backo 0.1.2

Back office low code

Back Office Low Code (backo)

Jump to Quickstart for a basic setup.

See Documentation for all details

What is backo

It aims at providing a simple and efficient way to create a backend application that exposes a set of RESTful API routes.

Installation

pip install backo

Quickstart

Here is a sample with a DB (storage full in yaml file) with users to adresses reference

Imagine this Entity Relationship (users ar living in exactly one address, but zero onr more user can live at the same address)

erDiagram
    direction LR
    classDef className fill:#f9f,stroke:#333,stroke-width:4px
    Users |o--o{ Addresses : live
    Users {
        String name
        String surname
        Bool male
    }
    Addresses {
        String name
        String address
    }

You only will add Ref and Refs to exprime the relationship.

erDiagram
    direction LR
    classDef className fill:#f9f,stroke:#333,stroke-width:4px
    Users |o--o{ Addresses : live
    Users {
        String name
        String surname
        Bool male
        Ref addr "link one-to-many to Addresses"
    }
    Addresses {
        String name
        String address
        Refs users "link many-to-one to Usres"
    }

It translates into Python code using Backo like following:

from flask import Flask
from backo import Item, DBYmlConnector, Backoffice, Collection, Item
from backo import Ref, RefsList, DeleteStrategy

# --- Storage for user
yml_users = DBYmlConnector(path="/path_to_users")
# --- Storage for addresses
yml_addr = DBYmlConnector(path="/path_to_addresses")

# -- Description of the backoffice
my_backoffice = Backoffice("myApp")

# -- Add the collection "users" into this backoffice (with a reference to the futur "addrs" collection)
coll_users = Collection( 
    "users",
    Item({
        "name": String(),
        "surname": String(),
        "addr": Ref(coll="addrs", field="$.users", required=True),
        "male": Bool(default=True),
    }),
    yml_users,
)
my_backoffice.add_collection(coll_users)

# -- Add the collection "addrs" into this backoffice (with reference to the "usres" collection)
coll_addrs = Collection(
    "addrs",
    Item(
        {
            "name": String(),
            "address": String(),
            "users": RefsList(
                coll="users", field="$.addr", ods=DeleteStrategy.MUST_BE_EMPTY
            ),
        },
    ),
    yml_addr,
)
my_backoffice.register_collection(coll_addrs)

# Your flask application
flask = Flask(__name__)

# -- Add CRUD routes for this application
my_backoffice.build_routes(flask)

Item

Item is the central class in Backo.

It represents a database entity and includes all the methods required for CRUD operations: Create, Read, Update, and Delete.

A generic object is a stricto Dict() object.

Item( description object )

example

# Describe what a 'cat' is
cat_item = Item(
        {
            "name": String( required=True, default='Felix'),
            "address": String(),
            "age" : Int()
            # And so on...
        })

# --- Storage for cats
db_for_cats = DBMongoConnector(
    connection_string="mongodb://localhost:27017/my_database", collection="Cate"
)


# Add the cat object into the Collection object by adding a name and a db connector
cats = Collection("cats", cat_item, db_for_cats)

# Then add collection to the backoffice
backoffice.register_collection( "cats", cats )

[!IMPORTANT]
no need to care about _id.

Methods

Method	Description
.create( data :dict )	Create a new Item in the database using the provided data dictionary.
.save()	saves the current Item to the database.
.load( _id :str )	loads an Item from the database by its _id.
.reload()	reloads the current Item from the database.
.delete()	deletes the current Item from the database.
.new()	creates a new empty Item (must be populated with .set() and then saved).
.select()	retrieves a selection of Item from the database based on the selection criteria.

For each function above, an error is triggered in case of something went wrong.

Cardinalities

Relations cardinalities are expressed by the mean of Ref() and RefsList():

• Ref(): for 0 or 1 or exactly 1 relations.
• RefsList(): for 0 or more or 1 or more relations.

Item A	Item B	description
Ref	Ref	0 or One to one
Ref	RefsList	0 or One to many
Ref(require=True)	RefsList	One to many
RefsList	RefsList	Many to many

RefsList

• ofs= On Fill Strategy : fill the value ?

  • FillStrategy.FILL (by default) fill it
  • FillStrategy.NOT_FILL just keep ref

• ods= On Delete Strategy : what append when delete this item ?

