Changeset 33604 for gs3-extensions/maori-lang-detection/src/org/greenstone/atea/Utility.java

Timestamp:

2019-10-24T23:22:30+13:00 (5 years ago)

Author:

ak19

Message:

1. Better output into possible-product-sites.txt including the overseas country code prefix to help decide whether the site is worth keeping or not. 2. Updated whitelisting and top-sites filters to grab the /mi/ subsections of sites that don't appear to be autotranslated. This is done in preparation for blocking out product sites hereafter

File:

• gs3-extensions/maori-lang-detection/src/org/greenstone/atea/Utility.java (modified) (1 diff)

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

@@ -54 +54 @@
      * version I'm using: https://github.com/maxmind/geoip-api-java
      * Newer version: https://maxmind.github.io/GeoIP2-java/
+     *
+     * @return 2 letter countrycode in uppercase or an exception
      */
+    public static String getCountryCodeOfDomain(String domainWithProtocol, File geoLiteCityDatFile)
+    throws Exception
+    {   
+    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);   
+    
+    // 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) {
+        throw new Exception("@@@@ No location info in DB for: " + domain);
+    } else {
+        return location.countryCode;
+    }
+    
+    }
+    
     public static boolean isDomainInCountry(String domainWithProtocol,
                         String countryCode, File geoLiteCityDatFile)