szndaogen 2.4.3

SZN Database Access Object Generator for MySQL. Generates Models and DataManagers from existing MySQL database structure.

SZN DAOGen

Database Access Object Generator by Seznam.cz

Is a simple utility which enables you to generate Models and Data Managers for your project based on MySQL database to access data easily.

Description

DAOGenerator enables you to generate Models and DataManagers from existing database structure. For each database table and view the python Model will be generated. Each model will contain all table attributes with its datatypes and other meta information. For each Model one base DataManager for database access will be generated too. For each base DataManager one DataManager will be generated if it does not exist from previous szndaogen run. If DataManager exists file generation of it will be skiped. DataManager is designed for future custom functionality extending. Models and base DataManagers should not be edited it will be overwiten by next szndaogen script execution.

Advantages

• changes are needed only on database side
• you can stop beeing a coding monkey to write python models
• autogenerated table meta information
• works with Tables and Views
• View performance issue solved
• it is not nessesary to have view defined on production database
• Models and DataManages templates are written in Jinja2
• PyCharm code completion works perfectly with generated Models and DataManagers
• fast and easy change of all Models and DataManager if DB structure changed (it takes less than second)

Disadvantages

• you can feel shamed how easy to use it could be

Usage

szndaogen --help

Usage of SZN DAOGen v2.3.0 - Database Access Object Generator by Seznam.cz
szndaogen [options] output_path
    example: szndaogen -a localhost -d my_database -u root -p pass /path/to/data_access
    example: szndaogen -a localhost -d my_database -u root -p pass ./data_access


Options:
  -h, --help            show this help message and exit
  -a DB_HOST, --host-address=DB_HOST
                        MySQL database host. (required)
  -r DB_PORT, --port=DB_PORT
                        MySQL database port.
  -d DB_NAME, --database=DB_NAME
                        MySQL database name. (required)
  -u DB_USER, --user=DB_USER
                        User name for MySQL DB authentication. (required)
  -p DB_PASS, --password=DB_PASS
                        Password for MySQL DB authentication.
  -t TEMPLATES_PATH, --templates-path=TEMPLATES_PATH
                        Path to custom templates of Models (model.jinja),
                        DataManagers (manager.jinja) and DataManagerBases
                        (manager_base.jinja).

Installation

python3 setup.py install

or

pip3 install szndaogen

Example

For our example we will use sample database from MySQLTutorial.org. Zipped database file is here.

Importing database:

mysql < mysqlsampledatabase.sql

Create custom application directory or use example directory from our repository.
Lets generate Models and DataManagers from imported database by typing command:

szndaogen

It will run simple szndaogen wizzard (if no arguments passed):

Required parameters are not satisfied. Would you like to run setup wizard? [Y/n] y
MySQL host address: localhost
MySQL port (default 3306): 3306
MySQL database name: classicmodels
MySQL username: mysql-user
MySQL password: mysql-user-password
Output path where all models and managers will be generated (default "./data_access"): ./example_dao
Before you proceed, would you like to save this configuration as a bash script in CWD for future use? [Y/n] y

Shortcut script 'szndaogen-localhost-classicmodels.sh' created in current working directory.
Writing model `Customers` into `example_dao/models/customers_model.py`
Writing manager `Customers` into `example_dao/managers/customers_manager.py`
Writing manager `Customers` into `example_dao/managers/base/customers_manager_base.py`
Writing model `Employees` into `example_dao/models/employees_model.py`
Writing manager `Employees` into `example_dao/managers/employees_manager.py`
Writing manager `Employees` into `example_dao/managers/base/employees_manager_base.py`
Writing model `Offices` into `example_dao/models/offices_model.py`
Writing manager `Offices` into `example_dao/managers/offices_manager.py`
Writing manager `Offices` into `example_dao/managers/base/offices_manager_base.py`
Writing model `Orderdetails` into `example_dao/models/orderdetails_model.py`
Writing manager `Orderdetails` into `example_dao/managers/orderdetails_manager.py`
Writing manager `Orderdetails` into `example_dao/managers/base/orderdetails_manager_base.py`
Writing model `Orders` into `example_dao/models/orders_model.py`
Writing manager `Orders` into `example_dao/managers/orders_manager.py`
Writing manager `Orders` into `example_dao/managers/base/orders_manager_base.py`
Writing model `Payments` into `example_dao/models/payments_model.py`
Writing manager `Payments` into `example_dao/managers/payments_manager.py`
Writing manager `Payments` into `example_dao/managers/base/payments_manager_base.py`
Writing model `Productlines` into `example_dao/models/productlines_model.py`
Writing manager `Productlines` into `example_dao/managers/productlines_manager.py`
Writing manager `Productlines` into `example_dao/managers/base/productlines_manager_base.py`
Writing model `Products` into `example_dao/models/products_model.py`
Writing manager `Products` into `example_dao/managers/products_manager.py`
Writing manager `Products` into `example_dao/managers/base/products_manager_base.py`

Now your project contains file structure as you can see bellow:

tree .

