHelper AddTablet(const std::string &tag, int num_samples) {
     return TabletHelper(

visualdl/backend/storage/storage.cc

@@ -2,62 +2,69 @@
 #include <fstream>
 
 #include "visualdl/backend/storage/storage.h"
+#include "visualdl/backend/utils/filesystem.h"
 
 namespace visualdl {
 
-storage::Tablet *Storage::Add(const std::string &tag, int num_samples) {
-  auto *tablet = &(*proto_.mutable_tablets())[tag];
-  tablet->set_num_samples(num_samples);
-  return tablet;
+std::string GenPathFromTag(const std::string &dir, const std::string &tag) {
+  return dir + "/" + tag;
 }
 
-storage::Tablet *Storage::Find(const std::string &tag) {
-  auto it = proto_.mutable_tablets()->find(tag);
-  if (it != proto_.tablets().end()) {
+const std::string StorageBase::meta_file_name = "storage.meta";
+
+storage::Tablet *MemoryStorage::NewTablet(const std::string &tag,
+                                          int num_samples) {
+  auto it = tablets_.find(tag);
+  if (it == tablets_.end()) {
+    // create new tablet
+    tablets_[tag] = storage::Tablet();
+    tablets_[tag].set_tag(tag);
+    *storage_.add_tags() = tag;
+  } else {
     return &it->second;
   }
-  return nullptr;
-}
-
-storage::Record *Storage::NewRecord(const std::string &tag) {
-  auto *tablet = Find(tag);
-  CHECK(tablet) << "Tablet" << tag << " should be create first";
-  auto *record = tablet->mutable_records()->Add();
-  // increase num_records
-  int num_records = tablet->total_records();
-  tablet->set_total_records(num_records + 1);
-  return record;
+  return &tablets_[tag];
 }
-storage::Record *Storage::GetRecord(const std::string &tag, int offset) {
-  auto *tablet = Find(tag);
-  CHECK(tablet) << "Tablet" << tag << " should be create first";
 
-  auto num_records = tablet->total_records();
-  CHECK_LT(offset, num_records) << "invalid offset";
-  return tablet->mutable_records()->Mutable(offset);
+storage::Tablet *MemoryStorage::tablet(const std::string &tag) {
+  auto it = tablets_.find(tag);
+  CHECK(it != tablets_.end()) << "tablet tagged as " << tag << " not exists";
+  return &it->second;
 }
 
-void Storage::Save(const std::string &path) const {
-  std::ofstream file(path, file.binary | file.out);
-  CHECK(file.is_open()) << "can't open path " << path;
-  auto str = Serialize();
-  file.write(str.c_str(), str.size());
-}
-
-void Storage::Load(const std::string &path) {
-  std::ifstream file(path, file.binary);
-  CHECK(file.is_open()) << "can't open path " << path;
-  size_t size = file.tellg();
-  std::string buffer(size, ' ');
-  file.seekg(0);
-  file.read(&buffer[0], size);
-  DeSerialize(buffer);
+void MemoryStorage::PersistToDisk() const {
+  VLOG(3) << "persist storage to disk path " << storage_.dir();
+  // make a directory if not exist
+  fs::TryMkdir(storage_.dir());
+  // write storage out
+  CHECK(!storage_.dir().empty()) << "storage's dir should be set first";
+  const auto meta_path = storage_.dir() + "/" + meta_file_name;
+  fs::Write(meta_path, fs::Serialize(storage_));
+  // write all the tablets
+  for (auto tag : storage_.tags()) {
+    auto path = GenPathFromTag(storage_.dir(), tag);
+    auto it = tablets_.find(tag);
+    CHECK(it != tablets_.end());
+    fs::Write(path, fs::Serialize(it->second));
+  }
 }
 
-std::string Storage::Serialize() const { return proto_.SerializeAsString(); }
+void MemoryStorage::LoadFromDisk(const std::string &dir) {
+  VLOG(3) << "load storage from disk path " << dir;
+  CHECK(!dir.empty()) << "dir is empty";
+  // load storage
+  const auto meta_path = dir + "/" + meta_file_name;
+  auto buf = fs::Read(meta_path);
+  CHECK(fs::DeSerialize(&storage_, buf))
+      << "failed to parse protobuf loaded from " << meta_path;
 
