first draft

fix build

fix ObjectInfo creation

another cleaned attempt

fix schema id cache population

correct splitting of URI into parts

fix storage schema normalization

store per-file object_storage_ptr in object_info

make new storages properly

remove cloneObjectStorage

parse s3 uri differently

tmp

Author: zvonand

src/Disks/ObjectStorages/AzureBlobStorage/AzureObjectStorage.cpp

@@ -71,7 +71,7 @@ class AzureIteratorAsync final : public IObjectStorageIteratorAsync
     }
 
 private:
-    bool getBatchAndCheckNext(RelativePathsWithMetadata & batch) override
+    bool getBatchAndCheckNext(PathsWithMetadata & batch) override
     {
         ProfileEvents::increment(ProfileEvents::AzureListObjects);
         if (client->IsClientForDisk())
@@ -84,7 +84,7 @@ class AzureIteratorAsync final : public IObjectStorageIteratorAsync
 
         for (const auto & blob : blobs_list)
         {
-            batch.emplace_back(std::make_shared<RelativePathWithMetadata>(
+            batch.emplace_back(std::make_shared<PathWithMetadata>(
                 blob.Name,
                 ObjectMetadata{
                     static_cast<uint64_t>(blob.BlobSize),
@@ -166,7 +166,7 @@ ObjectStorageIteratorPtr AzureObjectStorage::iterate(const std::string & path_pr
     return std::make_shared<AzureIteratorAsync>(path_prefix, client_ptr, max_keys ? max_keys : settings_ptr->list_object_keys_size);
 }
 
-void AzureObjectStorage::listObjects(const std::string & path, RelativePathsWithMetadata & children, size_t max_keys) const
+void AzureObjectStorage::listObjects(const std::string & path, PathsWithMetadata & children, size_t max_keys) const
 {
     auto client_ptr = client.get();
 
@@ -188,7 +188,7 @@ void AzureObjectStorage::listObjects(const std::string & path, RelativePathsWith
 
         for (const auto & blob : blobs_list)
         {
-            children.emplace_back(std::make_shared<RelativePathWithMetadata>(
+            children.emplace_back(std::make_shared<PathWithMetadata>(
                 blob.Name,
                 ObjectMetadata{
                     static_cast<uint64_t>(blob.BlobSize),

src/Disks/ObjectStorages/AzureBlobStorage/AzureObjectStorage.h

@@ -37,7 +37,7 @@ class AzureObjectStorage : public IObjectStorage
 
     bool supportsListObjectsCache() override { return true; }
 
-    void listObjects(const std::string & path, RelativePathsWithMetadata & children, size_t max_keys) const override;
+    void listObjects(const std::string & path, PathsWithMetadata & children, size_t max_keys) const override;
 
     /// Sanitizer build may crash with max_keys=1; this looks like a false positive.
     ObjectStorageIteratorPtr iterate(const std::string & path_prefix, size_t max_keys) const override;

src/Disks/ObjectStorages/Cached/CachedObjectStorage.cpp

@@ -193,7 +193,7 @@ void CachedObjectStorage::copyObject( // NOLINT
     object_storage->copyObject(object_from, object_to, read_settings, write_settings, object_to_attributes);
 }
 
-void CachedObjectStorage::listObjects(const std::string & path, RelativePathsWithMetadata & children, size_t max_keys) const
+void CachedObjectStorage::listObjects(const std::string & path, PathsWithMetadata & children, size_t max_keys) const
 {
     object_storage->listObjects(path, children, max_keys);
 }

src/Disks/ObjectStorages/Cached/CachedObjectStorage.h

@@ -64,7 +64,7 @@ class CachedObjectStorage final : public IObjectStorage
         IObjectStorage & object_storage_to,
         std::optional<ObjectAttributes> object_to_attributes = {}) override;
 
-    void listObjects(const std::string & path, RelativePathsWithMetadata & children, size_t max_keys) const override;
+    void listObjects(const std::string & path, PathsWithMetadata & children, size_t max_keys) const override;
 
     ObjectMetadata getObjectMetadata(const std::string & path) const override;
 

src/Disks/ObjectStorages/HDFS/HDFSObjectStorage.cpp

@@ -167,7 +167,7 @@ ObjectMetadata HDFSObjectStorage::getObjectMetadata(const std::string & path) co
     return metadata;
 }
 