├── szndaogen-localhost-classicmodels.sh
├── example_dao
│   ├── __init__.py
│   ├── managers
│   │   ├── base
│   │   │   ├── __init__.py
│   │   │   ├── customers_manager_base.py
│   │   │   ├── employees_manager_base.py
│   │   │   ├── offices_manager_base.py
│   │   │   ├── orderdetails_manager_base.py
│   │   │   ├── orders_manager_base.py
│   │   │   ├── payments_manager_base.py
│   │   │   ├── productlines_manager_base.py
│   │   │   └── products_manager_base.py
│   │   ├── __init__.py
│   │   ├── customers_manager.py
│   │   ├── employees_manager.py
│   │   ├── offices_manager.py
│   │   ├── orderdetails_manager.py
│   │   ├── orders_manager.py
│   │   ├── payments_manager.py
│   │   ├── productlines_manager.py
│   │   └── products_manager.py
│   └── models
│       ├── __init__.py
│       ├── customers_model.py
│       ├── employees_model.py
│       ├── offices_model.py
│       ├── orderdetails_model.py
│       ├── orders_model.py
│       ├── payments_model.py
│       ├── productlines_model.py
│       └── products_model.py
└── requirements.txt

Example of auto-generated Model class:

# !!! DO NOT MODIFY !!!
# Automatically generated Model class
# Generated by "szndaogen" tool


import typing
from szndaogen.data_access.model_base import ModelBase


class EmployeesModel(ModelBase):
    class Meta:
        TABLE_NAME: str = "employees"
        TABLE_TYPE: str = "BASE TABLE"
        # fmt: off
        SQL_STATEMENT: str = "SELECT {PROJECTION} FROM `employees` {WHERE} {ORDER_BY} {LIMIT} {OFFSET}"
        # fmt: on

        SQL_STATEMENT_WHERE_BASE: str = "1"
        SQL_STATEMENT_ORDER_BY_DEFAULT: str = ""

        PRIMARY_KEYS: typing.List = ["employeeNumber", ]
        ATTRIBUTE_LIST: typing.List = ["employeeNumber", "lastName", "firstName", "extension", "email", "officeCode", "reportsTo", "jobTitle", ]
        ATTRIBUTE_TYPES: typing.Dict = {
            "employeeNumber": int,
            "lastName": str,
            "firstName": str,
            "extension": str,
            "email": str,
            "officeCode": str,
            "reportsTo": int,
            "jobTitle": str,
        }
        MODEL_DATA_CONVERTOR: typing.Dict = {
        }

        # Class attribute to table attribute name conversion
        employeeNumber: str = "employeeNumber"
        lastName: str = "lastName"
        firstName: str = "firstName"
        extension: str = "extension"
        email: str = "email"
        officeCode: str = "officeCode"
        reportsTo: str = "reportsTo"
        jobTitle: str = "jobTitle"

    def __init__(self, init_data: typing.Dict = {}):
        self.employeeNumber: int = None
        """Type: int(11), Can be NULL: NO, Key: PRI"""
        self.lastName: str = None
        """Type: varchar(50), Can be NULL: NO"""
        self.firstName: str = None
        """Type: varchar(50), Can be NULL: NO"""
        self.extension: str = None
        """Type: varchar(10), Can be NULL: NO"""
        self.email: str = None
        """Type: varchar(100), Can be NULL: NO"""
        self.officeCode: str = None
        """Type: varchar(10), Can be NULL: NO, Key: MUL"""
        self.reportsTo: int = None
        """Type: int(11), Can be NULL: YES, Key: MUL"""
        self.jobTitle: str = None
        """Type: varchar(50), Can be NULL: NO"""
        super().__init__(init_data)

Example of auto-generated Base Manager class:

# !!! DO NOT MODIFY !!!
# Automatically generated Base Manager class
# Generated by "szndaogen" tool

import typing
from szndaogen.data_access.manager_base import TableManagerBase
from ...models.employees_model import EmployeesModel


class EmployeesManagerBase(TableManagerBase):
    MODEL_CLASS = EmployeesModel

    @classmethod
    def create_model_instance(cls, init_data: typing.Dict = None) -> EmployeesModel:
        if init_data is None:
            init_data = {}

        return super().create_model_instance(init_data)

    def select_one(self, employeeNumber: int, condition: str = "1", condition_params: typing.Tuple = (), projection: typing.Tuple = (), order_by: typing.Tuple = ()) -> EmployeesModel:
        return super().select_one(employeeNumber, condition=condition, condition_params=condition_params, projection=projection, order_by=order_by)

    def select_all(self, condition: str = "1", condition_params: typing.Tuple = (), projection: typing.Tuple = (), order_by: typing.Tuple = (), limit: int = 0, offset: int = 0) -> typing.List[EmployeesModel]:
        return super().select_all(condition=condition, condition_params=condition_params, projection=projection, order_by=order_by, limit=limit, offset=offset)

Example of auto-generated Manager class which could be modified/extended by your own functionality:

# This file can be modified. If file exists it wont be replaced by "szndaogen" any more.
# Automatically generated Manager class
# Generated by "szndaogen" tool

from .base.employees_manager_base import EmployeesManagerBase


class EmployeesManager(EmployeesManagerBase):
    pass

Example application

Now you can write application like this:

from example_dao.managers.employees_manager import EmployeesManager
from szndaogen.config import Config
from szndaogen.data_access.db import DBI
from szndaogen.tools.log import Logger, StdOutLogger


@DBI.transaction("dbi")
def update_employee_first_name(employee_id: int, new_first_name: str, dbi: DBI = None) -> int:
    manager = EmployeesManager(dbi=dbi)  # tell manager to work with passed DBI instance to keep transaction connection
    model_instance = manager.select_one(employee_id)
    model_instance.firstName = new_first_name
    return manager.update_one(model_instance)


if __name__ == '__main__':
    Config.MANAGER_AUTO_MAP_MODEL_ATTRIBUTES = True  # if disabled, you can control attributes mapping by `map_model_attributes()` method on model instance if needed to better performance
    Config.MYSQL_HOST = "localhost"
    Config.MYSQL_DATABASE = "classicmodels"
    Config.MYSQL_USER = "root"
    Config.MYSQL_PASSWORD = ""

    Logger.set_external_logger(logger_instance=StdOutLogger())

    employee_manager = EmployeesManager()
    employee_result = employee_manager.select_all(order_by=(f"{EmployeesManager.MODEL_CLASS.Meta.employeeNumber} ASC",))

    for employee_model_instance in employee_result:
        print(f"{employee_model_instance.firstName} {employee_model_instance.lastName} - {employee_model_instance.employeeNumber}")

    # autocommit update
    employee_result = employee_manager.select_all("lastName=%s", ("Thompson",))
    if len(employee_result) == 1:
        employee_model_instance = employee_result[0]
        print(f"Trying to update record id: {employee_model_instance.employeeNumber} - {employee_model_instance.firstName} {employee_model_instance.lastName}")
        employee_model_instance.firstName = "New Leslie"
        employee_manager.update_one(employee_model_instance)

    employee_result = employee_manager.select_all("lastName=%s", ("Thompson",))
    employee_model_instance = employee_result[0]
    print(f"Updated record id: {employee_model_instance.employeeNumber} - {employee_model_instance.firstName} {employee_model_instance.lastName}")

    # transaction update
    update_employee_first_name(1166, "Leslie forever")

    # new item
    new_employee = employee_manager.create_model_instance()
    new_employee.employeeNumber = 9999
    new_employee.firstName = "John"
    new_employee.lastName = "Doe"
    new_employee.extension = "xxx"
    new_employee.email = "a@b.c"
    new_employee.officeCode = 4
    new_employee.jobTitle = "Incognito"
    employee_manager.insert_one(new_employee)

    # delete item
    employee_manager.delete_one(new_employee)
    # OR
    employee_manager.delete_all(f"{EmployeesManager.MODEL_CLASS.Meta.employeeNumber}=%s", (9999,))
    # OR simply
    employee_manager.delete_all("employeeNumber=%s", (9999,))