  • DeleteStrategy.MUST_BE_EMPTY (by default) drop only if empty
  • DeleteStrategy.DELETE_REFERENCED_ITEMS drop all
  • DeleteStrategy.UNLINK_REFERENCED_ITEMS references are lost

Options

Option for Ref	Default	Description
coll=	None	the collection to make the ref
field=	None	The reverse field in the targeted collection (use selector to target it)
ods=	DeleteStrategy.MUST_BE_EMPTY	On Delete Strategy see ods
ofs=	FillStrategy.FILL	On Fill Strategy

And all options availables in stricto String() fields.

Deletion strategies (ods)

ods define the behaviour of the database when a delete occure and the object contain some RefList. For each RefList, you can define the strategy :

• DeleteStrategy.MUST_BE_EMPTY (by default) Requires the RefList to be empty before allowing the object to be deleted. If the RefList contains items, the deletion will be blocked and an error will be raised.

• DeleteStrategy.DELETE_REFERENCED_ITEMS When deleting the parent object, this strategy also deletes all objects referenced in the RefList. [!Warning] Use with caution, as this causes cascading deletes and may remove many related objects unintentionally.

• DeleteStrategy.UNLINK_REFERENCED_ITEMS Instead of deleting the referenced objects, this strategy removes the reference to the deleted object from each referenced item. This is commonly used for many-to-many relationships where you want to delete the parent object but keep the referenced objects intact, but simply cleaning up their links.

Example

Relationship example: Books and Authors

• An author can write 0 or many books.
• A book can be written only by 1 author.

# Authors write books
an_author = Item({
    'name' : String(),
    # An author may have written 0 or many books
    'books' : RefsList( coll='books', field="$.author" )
})

# A book is written by an author
a_book = Item({
    ... # Some attibutes
    # A book may have 1 or more authors
    author = RefsList( coll='authors', field="$.books", required=True )
})

Files objects

backo can manage files in a simple way : Like other stricto types. You can use File() or BlobFile().

object	Description
File()	Generic object to manage a file. You need to define a FileConnector to indicate the object File where to store the file
BlobFile()	File data is integrated literally into the datastructure, so there is no need for a FileConnector. Reserved for small files.

You can use all parameters for Files like other fields.

However there are extra specific parameters :

Option	Default	Description
mime_types=[ str ]	None	The list of allowed content types
max_size=8192	None	The maximum size of the file
work_connector=FileConnector	None	File working copy connector (main file location)
storage_connector=FileConnector	None	Second fileConnector used to store the file permanently after processing. If unset, file unique location is work_connector

Attributes and methods of the File object:

a = BlobFile()

a.set(b'Hello word')
# or
a.set_content(b'Hello word')

a.filename = "my_hello.txt"
a.content_type # -> 'text/plain'
a.size # -> 9
a.file_id # -> A random id

# Get the content of the file
a.get_content() # -> b'Hello word'

BlobFile()

Add a file with inlined data:

an_author = Item({
    'name' : String( require=True ),
    'surname' : String(),
    'pict' : BlobFile( require=True , mime_types=[ 'image/jpeg', 'image/png' ])
    'books' : RefsList( coll='books', field="$.author" )
})

File()

Add a file stored in a specific location on the file system:

an_author = Item({
    'name' : String( require=True ),
    'surname' : String(),
    'pict' : File( require=True , mime_types=[ 'image/jpeg', 'image/png' ], work=FileSystemConnector( path="/path/to/store/the/file" ) )
    'books' : RefsList( coll='books', field="$.author" )
})

routes

In a general context, routes are RESTful JSON API calls. But with files, every routes (CRUD, actions, ... ) can be called in two ways :

1. With files embedded in JSON in string format (for text files) or with base64 encoded (for other).

       # creating an author with a pict
       curl -X POST 'http://localhost/myApp/authors/' -d '{"surname":"John","name":"Rambo", "pict" : "base64:Sm9obiBwaWN0"}'
   

2. With a multipart route

   Each file is a part of the HTML multipart message. The JSON structure must be encoded in a _json multipart part.

