def get_column_types(self):
if getattr(self.args, 'blanks', None):
type_kwargs = {'null_values': ()}
else:
type_kwargs = {}
text_type = agate.Text(**type_kwargs)
if self.args.no_inference:
types = [text_type]
else:
number_type = agate.Number(locale=self.args.locale, **type_kwargs)
# See the order in the `agate.TypeTester` class.
types = [
agate.Boolean(**type_kwargs),
agate.TimeDelta(**type_kwargs),
agate.Date(date_format=self.args.date_format, **type_kwargs),
agate.DateTime(datetime_format=self.args.datetime_format,
**type_kwargs),
text_type,
]
# In order to parse dates like "20010101".
if self.args.date_format or self.args.datetime_format:
types.insert(-1, number_type)
else:
types.insert(1, number_type)
return agate.TypeTester(types=types)
def setUp(self):
self.rows = (
(1, 'a', True, '11/4/2015', '11/4/2015 12:22 PM'),
(2, u'👍', False, '11/5/2015', '11/4/2015 12:45 PM'),
(None, 'b', None, None, None),
)
self.column_names = [
'number',
'text',
'boolean',
'date',
'datetime',
]
self.user_provided_column_names = [
'number',
'text',
'boolean',
'date',
'datetime',
]
self.column_types = [
agate.Number(),
agate.Text(),
agate.Boolean(),
agate.Date(),
agate.DateTime(),
]
self.table = agate.Table(self.rows, self.column_names,
self.column_types)
def setUp(self):
self.rows = (
(1.123, 'a', True, '11/4/2015', '11/4/2015 12:22 PM'),
(2, u'👍', False, '11/5/2015', '11/4/2015 12:45 PM'),
(2, u'c', False, '11/5/2015', '11/4/2015 12:45 PM'),
(None, 'b', None, None, None),
)
self.column_names = [
'number',
'text',
'boolean',
'date',
'datetime',
]
self.column_types = [
agate.Number(),
agate.Text(),
agate.Boolean(),
agate.Date(),
agate.DateTime(),
]
self.table = agate.Table(self.rows, self.column_names,
self.column_types)
self.connection_string = 'sqlite:///:memory:'
def load_data(data):
text_type = agate.Text()
number_type = agate.Number()
boolean_type = agate.Boolean()
columns = (
('last_name', text_type),
('first_name', text_type),
('age', number_type),
('race', text_type),
('state', text_type),
('tags', text_type),
('crime', text_type),
('sentence', text_type),
('convicted', number_type),
('exonerated', number_type),
('dna', boolean_type),
('dna_essential', text_type),
('mistaken_witness', boolean_type),
('false_confession', boolean_type),
('perjury', boolean_type),
('false_evidence', boolean_type),
('official_misconduct', boolean_type),
('inadequate_defense', boolean_type),
)
with open('examples/realdata/exonerations-20150828.csv') as f:
# Create a csv reader
reader = csv.reader(f)
# Skip header
next(f)
# Create the table
data['exonerations'] = agate.Table(reader, columns)
def _overall_stats(self):
count_open_licenses = agate.Summary(
'license_id', agate.Number(),
lambda r: sum(license_id in utils.OPEN_LICENSES
for license_id in r.values()))
self.overall_package_stats = self._package_table().aggregate([
('open_data_count', count_open_licenses),
])
self.resource_stats = self._package_resource_table().compute([
('open_format', agate.Formula(agate.Boolean(),
open_formats_count)),
])
if len(self._package_resource_table()) > 0:
self.resource_stats = self.resource_stats.aggregate([
('open_format_count', agate.Count('open_format', True)),
('min_date', agate.Min('created')),
('max_date', agate.Max('created'))
])
format_table = self._package_resource_table().group_by(
"format").aggregate([
('count', agate.Count()),
])
count = format_table.aggregate([
('different_formats', agate.Count()),
])
self.open_datasets = self.overall_package_stats.get(
"open_data_count", 0)
self.open_format_count = self.resource_stats.get(
"open_format_count", 0)
self.format_count = count.get("different_formats", 0)
self.compute_dates()
def from_sql(cls, connection_or_string, table_name):
"""
Create a new :class:`agate.Table` from a given SQL table. Types will be
inferred from the database schema.
Monkey patched as class method :meth:`Table.from_sql`.
:param connection_or_string:
An existing sqlalchemy connection or connection string.
:param table_name:
The name of a table in the referenced database.
