Skip to content

Latest commit

 

History

History
55 lines (43 loc) · 2.72 KB

README.md

File metadata and controls

55 lines (43 loc) · 2.72 KB

BIG DATA PLATFORM - Schema Generator

Generate a schema dump of all entities inside a given path and write it into a file.

Short version

To generate a new schema, execute the shell script bdp-sqldump.sh.

Detailed version

This chapter is about the Java tool itself, the shell script above simplifies these steps.

The output patterns, that define how identifiers are written, are defined inside the implicit naming strategy class. For example, any unique constraint starts with uc_ and contains the current table and used columns, all lower case with underscores. For more details see code comments.

Compilation:

mvn clean package

Example execution:

Configure the PostgreSQL connection via environmental variables:

  • ODH_SG_SERVER - database server name (optional, default = localhost)
  • ODH_SG_DBNAME - database name
  • ODH_SG_USER - database user (optional, default = postgres)
  • ODH_SG_PASSWD - database password NB: This configuration is not needed, you can also leave it unset, but it will generate a warning. See Known issues for details.

Dump BDP/dal entity classes:

    java -cp '../dto/target/dto-2.0.0.jar:../dal/target/dal-2.0.0.jar:target/schemagenerator-1.0.0.jar' \
         -Dlog4j.configuration=src/main/resources/log4j.properties \
         com.opendatahub.timeseries.bdp.tools.SchemaGenerator \
         com.opendatahub.timeseries.bdp.writer.dal \
         com.opendatahub.timeseries.bdp.writer.dal.util.SchemaGeneratorImplicitNamingStrategy \
         /tmp/schema_dump1.sql

Hereby, ../dto/target/dto-2.0.0.jar:../dal/target/dal-2.0.0.jar are class path entries that contain the @Entity annotated classes, that should be dumped as SQL DDL commands. target/schemagenerator-1.0.0.jar is the library that contains the main method. -Dlog4j.configuration=src/main/resources/log4j.properties is to override log levels, which may be too verbose. com.opendatahub.timeseries.bdp.tools.SchemaGenerator is the class that contains the main method to be executed. The script scans com.opendatahub.timeseries.bdp.writer.dal for entity classes, and dumps the result configured by com.opendatahub.timeseries.bdp.writer.dal.util.SchemaGeneratorImplicitNamingStrategy into /tmp/schema_dump1.sql.

Known issues

You can provide PostgreSQL connection information. However, the script also works if they are wrong. That is, because of how the registry builder works. The registry builder needs a data source, however, also if the connection fails it succeeds in building a DDL SQL script. The drawback hereby is, that it generates a warning on stderr, which cannot be avoided with a try-catch...