       # creating an author with a pict
       curl -X POST -F _json='{"surname":"John","name":"Rambo"}' -F pict=@rambo.png http://localhost/myApp/authors/
   
       # Modifying the author picture
       curl -X PUT -F pict=@rambo1.png http://localhost/myApp/authors/id_of_rambo
   
       # Modifying the author picture and name
       curl -X PUT  -F _json='{ "name":"Rimbaud" }' -F pict=@rimbaud.png http://localhost/myApp/authors/id_of_rambo
   

3. link to the file are GET <my-app-name>/<collection name>/<_id>/<path>

See routes for more details

Collection

A collection is composed by :

1. an Item
2. a connector to a DB (DBConnector)
3. Some rights

[!IMPORTANT]
A collection must be registered to a backoffice (with .register_collection() call)

A collection can have :

1. Some Actions
2. Some Selections

See routes for resulting RESTful API routes.

example

from backo import Collection, DBMongoConnector, current

connector = DBMongoConnector(
    connection_string="mongodb://localhost:27017/media_library", collection="Books"
)
def can_delete(right_name: str, book: Item) -> bool:
    """Check if can delete a book"""
    if current_user.has_role(["ADMIN", "EMPLOYEE"]):
        return True
    return False

books = Collection(
    "books",
    books_item,
    connector,
    can_create=True,
    can_modify=True,
    can_delete=can_delete,
)

rights

A right is a function (or a lambda) with thoses parameters, and must return a boolean.

def your_function_name(right_name: str, o: Item) -> bool:

right	description
can_read	Check if the collection can be read
can_modify	Check if some elements in the collection can be modified
can_delete	Check if some elements in the collection can be deleted

Action

An action represent some modifications to an Item witch is more complex than just a PATCH/PUT (modification).

An action is compose by :

1. an Stricto Dict
2. a function
3. Some rights

[!IMPORTANT]
An action must be registered to a collection (with .register_action() call)

example

This is an action to borrow a book, with a start time at daytime. (all collections are not described)

def borrow(action: Action, book: Item, **kwargs) -> None:
    """borrow the book"""
    book.borrow.user = action.user_id
    book.borrow.date = datetime.now().replace(microsecond=0)
    book.borrow.return_date = action.return_date
    book.save(**kwargs)


def can_borrow(right_name: str, book: Item) -> bool:
    """Tel if current_user can execute the borrow action
    """
    if current_user.has_role("EMPLOYEE"):
        return True
    return False


def can_see_borrow_action(right_name: str, book: Item) -> bool:
    """Tel the borrow action is a nonsense ?
    """
    return not book.borrowed
#
# Definition of the action
borrow_action = Action(
    {
        "user_id": String(require=True), 
        "return_date": Datetime(require=True)
    },
    borrow,
    can_execute=can_borrow,
    exists=can_see_borrow_action,
)

# Add the action to the book collection
books.register_action("borrow", borrow_action)

rights for actions

A right is a function (or a lambda) with these parameters, and must return a boolean.

def your_function_name(right_name: str, o: Item) -> bool:

right	description
can_execute	Check if current_user can execute the action
can_see	exists

routes

See actions routes for resulting RESTful API routes.

Selection

A selection is a named list of Items with sub elements provided by a path.

An selection is compose by :

1. a list of selectors
2. a filter
3. Some rights

[!IMPORTANT]
A selection must be registered to a collection (with .register_selection() call)

example

This is a selection of borrowed books (all collections ar not described).

# Les livres empruntés
borrowed_book_select = Selection(
     [ "$.title", "$.borrow.user.login" ], 
     filter={ 'borrowed' : True }
     can_read=...
     )
books.register_selection("borrowed", borrowed_book_select)

And to call the selection...

curl -X GET 'http://localhost/media_library/coll/books/_selections/borrowed' 
  '{"result": [
     ["666", "Docker et plaisir", "Wallrich"], 
     ["1234", "Martine chez Epstein", "Lang"]
    ],
    "total": 2, "_skip": 0, "_page": 10}'
curl -X GET 'http://localhost/media_library/coll/books/_selections/borrowed?title.$reg=Martine'

rights for selections

A right is a function (or a lambda) with these parameters, and must return a boolean.

def your_function_name(right_name: str, o: Item) -> bool:

right	description
can_read	Check if current_user can do the selection

routes

See actions routes for resulting RESTful API routes.

Ref and RefsList

See Cardinalities

Authentication

General

Authentication is not managed by backo. Authentication is your side and you have to provide :

1. One or more authentication methode which set a JWT token
2. A verify function which read the JWT token and populate the current_user.

