blob-dict 0.1.16

Introduction

This library allows you to access blobs via Pythonic dict-like interface. Note that due to the nature of blob, not all dict methods are implemented. Therefore, it is not a subclass of MutableMapping type.

Supported Blob Dict Methods

Specifically, for a blob dict d, the following methods are available:

• len(d)
• key in d
• d.get(key, default)
• d[key]
• for key in d:
• d.pop(key, default)
• del d[key]
• d[key] = blob

Supported Blob Dict Types

There are different blob dict implementations:

• InMemoryBlobDict for in-memory storage, with optional TTL
• PathBlobDict for specified folder on file system as storage, with relative file path as key
  • It supports local file systems via PathLib
    • Ideally use provided LocalPath class (which is a subclass of Path) for full support
  • It also supports cloud file systems (like AWS S3 (and competible Cloudflare R2), Azure Blob Storage, and Google Cloud Storage) via CloudPathLib
• GitBlobDict for specified Git repo on file system as storage, with relative file path as key
  • It auto commits any add/update/delete
  • It auto syncs with remote if enabled
  • Optionally, key can be tuple where the second part is version:
    • str for version at specified commit/tag, like HEAD~2
    • int for relative version (same way of Python sequence index) of file, like 0 for first version and -2 for previous version
    • datetime for latest version before specified date/time
    • timedelta for latest version of specified duration before now
• VulkeyBlobDict for Vulkey/Redis-based storage, with optional TTL
• Specially, MultiReplicaBlobDict for utilizing multiple blob dicts underneath
  • For example, you can use in-memory or local file system blob dict as cache layer, while any cloud file system blob dict as storage

Supported Blob Types

There as many types of blobs with following class hierarchy relationship:

• BytesBlob for any data
  • StrBlob for any string
    • JsonDictBlob for any JSON dict object
    • JsonModelBlob for any Pydantic model object
  • AudioBlob for any SoundFile audio data
  • ImageBlob for any PIL/Pillow image data
  • VideoBlob for any MoviePy video clip data

Ultimately, all data are persisted as bytes.