Pulkit Aggarwal

This commit introduces a **deprecation warning** for the `generation_number` parameter within the **`AsyncMultiRangeDownloader`** constructor in the **`google-cloud-storage`** library. This **maintenance** change alerts users that `generation_number` will be removed in the next major release, guiding them to update their code. A corresponding unit test was updated to ensure the warning is correctly logged when the deprecated parameter is used. This helps prepare users for upcoming API changes and ensures a smoother transition.

This commit introduces a **new feature** by adding the `get_object` method to the **`asyncio` gRPC client** within the `google-cloud-storage` library. This new asynchronous capability allows users to **fetch object metadata** from Google Cloud Storage programmatically. The primary impact is on the **`async_grpc_client.py` module**, which now exposes the `get_object` function. Comprehensive unit tests were added in `test_async_grpc_client.py` to cover basic and parameterized calls, alongside a system test in `test_zonal.py` to verify `get_object`'s functionality after appendable write operations. This ensures the robust and reliable retrieval of object metadata for asynchronous workflows.

This commit **fixes a bug** in the **asyncio Google Cloud Storage client's write object stream** that prevented proper stream closure when `writer.close()` was called without `finalize_on_close`. Specifically, the `requests_done` method in `async_write_object_stream.py` is adjusted to ensure that both the `persisted_size` and final `EOF` responses are received, even if they arrive sequentially. This **bug fix** prevents premature stream termination and ensures reliable completion of write operations, particularly when the final object size needs to be confirmed. New unit tests have been added to `test_async_write_object_stream.py` to verify this corrected behavior.

This commit **enhances the `AsyncMultiRangeDownloader`** within the `google-cloud-storage` package by **updating its constructor and the `create_mrd` method to accept a `generation` parameter**. This **feature enhancement** modernizes the API to align with other public interfaces, making the parameter naming more consistent across the library. Crucially, it **maintains backward compatibility** for the older `generation_number` parameter, ensuring existing integrations continue to function without disruption. This change improves the usability and consistency of the **multi-range download functionality**.

This commit introduces a **new capability** by adding a default user agent string to the **gRPC client** within the `google-cloud-storage` library. Specifically, the `google/cloud/storage/_experimental/asyncio/async_grpc_client.py` module's constructor now sets a `gcloud-python/{__version__}` user agent in the client's `client_info`. This change ensures consistency with the existing JSON client's behavior, providing better identification for requests originating from the `gcloud-python` library. All gRPC requests made by this client will now include this default user agent, and new unit tests verify its correct application.

This commit introduces a **new capability** to the **experimental asyncio client** for Google Cloud Storage, integrating a comprehensive retry and resumption mechanism for **bidirectional write operations**, specifically targeting appendable objects. It implements a new `_WriteResumptionStrategy` within the `AsyncAppendableObjectWriter` to manage write state, generate requests with flush logic, and recover from failures, including server-side redirects. The `_AsyncWriteObjectStream` is updated to support this strategy by accepting routing tokens and metadata. This enhancement significantly **improves the reliability and fault tolerance** of write operations, ensuring more robust data consistency for users of the experimental client. Extensive unit and conformance tests have been added to validate the new retry and recovery logic.

This commit introduces a **new capability** to the **experimental asynchronous appendable object writer** by modifying its `append` method to **immediately flush the last data chunk**, rather than deferring this operation to the `close` method. This change primarily affects the `append`, `close`, and `finalize` methods of `AsyncAppendableObjectWriter` within the `google-cloud-storage` package, ensuring more prompt data persistence for partially written objects. **Comprehensive unit and system tests** were added and updated to validate this new flushing logic and state lookup behavior, including a new system test `test_append_flushes_and_state_lookup`.

This commit introduces a **new write resumption strategy** within the **`google-cloud-storage`** client library's experimental asyncio retry mechanisms. This **new capability** provides robust error handling for bidirectional write operations, allowing them to resume after transient failures. The core logic is implemented in the new `_WriteResumptionStrategy` class within `writes_resumption_strategy.py`, which defines methods like `generate_requests` and `recover_state_on_failure`. Comprehensive unit tests have been added in `test_writes_resumption_strategy.py` to validate its functionality and error recovery. This significantly enhances the reliability of **asynchronous write operations** by making them more resilient to network issues and other interruptions.

This commit introduces a **new capability** by adding a **bidi stream retry manager** (`_BidiStreamRetryManager`) to enhance the reliability of asynchronous operations within the `google-cloud-storage` library. It **integrates** this manager into the `AsyncMultiRangeDownloader` (MRD) by refactoring its `download_ranges()` method to leverage the new retry mechanism, and significantly refines the `_ReadResumptionStrategy` to handle smart resumption and redirects. This improves the resilience of **asynchronous multi-range downloads** and **bidi stream reads** by automatically managing retries, processing `BidiReadObjectRedirectedError`, and ensuring data integrity through checksum and offset verification. Comprehensive **conformance and unit tests** have been added to validate all error scenarios, retry logic, and redirect handling, ensuring robust behavior.