[!IMPORTANT]
The verify function must be passed as args to .build_routes() call

flowchart LR
    login[ /login ]
    route[At each route]
    setjwt[ set jwt ]
    decjwt[ decode jwt ]
    set[ current_user.set ]
    logout[ /logout ]
    dropjwt[drop jwt]

    login --> setjwt
    route --> decjwt
    decjwt --> set
    logout --> dropjwt

Example

This is an short and uncompleted example to authenticate and fill current_user.

First, the connection part. a /login route allows the login and returns a JWT.

app = Flask(__name__)
app.secret_key = "super secret key"
# Set the login route
@app.route("/login", methods=["POST"])
def login():
    """
    login load the user from the DB, verify the password and if
    Ok generate a JWT an return it to the client.
    At this point, current_user is not used.
    """
    d = request.json
    login = d["login"]
    password = d["password"]
    # find the user by your way
    user = load_from_db_by_login( login )
    # Do the check of the password
    if not user.check_password( password ):
      return jsonify({"message": "Invalid email or password"}), 401
 
    token = jwt.encode(
        {
            "exp": datetime.now(timezone.utc) + timedelta(hours=1),
            "user_id": user._id ,
        },
        "myappsecretkey",
        algorithm="HS256",
    )
    response = make_response(json.dumps({"login": "ok"}))
    response.set_cookie("jwt_token", token)
    return response

Second part : The decorator used to authenticate the route and to populate current_user

def token_required(f):
    @wraps(f)
    def decorated(*args, **kwargs):
        token = request.cookies.get("jwt_token")
        if not token:
            return jsonify({"message": "Token is missing!"}), 401
        try:
            data = jwt.decode(token, "myappsecretkey", algorithms=["HS256"])
        except:  # pylint: disable=bare-except
            return jsonify({"message": "Token is invalid!"}), 401

        # Load the user from the DB and fill current_user
        # we supose roles are store in your user_db
        user = load_from_db_by_id( data["user_id"] )
        current_user.set( { "_id" : user._id, "login" : user.login, "roles" : user.roles })

        return f(*args, **kwargs)
    return decorated

Then pass the function as args when you build routes.

myapp.build_routes(flask, "", check_user_token)

current_user

current_user represents the currently loggued user.

usage

# anywhere in your code
from backo import current_user

print(current_user._id) # The id of the current connected user
print(current_user.login) # A login name or whatever you store as "login"

# check if the user has the given role
if current_user.has_role('teamManager'):
  return True

How to set values to the current_user ?

backo doesnt manage authentication, because there is so much way to authenticate a user (or an app). So you have to do the auth by yourself.

However, using jwt is a goot solution. See Authentication for that.

current_user API

current_user is a very simple Stricto Dict (but can be extended). It contains :

key	type	usage
_id	Stricto String()	the _id of the user.
login	Stricto String()	the login or whatever you store .
roles	Stricto List( String() )	the list of roles for this users. A role is a kid of group the user belongs to
has_role()	function	return True or false to the role givent in param
reset()	function	change the current_user object type. See extend current_user for that

extend current_user

You can extend current_user, adding stuff in its schema, functions... etc.

from backo import current_user, CurrentUser, CurrentUserWrapper

class MyCurrentUser(CurrentUser):


      def __init__(self, obj: CurrentUser):

        CurrentUser.__init__(self, obj)
        self.add_to_model( 'email' , String() )


    """
    a new class child of CuurrentUser
    """
    def my_function(self):
        """
        a Dummy function
        """
        return(f"user {self.login} ({self._id})")

current_user.reset(MyCurrentUser())

# Later in the code
current_user.email = "[e-mail]"
current_user.my_function() # return the string "user ..."

Routes

When you call

mybackoffice.build_routes(flask)

Api routes are generated

CRUD Routes

GET <my-app-name>/<collection name>/<_id> ?_view=<view name>

_view are defined in stricto views

Return the object of this collection by _id.

curl -X GET 'http://localhost/myApp/users/123'

# Equivalent (by default _view=client)
curl -X GET 'http://localhost/myApp/users/123?_view=client'

# Another view
curl -X GET 'http://localhost/myApp/users/123?_view=otherviewname'

Answers can be :

code	data	Description
200	JSON object data	the requested item
401	None	you are not authorized to view this item
404	None	item not found
500	None	server-side error

GET <my-app-name>/<collection name>?<query_string>

Get a list of objects matching the query string. The query string can be with this format

key	value	description
<field>	<value>	matches items where <field> equals <value>. Example: surname=donald finds all items where surname equals to "donald".
<field>.<operator>	<value>	matches items where <field> satisfies <operator> with <value>. Example: age.$lt=12 finds items where age is less than 12.
<field>.<subfield>	<value>	Matches items where <field> is a nested dictionary containing <subfield> equal to <value>. Example: address.number=1 matches items where address.number equals to 1.

list of available operators

key	value	default	description
_view	string	"client"	selects the view (stricto views)
_page	int	-	sets the desired number of items per page in paginated data presentation
_skip	int	-	skips the n-first items of the result list in paginated data presentation.

