[
https://issues.apache.org/jira/browse/NIFI-6061?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17201413#comment-17201413
]
ZhangCheng commented on NIFI-6061:
----------------------------------
[~mattyb149]I didn't find problems about CLOB, and I found that the record
field type has been converted to Array[Number] not Array[Byte], so I only fix
the BLOB problem in new class AbstractDatabaseAdapter.java, if there are other
problems about record and sql type, we can add some code in switch-case below:
{code:java}
void psSetValue(PreparedStatement ps, int index, Object value, int sqlType, int
recordSqlType) throws SQLException, IOException {
if (null == value) {
ps.setNull(index, sqlType);
} else {
switch (sqlType) {
case Types.BLOB:
//resolve BLOB type for record
if (Types.ARRAY == recordSqlType) {
Object[] objects = (Object[]) value;
byte[] byteArray = new byte[objects.length];
for (int k = 0; k < objects.length; k++) {
Object o = objects[k];
if (o instanceof Number) {
byteArray[k] = ((Number) o).byteValue();
}
}
try (InputStream inputStream = new
ByteArrayInputStream(byteArray)) {
ps.setBlob(index, inputStream);
} catch (IOException e) {
throw new IOException("Unable to parse binary data
" + value.toString(), e.getCause());
}
} else {
try (InputStream inputStream = new
ByteArrayInputStream(value.toString().getBytes())) {
ps.setBlob(index, inputStream);
} catch (IOException e) {
throw new IOException("Unable to parse binary data
" + value.toString(), e.getCause());
}
}
break;
//add other Types here to resolve data
default:
ps.setObject(index, value, sqlType);
}
}
}
{code}
> PutDatabaseRecord does not properly handle BLOB/CLOB fields
> -----------------------------------------------------------
>
> Key: NIFI-6061
> URL: https://issues.apache.org/jira/browse/NIFI-6061
> Project: Apache NiFi
> Issue Type: Bug
> Components: Extensions
> Reporter: Matt Burgess
> Assignee: ZhangCheng
> Priority: Major
> Time Spent: 10m
> Remaining Estimate: 0h
>
> BLOB/CLOB fields in NiFi's Record API are returned from the record as
> Object[Byte], but when PutDatabaseRecord tries to insert Object[] via
> setObject(), the following error occurs:
> 2019-02-20 15:11:16,216 WARN [Timer-Driven Process Thread-10]
> o.a.n.p.standard.PutDatabaseRecord
> PutDatabaseRecord[id=0c84b9de-0169-1000-0164-3fbad7a17664] Failed to process
> StandardFlowFileRecord[uuid=d739f432-0871-41bb-a0c9-d6ceeac68a6d,claim=StandardContentClaim
> [resourceClaim=StandardResourceClaim[id=1550690599998-1, container=default,
> section=1], offset=1728,
> length=251],offset=0,name=d739f432-0871-41bb-a0c9-d6ceeac68a6d,size=251] due
> to org.postgresql.util.PSQLException: Can't infer the SQL type to use for an
> instance of [Ljava.lang.Object;. Use setObject() with an explicit Types value
> to specify the type to use.:
> Somewhere in the value conversion/representation, PutDatabaseRecord would
> likely need to create a java.sql.Blob object and transfer the bytes into it.
> One issue I see is that the record field type has been converted to
> Array[Byte], so the information that the field is a BLOB is lost by that
> point. If this requires DB-specific code, we'd likely need to add a Database
> Adapter property and delegate out to the various DB adapters.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
