[WiP] Allow to read Iceberg data from any location

src/Disks/ObjectStorages/AzureBlobStorage/AzureObjectStorage.cpp

@@ -65,7 +65,7 @@ class AzureIteratorAsync final : public IObjectStorageIteratorAsync
     }
 
 private:
-    bool getBatchAndCheckNext(RelativePathsWithMetadata & batch) override
+    bool getBatchAndCheckNext(PathsWithMetadata & batch) override
     {
         ProfileEvents::increment(ProfileEvents::AzureListObjects);
         if (client->IsClientForDisk())
@@ -78,7 +78,7 @@ 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),
@@ -160,7 +160,7 @@ 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();
 
@@ -182,7 +182,7 @@ 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

@@ -32,20 +32,20 @@ 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));
@@ -104,9 +104,15 @@ WriteSettings IObjectStorage::patchSettings(const WriteSettings & write_settings
     return write_settings;
 }
 
-RelativePathWithMetadata::RelativePathWithMetadata(const String & task_string, std::optional<ObjectMetadata> metadata_)
+PathWithMetadata::PathWithMetadata(
+    const String & task_string,
+    std::optional<ObjectMetadata> metadata_,
+    std::optional<String> absolute_path_,
+    std::optional<ObjectStoragePtr> object_storage_to_use_)
     : metadata(std::move(metadata_))
     , command(task_string)
+    , absolute_path(absolute_path_)
+    , object_storage_to_use(object_storage_to_use_)
 {
     if (!command.isParsed())
         relative_path = task_string;
@@ -119,14 +125,20 @@ RelativePathWithMetadata::RelativePathWithMetadata(const String & task_string, s
     }
 }
 
-RelativePathWithMetadata::RelativePathWithMetadata(const DataFileInfo & info, std::optional<ObjectMetadata> metadata_)
+PathWithMetadata::PathWithMetadata(
+    const DataFileInfo & info,
+    std::optional<ObjectMetadata> metadata_,
+    std::optional<String> absolute_path_,
+    std::optional<ObjectStoragePtr> object_storage_to_use_)
     : 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;
 }
 
-void RelativePathWithMetadata::loadMetadata(ObjectStoragePtr object_storage, bool ignore_non_existent_file)
+void PathWithMetadata::loadMetadata(ObjectStoragePtr object_storage, bool ignore_non_existent_file)
 {
     if (!metadata)
     {
@@ -143,7 +155,7 @@ void RelativePathWithMetadata::loadMetadata(ObjectStoragePtr object_storage, boo
     }
 }
 
-RelativePathWithMetadata::CommandInTaskResponse::CommandInTaskResponse(const std::string & task)
+PathWithMetadata::CommandInTaskResponse::CommandInTaskResponse(const std::string & task)
 {
     Poco::JSON::Parser parser;
     try
@@ -167,7 +179,7 @@ RelativePathWithMetadata::CommandInTaskResponse::CommandInTaskResponse(const std
     }
 }
 
-std::string RelativePathWithMetadata::CommandInTaskResponse::toString() const
+std::string PathWithMetadata::CommandInTaskResponse::toString() const
 {
     Poco::JSON::Object json;
 

src/Disks/ObjectStorages/IObjectStorage.h

@@ -106,7 +106,7 @@ struct DataFileInfo;
 class DataFileMetaInfo;
 using DataFileMetaInfoPtr = std::shared_ptr<DataFileMetaInfo>;
 
-struct RelativePathWithMetadata
+struct PathWithMetadata
 {
     class CommandInTaskResponse
     {
@@ -138,18 +138,30 @@ struct RelativePathWithMetadata
     std::optional<ObjectMetadata> metadata;
     CommandInTaskResponse command;
     std::optional<DataFileMetaInfoPtr> file_meta_info;
+    std::optional<String> absolute_path;
+    std::optional<ObjectStoragePtr> object_storage_to_use = std::nullopt;
 
-    RelativePathWithMetadata() = default;
+    PathWithMetadata() = default;
 
-    explicit RelativePathWithMetadata(const String & task_string, std::optional<ObjectMetadata> metadata_ = std::nullopt);
-    explicit RelativePathWithMetadata(const DataFileInfo & info, std::optional<ObjectMetadata> metadata_ = std::nullopt);
+    explicit PathWithMetadata(
+        const String & task_string,
+        std::optional<ObjectMetadata> metadata_ = std::nullopt,
+        std::optional<String> absolute_path_ = std::nullopt,
+        std::optional<ObjectStoragePtr> object_storage_to_use_ = std::nullopt);
 
-    virtual ~RelativePathWithMetadata() = default;
+    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);
+
+    virtual ~PathWithMetadata() = default;
 
