/**
* suffers from long time skipping to the offset "from", has to sort everything and then start
* skipping nodes.
*
* @param xp
* @param from
* @param count
* @return
*/
public List<XMLNode> getByXpathHolder(XpathHolder xp, long from, long count) {
return getSession()
.createSQLQuery(
"select * from xml_node_"
+ xp.getXmlObject().getDbID()
+ " where xpath_summary_id = :xpath and xml_object_id = :xo "
+ "order by xml_node_id asc")
.addEntity(XMLNode.class)
.setEntity("xpath", xp)
.setEntity("xo", xp.getXmlObject())
.setMaxResults((int) count)
.setFirstResult((int) from)
.list();
}
/**
* Retrieve the given node with subtree and all parent elements. No actual value wrapping is done
* in this method! Only empty element wrapping to adjust for straight xpaths.
*
* @param parent
* @return
*/
public XMLNode wrappedDOMTree(XMLNode parent) {
XMLNode result = getDOMTree(parent);
do {
XpathHolder parentPath = result.getXpathHolder().getParent();
if (parentPath == null) break; // shouldnt happen
if (parentPath.getParent() == null) break; // this should happen
XMLNode newParent = new XMLNode();
newParent.getChildren().add(result);
result.setParentNode(newParent);
newParent.setXmlObject(result.getXmlObject());
newParent.setXpathHolder(parentPath);
result = newParent;
} while (true);
return result;
}
public List<Object[]> getValues(XpathHolder xp, int start, int count, String filter) {
List<Object[]> l = Collections.emptyList();
String cond = null;
if (filter != null) {
cond = "and content like :filter ";
}
Query q =
getSession()
.createSQLQuery(
"select content, count(*) "
+ "from xml_node_"
+ xp.getXmlObject().getDbID()
+ " where xpath_summary_id = :xpath "
+ (cond != null ? cond : "")
+ "group by content "
+ "order by content ")
.setEntity("xpath", xp);
if (count > 0) {
q.setMaxResults(count).setFirstResult(start);
}
if (cond != null) {
q.setString("filter", "%" + filter + "%");
}
l = q.list();
return l;
}
public float getAvgLength(XpathHolder xpathHolder) {
Double val;
if (xpathHolder == null) return -1f;
if (!(xpathHolder.isTextNode() || xpathHolder.isAttributeNode())) {
xpathHolder = xpathHolder.getTextNode();
if (xpathHolder == null) return -1f;
}
val =
(Double)
getSession()
.createQuery(
"select avg( length (content) ) from XMLNode "
+ "where xmlObject = :xo and xpathHolder = :xp ")
.setEntity("xo", xpathHolder.getXmlObject())
.setEntity("xp", xpathHolder)
.uniqueResult();
return val.floatValue();
}
public List<XMLNode> getStatelessByXpathHolder(XpathHolder xp, long from, long count) {
List<XMLNode> l =
DB.getStatelessSession()
.createSQLQuery(
"select * from xml_node_"
+ xp.getXmlObject().getDbID()
+ " where xpath_summary_id = :xpath and xml_object_id = :xo "
+ "order by xml_node_id asc")
.addEntity(XMLNode.class)
.setEntity("xpath", xp)
.setEntity("xo", xp.getXmlObject())
.setMaxResults((int) count)
.setFirstResult((int) from)
.list();
for (XMLNode node : l) {
node.setXpathHolder(DB.getXpathHolderDAO().findById(node.getXpathHolder().getDbID(), false));
}
return l;
}
/**
* Quickly build a dom tree. Use per item, not for trees with more than 2000 nodes (or about
* that). The XMLNodes are not in the Hibernate session and don't lazy load their parent or
* anything else. They should not need to, though. The attached XpathHolders are in the session
* and behave normally.