The request returns a HTTP status 200 with that JSON object:

{
    "result": # list of dict containing objects matched
    "total": # (int) total number of object matched
    "_view": # the _view given in the request
    "_skip": # the _skip given in the request
    "_page": # the _page given in the request
}

Example

Select all users whose name includes 'do' and present the result list with 10 items per page.

curl -X GET 'http://localhost/myApp/users/?name.$re=do&_page=10'  

GET <my-app-name>/<collection name>/<_id>/<path>

Get the file content (if path refer to a file, otherwhise an error)

curl -X GET 'http://localhost/myApp/users/1234/pict'

# will probably return a content_type 'image/--'  and its content.

POST <my-app-name>/<collection name>

Create a new item for the collection collection name.

Example

curl -X POST 'http://localhost/myApp/users/' -d '{"name":"John","surname":"Rambo"}'

It returns the created user JSON object with a generated unique identifier _id and some _metadata or an error otherwise.

PUT /<my-app-name>/<collection name>/<_id>

Modify an existing object whose id is _id.

Example

curl -X PUT 'http://localhost/myApp/users/1234' -d '{"name":"Johnny"}'

Modify the users with _id 1234 and return the modified object.

DELETE /<my-app-name>/<collection name>/<_id>

Delete an existing object whose id is _id.

Example

curl -X DELETE 'http://localhost/myApp/users/1234'

Delete the user that has _id = 1234.

PATCH /<my-app-name>/<collection name>/<_id>

Partial change of an existing object whose id is _id. Please refer to the Stricto patch method.

Example

curl -X PATCH  'http://localhost/myApp/users/1234' -d '{"op": "replace", "path" : "$.name", "value": "Gilda"}'

Partial modification of the user with _id 1234 with the patch.

Patch content can be a list of patch operations.

Validation routes

POST /<my-app-name>/<collection name>/_check

Check the validity of an item field.

Please refer to stricto selectors for more details on selectors.

get must provide a json structure in the body :

field	type	descrimtion
item	dict	the data to check. It can be partial, see examples belowv
path	string	the field to check in the item. This is a selector. Please refer to stricto selectors for more details

The answer is a status 200 message with a json structure :

field	type	descrimtion
error	string or null	if null ther is no error, otherwise the error message

Examples :

curl -X POST 'http://localhost/myApp/users/_check' -d '{ "item" : { "name" : "John", "surname" : 32 }, "path" : "$.surname" }'
# will check surname an return a response.data like 
{
    'error' : "Must be a string"
}

The response is a status 200 even if the check returns an error. The request is correct.

curl -X POST 'http://localhost/myApp/users/_check' -d '{ "item" : { "surname" : "Johnny" }, "path" : "$.surname" }' 
# will check surname an return a response.data like 
{
    'error' : null
}


curl -X GET 'http://localhost/myApp/users/_check' -d '{ "item" : { "name" : 23, "surname" : "Johnny" }, "path" : "$.surname" }'
# (only the surname is tested. don't car if name is correct or not)
{
    'error' : null
}

Actions routes

Each collection has routes for its actions.

POST /<my-app-name>/<collection name>/_actions/<action_name>/<_id>

is used to call an action.

Selections routes

Method	Route	Description
GET	<my-app-name>/<collection name>/_selections/<selection_name>	do the selection
POST	<my-app-name>/<collection name>/_selections/<selection_name>	do the selection with complex filter

example

curl -X GET 'http://localhost/media_library/coll/books/_selections/borrowed' 
  '{"result": [
     ["666", "Docker as BDSM toy", "Wallrich"], 
     ["1234", "Milf forever", "Epstein"]
    ],
    "total": 2, "_skip": 0, "_page": 10}'
curl -X GET 'http://localhost/media_library/coll/books/_selections/borrowed?title.$reg=Dock'

Meta routes

Meta route are used to retrieve introspective informations about the application itself (structure, rights, ...).

