public static void analyzePbData(String pb) throws IOException {
DelimitedWriter dw = new DelimitedWriter(pb + ".analyze");
List<Relation> all = new ArrayList<Relation>();
int numCases = 0;
{
HashCount<String> labelDist = new HashCount<String>();
Relation r = null;
PbReader pr = new PbReader(pb);
while ((r = pr.read()) != null) {
all.add(r);
labelDist.add(r.getRelType());
numCases++;
}
pr.close();
dw.write("Label Distribution");
Iterator<Entry<String, Integer>> it = labelDist.iterator();
while (it.hasNext()) {
Entry<String, Integer> e = it.next();
dw.write(e.getKey(), e.getValue());
}
}
Collections.shuffle(all);
{
HashCount<String> hassampled = new HashCount<String>();
List<String[]> towrite = new ArrayList<String[]>();
for (int i = 0; i < all.size(); i++) {
Relation r = all.get(i);
String rel = r.getRelType();
if (hassampled.see(rel) < 10) {
towrite.add(new String[] {rel, r.getSourceGuid(), r.getDestGuid()});
hassampled.add(rel);
}
}
StringTable.sortByColumn(towrite, new int[] {0});
for (String[] w : towrite) {
dw.write(w);
}
}
dw.close();
}
public static void main(String[] args) throws IOException {
InputStream is = new FileInputStream(input1);
Relation r = null;
while ((r = Relation.parseDelimitedFrom(is)) != null) {
for (int i = 0; i < r.getMentionCount(); i++) {
RelationMentionRef rmf = r.getMention(i);
System.out.println(r.getSourceGuid() + "\t" + r.getDestGuid() + "\t" + r.getRelType());
}
}
is.close();
}