-void Storage::DeSerialize(const std::string &data) {
-  proto_.ParseFromString(data);
+  // load all the tablets
+  for (int i = 0; i < storage_.tags_size(); i++) {
+    std::string tag = storage_.tags(i);
+    auto path = GenPathFromTag(storage_.dir(), tag);
+    CHECK(tablets_[tag].ParseFromString(fs::Read(path)))
+        << "failed to parse protobuf text loaded from " << path;
+  }
 }
 
 }  // namespace visualdl

visualdl/backend/storage/storage.h

@@ -2,68 +2,84 @@
 #define VISUALDL_STORAGE_H
 
 #include <time.h>
+#include <map>
 #include <string>
 
 #include "visualdl/backend/storage/storage.pb.h"
 
 namespace visualdl {
 
-class Storage final {
+/*
+ * Generate a tablet path in disk from its tag.
+ */
+inline std::string GenPathFromTag(const std::string &dir,
+                                  const std::string &tag);
+
+/*
+ * Storage Interface. The might be a bunch of implementations, for example, a
+ * MemStorage that keep a copy of all the taplets in memory, can be changed with
+ * a higher performance; a DiskStorage that keep all the data in disk, apply to
+ * the scenerios where memory consumption should be considered.
+ */
+class StorageBase {
 public:
-  Storage() {
-    // set time stamp
-    time_t time0;
-    time(&time0);
-    proto_.set_timestamp(time0);
+  const static std::string meta_file_name;
+
+  enum Type { kMemory = 0, kDisk = 1 };
+
+  void SetStorage(const std::string &dir) {
+    time_t t;
+    time(&t);
+    storage_.set_timestamp(t);
+    storage_.set_dir(dir);
   }
 
   /*
-   * Add a new tablet named `tag`, the newly added instance will be returned.
+   * Create a new Tablet storage.
    */
-  storage::Tablet *Add(const std::string &tag, int num_samples);
+  virtual storage::Tablet *NewTablet(const std::string &tag,
+                                     int num_samples) = 0;
 
   /*
-   * Search the tablet named `tag`, if not exist, return nullptr.
+   * Get a tablet from memory, this can be viewed as a cache, if the storage is
+   * in disk, a hash map in memory will first load the corresponding Tablet
+   * Protobuf from disk and hold all the changes.
    */
-  storage::Tablet *Find(const std::string &tag);
+  virtual storage::Tablet *tablet(const std::string &tag) = 0;
 
   /*
-   * Append a new record to the tail of tablet.
+   * Persist the data from cache to disk. Both the memory storage or disk
+   * storage should write changes to disk for persistence.
    */
-  storage::Record *NewRecord(const std::string &tag);
+  virtual void PersistToDisk() const = 0;
 
   /*
-   * Get a record at `offset`, if the offset is not valid, yield a failed CHECK.
+   * Load data from disk.
    */
-  storage::Record *GetRecord(const std::string &tag, int offset);
+  virtual void LoadFromDisk(const std::string &dir) = 0;
 
-  /*
-   * Serialize this object to string and save it to a file.
-   */
-  void Save(const std::string &path) const;
+  storage::Storage *mutable_data() { return &storage_; }
+  const storage::Storage &data() { return storage_; }
 
-  /*
-   * Load the Protobuf message from a file.
-   */
-  void Load(const std::string &path);
+protected:
+  storage::Storage storage_;
+};
 
-  storage::Storage *mutable_data() { return &proto_; }
+/*
+ * Storage in Memory, that will support quick edits on data.
+ */
+class MemoryStorage final : public StorageBase {
+public:
+  storage::Tablet *NewTablet(const std::string &tag, int num_samples) override;
 
-  const storage::Storage &data() { return proto_; }
+  storage::Tablet *tablet(const std::string &tag) override;
 
-protected:
-  /*
-   * Serialize the Storage instance to string.
-   */
-  std::string Serialize() const;
+  void PersistToDisk() const override;
 
-  /*
-   * De-serialize from a string and update this Storage instance.
-   */
-  void DeSerialize(const std::string &data);
+  void LoadFromDisk(const std::string &dir) override;
 
 private:
-  storage::Storage proto_;
+  std::map<std::string, storage::Tablet> tablets_;
 };
 