GET /<my-app-name>/_meta

Return the structure of the application as a JSON with thoses keys :

key	type	description
name	string	The name of the application
collections	array of collection description	list of all collections description

collection description

Describe a collection

key	type	description
name	string	The name ov the collection
item	meta element description	description of an item

meta element description

Describe an element (an item, a key in an item)

key	type	description
type	string	the type of this element. For example "<class 'backo.item.Item'>" or "<class 'stricto.string.String'>"
type_short	string	the type of this element, but a short version like "Item" or "String"
description	string	a sort of comment. (optional) or null
require	boolean	mean this element is required or not
in	array of values	if the element must be in a list of value, or null if not.
constraints	boolean	means if there is one or more constraints on this value
default	--	the default value for this field (= null if no default value)
transform	boolean	say there is a transformation function
exists	boolean	false mean this field does not exist and must not be displayed
rights	meta rights description	the description of rights
sub_scheme	meta element description	reccusive description for childs if this object is a Dict or an Item
sub_type	meta element description	description of the content if this object is a List
sub_types	array of meta element description	description of the tuple content if this object is a Tuple

meta rights description

Example

curl -X GET  'http://localhost/myApp/_meta'
# Will return for example

{
  "name": "myApp",
  "collections": [
    {
      "name": "users",
      "item": {
        "type": "<class 'backo.item.Item'>",
        "type_short": "Item",
        "description": null,
        "required": false,
        "in": null,
        "constraints": [],
        "default": null,
        "transform": null,
        "exists": true,
        "rights": {
          "read": null,
          "modify": null
        },
        "sub_scheme": {
          "name": {
            "type": "<class 'stricto.string.String'>",
            "type_short": "String",
            "description": null,
            "required": false,
            "in": null,
            "constraints": [],
            "default": null,
            "transform": null,
            "exists": true,
            "rights": {
              "read": null,
              "modify": null
            },
            "regexp": []
          },
          "surname": {
            "type": "<class 'stricto.string.String'>",
            "type_short": "String",
            "description": null,
            "required": false,
            "in": null,
            "constraints": [],
            "default": null,
            "transform": null,
            "exists": true,
            "rights": {
              "read": null,
              "modify": null
            },
            "regexp": []
          },
          "age": {
            "type": "<class 'stricto.int.Int'>",
            "type_short": "Int",
            "description": null,
            "required": false,
            "in": null,
            "constraints": [
              "func"
            ],
            "default": null,
            "transform": null,
            "exists": true,
            "rights": {
              "read": null,
              "modify": null
            },
            "min": null,
            "max": null
          },
          "site": {
            "type": "<class 'backo.reference.Ref'>",
            "type_short": "Ref",
            "description": null,
            "required": false,
            "in": null,
            "constraints": [],
            "default": null,
            "transform": null,
            "exists": true,
            "rights": {
              "read": null,
              "modify": null
            },
            "regexp": []
          },
          "_id": {
            "type": "<class 'stricto.string.String'>",
            "type_short": "String",
            "description": null,
            "required": false,
            "in": null,
            "constraints": [],
            "default": null,
            "transform": null,
            "exists": true,
            "rights": {
              "read": null,
              "modify": null
            },
            "regexp": []
          }
        }
      }
    }
  ]
}

POST /<my-app-name>/<collection name>/_meta

Ask the backoffice the current meta informations for this collection object.

        "type": ty,
        "type_short" : re.sub(".*\.|\'>", '', ty),
        "description": self.get_as_string(self._description),
        "required": self.get_as_string(self._not_none),
        "in": self.get_as_string(self._union),
        "constraints": self.get_as_string(self._constraints),
        "default": self.get_as_string(self._default),
        "exists": self.exists( self.get_value() ),
        "rights": rights,

Example

For this structure :

def can_see_and_modify_salary(right_name, o, other):

    """
    return true if can read the salary
    """
    global current_user_name
    if current_user_name == o.name:
        return True
    return False

my_backoffice.add_collection(
    "users",
    Item(
        {
            "name" : String(),
            "salary" : Int( default=0, can_read=can_see_and_modify_salary, can_modify=can_see_and_modify_salary ),
        },
        yml_users,
    ),
)