"""
engine, connection = get_engine_and_connection(connection_or_string)
metadata = MetaData(connection)
sql_table = Table(table_name,
metadata,
autoload=True,
autoload_with=connection)
column_names = []
column_types = []
for sql_column in sql_table.columns:
column_names.append(sql_column.name)
if type(sql_column.type) in INTERVAL_MAP.values():
py_type = datetime.timedelta
else:
py_type = sql_column.type.python_type
if py_type in [int, float, decimal.Decimal]:
if py_type is float:
sql_column.type.asdecimal = True
column_types.append(agate.Number())
elif py_type is bool:
column_types.append(agate.Boolean())
elif issubclass(py_type, six.string_types):
column_types.append(agate.Text())
elif py_type is datetime.date:
column_types.append(agate.Date())
elif py_type is datetime.datetime:
column_types.append(agate.DateTime())
elif py_type is datetime.timedelta:
column_types.append(agate.TimeDelta())
else:
raise ValueError('Unsupported sqlalchemy column type: %s' %
type(sql_column.type))
s = select([sql_table])
rows = connection.execute(s)
try:
return agate.Table(rows, column_names, column_types)
finally:
if engine is not None:
connection.close()
engine.dispose()
def get_org_format_aggregates(self, package_table):
# format can not exist!?!
format_table = package_table.group_by("format").aggregate([
('count', agate.Count()),
])
open_format_table = package_table.compute([
('open_format', agate.Formula(agate.Boolean(), open_formats_count))
])
open_format_table_aggregates = open_format_table.aggregate([
('open_formats', agate.Count('open_format', True)),
])
new_table = format_table.compute([
('open_format', agate.Formula(agate.Boolean(), open_formats_count))
])
count = new_table.aggregate([
('different_formats', agate.Count()),
('open_formats', agate.Count('open_format', True)),
])
count["open_formats_datasets"] = open_format_table_aggregates[
"open_formats"]
return count
def test_distinct_values(self):
column_names: List = [
'id',
'name',
'dob',
'last seen',
'size',
'active',
]
column_types: List = [
agate.Number(),
agate.Text(),
agate.Date(),
agate.DateTime(),
agate.Text(),
agate.Boolean(),
]
rows = [(1, 'Alvin Cotton', '03-01-1980', '06-30-2019 12:12:00', 'L',
True),
(2, 'Usmaan Rojas', '01-12-1978', '06-30-2019 12:12:00', 'S',
False),
(3, 'Kingston Odling', '04-09-1990', '06-30-2019 12:12:00',
'M', True),
(3, 'Pooja Gillespie', '10-07-1985', '06-30-2019 12:12:00',
'S', True),
(4, 'Hal Blake', '08-17-1989', '06-30-2019 12:12:00', 'L',
True),
(5, 'Shannen Blevins', '06-10-1981', '06-30-2019 12:12:00',
'M', False),
(5, 'Courteney Weston', '04-23-1992', '06-30-2019 12:12:00',
'M', False),
(6, 'Conner Calhoun', '05-16-1977', '06-30-2019 12:12:00',
'XL', True),
(7, 'Susie Rasmussen', '02-08-1987', '06-30-2019 12:12:00',
'L', False),
(8, 'Cassie Beltran', '12-15-1982', '06-30-2019 12:12:00', 'M',
True)]
model = csvhound.core.BaseHound()
table = model.get_table_from_file('sample-data/test-distinct.csv')
distinct = model.distinct_values('size')
agate_table = agate.Table(rows, column_names, column_types)
distinct_agate = agate_table.select('size').distinct('size')
# now do the testing
self.assertColumnNames(distinct, ('size', ))
self.assertColumnTypes(distinct,
[type(c) for c in distinct.column_types])
self.assertRows(distinct, distinct_agate)
def test_grouping(self):
exonerations = agate.Table.from_csv(
'../../../data/exonerations-20150828.csv')
clean_state_data = exonerations.compute(
[('federal',
agate.Formula(agate.Boolean(),
lambda row: row['state'].startswith('F-'))),
('state',
agate.Formula(
agate.Text(), lambda row: row['state'][2:]
if row['state'].startswith('F-') else row['state']))],
replace=True)
by_state = clean_state_data.group_by('state')
state_totals = by_state.aggregate([('count', agate.Count())])
sorted_totals = state_totals.order_by('count', reverse=True)
sorted_totals.print_table(max_rows=10)
def get_column_types(self):
if getattr(self.args, 'blanks', None):
text_type = agate.Text(cast_nulls=False)
else:
text_type = agate.Text()
if self.args.no_inference:
return agate.TypeTester(types=[text_type])
else:
return agate.TypeTester(types=[
agate.Boolean(),
agate.Number(locale=self.args.locale),
agate.TimeDelta(),
agate.Date(date_format=self.args.date_format),
agate.DateTime(datetime_format=self.args.datetime_format),
text_type
])
def get_column_types(self):
if getattr(self.args, 'blanks', None):
type_kwargs = {'null_values': ()}
else:
type_kwargs = {}
types = [agate.Text(**type_kwargs)]
if not self.args.no_inference:
types = [
agate.Boolean(**type_kwargs),
agate.Number(locale=self.args.locale, **type_kwargs),
agate.TimeDelta(**type_kwargs),
agate.Date(date_format=self.args.date_format, **type_kwargs),
agate.DateTime(datetime_format=self.args.datetime_format, **type_kwargs),
] + types
return agate.TypeTester(types=types)
def get_column_types(self):
if getattr(self.args, 'blanks', None):
type_kwargs = {'null_values': ()}
else:
type_kwargs = {}
types = [agate.Text(**type_kwargs)]
if not self.args.no_inference:
types = [
agate.Boolean(**type_kwargs),
agate.TimeDelta(**type_kwargs),
agate.Date(date_format=self.args.date_format, **type_kwargs),
agate.DateTime(datetime_format=self.args.datetime_format,
**type_kwargs),
# This is a different order than agate's default, in order to parse dates like "20010101".
agate.Number(locale=self.args.locale, **type_kwargs),
] + types
return agate.TypeTester(types=types)
def setUp(self):
self.rows = ((1, 'a', True, '11/4/2015', '11/4/2015 12:22 PM', '4:15'),
(2, u'👍', False, '11/5/2015', '11/4/2015 12:45 PM',
'6:18'), (None, 'b', None, None, None, None))
self.column_names = [
'number', 'text', 'boolean', 'date', 'datetime', 'timedelta'
]
self.column_types = [
agate.Number(),
agate.Text(),
agate.Boolean(),
agate.Date(),
agate.DateTime(),
agate.TimeDelta()
]
self.table = agate.Table(self.rows, self.column_names,
self.column_types)
def setUp(self):
self.rows = (('1', 'a', 'True', '10/01/2015', '10/01/2015 12:30 PM',
'4h45m'), ('2', 'b', 'False', '11/01/2015',
'11/01/2015 12:45 PM', '3h25m'), ('', '', '',
'', '', ''))
self.number_type = agate.Number()
self.text_type = agate.Text()
self.boolean_type = agate.Boolean()
self.date_type = agate.Date()
self.datetime_type = agate.DateTime()
self.timedelta_type = agate.TimeDelta()
self.column_names = ('number', 'text', 'boolean', 'date', 'datetime',
'timedelta')
self.column_types = (self.number_type, self.text_type,
self.boolean_type, self.date_type,
self.datetime_type, self.timedelta_type)
self.table = agate.Table(self.rows,
zip(self.column_names, self.column_types))
title_rows = zip(sheet.row_values(4), sheet.row_values(5))
print(title_rows)
titles = [t[0] + ' ' + t[1] for t in title_rows]
titles = [t.strip() for t in titles]
titles
country_rows = [sheet.row_values(r) for r in range(6, 114)]
country_rows
from xlrd.sheet import ctype_text
import agate
text_type = agate.Text()
number_type = agate.Number()
boolean_type = agate.Boolean()
date_type = agate.Date()
example_row = sheet.row(6)
print example_row
print example_row[0].ctype
print example_row[0].value
print ctype_text
types = []
for v in example_row:
value_type = ctype_text[v.ctype]
if value_type == 'text':
types.append(text_type)
elif value_type == 'number':
print(impressions)
return None, None
INSERT_QUERY = "INSERT INTO creative_stats ({}) VALUES ({}) ON CONFLICT (ad_id) DO UPDATE SET {}".format(', '.join([k for k in KEYS]), ', '.join([":" + k for k in KEYS]), ', '.join([f"{k} = :{k}" for k in KEYS]))
def chunks(l, n):
"""Yield successive n-sized chunks from l."""