-void HDFSObjectStorage::listObjects(const std::string & path, RelativePathsWithMetadata & children, size_t max_keys) const
+void HDFSObjectStorage::listObjects(const std::string & path, PathsWithMetadata & children, size_t max_keys) const
 {
     initializeHDFSFS();
     LOG_TEST(log, "Trying to list files for {}", path);
@@ -203,7 +203,7 @@ void HDFSObjectStorage::listObjects(const std::string & path, RelativePathsWithM
         }
         else
         {
-            children.emplace_back(std::make_shared<RelativePathWithMetadata>(
+            children.emplace_back(std::make_shared<PathWithMetadata>(
                 String(file_path),
                 ObjectMetadata{
                     static_cast<uint64_t>(ls.file_info[i].mSize),

src/Disks/ObjectStorages/HDFS/HDFSObjectStorage.h

@@ -92,7 +92,7 @@ class HDFSObjectStorage : public IObjectStorage, public HDFSErrorWrapper
         const WriteSettings & write_settings,
         std::optional<ObjectAttributes> object_to_attributes = {}) override;
 
-    void listObjects(const std::string & path, RelativePathsWithMetadata & children, size_t max_keys) const override;
+    void listObjects(const std::string & path, PathsWithMetadata & children, size_t max_keys) const override;
 
     String getObjectsNamespace() const override { return ""; }
 

src/Disks/ObjectStorages/IObjectStorage.cpp

@@ -30,20 +30,20 @@ const MetadataStorageMetrics & IObjectStorage::getMetadataStorageMetrics() const
 
 bool IObjectStorage::existsOrHasAnyChild(const std::string & path) const
 {
-    RelativePathsWithMetadata files;
+    PathsWithMetadata files;
     listObjects(path, files, 1);
     return !files.empty();
 }
 
-void IObjectStorage::listObjects(const std::string &, RelativePathsWithMetadata &, size_t) const
+void IObjectStorage::listObjects(const std::string &, PathsWithMetadata &, size_t) const
 {
     throw Exception(ErrorCodes::NOT_IMPLEMENTED, "listObjects() is not supported");
 }
 
 
 ObjectStorageIteratorPtr IObjectStorage::iterate(const std::string & path_prefix, size_t max_keys) const
 {
-    RelativePathsWithMetadata files;
+    PathsWithMetadata files;
     listObjects(path_prefix, files, max_keys);
 
     return std::make_shared<ObjectStorageIteratorFromList>(std::move(files));
@@ -102,21 +102,14 @@ WriteSettings IObjectStorage::patchSettings(const WriteSettings & write_settings
     return write_settings;
 }
 
-RelativePathWithMetadata::RelativePathWithMetadata(const DataFileInfo & info, std::optional<ObjectMetadata> metadata_)
-    : metadata(std::move(metadata_))
-{
-    relative_path = info.file_path;
-    file_meta_info = info.file_meta_info;
-}
-
-std::string RelativePathWithMetadata::getPathOrPathToArchiveIfArchive() const
+std::string PathWithMetadata::getPathOrPathToArchiveIfArchive() const
 {
     if (isArchive())
         return getPathToArchive();
     return getPath();
 }
 
-RelativePathWithMetadata::CommandInTaskResponse::CommandInTaskResponse(const std::string & task)
+PathWithMetadata::CommandInTaskResponse::CommandInTaskResponse(const std::string & task)
 {
     Poco::JSON::Parser parser;
     try
@@ -136,7 +129,7 @@ RelativePathWithMetadata::CommandInTaskResponse::CommandInTaskResponse(const std
     }
 }
 
-std::string RelativePathWithMetadata::CommandInTaskResponse::to_string() const
+std::string PathWithMetadata::CommandInTaskResponse::to_string() const
 {
     Poco::JSON::Object json;
     if (retry_after_us.has_value())

src/Disks/ObjectStorages/IObjectStorage.h

@@ -114,7 +114,7 @@ using DataFileMetaInfoPtr = std::shared_ptr<DataFileMetaInfo>;
 
 struct DataLakeObjectMetadata;
 
-struct RelativePathWithMetadata
+struct PathWithMetadata
 {
     class CommandInTaskResponse
     {
@@ -143,28 +143,49 @@ struct RelativePathWithMetadata
     std::optional<DataFileMetaInfoPtr> file_meta_info;
     /// Retry request after short pause
     CommandInTaskResponse command;
+    std::optional<String> absolute_path;
+    std::optional<ObjectStoragePtr> object_storage_to_use = std::nullopt;
 
-    RelativePathWithMetadata() = default;
+    PathWithMetadata() = default;
 
-    explicit RelativePathWithMetadata(String command_or_path, std::optional<ObjectMetadata> metadata_ = std::nullopt)
+    explicit PathWithMetadata(
+        const String & command_or_path,
+        std::optional<ObjectMetadata> metadata_ = std::nullopt,
+        std::optional<String> absolute_path_ = std::nullopt,
+        std::optional<ObjectStoragePtr> object_storage_to_use_ = std::nullopt)
         : relative_path(std::move(command_or_path))
         , metadata(std::move(metadata_))
         , command(relative_path)
+        , absolute_path(absolute_path_)
+        , object_storage_to_use(object_storage_to_use_)
     {
         if (command.is_parsed())
             relative_path = "";
     }
 
