Dynamic class creation example

class_def.py

# written and discarded in favor of sqlalchemy methods - still a good example of how to dynamically create classes in Python

import glob
import sys
from sqlalchemy import Column, ForeignKey, Integer, String
from sqlalchemy.dialects.mysql import TIMESTAMP
from sqlalchemy.ext.declarative import declarative_base
from sqlalchemy.orm import relationship
from sqlalchemy import create_engine
import pandas as pd
from sqlalchemy.orm import sessionmaker
import pdb
import pprint

# stuff to setup ability to write to db in sqlalchemy
Base = declarative_base()
engine = create_engine('sqlite:///test.db')
Base.metadata.bind = engine
DBSession = sessionmaker(bind=engine)
session = DBSession()

# maps dtype to an sql-friendly type for writing
def convert_type(dtype):

    dtype = str(dtype) # hacky
    # this dict either doesnt need to exist or needs to be changed depending on where data is going to be put (sqlite3, etc). Current version is very basic for sqlalchemy.
    type_map_dict = {'int64': 'Integer', 'object' :'String', 'float64':'Float'}

    if dtype in type_map_dict:
        new_type = type_map_dict[dtype]
        return new_type
    else:
        print dtype
        raise Exception('not in type_map_dict')

# creates attribute dict to give to create_data_class based on csv input
def create_data_attributes(table_source):
    attributes = {}
    attributes['columns'] = []
    attributes['table_name'] = table_source.split('/')[-1].lower().replace('.csv','')
    attributes['class_name'] = attributes['table_name'].capitalize()
    df = pd.read_csv(table_source)
    for column in df:
        attributes_dict = {}
        attributes_dict['col_name'] = column
        attributes_dict['col_type'] = convert_type(df[column].dtype)
        attributes_dict['obj_name'] = column.lower()
        attributes['columns'].append(attributes_dict)

    pp.pprint(attributes) # for testing

# programmically creates a new python class. Establishes a tablename and attributes for writing to sql
def create_data_class(attributes):
    class_name = attributes['class_name']
    class_dict = {'__tablename__':attributes['table_name']}
    columns = attributes['columns']

    for col in columns:
        class_dict[col['obj_name']] = "Column({0})".format(col['col_type'])

    Obj = type(class_name, (object,), class_dict)


def grab_data_tables():
    tables  = glob.glob('data/*.csv') # hard code
    for table in tables:
        data_attributes = create_data_attributes(table)
        create_data_class(data_attributes)
        print("processed {0}").format(table)

if __name__ == '__main__':
    grab_data_tables()
    pdb.set_trace()

    # these are for instances of a class to save to db
    # session.add(o)
    # session.commit()