*
* @param parent
* @return
*/
public XMLNode getDOMTree(XMLNode parent) {
StatelessSession ss = DB.getStatelessSession();
List<XMLNode> l;
int maxNodes = 10000;
if (parent.getSize() < maxNodes) maxNodes = (int) parent.getSize();
Stack<XMLNode> stack = new Stack<XMLNode>();
HashMap<Long, XpathHolder> xpathCache = new HashMap<Long, XpathHolder>();
l =
ss.createSQLQuery(
"select * from xml_node_"
+ parent.getXmlObject().getDbID()
+ " where xml_node_id >= :parentId order by xml_node_id")
.addEntity(XMLNode.class)
.setLong("parentId", parent.getNodeId())
.setMaxResults(maxNodes)
.list();
// now every node has the wrong parent and XpathHolder and no Children..
for (XMLNode x : l) {
// find the right place in stack
x.setChildren(new ArrayList<XMLNode>());
while (!stack.isEmpty()) {
if (stack.peek().getNodeId() != x.getParentNode().getNodeId()) stack.pop();
else break;
}
if (!stack.isEmpty()) {
x.setParentNode(stack.peek());
stack.peek().getChildren().add(x);
}
stack.push(x);
// now the xpathholder
XpathHolder path = xpathCache.get(x.getXpathHolder().getDbID());
if (path == null) {
path = DB.getXpathHolderDAO().findById(x.getXpathHolder().getDbID(), false);
xpathCache.put(path.getDbID(), path);
}
x.setXpathHolder(path);
x.setXmlObject(parent.getXmlObject());
}
if (l.size() > 0) return l.get(0);
else return null;
}
/**
* Cursor over all nodes. (You still have to get XpathHolders the normal way) Call the index
* function on given object. Proceeds in node order.
*
* @param xo
* @param ni
*/
public void indexNodes(XmlObject xo, NodeIndexer ni) {
StatelessSession ss = DB.getStatelessSession();
Stack<XMLNode> stack = new Stack<XMLNode>();
HashMap<Long, XpathHolder> xpathCache = new HashMap<Long, XpathHolder>();
ScrollableResults sr = null;
try {
sr =
ss.createSQLQuery("select * from xml_node_" + xo.getDbID() + " order by xml_node_id")
.addEntity(XMLNode.class)
.scroll(ScrollMode.FORWARD_ONLY);
while (sr.next()) {
XMLNode x = (XMLNode) sr.get()[0];
while (!stack.isEmpty()) {
if ((x.getParentNode() == null)
|| (stack.peek().getNodeId() != x.getParentNode().getNodeId())) stack.pop();
else break;
}
stack.push(x);
// now the xpathholder
XpathHolder path = xpathCache.get(x.getXpathHolder().getDbID());
if (path == null) {
path = DB.getXpathHolderDAO().findById(x.getXpathHolder().getDbID(), false);
xpathCache.put(path.getDbID(), path);
}
x.setXpathHolder(path);
x.setXmlObject(xo);
// node ready to index
ni.index(x);
}
ni.index(null);
} catch (Exception e) {
log.error("Error while scrolling XMLNodes for indexing.", e);
} finally {
if (sr != null) sr.close();
}
}
/**
* Looks at content for simple nodes and checksums for subtrees.
*
* @param xp
* @return
*/
public long countDistinct(XpathHolder xp) {
Long val;
if (xp.name.equals("text()") || xp.name.startsWith("@"))
val =
(Long)
getSession()
.createQuery(
"select count( distinct content ) from XMLNode where xpathHolder = :xpath and xmlObject = :xo")
.setEntity("xpath", xp)
.setEntity("xo", xp.getXmlObject())
.uniqueResult();
else
val =
(Long)
getSession()
.createQuery(
"select count( distinct checksum ) from XMLNode where xpathHolder = :xpath and xmlObject = :xo ")
.setEntity("xpath", xp)
.setEntity("xo", xp.getXmlObject())
.uniqueResult();
return val.longValue();
}
/**
* Ordered List of value, frequency for given xpath. Only given number of values listed.
*
* @param xp
* @param limit
* @return
*/
public List<Object[]> getCountByValue(XpathHolder xp, int limit) {
if (xp.name.equals("text()") || xp.name.startsWith("@")) {
List<Object[]> l =
(List<Object[]>)
getSession()
.createQuery(
"select content, count( * ) "
+ "from XMLNode where xpathHolder = :xpath "
+ "and xmlObject = :xo "
+ "group by content "
+ "order by count(*) desc")
.setEntity("xpath", xp)
.setEntity("xo", xp.getXmlObject())
.setMaxResults(limit)
.list();
return l;
} else return Collections.emptyList();
}
