@Override
public void buildEdgeRelation(CategoryCache categoryCache) throws WikitException {
if (!categoryCache.hasDone()) {
throw new WikitException("Please create category cache first.");
}
String[] skippedCategories = new String[] {"跨学科领域", "总类", "词汇列表"};
Set<Integer> skippedCatIds = new HashSet<>();
for (String c : skippedCategories) {
skippedCatIds.add(categoryCache.getIdByName(c));
}
ProgressCounter counter = new ProgressCounter();
try {
String root = conf.getWikiRootCategoryName();
int id = categoryCache.getIdByName(root);
this.saveNameIdMapping(root, id);
this.rootId = id;
this.saveDepth(id, 0);
Queue<Integer> queue = new LinkedList<>();
queue.add(id);
while (!queue.isEmpty()) {
int pid = queue.poll();
int depth = this.getDepth(pid);
Set<Integer> childIds = categoryCache.getChildIds(pid);
for (Integer childId : childIds) {
if (skippedCatIds.contains(childId)) {
continue;
}
int childDepth = categoryCache.getDepth(childId, -1);
if (childDepth != (depth + 1)) {
continue; // skip
}
if (!idExist(childId)) {
String name = categoryCache.getNameById(childId);
this.saveNameIdMapping(name, childId);
this.saveDepth(childId, depth + 1);
queue.add(childId);
}
this.saveParents(childId, pid);
this.saveChildren(pid, childId);
}
counter.increment();
}
this.finishNameIdMapping();
this.edgeRelationCreated();
} catch (MissedException e) {
LOG.error(e.toString());
throw new WikitException(e);
}
counter.done();
LOG.info("Edge relation has created.");
}
private void saveParents(int catId, Collection<String> parents) {
byte[] key = (prefix + "id2pids").getBytes(ENCODING);
byte[] hkey = NumberUtils.int2Bytes(catId);
byte[] value = jedis.hget(key, hkey);
Set<Integer> ids = NumberUtils.bytes2IntSet(value);
boolean changed = false;
for (String p : parents) {
if (nameExist(p)) {
try {
ids.add(getIdByName(p));
changed = true;
} catch (MissedException e) {
LOG.warn(e.toString());
}
}
}
if (changed) jedis.hset(key, hkey, NumberUtils.intSet2Bytes(ids));
}
private void saveChildren(int catId, String... children) {
byte[] key = (prefix + "id2cids").getBytes(ENCODING);
byte[] hkey = NumberUtils.int2Bytes(catId);
byte[] value = jedis.hget(key, hkey);
Set<Integer> ids = NumberUtils.bytes2IntSet(value);
boolean changed = false;
for (String c : children) {
if (nameExist(c)) {
try {
ids.add(getIdByName(c));
changed = true;
} catch (MissedException e) {
e.printStackTrace();
}
}
}
if (changed) jedis.hset(key, hkey, NumberUtils.intSet2Bytes(ids));
}