# Logged as "Hector"
curl -X POST 'http://localhost/myApp/users/_meta' -d { 'name' : "John" }
# Will return this structure.
# rights "read" and "modify" are set to false for the salary
{
      "name": "users",
      "item": {
        "type": "<class 'backo.item.Item'>",
        "type_short": "Item",
        "description": null,
        "required": false,
        "in": null,
        "constraints": [],
        "default": null,
        "transform": null,
        "exists": true,
        "rights": {
          "read": null,
          "modify": null
        },
        "sub_scheme": {
          "name": {
            "type": "<class 'stricto.string.String'>",
            "type_short": "String",
            "description": null,
            "required": false,
            "in": null,
            "constraints": [],
            "default": null,
            "transform": null,
            "exists": true,
            "rights": {
              "read": true,
              "modify": true
            },
            "regexp": []
          },
          "salary": {
            "type": "<class 'stricto.string.Int'>",
            "type_short": "Int",
            "description": null,
            "required": false,
            "in": null,
            "constraints": [],
            "default": 0,
            "transform": null,
            "exists": true,
            "rights": {
              "read": false,
              "modify": false
            },
            "regexp": []
          }
        }
    }
}

Internal usage

Typical use cases for users and related addresses.

# Create an adress, save it in the DB and provide the object with a unique id.
moon_address = backoffice.addrs.create({"name": "moon", "address": "far"})

# Create a user with this address
astro = backoffice.users.create({"name": "neil", "surname": "amstrong", "addr": moon_address._id})

moon_address.users # -> return []
moon_address.reload() # reload data from the DB so that crossed references are updated
moon_address.users # -> return [ astro._id ]

_id

You dont't have to care about _ids in your item description. Backo will alter schema to add _id for each Item (see stricto schemas for details).

_meta

the db_connector adds meta data to each item by altering its schema.

The provided meta_data_handler give this Dict() :

Dict(
    {
        "ctime": Datetime(),
        "mtime": Datetime(),
        "created_by": Dict(
            {"_id": String(), "login": String()}
        ),
        "modified_by": Dict(
            {"_id": String(), "login": String()}
        ),
    },
),

Workflow and events

Each Item has a specific workflow and triggers specific events.

workflow

The defined states for each item are following:

State	descripion
UNSET	The Item result of a .new() function. It can't be saved in this state
UNSAVED	The Item has been modified and must be saved
SAVED	The Item is saved in the DB and has not been changed since last save

stateDiagram
    [*] --> UNSET : .new()

    UNSET --> SAVED : .load()
    UNSET --> UNSAVED : .set()
    UNSAVED --> SAVED : .save()
    SAVED --> UNSAVED : .set()
    SAVED --> UNSET : .delete()
    UNSAVED --> UNSET : .delete()

Events

The following events are triggered when the functions above are called:

function	event before	event after
.load()		"loaded"
.save()	"before_save"	"saved"
.delete()	"before_delete"
.create()	None	"created"

Example

Below is a simple use case on how to use these events:

        def rip( event_name, root, me, **kwargs ):
            """
            event_name = "before_delete"
            root = cat Item
            me = cat Item too
            """
            # Do what you want

        cat = Item( {
                'name' : String()
                'birth' : Datetime()
            }
            on=[ ( "before_delete", rip ) ]
        )

Transactions

Soon

Logs

The log system is based on logging Python module

You must first design your logging system with handlers. Then you can write the logs outputs.

Example

from backo import log_system, LogLevel

# To write all file to stderr
log_system.add_handler( log_system.set_streamhandler() )

# To write in a file
log_system.add_handler( log_system.set_filehandler("/var/log/mylog.log") )

# Set the level 
log_system.setLevel( LogLevel.INFO )

# create your own sub logger with its specific logging level
log = log_system.get_or_create_logger("custom")
log.setLevel(LogLevel.DEBUG)

log.debug("hey this is my first debug message")

Advanced use

You can select a specific logger and modify it by adding/removing handlers and changing its level.

log = log_system.get_or_create_logger("custom")
log.setLevel(loggind.DEBUG)
log.addHandler ( my_own_handler )
# ...

stack()

If you need, you can had the stack() into your message, to get a nice stack call:

log = log_system.get_or_create_logger("custom")
log.setLevel(loggind.DEBUG)

# somewhere in you code
log.debug(f'this is a debug message with stack {stack()}')
# ...

Will output something like :

