Changeset 33603 for gs3-extensions/maori-lang-detection/src/org/greenstone/atea/Utility.java

Timestamp:

2019-10-24T22:04:37+13:00 (4 years ago)

Author:

ak19

Message:

Incorporating Dr Nichols suggestion to help weed out product sites: if tld of seed URL addresses containing /mi/ is outside NZ, add to list of possible-product-sites.txt. This should be a smaller number hopefully than all urls containing /mi and, because they're located outside nz, more likely to be a product site than not.

File:

• gs3-extensions/maori-lang-detection/src/org/greenstone/atea/Utility.java (modified) (2 diffs)

gs3-extensions/maori-lang-detection/src/org/greenstone/atea/Utility.java

@@ -2 +2 @@
 
 import java.io.*;
+import java.net.InetAddress;
 import java.util.zip.GZIPInputStream;
+
+import com.maxmind.geoip.*; // for LookupService and Location
 
 import org.apache.log4j.Logger;
@@ -41 +44 @@
     }
 
+    /**
+     * Attribution following below is as per https://dev.maxmind.com/geoip/geoip2/geolite2/
+     *
+     * This product includes GeoLite2 data created by MaxMind, available from
+     * <a href="https://www.maxmind.com">https://www.maxmind.com</a>.
+     *
+     * Usage:
+     * https://stackoverflow.com/questions/1415851/best-way-to-get-geo-location-in-java
+     * version I'm using: https://github.com/maxmind/geoip-api-java
+     * Newer version: https://maxmind.github.io/GeoIP2-java/
+     */
+    public static boolean isDomainInCountry(String domainWithProtocol,
+                        String countryCode, File geoLiteCityDatFile)
+    {
+    countryCode = countryCode.toUpperCase();
+    
+    int startIndex = domainWithProtocol.indexOf("//"); // http:// or https:// prefix
+    startIndex = (startIndex == -1) ? 0 : (startIndex+2); // skip past the protocol's // portion
+    String domain = domainWithProtocol.substring(startIndex);   
+    
+    boolean result = false;
+    try {
+        // pass in the GeoLiteCity.dat file to be able to do the location lookup for domain's IP
+        LookupService cl = new LookupService(geoLiteCityDatFile, LookupService.GEOIP_MEMORY_CACHE);
+        
+        // get IP for domain
+        InetAddress inetAddress = InetAddress.getByName(domain);
+        String ipAddress = inetAddress.getHostAddress();
+        
+        // get location object for IP
+        Location location = cl.getLocation(ipAddress);
+
+        if(location != null) {
+        // compare country code with tld parameter
+        System.err.println("@@@@ Got country code: " + location.countryCode);
+        result = location.countryCode.equals(countryCode);
+        } else {
+        System.err.println("@@@@ No location info in DB for: " + domainWithProtocol);
+        }
+    } catch(Exception e) {
+        e.printStackTrace();
+        System.err.println("Could not check if domain " + domain + " was in country: " + countryCode);
+    } finally {
+        return result;
+    }
+    }
 }