-    explicit RelativePathWithMetadata(const DataFileInfo & info, std::optional<ObjectMetadata> metadata_ = std::nullopt);
+    explicit PathWithMetadata(
+        const DataFileInfo & info,
+        std::optional<ObjectMetadata> metadata_ = std::nullopt,
+        std::optional<String> absolute_path_ = std::nullopt,
+        std::optional<ObjectStoragePtr> object_storage_to_use_ = std::nullopt)
+        : metadata(std::move(metadata_))
+        , absolute_path(absolute_path_)
+        , object_storage_to_use(object_storage_to_use_)
+    {
+        relative_path = info.file_path;
+        file_meta_info = info.file_meta_info;
+    }
 
-    RelativePathWithMetadata(const RelativePathWithMetadata & other) = default;
+    PathWithMetadata(const PathWithMetadata & other) = default;
 
-    virtual ~RelativePathWithMetadata() = default;
+    virtual ~PathWithMetadata() = default;
+//>>>>>>> 53a743ff8c6 (first draft)
 
     virtual std::string getFileName() const { return std::filesystem::path(relative_path).filename(); }
     virtual std::string getFileNameWithoutExtension() const { return std::filesystem::path(relative_path).stem(); }
 
     virtual std::string getPath() const { return relative_path; }
+    virtual std::optional<std::string> getAbsolutePath() const { return absolute_path; }
     virtual bool isArchive() const { return false; }
     virtual std::string getPathToArchive() const { throw Exception(ErrorCodes::LOGICAL_ERROR, "Not an archive"); }
     virtual size_t fileSizeInArchive() const { throw Exception(ErrorCodes::LOGICAL_ERROR, "Not an archive"); }
@@ -174,6 +195,8 @@ struct RelativePathWithMetadata
     std::optional<DataFileMetaInfoPtr> getFileMetaInfo() const { return file_meta_info; }
 
     const CommandInTaskResponse & getCommand() const { return command; }
+
+    std::optional<ObjectStoragePtr> getObjectStorage() const { return object_storage_to_use; }
 };
 
 struct ObjectKeyWithMetadata
@@ -189,8 +212,8 @@ struct ObjectKeyWithMetadata
     {}
 };
 
-using RelativePathWithMetadataPtr = std::shared_ptr<RelativePathWithMetadata>;
-using RelativePathsWithMetadata = std::vector<RelativePathWithMetadataPtr>;
+using PathWithMetadataPtr = std::shared_ptr<PathWithMetadata>;
+using PathsWithMetadata = std::vector<PathWithMetadataPtr>;
 using ObjectKeysWithMetadata = std::vector<ObjectKeyWithMetadata>;
 
 class IObjectStorageIterator;
@@ -231,7 +254,7 @@ class IObjectStorage
     virtual bool existsOrHasAnyChild(const std::string & path) const;
 
     /// List objects recursively by certain prefix.
-    virtual void listObjects(const std::string & path, RelativePathsWithMetadata & children, size_t max_keys) const;
+    virtual void listObjects(const std::string & path, PathsWithMetadata & children, size_t max_keys) const;
 
     /// List objects recursively by certain prefix. Use it instead of listObjects, if you want to list objects lazily.
     virtual ObjectStorageIteratorPtr iterate(const std::string & path_prefix, size_t max_keys) const;

src/Disks/ObjectStorages/Local/LocalObjectStorage.cpp

@@ -151,7 +151,7 @@ ObjectMetadata LocalObjectStorage::getObjectMetadata(const std::string & path) c
     return object_metadata;
 }
 
-void LocalObjectStorage::listObjects(const std::string & path, RelativePathsWithMetadata & children, size_t/* max_keys */) const
+void LocalObjectStorage::listObjects(const std::string & path, PathsWithMetadata & children, size_t/* max_keys */) const
 {
     if (!fs::exists(path) || !fs::is_directory(path))
         return;
@@ -164,7 +164,7 @@ void LocalObjectStorage::listObjects(const std::string & path, RelativePathsWith
             continue;
         }
 
-        children.emplace_back(std::make_shared<RelativePathWithMetadata>(entry.path(), getObjectMetadata(entry.path())));
+        children.emplace_back(std::make_shared<PathWithMetadata>(entry.path(), getObjectMetadata(entry.path())));
     }
 }
 

src/Disks/ObjectStorages/Local/LocalObjectStorage.h

@@ -62,7 +62,7 @@ class LocalObjectStorage : public IObjectStorage
 
     ObjectMetadata getObjectMetadata(const std::string & path) const override;
 
-    void listObjects(const std::string & path, RelativePathsWithMetadata & children, size_t max_keys) const override;
+    void listObjects(const std::string & path, PathsWithMetadata & children, size_t max_keys) const override;
 
     bool existsOrHasAnyChild(const std::string & path) const override;