 }  // namespace visualdl

visualdl/backend/storage/storage.proto

@@ -130,7 +130,9 @@ The Storage stores all the records.
 */
 message Storage {
   // tags to Tablet, should be thread safe if fix the keys after initialization.
+  // TODO to delete in the new storage interface.
   map<string, Tablet> tablets = 1;
+  repeated string tags = 4;
   string dir = 2;
   int64 timestamp = 3;
-}
+}
\ No newline at end of file

visualdl/backend/storage/storage_test.cc

+#include "visualdl/backend/storage/storage.h"
+
+#include <glog/logging.h>
+#include <gtest/gtest.h>
+
+namespace visualdl {
+using namespace std;
+
+class MemoryStorageTest : public ::testing::Test {
+public:
+  void SetUp() override { storage_.SetStorage("./tmp"); }
+
+  MemoryStorage storage_;
+};
+
+TEST_F(MemoryStorageTest, SetStorage) {
+  string dir = "./tmp";
+  storage_.SetStorage(dir);
+
+  ASSERT_EQ(storage_.data().dir(), dir);
+}
+
+TEST_F(MemoryStorageTest, AddTablet) {
+  // TODO need to escape tag as name
+  string tag = "add%20tag0";
+  storage_.NewTablet(tag, -1);
+
+  auto* tablet = storage_.tablet(tag);
+
+  ASSERT_TRUE(tablet != nullptr);
+  ASSERT_EQ(tablet->tag(), tag);
+}
+
+TEST_F(MemoryStorageTest, PersistToDisk) {
+  string tag = "add%20tag0";
+  storage_.NewTablet(tag, -1);
+
+  storage_.PersistToDisk();
+
+  MemoryStorage other;
+  other.LoadFromDisk("./tmp");
+  ASSERT_EQ(other.data().SerializeAsString(),
+            storage_.data().SerializeAsString());
+}
+
+}  // namespace visualdl

visualdl/backend/utils/filesystem.h

+#ifndef VISUALDL_BACKEND_UTILS_FILESYSTEM_H
+#define VISUALDL_BACKEND_UTILS_FILESYSTEM_H
+
+#include <google/protobuf/text_format.h>
+#include <sys/stat.h>
+#include <sys/types.h>
+#include <unistd.h>
+#include <fstream>
+
+namespace visualdl {
+
+namespace fs {
+
+template <typename T>
+std::string Serialize(const T& proto, bool human_readable = false) {
+  if (human_readable) {
+    std::string buffer;
+    google::protobuf::TextFormat::PrintToString(proto, &buffer);
+    return buffer;
+  }
+  return proto.SerializeAsString();
+}
+
+template <typename T>
+bool DeSerialize(T* proto, const std::string buf, bool human_readable = false) {
+  // NOTE human_readable not valid
+  if (human_readable) {
+    return google::protobuf::TextFormat::ParseFromString(buf, proto);
+  }
+  return proto->ParseFromString(buf);
+}
+
+void TryMkdir(const std::string& dir) {
+  VLOG(1) << "try to mkdir " << dir;
+  struct stat st = {0};
+  if (stat(dir.c_str(), &st) == -1) {
+    ::mkdir(dir.c_str(), 0700);
+  }
+}
+
+inline void Write(const std::string& path,
+                  const std::string& buffer,
+                  std::ios::openmode open_mode = std::ios::binary) {
+  VLOG(1) << "write to path " << path;
+  std::ofstream file(path, open_mode);
+  CHECK(file.is_open()) << "failed to open " << path;
+  file.write(buffer.c_str(), buffer.size());
+  file.close();
+}
+
+inline std::string Read(const std::string& path,
+                        std::ios::openmode open_mode = std::ios::binary) {
+  VLOG(1) << "read from path " << path;
+  std::string buffer;
+  std::ifstream file(path, open_mode | std::ios::ate);
+  CHECK(file.is_open()) << "failed to open " << path;
+  size_t size = file.tellg();
+  file.seekg(0);
+  buffer.resize(size);
+  file.read(&buffer[0], size);
+  return buffer;
+}
+
+}  // namespace fs
+
+}  // namespace visualdl
+
+#endif  // VISUALDL_BACKEND_UTILS_FILESYSTEM_H