This project has moved to be a sub-project of Framework Libraries and is located here
Pull requests against this project have been disabled. Please contact one of the project owners for emergency bug fixes on this version
A simple drop in library for storing binary files in Postgres. This library assumes that it is running inside a JEE container with a datasource configured to point at Postgres
<dependency>
<groupId>uk.gov.justice.services</groupId>
<artifactId>file-service-persistence</artifactId>
<version>1.5.0</version>
</dependency>
final JsonObject metadata = ...
final InputStream fileContentStream = ...
final UUID fileId = fileStorer.store(metadata, fileContentStream);
final Optional<FileReference> fileReference = fileRetriever.retrieve(fileId);
if(fileReference.isPresent()) {
// always remember to close the input streams
// as this will also close the database Connection
try(final InputStream inputStream = fileReference.get().getContentStream()) {
// do stuff...
}
}
There needs to be a Postgres DataSource configured in your container with the JNDI name
java:/app/fileservice/DS.fileservice
There is a liquibase script configured to run against a local database. This database should be created as follows:
database name: fileservice
username: fileservice
password: fileservice
port: 5432
To run liquibase there is a bash script in the root folder which will create the tables for you. To run:
./runLiquibase.sh
As of writing, files are no longer deleted when calling 'delete(...)' but instead are marked as deleted (there is a new column in the content table 'deleted' which would be set to true). There is no effect on retrieval of files: you can still get a 'deleted' file, just that FileReference.isDeleted() would now return true.
There are few integration tests in the file-service-it module. To allow for the integration tests to be run as Continuous Integration these point to an in memory instance of H2. However, as we use Postgres specific features (namely JsonB column type) we override the json insert for the tests. There is one integration test 'ContentJdbcRepositoryPostgresIT' which points at a local instance of Postgres and will fail if none is found. This test is marked as @Ignored by default. To test against a Postgres instance just comment out the @Ignored annotation.
- This implementation is currently Postgres specific: to take advantage of Postgres' JsonB data type.
- The integration tests use H2 as an in memory database.
- This implementation currently stores the binary file data in Postgres using it's bytea data type. Although bytea will allow storage of up to 1 Gb of data, it isn't recommended to store anything larger than 100 Mb as larger files will unreasonably slow down the database.
- There is an issue with returning the InputStream of the binary content stored in the database. Although closing the SQL Connection before the InputStream is read and closed doesn't cause any issues by itself, using a pooled Connection can cause problems. If the Connection is given to a new request then the InputStream would be null and the previous request would fail. For this reason, we wrap the InputStream in our own InputStream that contains a reference to the database Connection used to read the file. When close() is called on the InputStream then the corresponding SQL Connection is also closed. For this reason it is vitally important to remember to close the InputStream in FileReference after use.