import geocoder
import csv
import sys
import numpy as np
from collections import Counter
import re
import time

#file paths
filepath1 = "diocese_correct.csv"
filepath2 = "dachau_initial_file.csv"

#list to store diocese names
diocese = []

with open(filepath1, 'rU') as f1:

        #create csv reader object
        reader1 = csv.reader(f1)
        for row in reader1:
                diocese.append(row[1])

#zero'd numpy array to contain index, latitude, longitude
coords = np.zeros((len(diocese), 3))

for i in range(0, len(diocese)):

	#if the string field for the diocese is not empty (e.g., contains letters), then find latitude and longitude using geocoder
	if re.search('[a-zA-Z]', diocese[i]):

		g = geocoder.geonames(diocese[i], username='benjaminlee')

		#g is returned in JSON form, so easy to query for latitude (lat) and longitude (lng)
		coords[i][0] = i
		coords[i][1] = g.lat
		coords[i][2] = g.lng


#opens full Dachau file for reading and appending lat, lng
with open(filepath2, 'rU') as f2:

        reader2 = csv.reader(f2)

        #opens output CSV file for writing
        with open('dachau_with_geolocations.csv', 'w') as outfile:

                writer = csv.writer(outfile)

		i = 0

                #iterate over the rows of the bio CSV
                for row in reader2:

			#if there is a coordinate
			if coords[i][1] != 0:
#				row.insert(0, str(coords[i][1]) + ',' + str(coords[i][2]))
				row.insert(0, coords[i][1])
				row.insert(1, coords[i][2])
				writer.writerow(row)

			else:
				row.insert(0, '')
				row.insert(1, '')
				writer.writerow(row)


			i= i + 1
			if i == 2679:
				break

sys.exit()