X-Git-Url: https://code.wpia.club/?p=gigi.git;a=blobdiff_plain;f=util%2Forg%2Fcacert%2Fgigi%2Futil%2FHighFinancialValueFetcherUmbrella.java;fp=util%2Forg%2Fcacert%2Fgigi%2Futil%2FHighFinancialValueFetcherUmbrella.java;h=ef96a0fe83127bef0daf5b83a7a6252a72ec2111;hp=0000000000000000000000000000000000000000;hb=4adc67e31c14c5f1192eec188a78595e044d92f8;hpb=d8d62e3c43cde58c8b1edbd875562ea863cce496 diff --git a/util/org/cacert/gigi/util/HighFinancialValueFetcherUmbrella.java b/util/org/cacert/gigi/util/HighFinancialValueFetcherUmbrella.java new file mode 100644 index 00000000..ef96a0fe --- /dev/null +++ b/util/org/cacert/gigi/util/HighFinancialValueFetcherUmbrella.java @@ -0,0 +1,29 @@ +package org.cacert.gigi.util; + +import java.io.File; +import java.io.PrintWriter; +import java.util.HashSet; + +public class HighFinancialValueFetcherUmbrella extends HighFinancialValueFetcher { + + public HighFinancialValueFetcherUmbrella(File f, int max) { + super(f, max, "https://s3-us-west-1.amazonaws.com/umbrella-static/top-1m.csv.zip"); + } + + private HashSet printed = new HashSet<>(); + + @Override + public void handle(String line, PrintWriter fos) { + String[] parts = line.split(","); + // Assert that the value before the "," is an integer + Integer.parseInt(parts[0]); + + String registrablePart = PublicSuffixes.getInstance().getRegistrablePart(parts[1]); + if (registrablePart != null && printed.add(registrablePart)) { + emit(fos, registrablePart); + System.out.println(registrablePart); + } + + } + +}