private Integer executeDocInsert(JCas jCas) throws SQLException, BaleenException {
DocumentAnnotation da = getDocumentAnnotation(jCas);
String documentId = ConsumerUtils.getExternalId(da, contentHashAsId);
insertDocStatement.clearParameters();
insertDocStatement.setString(1, documentId);
insertDocStatement.setString(2, da.getDocType());
insertDocStatement.setString(3, da.getSourceUri());
insertDocStatement.setString(4, jCas.getDocumentText());
insertDocStatement.setString(5, jCas.getDocumentLanguage());
insertDocStatement.setTimestamp(6, new Timestamp(da.getTimestamp()));
insertDocStatement.setString(7, da.getDocumentClassification());
insertDocStatement.setArray(
8, createVarcharArray(postgresResource.getConnection(), da.getDocumentCaveats()));
insertDocStatement.setArray(
9, createVarcharArray(postgresResource.getConnection(), da.getDocumentReleasability()));
insertDocStatement.executeUpdate();
Integer docKey = getKey(insertDocStatement);
if (docKey == null) {
throw new BaleenException("No document key returned");
}
return docKey;
}
/** If the tables don't already exist, then create them. */
private void createTables() throws ResourceInitializationException {
try (Statement s = postgresResource.getConnection().createStatement(); ) {
s.execute(
CREATE_TABLE_PREFIX
+ getTableName(DOC_ROOT)
+ " (key serial primary key, externalId character varying, type character varying, source character varying, content character varying, language character varying, processed timestamp, classification character varying, caveats character varying[], releasability character varying[])");
s.execute(
CREATE_TABLE_PREFIX
+ getTableName(DOC_METADATA_ROOT)
+ " (key serial primary key, doc_key integer references "
+ getTableName(DOC_ROOT)
+ "(key), name character varying, value character varying)");
s.execute(
CREATE_TABLE_PREFIX
+ getTableName(ENTITY_ROOT)
+ " (key serial primary key, doc_key integer references "
+ getTableName(DOC_ROOT)
+ "(key), externalId character varying[], type character varying, value character varying[])");
s.execute(
CREATE_TABLE_PREFIX
+ getTableName(ENTITY_GEO_ROOT)
+ " (key serial primary key, entity_key integer references "
+ getTableName(ENTITY_ROOT)
+ "(key), geo geometry(Geometry, 4326))");
postgresResource.getConnection().setAutoCommit(false);
} catch (SQLException e) {
throw new ResourceInitializationException(e);
}
}
@Override
protected void doProcess(JCas jCas) throws AnalysisEngineProcessException {
Connection conn = postgresResource.getConnection();
try {
// Insert document and metadata into database
Integer docKey = executeDocInsert(jCas);
for (Metadata md : JCasUtil.select(jCas, Metadata.class)) {
executeDocMetadataInsert(docKey, md);
}
processEntities(jCas, docKey);
conn.commit();
} catch (SQLException | BaleenException e) {
getMonitor().error("Unable to insert document into Postgres database", e);
if (conn != null) {
try {
conn.rollback();
} catch (SQLException e2) {
getMonitor()
.error(
"Unable to rollback insertion - state of the database may have been left inconsistent",
e2);
}
}
}
}
@Override
public void doInitialize(UimaContext aContext) throws ResourceInitializationException {
checkVersion();
createTables();
try {
insertDocStatement =
postgresResource
.getConnection()
.prepareStatement(
INSERT_INTO_PREFIX
+ getTableName(DOC_ROOT)
+ " (externalId, type, source, content, language, processed, classification, caveats, releasability) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)",
Statement.RETURN_GENERATED_KEYS);
insertDocMetadataStatement =
postgresResource
.getConnection()
.prepareStatement(
INSERT_INTO_PREFIX
+ getTableName(DOC_METADATA_ROOT)
+ " (doc_key, name, value) VALUES (?, ?, ?)");
insertEntityStatement =
postgresResource
.getConnection()
.prepareStatement(
INSERT_INTO_PREFIX
+ getTableName(ENTITY_ROOT)
+ " (doc_key, externalId, type, value) VALUES (?, ?, ?, ?)",
Statement.RETURN_GENERATED_KEYS);
insertEntityGeoStatement =
postgresResource
.getConnection()
.prepareStatement(
INSERT_INTO_PREFIX
+ getTableName(ENTITY_GEO_ROOT)
+ " (entity_key, geo) VALUES (?, ST_GeomFromGeoJSON(?))");
} catch (SQLException e) {
throw new ResourceInitializationException(e);
}
}
private Integer executeEntityInsert(
Integer docKey, Collection<String> values, Collection<String> externalIds, String type)
throws SQLException {
insertEntityStatement.clearParameters();
insertEntityStatement.setInt(1, docKey);
insertEntityStatement.setArray(
2,
postgresResource
.getConnection()
.createArrayOf(VARCHAR, externalIds.toArray(new String[0])));
insertEntityStatement.setString(3, type);
insertEntityStatement.setArray(
4, postgresResource.getConnection().createArrayOf(VARCHAR, values.toArray(new String[0])));
insertEntityStatement.executeUpdate();
Integer entityKey = getKey(insertEntityStatement);
if (entityKey == null) {
getMonitor().error("No entity key returned - Geo insertion, if applicable, will be skipped");
}
return entityKey;
}
/** Check that Postgres has at least version 2 of PostGIS installed */
private void checkVersion() throws ResourceInitializationException {
try (Statement s = postgresResource.getConnection().createStatement(); ) {
ResultSet rs = s.executeQuery("SELECT PostGIS_Lib_Version() AS version");
rs.next();
String version = rs.getString("version");
String[] versionParts = version.split("\\.");
Integer majorVersion = Integer.parseInt(versionParts[0]);
if (majorVersion < 2) {
throw new BaleenException("Unsupported PostGIS Version");
}
} catch (SQLException | NumberFormatException | NullPointerException e) {
getMonitor().error("Unable to retrieve PostGIS version");
throw new ResourceInitializationException(e);
} catch (BaleenException e) {
throw new ResourceInitializationException(e);
}
}