Working with Views

szndaogen could process defined complicated database views too. There is no performance issue with MySQL views. Because view is parsed by szndaogen analyser and stored into Model definition. View declaration is executed on python application side. So all indexes and database optimalisations are used. Lets define sample view for out application defined as bellow select with a few joins:

SELECT o.`orderNumber`, od.`productCode`, od.`quantityOrdered`, p.`productName`, p.`quantityInStock`,
IF(p.`quantityInStock` > od.`quantityOrdered`, "enough", "to_stock_needed") AS productInStockStatus
FROM orders AS o
LEFT JOIN orderdetails AS od ON od.`orderNumber`=o.`orderNumber`
LEFT JOIN products AS p ON p.`productName`
WHERE o.`status`="In Process"

create view view_orders_to_be_processed from it:

DELIMITER $$

CREATE ALGORITHM=UNDEFINED DEFINER=`root`@`%` SQL SECURITY DEFINER VIEW `view_orders_to_be_processed` AS (
SELECT
  `o`.`orderNumber`      AS `orderNumber`,
  `od`.`productCode`     AS `productCode`,
  `od`.`quantityOrdered` AS `quantityOrdered`,
  `p`.`productName`      AS `productName`,
  `p`.`quantityInStock`  AS `quantityInStock`,
  IF((`p`.`quantityInStock` > `od`.`quantityOrdered`),'enough','to_stock_needed') AS `productInStockStatus`
FROM ((`orders` `o`
    LEFT JOIN `orderdetails` `od`
      ON ((`od`.`orderNumber` = `o`.`orderNumber`)))
   LEFT JOIN `products` `p`
     ON (`p`.`productName`))
WHERE (`o`.`status` = 'In Process'))$$

DELIMITER ;

run szndaogen again by stored bash script:

szndaogen-localhost-classicmodels.sh

Writing model `Customers` into `example_dao/models/customers_model.py`
Skipping manager `Customers` exists `example_dao/managers/customers_manager.py`
Writing manager `Customers` into `example_dao/managers/base/customers_manager_base.py`
Writing model `Employees` into `example_dao/models/employees_model.py`
Skipping manager `Employees` exists `example_dao/managers/employees_manager.py`
Writing manager `Employees` into `example_dao/managers/base/employees_manager_base.py`
Writing model `Offices` into `example_dao/models/offices_model.py`
Skipping manager `Offices` exists `example_dao/managers/offices_manager.py`
Writing manager `Offices` into `example_dao/managers/base/offices_manager_base.py`
Writing model `Orderdetails` into `example_dao/models/orderdetails_model.py`
Skipping manager `Orderdetails` exists `example_dao/managers/orderdetails_manager.py`
Writing manager `Orderdetails` into `example_dao/managers/base/orderdetails_manager_base.py`
Writing model `Orders` into `example_dao/models/orders_model.py`
Skipping manager `Orders` exists `example_dao/managers/orders_manager.py`
Writing manager `Orders` into `example_dao/managers/base/orders_manager_base.py`
Writing model `Payments` into `example_dao/models/payments_model.py`
Skipping manager `Payments` exists `example_dao/managers/payments_manager.py`
Writing manager `Payments` into `example_dao/managers/base/payments_manager_base.py`
Writing model `Productlines` into `example_dao/models/productlines_model.py`
Skipping manager `Productlines` exists `example_dao/managers/productlines_manager.py`
Writing manager `Productlines` into `example_dao/managers/base/productlines_manager_base.py`
Writing model `Products` into `example_dao/models/products_model.py`
Skipping manager `Products` exists `example_dao/managers/products_manager.py`
Writing manager `Products` into `example_dao/managers/base/products_manager_base.py`
Writing model `ViewOrdersToBeProcessed` into `example_dao/models/vieworderstobeprocessed_model.py`
Writing manager `ViewOrdersToBeProcessed` into `example_dao/managers/vieworderstobeprocessed_manager.py`
Writing manager `ViewOrdersToBeProcessed` into `example_dao/managers/base/vieworderstobeprocessed_manager_base.py`

and all work is done in a second. Now you can use ViewOrdersToBeProcessedManager with custom WHERE conditions, ORDER BY sorts, limits, etc.

from example_dao.managers.view_orders_to_be_processed_manager import ViewOrdersToBeProcessedManager

manager = ViewOrdersToBeProcessedManager()
results = manager.select_all(order_by=("`od`.`quantityOrdered` DESC",), limit=10)

print("Top 10 ordered quauntities waiting for processing")
for item in results:
    print(f"{item.orderNumber} - {item.productCode}: {item.productName}, {item.quantityOrdered}/{item.quantityInStock}")

Grouping tools

Packageszndaogen also comes with a set of helpful auto grouping tools placed in szndaogen/tools/auto_group.py.

