public static void getSentence(String input, String output) throws IOException {
DelimitedWriter dw = new DelimitedWriter(output);
HashMap<Integer, String> targetArticles = new HashMap<Integer, String>();
HashCount<String> mid2numSen = new HashCount<String>();
{
DelimitedReader dr = new DelimitedReader(input);
String[] l;
while ((l = dr.read()) != null) {
int x = Integer.parseInt(l[1]);
targetArticles.put(x, l[0]);
}
dr.close();
}
{
DelimitedReader dr = new DelimitedReader(Main.file_wiki_sentence + ".tokens");
String[] l;
while ((l = dr.read()) != null) {
String mid = l[1];
int artid = Integer.parseInt(mid);
if (targetArticles.containsKey(artid)) {
// the first sentence of this article
if (mid2numSen.see(mid) < 5) {
dw.write(mid, l[3]);
mid2numSen.add(mid);
}
}
}
dr.close();
}
dw.close();
}
public static void getWidOfNobleMen(String input, String output) throws IOException {
HashMap<String, Integer> mid2wid = new HashMap<String, Integer>();
{
DelimitedReader dr = new DelimitedReader(Main.dir_freebase + "/mid2wid");
String[] l;
while ((l = dr.read()) != null) {
mid2wid.put(l[0], Integer.parseInt(l[1]));
}
dr.close();
}
{
DelimitedReader dr = new DelimitedReader(input);
DelimitedWriter dw = new DelimitedWriter(output);
String[] l;
while ((l = dr.read()) != null) {
String mid = l[0];
Integer wid = mid2wid.get(mid);
if (wid != null) {
dw.write(mid, wid);
}
}
dr.close();
dw.close();
}
}
private static void sampleSql2isntance4DebugPurpose() throws IOException {
HashMap<Integer, String> wid2title = new HashMap<Integer, String>();
{
// load wid 2 title
DelimitedReader dr = new DelimitedReader(Main.file_gnid_mid_wid_title);
String[] l;
while ((l = dr.read()) != null) {
wid2title.put(Integer.parseInt(l[2]), l[3]);
}
}
HashCount<String> hc = new HashCount<String>();
{
DelimitedReader dr = new DelimitedReader(Main.file_sql2instance);
DelimitedWriter dw = new DelimitedWriter(Main.file_sql2instance + ".sample10");
String[] l;
while ((l = dr.read()) != null) {
String key = l[2] + "\t" + l[3];
int a = hc.see(key);
if (a < 10) {
int wid1 = Integer.parseInt(l[0]);
int wid2 = Integer.parseInt(l[1]);
String title1 = wid2title.get(wid1);
String title2 = wid2title.get(wid2);
if (title1 != null && title2 != null) {
dw.write(wid1, wid2, title1, title2, l[2], l[3]);
hc.add(key);
}
}
}
dw.close();
dr.close();
}
}
static void step2() throws IOException {
{
HashMap<String, String[]> mid2other = new HashMap<String, String[]>();
DelimitedWriter dw = new DelimitedWriter(file_targetrelation + ".2");
{
DelimitedReader dr = new DelimitedReader(Main.file_mid2typename);
String[] l;
while ((l = dr.read()) != null) {
String mid = l[0];
mid2other.put(mid, l);
}
dr.close();
}
{
DelimitedReader dr = new DelimitedReader(file_targetrelation + ".1");
String[] l;
HashSet<String> avoidduplicate = new HashSet<String>();
while ((l = dr.read()) != null) {
String mid1 = l[1];
String mid2 = l[3];
String[] info1 = mid2other.get(mid1);
String[] info2 = mid2other.get(mid2);
if (!avoidduplicate.contains(mid1 + "\t" + mid2)
&& info1 != null
&& info2 != null
&& !mid1.equals(mid2)) {
dw.write(mid1, mid2, info1[3], info2[3], info1[2], info2[2]);
avoidduplicate.add(mid1 + "\t" + mid2);
}
}
dr.close();
}
dw.close();
}
{
HashSet<String> typeOfArg2 = new HashSet<String>();
List<String[]> all = (new DelimitedReader(file_targetrelation + ".2")).readAll();
for (String[] a : all) {
typeOfArg2.add(a[5]);
}
for (String a : typeOfArg2) {
D.p(a);
}
}
}
public static void getAllNobleMen(String output) throws IOException {
DelimitedReader dr = new DelimitedReader(Main.dir_freebase + "/visible");
DelimitedWriter dw = new DelimitedWriter(output);
String[] l;
/// j0 /m/04l4q52 /type/type/instance /m/02rg_
while ((l = dr.read()) != null) {
if (l[1].equals(MIDOFNOBLEPERSON) && l[2].equals("/type/type/instance")) {
dw.write(l[3]);
}
}
dr.close();
dw.close();
}
static void step1() throws IOException {
Set<String> usedrel = new HashSet<String>();
for (String r : mappedrelation) usedrel.add(r);
DelimitedReader dr = new DelimitedReader(Main.file_fbvisibledump);
DelimitedWriter dw = new DelimitedWriter(file_targetrelation + ".1");
String[] l;
while ((l = dr.read()) != null) {
if (usedrel.contains(l[2])) {
dw.write(l);
}
}
dr.close();
dw.close();
}