def test_organization_relationships_built(self):
org = Organization(TEST_DATA)
self.assertIsNotNone(org.past_team)
self.assertEqual('First', org.past_team.get(0).first_name)
self.assertFalse(org.news)
self.assertFalse(org.news.get(0))
self.assertIsNone(org.news.get(0).first_name)
def test_organization_built(self):
org = Organization(TEST_DATA)
self.assertEqual(org.description, 'Description')
self.assertEqual(org.short_description, 'short description')
self.assertEqual(org.founded_on, datetime(2012, 1, 2))
self.assertEqual(org.permalink, "testpermalink")
self.assertEqual(org.homepage_url, "http://www.example.com")
self.assertEqual(org.name, "Organization")
self.assertEqual(org.total_funding_usd, 1234567)
self.assertEqual(org.number_of_investments, 1)
self.assertEqual(org.stock_symbol, "COY")
self.assertEqual(org.number_of_employees, 1234)
def test_organization_built(self):
org = Organization(TEST_DATA)
self.assertEqual(org.permalink, "testpermalink")
self.assertEqual(org.name, "Organization")
self.assertEqual(org.short_description, 'short description')
self.assertEqual(org.description, 'Description')
self.assertEqual(org.primary_role, 'company')
self.assertEqual(org.role_company, True)
self.assertEqual(org.role_investor, False)
self.assertEqual(org.role_group, False)
self.assertEqual(org.role_school, False)
self.assertEqual(org.founded_on, datetime(2012, 1, 2))
self.assertEqual(org.is_closed, False)
self.assertEqual(org.closed_on, None)
self.assertEqual(org.num_employees_min, 1234)
self.assertEqual(org.num_employees_max, 2345)
self.assertEqual(org.total_funding_usd, 1234567)
self.assertEqual(org.number_of_investments, 1)
self.assertEqual(org.homepage_url, "http://www.example.com")
self.assertEqual(org.stock_symbol, "COY")
def test_organization_relationships_built(self):
org = Organization(TEST_DATA)
self.assertIsNotNone(org.current_team)
self.assertEqual('Director of Title', org.current_team.get(0).title)
def enrich_company(self):
try:
name = self.enriched_entity.deduced['name']
if not self.enrich_key:
# Search for all the companies with this name
response = self._make_request(CrunchBaseEngager.PEOPLE_URL,
{'name': name})
data = response.json().get('data')
if not data or data.get('error'):
raise EngagementException(
"CrunchBaseEngager: error in retrieving company %s." %
name)
if len(data["items"]) > 1:
raise EngagementException(
"CrunchBaseEngager: company %s not ambiguous. Found %d people with this name."
% (name, len(data["items"])))
if len(data["items"]) == 0:
raise EngagementException(
"CrunchBaseEngager: company %s not found." % name)
# TODO: Future: Iterate over the returned people and check if there's another matching attribute (like social url) we can use to choose the right person
permalink = data['items'][0]['properties']['permalink']
else:
permalink = self.enrich_key
response = self.get_node('organizations', permalink)
if hasattr(response, 'from_cache'):
self.set_data("from_cache", response.from_cache)
if not response.from_cache:
pass # code for debugging purposes
data = response.json().get('data')
org = Organization(data)
# Name
if org.name:
self.set_data(C.NAME, org.name)
# Get company logo
if org.primary_image and len(org.primary_image) > 0:
logo_url = org.primary_image[0].asset_path
self.add_data(C.LOGOS, {
C.LOGO_URL: logo_url,
C.LOGO_SOURCE: 'crunchbase'
})
# Get overview stats (acquisitions, total funds, etc.)
if org.acquired_by and hasattr(org.acquired_by, 'acquirer'):
acquiring_company = org.acquired_by.acquirer.name
self.set_data(C.ACQUIRED_BY, acquiring_company)
# Get headquarters
if org.headquarters and len(org.headquarters) > 0:
headquarters = '%s, %s' % (org.headquarters[0].city,
org.headquarters[0].country)
self.set_data(C.HEADQUARTERS, headquarters)
# Get description
if org.short_description:
description = org.short_description
self.set_data(C.DESCRIPTION, description)
# Get founders
if org.founders and len(org.founders) > 0:
founders = []
for founder in org.founders:
full_name = '%s %s' % (founder.first_name,
founder.last_name)
founders.append(full_name)
self.set_data(C.FOUNDERS, founders)
# Get categories
if org.categories and len(org.categories) > 0:
for category in org.categories:
self.add_data(C.CATEGORIES, category.name)
# Grab aliases
if org.also_known_as:
self.set_data(C.ALIASES, org.also_known_as)
