same_count += count_matching_rejected_issuers_for_company(name)
similar_count += count_similar_rejected_issuers_for_company(name)
similar_count += count_similar_rejected_issuers_for_company(name)
print "%s same names" % same_count
print "%s similar but different names" % similar_count
print "Companies to issuer names:"
with open("companies_to_issuers.csv", "w") as csv_output:
writer = csv.DictWriter(csv_output, ['COMPANY', 'ISSUER'])
writer.writeheader()
company_names = companies_to_issuers.keys()
company_names.sort()
for name in company_names:
issuers = list(set(companies_to_issuers[name]))
issuers.sort()
has_different_names = ([issuer != name
for issuer in issuers]).count(True) > 0
if not args.exclude or has_different_names:
print name
for issuer in issuers:
print "\t%s (Parent Co? %s)" % (
issuer, "Y" if is_parent_company(name, issuer) else "N")
if not args.parents or is_parent_company(name, issuer):
row = {}
row['COMPANY'] = name
row['ISSUER'] = issuer
writer.writerow(row)
print "Wrote to companies_to_issuers.csv."
same_count += count_matching_valid_issuers_for_company(name)
same_count += count_matching_rejected_issuers_for_company(name)
similar_count += count_similar_rejected_issuers_for_company(name)
similar_count += count_similar_rejected_issuers_for_company(name)
print "%s same names" % same_count
print "%s similar but different names" % similar_count
print "Companies to issuer names:"
with open("companies_to_issuers.csv", "w") as csv_output:
writer = csv.DictWriter(csv_output, ['COMPANY', 'ISSUER'])
writer.writeheader()
company_names = companies_to_issuers.keys()
company_names.sort()
for name in company_names:
issuers = list(set(companies_to_issuers[name]))
issuers.sort()
has_different_names = ([issuer != name for issuer in issuers]).count(True) > 0
if not args.exclude or has_different_names:
print name
for issuer in issuers:
print "\t%s (Parent Co? %s)" % (issuer, "Y" if is_parent_company(name, issuer) else "N")
if not args.parents or is_parent_company(name, issuer):
row = {}
row['COMPANY'] = name
row['ISSUER'] = issuer
writer.writerow(row)
print "Wrote to companies_to_issuers.csv."
def test_is_parent_company(self):
self.assertTrue(is_parent_company("Exxon Corp", "Exxon Corp DE"))
def test_is_not_parent_company(self):
self.assertFalse(is_parent_company("Coca Cola", "Coca Cola Bottling Company Inc"))
def test_is_parent_company_exact_match(self):
self.assertTrue(is_parent_company("ARCHER DANIELS MIDLAND", "ARCHER DANIELS MIDLAND"))
def test_adm_co(self):
self.assertTrue(is_parent_company("ARCHER DANIELS MIDLAND", "ARCHER DANIELS MIDLAND CO"))
def test_wells_fargo(self):
self.assertTrue(is_parent_company("WELLS FARGO COMPANY", "WELLS FARGO COMPANY MN"))
def test_apc(self):
self.assertTrue(is_parent_company("AIR PRODUCTS AND CHEMICALS INC", "AIR PRODUCTS CHEMICALS INC DE"))
def test_mcdonalds(self):
self.assertTrue(is_parent_company("MCDONALD'S CORP", "MCDONALDS CORP"))
print query_template
con = sqlite3.connect("ingest.db")
cur = con.cursor()
valid_common_stocks_count = 0
for name in company_names:
#print name
query = query_template % 'valid_items'
cur.execute(query, ['%' + db_escape_name(escape_name(name)) + '%'])
#print "Looking for valid issue name: " + db_escape_name(escape_name(name))
items = cur.fetchall()
has_common_stock = False
for item in items:
if is_parent_company(name, item[0]):
has_common_stock = True
break
else:
print "Comparing %s with %s" % (name, item[0])
print "NOT MATCH!!!"
if not has_common_stock:
no_valid_common_stock_companies.append(name)
else:
companies_with_valid_or_rejected_issues.append(name)
valid_common_stocks_count += 1
if len(items) == 0:
print "NO VALID ITEMS FOR %s" % name
print "# of companies with no valid common stock filings: %d" % len(
print query_template
con = sqlite3.connect("ingest.db")
cur = con.cursor()
valid_common_stocks_count = 0
for name in company_names:
#print name
query = query_template % 'valid_items'
cur.execute(query, ['%' + db_escape_name(escape_name(name)) + '%'])
#print "Looking for valid issue name: " + db_escape_name(escape_name(name))
items = cur.fetchall()
has_common_stock = False
for item in items:
if is_parent_company(name, item[0]):
has_common_stock = True
break
else:
print "Comparing %s with %s" % (name, item[0])
print "NOT MATCH!!!"
if not has_common_stock:
no_valid_common_stock_companies.append(name)
else:
companies_with_valid_or_rejected_issues.append(name)
valid_common_stocks_count += 1
if len(items) == 0:
print "NO VALID ITEMS FOR %s" % name
print "# of companies with no valid common stock filings: %d" % len(no_valid_common_stock_companies)