ditat-etl 1.5.7

Multiple tools and utilities for ETL pipelines and others.

Ditat ETL

Multiple tools and utilities for ETL pipelines and others.

Databases

Useful wrappers for databases and methods to execute queries.

Postgres

It is compatible with pandas.DataFrame interaction, either reading as dataframes and pushing to the db.

from ditat_etl.databases import Postgres

config = {
    "database": "xxxx",
    "user": "xxxx",
    "password": "xxxx",
    "host": "xxxxx",
    "port": "xxxx"
}
p = Postgres(config)

The main base function is query.

p.query(
  query_statement: list or str,
  df: bool=False,
  as_dict: bool=False,
  commit: bool=True,
  returning: bool=True,
  mogrify: bool=False,
  mogrify_tuple: tuple or list=None,
  verbose=False
)

This function is a workaround of pandas.to_sql() which drops the table before inserting. It really works like an upsert and it gives you the option to do nothing or update on the column(s) constraint.

p.insert_df_to_sql(
  df: pd.DataFrame,
  tablename: str,
  commit=True,
  conflict_on: str or list=None,
  do_update_columns: bool or list=False,
  verbose=False
):

This one is similar, it lets you "upsert" without necessarily having a primary key or constraint. Ideally use the previous method.

p.update_df_to_sql(
  df: pd.DataFrame,
  tablename: str,
  on_columns: str or list,
  insert_new=True,
  commit=True,
  verbose=False,
  overwrite=False
):

This class also has some smaller utitlies for describing tables, getting information, creating tables and indexes among others.

Salesforce

SalesforceObj

Salesforce wrapper adding functionalities to Simple Salesforce.

You can connect with login crendentials or through the Oauth2 protocol with a refresh token and client crendentials.

from ditat_etl.salesforce import SalesforceObj

sf_config = {
  "organizationId": "xxx",
  "password": "xxx",
  "security_token": "xxx",
  "username": "xxx",
  # Session Id for refresh token
  "session_id": "xxx"
}

# With login credentials
sf = Salesforce(sf_config)

# With Oauth2 refresh token and client credentials
sf = SalesforceObj(
  sf_config,
  client_id='xxx',
  client_secret='xxx',
)

The core of this class is the query method.

sf.query(
  tablename='xxx',
  columns=None,
  limit=None,
  df=False,
  date_from=None,
  date_to=None,
  date_window=None,
  date_window_variable='LastModifiedDate',
  verbose=False,
  timeout=None,
  max_columns=100,
)

Most times you can directly use the parallel execution.

sf.query_parallel(
  tablename='xxx',
  columns=None,
  limit=None,
  df=False,
  date_window=None,
  date_window_variable='LastModifiedDate',
  verbose=False,
  n_chunks=4
)

Update records

sf.update(
  tablename="xxx",
  record_list=[{'Id': '11111', 'field1': '1111'}],
  batch_size=1000
)

Insert records (update as an option)

sf.insert_df(
  tablename='xxx',
  dataframe=pd.DataFrame,
  batch_size=10000,
  update=True,
  conflict_on='Name',
  return_response=False,
)

Entity Resolution and Deduplication

The Matcher allows you to match two datasets on certain criteria and also find duplicates.

Matcher

from ditat_etl.utils.entity_resolution import Matcher

matcher = Matcher(exact_domain=False)

Entity Resolution

Set both datasets as dataframes. Specify the names of the columns to use as criteria for matching

# Set df 1
matcher.set_df(
  data=pd.DataFrame,
  name='df_1',
  index='id',
  domain=None,
  address=None,
  phone=None,
  country=None,
  entity_name=None
)
# Set df 2 
matcher.set_df(
  data=pd.DataFrame,
  name='df_2',
  index='id',
  domain=None,
  address=None,
  phone=None,
  country=None,
  entity_name=None
)

# Run
matches = matches.run(
  save=False,
  verbose=True,
  deduping=False,
  match_type_included=None,
  match_type_excluded=None,
  match_count_th=3
)

Deduplication

results = matcher.dedupe(
  data=pd.DataFrame,
  index='id',
  domain=None,
  address=None,
  phone=None,
  country=None,
  entity_name=None,
  save=True,
  match_type_included=None,
  match_type_excluded=None,
  match_count_th=3,
  include_self=True
)

TimeIt

The TimeIt class helps you time functions, methods, blocks of code.

from ditat_etl.time import TimeIt

A)
	@TimeIt
	def f():
		pass

	f()

	>> f takes: 0.0006 sec.

B) 
	@TimeIt(decimals=8)
	def f():
		pass

	f()

	>> f takes: 0.00000003 sec.

C) 
	with TimeIt():
		time.sleep(0.3)

	>> indented block takes: 0.3005 sec.

D) 
	class Foo:
	@TimeIt() # It has to be a callable for classes
	def m(self):
		pass

	Foo().m()

	>> m takes: 0.0 sec.

Enrichment

PeopleDataLabs

from ditat_etl.utils.enrichment import PeopleDataLabs

pdl = PeopleDataLabs(
    api_key='xxx',
    check_existing_method='s3',
    bucket_name='xxx',
)

You can:

• Enrich company
• Search company
• Enrich person
• Search person

r = pdl.enrich_company(
	min_likelihood=5,
	required='website',
	save=True,
	check_existing=True,
	s3_recalculate=False,
	company_name='companyx'
)

'''
PLEASE CHECK https://docs.peopledatalabs.com/docs 
FOR MORE DETAILS

'''

Url

functions

from ditat_etl.url.functions import eval_url, extract_domain

data = [
    'laskdj',
    'www.google.com',
    'accounts.google.com',
    'https://www.emol.com/economia/'
]

for d in data:
    r = extract_domain(d)
    print(r)

print('####')
r = eval_url(data)

print(r)

None
google.com
accounts.google.com
emol.com
####
Initializing 4 workers.
eval_url takes: 1.3309 sec.
{'laskdj': None, 'www.google.com': 'google.com', 'accounts.google.com': 'accounts.google.com', 'https://www.emol.com/economia/': 'emol.com'}

Url

Extension of module requests/urllib3 for Proxy usage and Bulk usage. High-level usage

from ditat_etl import url

response = url.get('https://google.com')
# You can pass the same parameters as the library requests and other special parameters.

# Check low level usage for more details.

Low-level usage

from ditat_etl.url import Url

u = Url()

We use the logging module and it is set by default with 'DEBUG'. You can change this parameter to any allowed level

u = Url(debug_level='WARNING') # Just an example

Manage your proxies

u.add_proxies(n=3) # Added 3 new proxies (not necessarily valid) to self.proxies

u.clean_proxies() # Multithreaded to validate and keep only valid proxies.

print(u.proxies)
# You can also u.proxies = [], set them manually but this is not recommended.

Main functionality

def request(
    queue: str or list,
    expected_status_code: int=200,
    n_times: int=1,
    max_retries: int=None,
    use_proxy=False,
    _raise=True,
    ***kwargs
    ):

Examples

result = u.request('https://google.com')

result = u.request(queue=['https://google.com', 'htttps://facebook.com'], use_proxy=True)

# You can also pass optional parameter valid por a requests "Request"
import json
result = u.request(queue='https://example.com', method='post', data=json.dumps({'hello': 'world'}))