added file: search_GeoNames_in_OpenRefine.py

search_GeoNames_in_OpenRefine.py

+import json
+import urllib2
+
+###
+# This script is used for the annotating geographical locations with the GeoNames IDs. It is run within OpenRefine. 
+# This script is highly specific for our usecase, i.e. the hardcoded column names, but we decided to share it nontheless, so it might be helpful for others with similar problems.
+# License: Mozilla Public License 2.0 (MPL 2.0)
+###
+
+
+
+USERNAME = '' #enter your geonames username here, otherwise the API calls will not work
+
+
+
+DOMAIN = 'http://api.geonames.org/'
+DICTIONARY = 'locations.json'
+
+#Dictionary providing entries already handled with possible cases:
+# Case 'not found'
+# Case 'ambiguous'
+# Case 'URI|Latitude|Longitude'
+with open(DICTIONARY) as json_file:  
+  data = json.load(json_file)
+
+#Current location given by the data table
+cellVal = cells['Fundort']['value']
+#return(cellVal)
+
+#Coordinates of the current entry
+latitude = float(cells['Latitude']['value'].replace(',', '.'))
+longitude = float(cells['Longitude']['value'].replace(',', '.'))
+coordinatesValue = latitude + longitude
+
+#If the current entry exists in dictionary and decide what to return
+if len(data) > 0 and cellVal in data:
+  if data[cellVal] == "not found":
+    return "not found"
+  if data[cellVal] == "ambiguous":
+    return "ambiguous"
+  elif data[cellVal] != " ":
+    dataEntry = data[cellVal].split('|')
+    latitudeData = float(dataEntry[1])
+    longitudeData = float(dataEntry[2])
+    diffCoordinates = abs(latitude - latitudeData) + abs(longitude - longitudeData)
+    if diffCoordinates < 0.06: 
+      return dataEntry[0]
+    else:
+      return "too distant"
+
+#If the current entry is unknown
+else:
+  method = 'searchJSON'
+  name = cells['Fundort']['value']
+  name = urllib2.quote(name.encode('utf-8'))
+  name = name.replace(" ", "+")
+  country = cells['fkIsoCode']['value']
+  adm2 = cells['Regional Unit']['value']
+  adm2 = urllib2.quote(adm2.encode('utf-8'))
+  adm2 = adm2.replace(" ", "+")
+  uri = "http://api.geonames.org/searchJSON?name=" + name + "&country=" + country + "&fuzzy=0.6" +  "&username=" + USERNAME
+  print("Abfrage!")
+  #resource = urllib2.urlopen(uri)
+  resourceString = resource.read()
+  js = json.loads(resourceString)
+  numberOfResults = js['totalResultsCount']
+  if len(js['geonames']) == 0:
+    data[cells['Fundort']['value']] = "not found"
+    with open(DICTIONARY, 'w') as outfile:  
+        json.dump(data, outfile)
+    return "not found"
+  if numberOfResults == 1:
+    latitudeResult = float(js['geonames'][0]['lat'])
+    longitudeResult = float(js['geonames'][0]['lng'])
+    diffCoordinates = abs(latitude-latitudeResult) + abs(longitude-longitudeResult)
+    if diffCoordinates < 0.06:
+      data[cells['Fundort']['value']] = "http://www.geonames.org/" + str(js['geonames'][0]['geonameId']) + "|" + str(latitudeResult) + "|" + str(longitudeResult)
+      with open(DICTIONARY, 'w') as outfile:  
+        json.dump(data, outfile)
+        return data[cellVal].split("|")[0]
+    else:
+      data[cells['Fundort']['value']] = "http://www.geonames.org/" + str(js['geonames'][0]['geonameId']) + "|" + str(latitudeResult) + "|" + str(longitudeResult)
+      with open(DICTIONARY, 'w') as outfile:  
+        json.dump(data, outfile)
+      return "too distant"
+  nearestPlace = []
+  for entry in range(0, numberOfResults):
+    latitudeResult = float(js['geonames'][entry]['lat'])
+    longitudeResult = float(js['geonames'][entry]['lng'])
+    diffCoordinates = abs(latitude-latitudeResult) + abs(longitude-longitudeResult)
+    if len(nearestPlace) > 0:
+      if diffCoordinates < nearestPlace[3]:
+        nearestPlace[0] = str(js['geonames'][entry]['geonameId'])
+        nearestPlace[1] = latitudeResult
+        nearestPlace[2] = longitudeResult
+        nearestPlace[3] = diffCoordinates
+    else: 
+      nearestPlace.append(str(js['geonames'][entry]['geonameId']))
+      nearestPlace.append(latitudeResult)
+      nearestPlace.append(longitudeResult)
+      nearestPlace.append(diffCoordinates) 
+  if nearestPlace[3] < 0.06:
+    data[cells['Fundort']['value']] = "http://www.geonames.org/" + str(nearestPlace[0]) + "|" + str(nearestPlace[1]) + "|" + str(nearestPlace[2])
+    nearestPlace = []
+    with open(DICTIONARY, 'w') as outfile:  
+      json.dump(data, outfile)
+    return data[cellVal].split("|")[0]
+  else:
+    data[cells['Fundort']['value']] = "http://www.geonames.org/" + str(js['geonames'][0]['geonameId']) + "|" + str(latitudeResult) + "|" + str(longitudeResult)
+    with open(DICTIONARY, 'w') as outfile:  
+      json.dump(data, outfile)
+    return "too distant"
\ No newline at end of file