for i in range(0, len(l), n):
yield l[i:i + n]
# specifying column types saves 50% of time in loading the CSV! (30min w/o, 15min w/)
CREATIVE_STATS_COLUMN_TYPES = {'Ad_ID': agate.Text(), 'Ad_URL': agate.Text(), 'Ad_Type': agate.Text(),
'Regions': agate.Text(), 'Advertiser_ID': agate.Text(), 'Advertiser_Name': agate.Text(),
'Ad_Campaigns_List': agate.Boolean(), 'Date_Range_Start': agate.Date(), 'Date_Range_End': agate.Date(),
'Num_of_Days': agate.Number(), 'Impressions': agate.Text(), 'Spend_USD': agate.Text(),
'First_Served_Timestamp': agate.DateTime(), 'Last_Served_Timestamp': agate.DateTime(),
'Age_Targeting': agate.Text(), 'Gender_Targeting': agate.Text(), 'Geo_Targeting_Included': agate.Text(), 'Geo_Targeting_Excluded': agate.Text(),
'Spend_Range_Min_USD': agate.Number(), 'Spend_Range_Max_USD': agate.Number(), 'Spend_Range_Min_EUR': agate.Number(), 'Spend_Range_Max_EUR': agate.Number(), 'Spend_Range_Min_INR': agate.Number(), 'Spend_Range_Max_INR': agate.Number(), 'Spend_Range_Min_BGN': agate.Number(), 'Spend_Range_Max_BGN': agate.Number(), 'Spend_Range_Min_HRK': agate.Number(), 'Spend_Range_Max_HRK': agate.Number(), 'Spend_Range_Min_CZK': agate.Number(), 'Spend_Range_Max_CZK': agate.Number(), 'Spend_Range_Min_DKK': agate.Number(), 'Spend_Range_Max_DKK': agate.Number(), 'Spend_Range_Min_HUF': agate.Number(), 'Spend_Range_Max_HUF': agate.Number(), 'Spend_Range_Min_PLN': agate.Number(), 'Spend_Range_Max_PLN': agate.Number(), 'Spend_Range_Min_RON': agate.Number(), 'Spend_Range_Max_RON': agate.Number(), 'Spend_Range_Min_SEK': agate.Number(), 'Spend_Range_Max_SEK': agate.Number(), 'Spend_Range_Min_GBP': agate.Number(), 'Spend_Range_Max_GBP': agate.Number(), 'Spend_Range_Min_NZD': agate.Number(), 'Spend_Range_Max_NZD': agate.Number()}
OLD_CREATIVE_STATS_COLUMN_TYPES = {'Ad_ID': agate.Text(), 'Ad_URL': agate.Text(), 'Ad_Type': agate.Text(),
'Regions': agate.Text(), 'Advertiser_ID': agate.Text(), 'Advertiser_Name': agate.Text(),
'Ad_Campaigns_List': agate.Text(), 'Date_Range_Start': agate.Date(), 'Date_Range_End': agate.Date(),
'Num_of_Days': agate.Number(), 'Impressions': agate.Text(), 'Spend_USD': agate.Text(),
}
CREATIVE_STATS_SCHEMA_CHANGE_DATE = date(2020, 7, 1) # it's sometime around here, I don't know for sure, that the schema changes
# Name: Shani Kumar
# Date: 02/23/2020
# Course: DSC-540 - Data Preparation
# Desc: Practice joining numerous datasets – an activity you will likely run into frequently. Following the example
# in your text that starts on page 229 – 233 of Data Wrangling with Python, work through the example to bring
# two datasets together.
# Usage: This program is to complete assignment 11.2 requirements
#
# Import required packages
import xlrd
import agate
from xlrd.sheet import ctype_text
text_type = agate.Text() # define text type
number_type = agate.Number() # define number type
boolean_type = agate.Boolean() # define boolean type
date_type = agate.Date() # define date type
def remove_bad_chars(val):
""" This method remove bad character from data. If it is '-' it returns none
:param val: input string data
:return: input string or none
"""
if val == '-':
return None
return val
def get_types(example_row):
"""
#!/usr/bin/env python
"""
This module contains the XLS extension to :class:`Table <agate.table.Table>`.
"""
import datetime
from collections import OrderedDict
import agate
import six
import xlrd
EXCEL_TO_AGATE_TYPE = {
xlrd.biffh.XL_CELL_EMPTY: agate.Boolean(),
xlrd.biffh.XL_CELL_TEXT: agate.Text(),
xlrd.biffh.XL_CELL_NUMBER: agate.Number(),
xlrd.biffh.XL_CELL_DATE: agate.DateTime(),
xlrd.biffh.XL_CELL_BOOLEAN: agate.Boolean(),
xlrd.biffh.XL_CELL_ERROR: agate.Text(),
xlrd.biffh.XL_CELL_BLANK: agate.Boolean(),
}
def from_xls(cls,
path,
sheet=None,
skip_lines=0,
header=True,
encoding_override=None,
**kwargs):
"""
'New Zealand',
'Norway',
'Poland',
'Portugal',
'Slovakia',
'Slovenia',
'Spain',
'Sweden',
'Switzerland',
'Turkey',
'UK',
'USA'
]
NULL_VALUES = ['', '-999']
boolean = agate.Boolean(null_values=NULL_VALUES)
number = agate.Number(null_values=NULL_VALUES)
text = agate.Text(null_values=NULL_VALUES)
def load_data(data):
tester = agate.TypeTester(types=[
boolean,
number,
text
])
data['dpi'] = agate.Table.from_csv('DPI2015_basefile.v5.csv', column_types=tester)
def add_value(data):
def get_computed_data_type(self, table):
return agate.Boolean()
