def get_db_connection(): if MONGO: return MongoDatabase('va_court_search', COURT_TYPE) if POSTGRES: return PostgresDatabase(COURT_TYPE) return None
fields_to_anonomize = [ 'Defendant', 'Name', 'AKA', 'AKA1', 'AKA2' ] court_type = sys.argv[1] if court_type != 'circuit' and court_type != 'district': raise ValueError('Unknown court type') anonymize = True if len(sys.argv) > 4 and sys.argv[4] == 'original': anonymize = False print 'DATA NOT ANONYMIZED' # connect to database db = MongoDatabase('va_court_search', court_type) courts = list(db.get_courts()) courts_by_fips = {court['fips_code']:court for court in courts} s3 = boto3.resource('s3') def simplify_time(time_string): time_string= time_string.replace(' Year(s)', 'Years ')\ .replace(' Month(s)', 'Months ')\ .replace(' Day(s)', 'Days ') days = 0 string_parts = time_string.split(' ') for string_part in string_parts: if 'Years' in string_part: days += int(string_part.replace('Years','')) * 365
from courtreader import readers MONGO = False POSTGRES = True if MONGO: import pymongo from courtutils.databases.mongo import MongoDatabase if POSTGRES: from courtutils.databases.postgres import PostgresDatabase #geolocator = GoogleV3(api_key=os.environ['GOOGLE_API_KEY']) print 'CIRCUIT COURT' circuit_db = None if MONGO: circuit_db = MongoDatabase('va_court_search', 'circuit') if POSTGRES: circuit_db = PostgresDatabase('circuit') circuit_db.drop_courts() reader = readers.CircuitCourtReader() courts = reader.connect() court_names = [] for fips, court in courts.iteritems(): print court['name'] court_locality = court['name'].replace(' Circuit Court', '') #location = geolocator.geocode(court_locality + ', Virginia, USA') circuit_db.add_court(court['name'], fips, None) court_names.append(court['name'] + ' ' + fips) circuit_db.add_court_location_index() circuit_db.commit() ''' court_names.sort()
start_date = datetime.strptime(sys.argv[1], '%m/%d/%Y') end_date = datetime.strptime(sys.argv[2], '%m/%d/%Y') if start_date < end_date: raise ValueError('Start Date must be after End Date so they decend') court_type = sys.argv[3] if court_type != 'circuit' and court_type != 'district': raise ValueError('Unknown court type') case_type = sys.argv[4] if case_type != 'criminal' and case_type != 'civil': raise ValueError('Unknown case type') # connect to database db = None if MONGO: db = MongoDatabase('va_court_search', court_type) if POSTGRES: db = PostgresDatabase(court_type) # get the courts to create tasks for # check command line args for a specific court courts = list(db.get_courts()) if len(sys.argv) > 5: courts = [court for court in courts if court['fips'] == sys.argv[5]] # create the tasks tasks = [] for court in courts: tasks.append({ 'fips': court['fips'], 'start_date': start_date, 'end_date': end_date,
'OperatorLicenseSuspensionTime', 'ProbationTime' ] fields_to_anonomize = ['Defendant', 'Name', 'AKA', 'AKA1', 'AKA2'] court_type = sys.argv[1] if court_type != 'circuit' and court_type != 'district': raise ValueError('Unknown court type') anonymize = True if len(sys.argv) > 4 and sys.argv[4] == 'original': anonymize = False print 'DATA NOT ANONYMIZED' # connect to database db = MongoDatabase('va_court_search', court_type) courts = list(db.get_courts()) courts_by_fips = {court['fips_code']: court for court in courts} s3 = boto3.resource('s3') def simplify_time(time_string): time_string= time_string.replace(' Year(s)', 'Years ')\ .replace(' Month(s)', 'Months ')\ .replace(' Day(s)', 'Days ') days = 0 string_parts = time_string.split(' ') for string_part in string_parts: if 'Years' in string_part:
if MONGO: from courtutils.databases.mongo import MongoDatabase if POSTGRES: from courtutils.databases.postgres import PostgresDatabase # get command line args start_date = datetime.strptime(sys.argv[1],'%m/%d/%Y') end_date = datetime.strptime(sys.argv[2],'%m/%d/%Y') if start_date < end_date: raise ValueError('Start Date must be after End Date so they decend') court_type = sys.argv[3] if court_type != 'circuit' and court_type != 'district': raise ValueError('Unknown court type') # connect to database db = None if MONGO: db = MongoDatabase('va_court_search', 'circuit') if POSTGRES: db = PostgresDatabase('va_court_search', 'circuit') # get the courts to create tasks for # check command line args for a specific court courts = list(db.get_courts()) if len(sys.argv) > 4: courts = [court for court in courts if court['fips'] == sys.argv[4]] # create the tasks tasks = [] for court in courts: tasks.append({ 'fips': court['fips'], 'start_date': start_date, 'end_date': end_date
import os import pymongo from geopy.geocoders import GoogleV3 from courtreader import readers MONGO = False POSTGRES = True if MONGO: from courtutils.databases.mongo import MongoDatabase if POSTGRES: from courtutils.databases.postgres import PostgresDatabase geolocator = GoogleV3(api_key=os.environ['GOOGLE_API_KEY']) print 'CIRCUIT COURT' circuit_db = None if MONGO: circuit_db = MongoDatabase('va_court_search', 'circuit') if POSTGRES: circuit_db = PostgresDatabase('va_court_search', 'circuit') circuit_db.drop_courts() reader = readers.CircuitCourtReader() courts = reader.connect() court_names = [] for fips, court in courts.iteritems(): print court['name'] court_locality = court['name'].replace(' Circuit Court', '') location = geolocator.geocode(court_locality + ', Virginia, USA') circuit_db.add_court(court['name'], fips, location) court_names.append(court['name'] + ' ' + fips) circuit_db.add_court_location_index() circuit_db.commit() '''