# Grab websites --> homepage_url ?
if org.homepage_url and len(org.homepage_url) > 0:
self.set_data(C.WEBSITE, org.homepage_url)
# Is it a VC company
if org.role_investor:
self.set_data(C.INVESTMENT_COMPANY_TYPE,
C.ORGANIZATION_TYPE_VENTURE_CAPITAL)
# Is it an educational organization
if org.role_school:
self.set_data(C.ORGANIZATION_TYPE, C.ORGANIZATION_TYPE_SCHOOL)
# Get socials
if org.websites and len(org.websites) > 0:
for url in org.websites:
url_type = url.website_type.lower()
if url_type == 'twitter':
self.set_data(C.TWITTER_URL, url.url)
elif url_type == 'facebook':
self.set_data(C.FACEBOOK_URL, url.url)
elif url_type == 'linkedin':
self.set_data(C.LINKEDIN_URL, url.url)
elif url_type == 'angellist':
self.set_data(C.ANGELLIST_URL, url.url)
else:
pass
# Get investments
if org.investments and len(org.investments) > 0:
all_investments = set()
for investment in org.investments:
investment_name = investment.invested_in.name
all_investments.add(investment_name)
self.set_data(C.PORTFOLIO_COMPANIES, list(all_investments))
# Get founding year
if org.founded_on:
founding_year = org.founded_on.year
self.set_data(C.FOUNDING_YEAR, founding_year)
# Get contact email - for emails-domain info
# Get number of employees
if org.num_employees_min and org.num_employees_max:
employees_range_str = '%s|%s' % (org.num_employees_min,
org.num_employees_max)
self.set_data(C.EMPLOYEES_RANGE, employees_range_str)
# Go over all investors
if org.investors and len(org.investors) > 0:
investors = []
for investor in org.investors:
investor_dict = investor.data
investor_type = investor_dict['type'].lower()
if investor_type == 'person':
investor_name = '%s %s' % (
investor_dict['properties']['first_name'],
investor_dict['properties']['last_name'])
elif investor_type == 'organization':
investor_name = investor_dict['properties']['name']
else:
pass
str = 'partner/round'
investors.append((investor_name, investor_type, str))
self.set_data(C.INVESTORS, investors)
# Go over all board members
if org.board_members_and_advisors and len(
org.board_members_and_advisors) > 0:
board_members = []
for board_member in org.board_members_and_advisors:
# Do we need this: board_member.person.role_investor:
full_name = '%s %s' % (board_member.person.first_name,
board_member.person.last_name)
board_members.append(full_name)
self.set_data(C.ADVISORS, board_members)
if org.founders and len(org.founders) > 0:
founders = []
for founder in org.founders:
# Do we need thjs: founder.role_investor:
full_name = founder.first_name + " " + founder.last_name
founders.append(full_name)
self.set_data(C.FOUNDERS, founders)
team_members = []
if org.past_team and len(org.past_team) > 0:
for team_member in org.past_team:
full_name = team_member.person.first_name + " " + team_member.person.last_name
team_members.append(full_name)
if org.current_team and len(org.current_team) > 0:
for team_member in org.current_team:
full_name = team_member.person.first_name + " " + team_member.person.last_name
team_members.append(full_name)
if len(team_members) > 0:
self.set_data(C.TEAM, team_members)
pass
# Only if data was not found, get the companies by names
# data = self.cb.organizations(company_name)
# if 'items' in data:
# permalink = data.items[0].permalink
# self.set_data("permalink", permalink)
# response = self.get_node('organizations', permalink)
# node_data = response.json().get('data')
# # Add the company name and other information (and then I can go to sleep!
# pass
except Exception as e:
print(
"CrunchBasengager::enrich_company - failed to enrich company %s (%s)"
% (name, e))
raise EngagementException(e)
return [C.NAME]