def main():
define("console", default=False, type=bool)
define("mongo_host", default='localhost')
parse_command_line()
basicConfig(options=options)
country_db = CountryDB(options.mongo_host)
company_db = CompanyDB(options.mongo_host)
stats_db = StatsDB(options.mongo_host)
#companies = company_db.find({}, fields={'offices': 1}, count=None)
##for company in adv_enumerate(companies):
## for office in company['offices']:
## country_db.increment(office['country_code'])
##print company_db
#_create_stats(stats_db, company_db)
_create_csv(stats_db, company_db)
return -1
categories = defaultdict(int)
year_month = defaultdict(int)
year = defaultdict(int)
for stats in adv_enumerate(stats_db.find()):
categories[get_dotted(stats, 'data.category_code')] += 1
founded_at = get_dotted(stats, 'data.founded_at')
if not founded_at:
continue
#if founded_at.year < 1995:
# print stats
year_month[(founded_at.year, founded_at.month)] += 1
year[founded_at.year] += 1
print sum(categories.values())
print year
print sorted(year_month.items())
def _create_csv(stats_db, company_db):
stats_cursor = stats_db.find({}, None)
print 'name,num_countries,funding,funding_capped,days_to_funding,num_rounds,employees,year,category,country'
for stats in adv_enumerate(stats_cursor, frequency=1000):
name = stats.get('_id').replace(' ', '_')
stats = stats['data']
countries = stats.get('countries')
country = countries[0] if len(countries) else None
country = country if country in COUNTRIES else 'other'
num_rounds = len(stats.get('funding_rounds'))
founded_at = stats.get('founded_at')
days_to_funding = 0
funding_rounds = stats.get('funding_rounds')
if funding_rounds and funding_rounds[0]['funded_at'] and founded_at.year > 1995:
days_to_funding = (funding_rounds[0]['funded_at'] - founded_at).days
funding_capped = min(float(stats.get('total_money_raised')) / 10**6, 100)
funding = float(stats.get('total_money_raised')) / 10**6
employees = stats.get('number_of_employees')
year = founded_at.year
category = stats.get('category_code')
try:
print ','.join(map(lambda k: str(k), [name, len(countries), funding, funding_capped, days_to_funding, num_rounds, employees, year, category, country]))
except Exception, e:
pass
def _create_stats(stats_db, company_db):
companies = company_db.find({},{}, count=None)
for company in adv_enumerate(companies, frequency=1000):
stats = {}
name = company.get('name')
countries = []
for office in company['offices']:
countries.append(office['country_code'])
stats['countries'] = countries
for fn in STATS:
stat_name, val = fn(company)
stats[stat_name] = val
stats_db.save(name, stats)