     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"); }
@@ -160,6 +172,8 @@ struct RelativePathWithMetadata
 
     void loadMetadata(ObjectStoragePtr object_storage, bool ignore_non_existent_file);
     const CommandInTaskResponse & getCommand() const { return command; }
+
+    std::optional<ObjectStoragePtr> getObjectStorage() const { return object_storage_to_use; }
 };
 
 struct ObjectKeyWithMetadata
@@ -175,8 +189,8 @@ 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;
@@ -217,7 +231,7 @@ 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;
 

src/Disks/ObjectStorages/MetadataStorageFromPlainObjectStorage.cpp

@@ -123,7 +123,7 @@ std::vector<std::string> MetadataStorageFromPlainObjectStorage::listDirectory(co
 {
     auto key_prefix = object_storage->generateObjectKeyForPath(path, std::nullopt /* key_prefix */).serialize();
 
-    RelativePathsWithMetadata files;
+    PathsWithMetadata files;
     std::string absolute_key = key_prefix;
     if (!absolute_key.ends_with('/'))
         absolute_key += '/';

src/Disks/ObjectStorages/ObjectStorageIterator.cpp

@@ -9,7 +9,7 @@ namespace ErrorCodes
     extern const int LOGICAL_ERROR;
 }
 
-RelativePathWithMetadataPtr ObjectStorageIteratorFromList::current()
+PathWithMetadataPtr ObjectStorageIteratorFromList::current()
 {
     if (!isValid())
         throw Exception(ErrorCodes::LOGICAL_ERROR, "Trying to access invalid iterator");

src/Disks/ObjectStorages/ObjectStorageIterator.h

@@ -16,10 +16,10 @@ class IObjectStorageIterator
     virtual bool isValid() = 0;
 
     /// Return the current element.
-    virtual RelativePathWithMetadataPtr current() = 0;
+    virtual PathWithMetadataPtr current() = 0;
 
     /// This will initiate prefetching the next batch in background, so it can be obtained faster when needed.
-    virtual std::optional<RelativePathsWithMetadata> getCurrentBatchAndScheduleNext() = 0;
+    virtual std::optional<PathsWithMetadata> getCurrentBatchAndScheduleNext() = 0;
 
     /// Returns the number of elements in the batches that were fetched so far.
     virtual size_t getAccumulatedSize() const = 0;
@@ -36,7 +36,7 @@ class IObjectStorageIterator
     /// Return the current batch of elements.
     /// It is unspecified how batches are formed.
     /// But this method can be used for more efficient processing.
-    virtual RelativePathsWithMetadata currentBatch() = 0;
+    virtual PathsWithMetadata currentBatch() = 0;
 };
 
 using ObjectStorageIteratorPtr = std::shared_ptr<IObjectStorageIterator>;
@@ -45,7 +45,7 @@ class ObjectStorageIteratorFromList : public IObjectStorageIterator
 {
 public:
     /// Everything is represented by just a single batch.
-    explicit ObjectStorageIteratorFromList(RelativePathsWithMetadata && batch_)
+    explicit ObjectStorageIteratorFromList(PathsWithMetadata && batch_)
         : batch(std::move(batch_))
         , batch_iterator(batch.begin()) {}
 
@@ -59,11 +59,11 @@ class ObjectStorageIteratorFromList : public IObjectStorageIterator
 
     bool isValid() override { return batch_iterator != batch.end(); }
 
-    RelativePathWithMetadataPtr current() override;
+    PathWithMetadataPtr current() override;
 
-    RelativePathsWithMetadata currentBatch() override { return batch; }
+    PathsWithMetadata currentBatch() override { return batch; }
 
-    std::optional<RelativePathsWithMetadata> getCurrentBatchAndScheduleNext() override
+    std::optional<PathsWithMetadata> getCurrentBatchAndScheduleNext() override
     {
         if (batch.empty())
             return {};
@@ -76,8 +76,8 @@ class ObjectStorageIteratorFromList : public IObjectStorageIterator
     size_t getAccumulatedSize() const override { return batch.size(); }
 
 private:
-    RelativePathsWithMetadata batch;
-    RelativePathsWithMetadata::iterator batch_iterator;
+    PathsWithMetadata batch;
+    PathsWithMetadata::iterator batch_iterator;
 };
 
 }

src/Disks/ObjectStorages/ObjectStorageIteratorAsync.cpp

@@ -112,7 +112,7 @@ bool IObjectStorageIteratorAsync::isValid()
     return !is_finished;
 }
 
-RelativePathWithMetadataPtr IObjectStorageIteratorAsync::current()
+PathWithMetadataPtr IObjectStorageIteratorAsync::current()
 {
     std::lock_guard lock(mutex);
 
@@ -123,7 +123,7 @@ RelativePathWithMetadataPtr IObjectStorageIteratorAsync::current()
 }
 
 
-RelativePathsWithMetadata IObjectStorageIteratorAsync::currentBatch()
+PathsWithMetadata IObjectStorageIteratorAsync::currentBatch()
 {
     std::lock_guard lock(mutex);
 
@@ -133,7 +133,7 @@ RelativePathsWithMetadata IObjectStorageIteratorAsync::currentBatch()
     return current_batch;
 }
 
-std::optional<RelativePathsWithMetadata> IObjectStorageIteratorAsync::getCurrentBatchAndScheduleNext()
+std::optional<PathsWithMetadata> IObjectStorageIteratorAsync::getCurrentBatchAndScheduleNext()
 {
     std::lock_guard lock(mutex);