we geocoding

1 files changed, 51 insertions, 0 deletions

diff --git a/s2-geocode.py b/s2-geocode.py
new file mode 100644
index 00000000..bd0885c3
--- /dev/null
+++ b/s2-geocode.py
@@ -0,0 +1,51 @@
+import os
+import glob
+import time
+import simplejson as json
+import click
+from urllib.parse import urlparse
+import operator
+from util import *
+from geopy.geocoders import Nominatim
+import random
+
+@click.command()
+@click.option('--fn', '-f', default='reports/institution_names.txt', help='List of institution names, to be geocoded :)')
+def s2_geocode(fn):
+  geolocator = Nominatim(user_agent="cool geocoding service")
+  print(fn)
+  rows = read_csv(fn, keys=False)
+  valid = read_csv('./reports/institutions_geocoded.csv', create=True)
+  invalid = read_csv('./reports/institutions_not_found.csv', create=True)
+  valid_names = []
+  invalid_names = []
+  random.shuffle(rows)
+  for row, i in rows:
+    name = row[0]
+    if name in invalid_names:
+      continue
+    if name in valid_names:
+      continue
+    location = geolocator.geocode(name)
+    if location:
+      print("found: {}".format(name))
+      valid.append([
+        name,
+        location.latitude,
+        location.longitude,
+        location.address,
+      ])
+      valid_names.append(name)
+    else:
+      print("not found: {}".format(name))
+      invalid.append(row)
+      invalid_names.append(name)
+    if (i % 20) == 0:
+      write_csv('./reports/institutions_geocoded.csv', keys=None, rows=valid)
+      write_csv('./reports/institutions_not_found.csv', keys=None, rows=invalid)
+    if (i % 100) == 0:
+      print("{}...".format(i))
+    time.sleep(5)
+
+if __name__ == '__main__':
+  s2_geocode()