This question has been flagged
17 Replies
9610 Views

When doing csv imports, how to force a column mapping to an id of an OpenERP element?

I know it can be done manually using the menus by opening the mapping menu and selecting "OpenERP element / BD Id" instead of "OpenERP element" (which tends to map itself to the name of that entity and not the id).

How can this be automated?

Avatar
Discard

mapping menu????

Author

A menu is by essence not automatable, although there might be solutions for automating browsers.... but anyway now am having a policy of by no means importing CSV documents by browser, doing it programmatically through XML-RPC 'creates' issued using client libraries.

« by opening the mapping menu » .... Where is this menu? I don't see it in Odoo

Author

By « by opening the mapping menu » I was referring the columns selection that appears in OpenERP 7 after selecting a file for importing, possibly after the user has explicited a file encoding, data separator, and quoting. If everything went fine, OpenERP 7 will open a table (I'm sorry now, it is really more a table than a menu) when the user would possibly do manual mapping from CSV columns to OpenERP fields.

Best Answer

provide the technical name inside the csv file for which field you are importing data. give that column as name it will automatically map to name field inside openerp.

Avatar
Discard
Author

Do you call _technical name_ to the OpenERP source code field name, or the PostgreSQL column name? In the later case, is not working for me. Mapping works but maps to _entity name_ instead of the exact PostgreSQL column name!

Name which is given inside the model is the table name it self. so whatever field name are define in spefice table in _columns {} dictionary you need to use that field name as column header to import data.

Author

Ah ja, but the problem comes when... I'm actually doing manual mappings to database id of an entity (which is not a field but an ORM elsething) because I can not use a mapping based on field. We are talking about detecting a type of account, but there are two instance rows for this type of account, if you consider its name ('Capital' type of account), each for a different chart of accounts (in a multi-company environment)! I solve it doing a mapping to the actual row id in the CSV, but that way, when doing the import, I have to manually specify it as a "Entity BD Id" where has been mapped to "Entity", which btw is containing the name.

Author

I finally solved these and all other issues performing imports using XML-RPC against OpenERP (hence committing after each import, I suppose, BTW), instead of using the browser CSV import, that makes a single commit if the import of all rows is successful and hence more expensive in terms of developer time.

Author

Aha, understood. I guess the browser import is actually far more complex than the approach I used. However, at this point of expertise, to me it seems easier not to import using browser but using the 'create' method of the 'osv.Model' class, using a small script that reads a file and creates an entity per line. All the comlexity of calculating column lines is distributed among previous scripts that query PostgreSQL e.g. for row ids retrieving.

Author

'column lines'* i meant 'values for each field in a row'.

Author

John Doe, what you pointed out in your comment, as an answer, might be the most correct answer possible, given the literal question.

Best Answer

Although you have solved your problem, allow me to provide a different solution. You can specify XML ID as a replacement of that field. The column into which this value need to be imported is field/id (e.g. parent_id/id). To obtain the XML ID, first perform export first for that particular model and make sure that you exported the id field.

Avatar
Discard
Best Answer

If you want the ids to map existing objects you can generate one file ir.model.data.csv and associate the xml_ids that you are going to import to existing objects (model, res_id).

For example if your database has already the stock locations (110, 111, 112) and you want to reference them :

After importing the new xml_ids you can import your csv with those xml_id references.

Avatar
Discard
Best Answer

There is some information about importing CVS files in odoo's developers documentation.

Avatar
Discard
Author

That page prompts "country_id:id" to force mapping to id. John Doe prompted to use "parent_id/id". Do you know which of either is older or which of either's deprecation is more probable?

Author Best Answer

Mapping to ids instead of names is also easy when importing using 'osv.Model' 'create(cr, uid, values, context=None)' method through XML-RPC or JSON-RPC.

Avatar
Discard