public static void indexTable(
SecureTable t, DboTableMeta tableMeta, Collection<SolrInputDocument> solrDocs)
throws IOException, SAXException, ParserConfigurationException, SolrServerException {
// Solr server instance
SolrServer solrServer = Search.getSolrServer();
// Add this new table to the meta index...
createCoreIfNeeded("databusmeta", "databusmeta", solrServer);
SolrInputDocument doc = new SolrInputDocument();
// Add special fields to track the type of record an the primary
// key.
doc.addField("id", t.getTableName());
doc.addField("type", "table");
if (t.getCreator() != null) doc.addField("creator_texts", t.getCreator().getUsername());
doc.addField("description_texts", t.getDescription());
if (t.getSchema() != null) {
doc.addField("database_texts", t.getSchema().getSchemaName());
doc.addField("databaseDescription_texts", t.getSchema().getDescription());
}
if (t.isSearchable()) doc.addField("isSearchable_texts", "true");
else doc.addField("isSearchable_texts", "false");
Set<String> allTermsSet = new HashSet<String>();
Set<String> columnsSet = new HashSet<String>();
for (DboColumnMeta m : tableMeta.getAllColumns()) {
SdiColumn sdicol = t.getNameToField().get(m.getColumnName());
columnsSet.add(sdicol.getColumnName());
}
doc.addField("column_texts", columnsSet);
allTermsSet.add((String) doc.getField("id").getValue());
allTermsSet.add((String) doc.getField("type").getValue());
allTermsSet.add((String) doc.getField("description_texts").getValue());
allTermsSet.addAll(columnsSet);
if (doc.getField("database_texts") != null) {
allTermsSet.add((String) doc.getField("database_texts").getValue());
allTermsSet.add((String) doc.getField("databaseDescription_texts").getValue());
}
doc.addField("allTerms_texts", allTermsSet);
solrDocs.add(doc);
// Create the core for data from this table to be indexed into when data is posted to this table
if (t.isSearchable()) createCoreIfNeeded(t.getTableName(), t.getTableName(), solrServer);
}
@Util
public static void reindex()
throws IOException, SAXException, ParserConfigurationException, SolrServerException {
deleteExistingCores();
indexSchemas();
Cursor<KeyValue<SecureTable>> tablesCursor = SecureTable.findAllCursor(NoSql.em());
int i = 0;
// very important to ignore any error and continue indexing. If an index has gotten
// corrupted (which happens...) this will throw com.alvazan.orm.api.exc.RowNotFoundException
// and will kill the entire reindex.
long docsindexed = 0;
long startTime = System.currentTimeMillis() - 1;
Collection<SolrInputDocument> solrDocs = new ArrayList<SolrInputDocument>();
while (tablesCursor.next()) {
if (++i % 200 == 0) NoSql.em().clear();
KeyValue<SecureTable> kv = tablesCursor.getCurrent();
try {
if (kv.getValue() == null) continue;
} catch (RowNotFoundException rnfe) {
if (log.isInfoEnabled())
log.error(
"got a corrupt index while reindexing, ignoring the error and continuing with indexing of other data.");
// rnfe.printStackTrace();
continue;
}
SecureTable table = kv.getValue();
DboTableMeta meta = table.getTableMeta();
SearchUtils.indexTable(table, meta, solrDocs);
if (table.isSearchable()) {
log.info("found a searchable table " + table.getName() + " indexing it.");
String sql = "select c from " + table.getTableName() + " as c";
Collection<SolrInputDocument> tablesolrDocs = new ArrayList<SolrInputDocument>();
try {
QueryResult result =
NoSql.em().getTypedSession().createQueryCursor(sql, SqlPullProcessor.BATCH_SIZE);
Iterator<List<TypedRow>> cursor = result.getAllViewsIter().iterator();
while (true) {
// I hate this, but cursor.hasNext() can throw an exception which means we need to skip
// over
// that item but continue on with the cursor till it runs out:
List<TypedRow> typedRows = getNext(cursor);
if (typedRows == null) break;
for (TypedRow prow : typedRows) {
SearchPosting.addSolrDataDoc(prow, table, tablesolrDocs);
}
if (tablesolrDocs.size() > REINDEX_BATCH_SIZE) {
docsindexed += solrDocs.size();
System.out.println(
"hit solr doc batch size in a searchable table, "
+ docsindexed
+ " docs so far, "
+ (System.currentTimeMillis() - startTime)
+ " millis elapsed "
+ (docsindexed / ((System.currentTimeMillis() - startTime) / 1000))
+ " docs per sec.");
SearchPosting.saveSolr("reindex", tablesolrDocs, null);
tablesolrDocs = new ArrayList<SolrInputDocument>();
}
}
SearchPosting.saveSolr("reindex", tablesolrDocs, null);
docsindexed += solrDocs.size();
} catch (Exception e) {
System.out.println(
"got an exception while indexing a searchable table with the query (probably a corrupt index in playorm):");
System.out.println(sql);
// e.printStackTrace();
}
}
if (solrDocs.size() > REINDEX_BATCH_SIZE) {
docsindexed += solrDocs.size();
System.out.println(
"hit solr doc batch size in metadata, "
+ docsindexed
+ " docs so far, "
+ (System.currentTimeMillis() - startTime)
+ " millis elapsed "
+ (docsindexed / ((System.currentTimeMillis() - startTime) / 1000))
+ " docs per sec.");
SearchPosting.saveSolr("reindex", solrDocs, "databusmeta");
solrDocs = new ArrayList<SolrInputDocument>();
}
}
if (solrDocs.size() > 0) {
docsindexed += solrDocs.size();
System.out.println(
"hit solr doc batch size during finalization, "
+ docsindexed
+ " docs so far, "
+ (System.currentTimeMillis() - startTime)
+ " millis elapsed "
+ (docsindexed / ((System.currentTimeMillis() - startTime) / 1000))
+ " docs per sec.");
SearchPosting.saveSolr("reindex", solrDocs, "databusmeta");
solrDocs = new ArrayList<SolrInputDocument>();
}
}