def write_to_mongodb(names, entry_id):
if names != []:
mycol = refer_cleaned_collection()
mycol.update_one({'_id': ObjectId(entry_id)},
{'$set': {
'important_person_company': names
}})
elif names == []:
mycol = refer_cleaned_collection()
mycol.update_one({'_id': ObjectId(entry_id)}, {
'$set': {
'important_person_company': 'No important persons found'
}
})
def extract_data(def_entry_id):
mycol = refer_cleaned_collection()
comp_data_entry = mycol.find({'_id': ObjectId(def_entry_id)})
data = [i for i in comp_data_entry]
print('data:', data)
# extracted_data = data[0]['header_text'] + [data[0]["description"]] + data[0]["paragraph_text"]
try:
extracted_data = data[0]['header_text'] + data[0]["paragraph_text"]
except KeyError:
extracted_data = None
return extracted_data
def get_persons_dnb(def_entry_id):
mycol = refer_cleaned_collection()
comp_data_entry = mycol.find({'_id': ObjectId(def_entry_id)})
data = [i for i in comp_data_entry]
try:
person_names = data[0]['dnb_cp_info']
if len(person_names) > 1:
person_names = person_names[-1]
else:
person_names = None
except KeyError:
return None
return person_names
def get_company(def_entry_id, key):
type = None
flag = False
check_link_list = ['www.', '.com', 'https', 'http']
mycol = refer_cleaned_collection()
comp_data_entry = mycol.find({'_id': ObjectId(def_entry_id)})
data = [i for i in comp_data_entry]
comp_name = data[0][key]
for i in range(len(check_link_list)):
if check_link_list[i] in comp_name:
flag = True
type = 'link'
break
if flag == False:
type = 'text'
return comp_name, type
def predict_emails(entry_id):
# Get the data for important_person_company for the given entry id
mycol = refer_cleaned_collection()
comp_data_entry = mycol.find({"_id": ObjectId(entry_id)})
data = [i for i in comp_data_entry]
important_person_company = data[0]['important_person_company']
comp_name, type = get_company(entry_id, 'search_text')
if type == 'link':
org = get_org(comp_name, type)
elif type == 'text':
if len(data[0]['search_text']) <= 15:
comp_name, type = get_company(entry_id, 'search_text')
org = get_org(comp_name, type)
elif len(data[0]['search_text']) > 15:
comp_name, type = get_company(entry_id, 'comp_name')
org = get_org(comp_name, type)
if important_person_company != 'No important persons found':
print(important_person_company)
# Iterating through the list of persons
for i in range(len(important_person_company)):
email = important_person_company[i]['email']
if email == None:
# Generate all possible combinations of emails and verify
email_list = generate_validate_email(
important_person_company[i]['name'], org)
important_person_company[i]['email'] = email_list
elif email != None:
continue
# Write back to the MongoDB record, the updated emails
write_to_mongodb(important_person_company, entry_id)