sqlalchemy-media

Documentation

See the documentation for full description.

Why ?

Nowadays, most of the database applications are used to allow users to upload and attach files of various types to ORM models.

Handling those jobs is not simple if you have to care about Security, High-Availability, Scalability, CDN and more things you may have already been concerned. Accepting a file from public space, analysing, validating, processing(Normalizing) and making it available to public space again is the main goal of this project.

Sql-Alchemy is the best platform for implementing this stuff. It has the SqlAlchemyMutable types facility to manipulate the objects with any type in-place. why not ?

Note

The main idea comes from dahlia's SQLAlchemyImageAttach.

Overview

• Storing and locating any file, tracking it by sqlalchemy models.
• Storage layer is completely separated from data model, with a simple api: (put, delete, open, locate)
• Using any SqlAlchemy data type which interfaces Python dictionary. This is achieved by using the SqlAlchemyTypeDecorators and SqlAlchemyMutable.
• Offering delete_orphan flag to automatically delete files which orphaned via attribute set or delete from collections, or objects leaved in memory alone! by setting it's last pointer to None.
• Attaching files from Url, LocalFileSystem and Streams.
• Extracting the file's mimetype from the backend stream if possible, using magic module.
• Limiting file size(min, max), to prevent DOS attacks.
• Adding timestamp in url to help caching.
• Using python type hinting to annotate arguments. So currently python3.5 and higher is supported.
• Auto generating thumbnails, using width, height and or ratio.
• Analyzing files & images using magic and wand.
• Validating mimetype, width, height and image ratio.
• Automatically resize & reformat images before store.
• Thanks @YukSeungChan, for implementing the S3Store and OS2Store.
• SSH/SFTP Store

Roadmap

The next two milestones which can develop simultaneously are, SSSHStore and Video Processing.

Quick Start

Here is a simple example to see how to use this library:

import functools
import json
from os.path import exists, join

from sqlalchemy import create_engine, TypeDecorator, Unicode, Column, Integer
from sqlalchemy.orm import sessionmaker
from sqlalchemy.ext.declarative import declarative_base

from sqlalchemy_media import StoreManager, FileSystemStore, Image, WandAnalyzer, ImageValidator, ImageProcessor


TEMP_PATH = '/tmp/sqlalchemy-media'
Base = declarative_base()
engine = create_engine('sqlite:///:memory:', echo=False)
session_factory = sessionmaker(bind=engine)


StoreManager.register(
    'fs',
    functools.partial(FileSystemStore, TEMP_PATH, 'http://static.example.org/'),
    default=True
)


class Json(TypeDecorator):
    impl = Unicode

    def process_bind_param(self, value, engine):
        return json.dumps(value)

    def process_result_value(self, value, engine):
        if value is None:
            return None

        return json.loads(value)


class ProfileImage(Image):
    __pre_processors__ = [
        WandAnalyzer(),
        ImageValidator(
            minimum=(80, 80),
            maximum=(800, 600),
            min_aspect_ratio=1.2,
            content_types=['image/jpeg', 'image/png']
        ),
        ImageProcessor(
            fmt='jpeg',
            width=120,
            crop=dict(
                left='10%',
                top='10%',
                width='80%',
                height='80%',
            )
        )
    ]


class Person(Base):
    __tablename__ = 'person'

    id = Column(Integer, primary_key=True)
    name = Column(Unicode(100))
    image = Column(ProfileImage.as_mutable(Json))

    def __repr__(self):
        return "<%s id=%s>" % (self.name, self.id)


Base.metadata.create_all(engine, checkfirst=True)

if __name__ == '__main__':
    session = session_factory()

    with StoreManager(session):
        person1 = Person()
        person1.image = ProfileImage.create_from('https://www.python.org/static/img/python-logo@2x.png')
        session.add(person1)
        session.commit()

        print('Content type:', person1.image.content_type)
        print('Extension:', person1.image.extension)
        print('Length:', person1.image.length)
        print('Original filename:', person1.image.original_filename)

        thumbnail = person1.image.get_thumbnail(width=32, auto_generate=True)
        print(thumbnail.height)
        assert exists(join(TEMP_PATH, thumbnail.path))

        thumbnail = person1.image.get_thumbnail(ratio=.3, auto_generate=True)
        print(thumbnail.width, thumbnail.height)
        assert exists(join(TEMP_PATH, thumbnail.path))

        person1.image.attach('https://www.python.org/static/img/python-logo.png')
        session.commit()

        print('Content type:', person1.image.content_type)
        print('Extension:', person1.image.extension)
        print('Length:', person1.image.length)
        print('Original filename:', person1.image.original_filename)

    with StoreManager(session, delete_orphan=True):
        deleted_filename = join(TEMP_PATH, person1.image.path)
        person1.image = None
        session.commit()

        assert not exists(deleted_filename)

        person1.image = ProfileImage.create_from('https://www.python.org/static/img/python-logo.png')
        session.commit()

        print('Content type:', person1.image.content_type)
        print('Extension:', person1.image.extension)
        print('Length:', person1.image.length)
        print('Original filename:', person1.image.original_filename)

Will produce:

Content type: image/jpeg
Extension: .jpg
Length: 2020
Original filename: https://www.python.org/static/img/python-logo@2x.png
8
28 7
Content type: image/jpeg
Extension: .jpg
Length: 2080
Original filename: https://www.python.org/static/img/python-logo.png
Content type: image/jpeg
Extension: .jpg
Length: 2080
Original filename: https://www.python.org/static/img/python-logo.png

Examples

Checkout the examples directory in the root of the repo.

Changelog

Here you can see the full list of changes made on each sqlalchemy-media release.

0.11.0

• ImageList
• Observe AttachmentList items for change.

0.10.0

• OS2 Store

0.9.7

• Fixed: get_thumbnail() does not return a Thumbnail: #72

0.9.6

• Support rrs(ReducedRedundancyStorage) and prefix of S3Store.

0.9.5

• BUG-FIX: #68, #69

0.9.4

• BUG-FIX: Using Image.__thumbnail_type__ for creating thumbnails, it was missed: #67

0.9.0

• Aws S3 Store.

0.8.0

• Automatically coerce :obj:.Attachable objects. So if True, you can set the models attribute by a file, filename or cgi.FieldStorage.

0.6.2

• Fixing a bug in optionals module.

0.6.1

• Fixing some problems in documents.

0.6.0

• Image crop feature: #16.

0.5.0

• #17, #55. Merge analyzers, validators and processors as processors. for simplicity.

0.4.1 (2016-10-06)

• #54 Fixed.

0.4.0 (2016-10-05)

• ImageDimensionValidator: #14
• WandAnalyzer: #52

0.3.0 (2016-10-05)

• Thumbnail auto generation implemented: #11, See doc.
• Not using python's built-in mimetype module, due the bug: https://bugs.python.org/issue4963

0.2.0 (2016-10-05)

• Added two tutorials in documentation.
• Restricting Content-type: #28
• MagicAnalyzer
• Including all requirements*.txt in distribution: #49
• Including test stuff in distribution: #36
• Descriptive error message when an optional package is missing: #48
• Analyser: #30
• Validation: #31
• Fixed two bugs: #42, #41

0.1.1 (2016-10-03)

• Improving coverage