This commit introduces a **new sample** for the **Google Cloud Storage Python client library**, specifically demonstrating the `list_buckets_with_partial_success` functionality. It provides a practical example of how to handle scenarios where listing buckets might result in **partial success**, such as when certain locations are unreachable. This **new capability** enhances the **developer experience** by offering clear guidance on implementing robust bucket listing operations and managing potential failures gracefully. The new snippet, found in `packages/google-cloud-storage/samples/snippets/storage_list_buckets_partial_success.py`, helps users effectively utilize this advanced feature.

This commit introduces a **new feature** to the **Google Cloud Storage client library**, enabling the `list_buckets` method to handle **partial success** scenarios. Previously, an error might have prevented a full list; now, the method can return available buckets along with information about any unreachable ones. This enhancement, implemented by modifying `list_buckets` and adding the `_buckets_page_start` helper in `client.py`, improves the robustness and informativeness of bucket listing operations. A new unit test `test_list_buckets_w_partial_success` verifies this behavior, ensuring users receive more comprehensive results even when some buckets are inaccessible.

This commit introduces a **new feature** to the **experimental asynchronous I/O retry logic** within the `google-cloud-storage` library. It specifically enhances the `recover_state_on_failure` method in `reads_resumption_strategy.py` to **handle `BidiReadObjectRedirectedError`** during bidirectional reads. By extracting and storing the routing token from this error, the system can now **gracefully resume read operations** after a redirection. This significantly improves the overall **resilience and reliability of asynchronous storage reads** by ensuring continuity even when facing service-side redirects. Unit tests were added to verify this new error handling.

This commit introduces an **experimental read resumption strategy** for **asynchronous I/O** within the `google-cloud-storage` library. It adds the new `_ReadResumptionStrategy` class and a helper `_DownloadState` to manage the state required for resumable read operations, enabling them to continue from where they left off after interruptions. The abstract method `update_state_from_response` in `base_strategy.py` was also updated to accept a `response` parameter, supporting this **new capability**. This **enhances the resilience** of **async read operations** by providing a mechanism for graceful recovery, with comprehensive unit tests ensuring its functionality.

This commit introduces a **new feature** by adding the abstract base class `_BaseResumptionStrategy` to the **`google-cloud-storage`** library's **experimental asynchronous retry module**. This foundational class defines the core interface for managing the **resumption of bidirectional streams**, including abstract methods such as `generate_requests`, `update_state_from_response`, and `recover_state_on_failure`. This work establishes a critical building block for implementing robust and reliable **bidi stream resumption strategies**, significantly enhancing the client's ability to recover from transient failures within its **`asyncio` operations**.

This commit **fixes flaky system tests** for the **`resumable_media` component** within the `google-cloud-storage` client library. It addresses test instability by updating how MD5 hashes are calculated in test utilities, specifically in `test_download.py` and `test_upload.py`. The **bug fix** conditionally adds the `usedforsecurity=False` parameter to `hashlib.md5` calls for Python 3.9 and newer, ensuring compatibility with changes in the standard library. This change improves the reliability of the **system tests** by preventing spurious failures related to Python version differences.

This commit introduces a **new experimental feature** to the **`google-cloud-storage`** library, implementing **CRC32C checksum verification** for **asynchronous multi-range downloads**. This enhancement, primarily within the `async_multi_range_downloader.py` module, ensures **data integrity** during these complex operations and validates the `google-crc32c` C extension. The work includes updates to `__init__` and `_read_from_stream` to incorporate checksum logic. Comprehensive unit tests have also been added to validate checksum functionality during initialization and detect data corruption.

This commit introduces a **new experimental feature** to the **asynchronous read object stream** within the `google-cloud-storage` library. The `recv` method in `async_read_object_stream.py` is now capable of **refreshing its internal `read_handle`** if a new handle is provided by the server during an ongoing read operation. This enhancement significantly improves the **resilience and reliability** of long-lived asynchronous object reads by allowing the stream to adapt to server-side handle changes without interruption. Unit tests were also added to ensure the correct behavior of this new handle refresh mechanism.

This commit introduces an **experimental new feature** by adding an **asynchronous gRPC client** to the **Google Cloud Storage** library. It provides the foundational `AsyncGrpcClient` class within the `_experimental` module, enabling future asynchronous interactions with the storage service. This **new capability** includes the client's initialization via `__init__` and a `grpc_client` property for accessing the underlying gRPC client. Comprehensive unit tests have been added to validate its construction and basic functionality, laying the groundwork for more performant and responsive storage operations in asynchronous Python applications.

This commit **introduces an experimental gRPC client** for Google Cloud Storage, providing a new and potentially more performant way to interact with the service. It adds the `GrpcClient` class within `google.cloud.storage._experimental/grpc_client.py`, which manages client initialization, DirectPath configuration, and API key handling, including a getter for the underlying gRPC client. This **new capability** significantly expands the client's connectivity options, with comprehensive unit tests added in `tests/unit/test_grpc_client.py` to validate its construction and configuration.

This commit **fixes** a critical issue that caused **GAPIC generation** to fail for the **`google-cloud-storage`** client library. It modifies the `owlbot.py` configuration to explicitly remove the target directory before renaming, thereby **preventing "Directory not empty" errors** during the build process. This **maintenance fix** ensures the **automated generation pipeline** for the **`google-cloud-storage`** package operates reliably, allowing for consistent and successful client library updates.