DEBUG-custom- "/my_path/myfile.py", line 196 : this is a debug message with stack
  / 
  \ called by "/my_path/myfile.py", line 128, in sub_test_set_get_sample
   \ called by "/my_pathtests/test_file.py", line 140, in test_set_get_sample
    \ called by "/usr/lib/python3.12/unittest/case.py", line 589, in _callTestMethod
     \ called by "/usr/lib/python3.12/unittest/case.py", line 634, in run
      \ called by "/usr/lib/python3.12/unittest/case.py", line 690, in __call__
       \ called by "/usr/lib/python3.12/unittest/suite.py", line 122, in run
        \ called by "/usr/lib/python3.12/unittest/suite.py", line 84, in __call__
         \ called by "/usr/lib/python3.12/unittest/suite.py", line 122, in run
          \__

current loggers

Currently available loggers are :

logger	description
backoffice	The main Backoffice system
Item	The database itself (CRUD operations )
ref	Ref and RefsList objects
transaction	transactions and roolback
yml	yaml database connector
mongo	mongo database connector
select	selections
migration	see migration

migration

If you change an Item of a Collection, data currently stored in the database can be rejected by the model. You need to migrate them and there is a tool for that.

How it works

You have to write a script importing your backoffice object call the .migrate() method of the backoffice, and provide a function to make change to the object.

Below a migration function example:

def transform_function( o: dict) -> dict:
    """ Do what you want and return o"""
    # Adding a new key
    if 'new_key' not in o:
        o['new_key'] = ...
    # removing
    del o['unwanted_key']
    # ...
    return o

By default, the flag dry_run is True. so you will broke nothing. dry_run will do all work except saving the object in the DB.

[!IMPORTANT]
there is no check of rights and current_user during migration. metadatas (see _meta) are not updated.

example

Imagine a books collection. You modify the book Item to add a note field which is a float and is required. If you do nothing, the load() of a old item in the DB will failed on the new field note (required). This is an example for migrating data.

from media_library import mybackoffice
from backo import log_system, LogLevel

# Set migration level to debug
log_migration = log_system.get_or_create_logger("migration", LogLevel.DEBUG)

# ---------------------------------------------
# Check if a data match the model
# it raise an error at first _id 
# ---------------------------------------------
report = myapp.migrate("books", _id="my_book_id") # check a specific _Id
# or
report = myapp.migrate("books", _ids=["my_book_id1", "my_book_id2"]) # check a list of _Id
# or
report = myapp.migrate("books")  # check all ids

# --------------------------------------------
# Do a changement
# example you must now have a note field with a float value
# --------------------------------------------
def update_with_note(o: dict) -> dict:
    """
    this the function for doing operation on objects before setting them into the Item
    You can do what you want.
    """
    if "note" not in o:
        o["note"] = 10.0
    return o


# Check if OK (dry_run is True by default)
report = mybackoffice.migrate("books", update_with_note, _id="my_book_id")
# or
report = mybackoffice.migrate("books", update_with_note, _ids=["my_book_id1", "my_book_id2"])
# or
report = mybackoffice.migrate("books", update_with_note)  # All ids

# do it for real
report = mybackoffice.migrate("books", update_with_note, _id="my_book_id", dry_run=False)
# or
report = mybackoffice.migrate(
    "books", update_with_note, _ids=["my_book_id1", "my_book_id2"], dry_run=False
)
# or
report = mybackoffice.migrate("books", update_with_note, dry_run=False)  # All ids

report

The migration returns a report structure :

• report.nochanges._ids The list (array) of unchanged _ids
• report.total The number of unchanged _id durion this migration
• report.changes._ids The list (array) of _ids which are changed during this migration
• report.changes.total the number of changed _ids.
• report.changes.diff The list (array) of diff for each _id. The format is DeepDiff

Tests & co

For personal use only.

# all tests
python -m unittest tests
# or for a specific test
python -m unittest tests.TestDict.test_simple_type

# reformat
python -m black .

# pylint
pylint $(git ls-files '*.py')

# coverage
coverage run -m unittest tests
coverage html # report under htmlcov/index.html
firefox htmlcov/index.html

# docs
cd docs
make html

Building a new release

For personal use only.

# Modify changelog
# modify pyproject.toml
git add -u
git commit -am 'preparing 0.0.x'
git push
git tag -a 0.0.x -m '0.0.x'
git push origin tag 0.0.x

# publish a new relase in github interface, based on tag