How to programatically dump a subset of records from a table - Forum - OpenEdge Development - Progress Community

How to programatically dump a subset of records from a table


How to programatically dump a subset of records from a table

This question is not answered

What: Looking for a way to dynamically dump a subset of records from a table. The EXPORT statement does not support dynamic buffers, and the tables have too many fields to try putting all the data into a single variable to then EXPORT. Additionally, I can't just EXPORT each BUFFER-FIELD individually because that would add a new line after each field, and it needs to look like a normal .d file (fields separated by spaces, records separated by new lines) in order to load properly.

Why: I want to generate a set of .d files that can be used to create a test database for testing the application against a localhost. Obviously, dumping the entire database is not realistic because it is too large. So, for tables greater than a certain size, I want to only dump a portion of the records (using prodict/dump_d.p works great for whole table dumps).

Any ideas? How would you typically handle the maintenance of a localized test database?

- Jon Hutchens

All Replies
  • Use the "BUFFER <name>:HANDLE" to get handles to static buffers. Then you can add those static buffers to a dynamic(ally prepared) query to fetch the records you need, and use the EXPORT statement to write them to file.

    Alternatively, if performance of the dump isn't much of a concern, start with the code in  and adapt it to your needs.

  • I had come across that article in initial searches but wasn't able to get it working - I'll take a closer look.

    On your solution using static buffers, that would require writing out the name for each table, right? I want to use this for multiple databases with a combined hundreds of tables, so I want to avoid hard coding any direct references to tables (this would also help if we drop or add tables later, for example).

    - Jon Hutchens

  • Static buffers would require writing or generating code, and compiling it (so need the appropriate license). That may be a factor in your decision.
  • Way back Tony Lavinio and Peter van Dam wrote a dynamic export program. And would you believe it, here it is

    You will need to adapt it to your needs and your current OpenEdge version because it was written when Progress 9.1 was out!

  • The DataDigger can do just this for you.

    In addition, the DataDigger can also generate a local database for you that has the same schema as your remote database.
    Right-click on the table browse and choose "clone this database":

    Choose the folder where the database should be created:

    You end up with a local, empty database with the same structure as your production db. 

    You can dump data from your database in various formats. Make your selection of the records and choose the download button:

    Then choose the format you like:

    If you choose XML format, you can simply dump them, switch to your local database and drag the XML file onto the DataDigger window. It will then import the XML file:

    If you have a lot of tables, this may not be the best way to do it, since you have to do it one by one. If the number of tables is limited, I think this is the smoothest solution. 

    The DataDigger is free and open source, and available here

    (disclaimer: I am the author)

  • Thanks Richard! I will take a look and see if I can modernize it a bit ;)

    - Jon Hutchens

  • Patrick, we do have hundreds of tables, so probably not a full solution for us. However, it might be useful as a supplemental tool for when you need to test against a specific set of data.

    - Jon Hutchens

  • I think eventually we will have to create ProDataSets, etc., for all the tables so we end up with meaningful data (parent/child relationships, some of this type of customer, some of that type, etc.). I saw a suggestion elsewhere to use the JSON methods for a TEMP-TABLE for import/export. It sounds like a good idea except for the extra buffering overhead (which might be inevitable anyway if we're going to ProDataSets).

    For now just to get started, I modified the prodict/dump_d.p source to allow just pulling X number of records off the top of the table as an option. That way I don't have to worry about proper formatting or the "PSC filename=table..." stuff at the end of the .d file.

    - Jon Hutchens