public void endDocument() throws SAXException {
// time to actually set up the edges
IntIterator rows = m_edges.rows();
while (rows.hasNext()) {
int r = rows.nextInt();
String src = m_edges.getString(r, SRCID);
if (!m_nodeMap.containsKey(src)) {
throw new SAXException(
"Tried to create edge with source node id=" + src + " which does not exist.");
}
int s = ((Integer) m_nodeMap.get(src)).intValue();
m_edges.setInt(r, SRC, s);
String trg = m_edges.getString(r, TRGID);
if (!m_nodeMap.containsKey(trg)) {
throw new SAXException(
"Tried to create edge with target node id=" + trg + " which does not exist.");
}
int t = ((Integer) m_nodeMap.get(trg)).intValue();
m_edges.setInt(r, TRG, t);
}
m_edges.removeColumn(SRCID);
m_edges.removeColumn(TRGID);
// now create the graph
m_graph = new Graph(m_nodes, m_edges, m_directed);
if (m_graphid != null) m_graph.putClientProperty(ID, m_graphid);
}
// this mutates nodeTable
private Table makeMergeTable(Table nodeTable, Graph mergeGraph, StringBuffer mergeLog) {
// get all the clusters in our merge graph (where each cluster is a group of nodes to be merged)
List clusters = extractWeakComponentClusters(mergeGraph);
// for each cluster...
for (Iterator clusterIt = clusters.iterator(); clusterIt.hasNext(); ) {
LinkedHashSet cluster = (LinkedHashSet) clusterIt.next();
// mark that we will merge every node into a single node
// (this step could be made smarter, but is ok for now. Will need user intervention in some
// cases)
Integer[] eachNodeInCluster = (Integer[]) cluster.toArray(new Integer[cluster.size()]);
Integer firstNode = null;
if (eachNodeInCluster.length <= 1) continue;
for (int ii = 0; ii < eachNodeInCluster.length; ii++) {
Integer node = eachNodeInCluster[ii];
if (firstNode == null) {
// (we arbitrarily choose the first node as the node to merge other nodes into)
firstNode = node;
String nodeOneAttribute =
(String) nodeTable.getString(firstNode.intValue(), this.compareAttributeName);
mergeLog.append("for \"" + nodeOneAttribute + "\"..." + "\r\n");
} else {
Integer nodeBeyondFirst = node;
// (we merge nodes beyond the first into the first node)
// (last value is off by one, because unique indices are 1-based instead of 0-based, but
// otherwise correlate with row number.
nodeTable.setInt(
nodeBeyondFirst.intValue(), UNIQUE_INDEX_COLUMN_NAME, firstNode.intValue() + 1);
String iAmNotThePrimaryNodeInCluster = "";
nodeTable.setString(
nodeBeyondFirst.intValue(),
COMBINE_VALUES_COLUMN_NAME,
iAmNotThePrimaryNodeInCluster);
String nodeOneAttribute =
(String) nodeTable.getString(nodeBeyondFirst.intValue(), this.compareAttributeName);
mergeLog.append(" merging in \"" + nodeOneAttribute + "\"" + "\r\n");
}
}
}
return nodeTable;
}
public Graph toGraph() {
Table edgeTable = new Table();
Table nodeTable = new Table();
HashMap<String, Integer> unique = new HashMap<String, Integer>();
edgeTable.addColumn("Node1", int.class);
edgeTable.addColumn("Node2", int.class);
nodeTable.addColumn("key", int.class);
nodeTable.addColumn("name", String.class);
nodeTable.addColumn("type", String.class);
nodeTable.addColumn("indeterminate", int.class);
int idx = 0;
for (Protein prot : minProteins.values()) {
int row = nodeTable.addRow();
unique.put(prot.getName(), idx);
nodeTable.setInt(row, "key", idx++);
nodeTable.setString(row, "name", prot.getName());
nodeTable.setString(row, "type", "protein");
nodeTable.setInt(row, "indeterminate", 0);
}
for (Peptide pep : minPeptides.values()) {
int row = nodeTable.addRow();
unique.put(pep.getSequence(), idx);
nodeTable.setInt(row, "key", idx++);
nodeTable.setString(row, "name", pep.getSequence());
nodeTable.setString(row, "type", "peptide");
if (pep.getIndeterminateType() == PeptideIndeterminacyType.NONE) {
nodeTable.setInt(row, "indeterminate", 0);
} else {
nodeTable.setInt(row, "indeterminate", 1);
}
}
for (Protein prot : minProteins.values()) {
int id1 = unique.get(prot.getName());
for (String pep : prot.getPeptides()) {
int id2 = unique.get(pep);
int row = edgeTable.addRow();
edgeTable.setInt(row, "Node1", id1);
edgeTable.setInt(row, "Node2", id2);
}
}
Graph g = new Graph(nodeTable, edgeTable, false, "key", "Node1", "Node2");
// System.err.println(g.getEdgeCount());
return g;
}