These tools solve a problem where you have M:N relations between two entities and need to join them together. Their functionality is also helpful when working with SQL views.

SELECT name, surname, job.name as job___name
FROM person
JOIN
    works ON works.person_id = person.id
    job ON job.id = works.job_id
WHERE
    person.id = 2;

For a single person, who has two jobs, DB might output something like this:

name	surname	job___name
Jane	Doe	Accountant
Jane	Doe	Developer

Which might be OK when fetching only one person and his jobs. There are use cases, when you need to fetch more and more people. Output will be much larger. This is the place, where auto grouping tools come handy.

auto_group_dict

This function groups dict keys with same prefix under one dict key. Groups used as group keys are identified by group separator ___.

person = {
    "name": "Jane",
    "surname": "Doe",
    "job___name": "Accountant",
    "job___established": 2001
}

ret = auto_group_dict(person)

# Returns in
ret = {
    "name": "Jane",
    "surname": "Doe",
    "job": {
        "name": "Accountant",
        "established": 2001,
    }
}

auto_group_list

IMPORTANT: All items which are inside lists are sorted exactly the same as they came from the DB.

Let's say that we have want to retrieve a new person from our DB. Jane Doe now has two jobs: an accountant and a developer.

Database returns two rows as specified above. But in object oriented world, it would be better for us to have it in one dict. This is where auto_group_list comes handy.

SELECT person.name, person.surname, job.name as jobs__name
FROM person
JOIN
    works ON works.person_id = person.id
    job ON job.id = works.job_id
WHERE
    person.id = 2;

Assuming our SQL query returns 2 rows like this:

rows = [
    {
        "name": "Jane",
        "surname": "Doe",
        "jobs__name": "Accountant",
    },
    {
        "name": "Jane",
        "surname": "Doe",
        "jobs__name": "Developer",
    }
]

ret = auto_group_list(rows)

# Returns in
ret = {
    "name": "Jane",
    "surname": "Doe",
    "jobs": [
        {
            "name": "Accountant"
        },
        {
            "name": "Developer"
        }
    ]
}

This is kind of handy, isn't it? But what if we want to omit our WHERE statement? This is where auto_group_list_by_pkeys comes in place.

auto_group_list_by_pkeys

Next and the last useful is handy when you want to for example fetch multiple people from DB, keep m..n relations and have everything grouped nicely. Like so:

SELECT person.id as _id, person.name, person.surname, job.name as jobs__name
FROM person
JOIN
    works ON works.person_id = person.id
    job ON job.id = works.job_id
WHERE
    person.id IN (2, 3);

Our person no. 2 is Jane Doe, who works as an accountant and a developer. Person no. 3 is John Doe, works as an DevOps Engineer and a developer.

Let's say our grouping key is _id.

Our fetched data converted to python might look something like this:

rows = [
    {
        "_id": 2,
        "name": "Jane",
        "surname": "Doe",
        "jobs__name": "Accountant"
    },
    {
        "_id": 2,
        "name": "Jane",
        "surname": "Doe",
        "jobs__name": "Developer"
    },
    {
        "_id": 3,
        "name": "John",
        "surname": "Doe",
        "jobs__name": "DevOps Engineer"
    },
    {
        "_id": 3,
        "name": "John",
        "surname": "Doe",
        "jobs__name": "Developer"
    }
]

Let's make it prettier!

ret = auto_group_list_by_pkeys(("_id",), rows, use_auto_group_dict=True)

# Returns dict with 2 items, grouped by key "_id"
ret = {
    "2": {
        "_id": 2,
        "name": "Jane",
        "surname": "Doe",
        "jobs": [
            {
                "name": "Accountant"
            },
            {
                "name": "Developer"
            }
        ]
    },
    "3": {
        "_id": 3,
        "name": "John",
        "surname": "Doe",
        "jobs": [
            {
                "name": "DevOps Engineer"
            },
            {
                "name": "Developer"
            }
        ]
    }
}

Now we have all our cases covered, ready to go.