Pushing data into OpenERP

The PushSQL module allows external apps to export data to OpenERP properly (through the ORM) just by using INSERT SQL statements.

The motivation for this came from a project where an ETL service was responsible for handling synchronization of data (such as departments, customers and contracts) across several databases and applications.

The ETL tool didn't have a simple way to make XML-RPC calls, and writing directly to OpenERP's tables is not a good solution.

This was solved using a relatively simple "hybrid" solution. The PushSQL module provides a database table for external apps to write data to imported to OpenERP, using plain SQL INSERT statements. A scheduled job regularly looks for new rows in this table and processes them, making the intended changes to OpenERP data.

The table , import_pushsql, has two mandatory columns: model and data. The first is the target model, such as res.partner. The second contains the data to import, using the format for CSV files, but using TAB instead of comma as the column separator.

For example, using PSQL:
INSERT INTO import_pushsql (model, data) VALUES ('res.partner', E'id\tname\tref\nres_partner_XXX\tDemo Customer\tXXX') 

Or, if using SQL Server:
INSERT INTO import_pushsql (model, data) VALUES ('res.partner',
'id' +char(9)+ 'name' +char(9)+ 'ref' +char(10)+ 'res_partner_XXX' +char(9)+ 'Demo Customer' +char(9)+ 'XXX'
)

The scheduled action uses this to run a standard import_data ORM method,  writes the result into the log column, and changes the state to 'done' if successful, or 'cancel' if an error was found.

Compared to the "pull" approach used by the import_odbc module, this "push" approach can provide a few advantages:

•  No need to install other database clients and middleware in OpenERP's server.
• It uses less resources on OpenERP server, because there is no need to connect to other servers.
• Event driven data sync, where an action in another system immediately triggers a write on OpenERP that can be processed within a few minutes.

There is also an obvious disadvantage: the process exporting the data does not have an immediate visibility on the outcome of the OpenERP transaction . There are ways to provide this feedback, but it's not immediate.

Overall, the PushSQL module can provide a very simple and straighforward way for business apps to safely send data to OpenERP, in scenarios where using the API is not an option or would be too cumbersome.

The import_pushsql can be downloaded either from OpenERP Apps or from Launchpad using: bzr branch lp:reis-openerp-addons

Import ODBC module steps up

The import_odbc module for OpenERP was recently improved and moved  to the OpenObject Extension community project. the original module is now split in two: base_external_dbsource and import_odbc.

The import_odbc code was refactored, but keeps the same funcionality as before. Documentation was improved and some examples are provided out of the box.

The base_external_dbsource was created to isolate the logic for connecting to external databases and executing SQL queries. It's used by import_odbc, but it also enables other modules to fetch foreign data on the fly.

Additionally, the connection types supported were extended. Support for ODBC and Oracle native connections is kept, and support for other native connections was added:

• MS SQL Server
• MySQL
• PostgreSQL
• SQLite

With this evolution, it would be more fair to name the module as "import_sql" instead of the "import_odbc", but the name was kept unchanged to avoid confusing current users.

You can get the latest version of these modules from the OpenObject Extension trunk branch:

bzr branch lp:openobject-extension

Share you experiences and feature wishlists with us!

Using codes to search and enter data in OpenERP forms

Using codes to refer to entities, such as customers or employees, is a common practice, and widely used in ERPs. To users familiar with this method it's natural to have the code displayed alongside with the description, and to enter data in a form field by directly typing a code.

This feature can be easilly added to OpenERP using the refcodes family of modules, available in Launchpad. For example, in the HR module an employee selection list might look like this:

Have fun.

Trigger-like Automated Actions

OpenERP's Automated Actions is a module usually associated with the CRM modules, allowing to automate sales process steps, without the need to write Python code.

Each automated action rule operates on an OpenERP model, any model. It's  behaviour is defined through three tabs:

• Conditions tell when the rule is fired.
• Actions tell what operations should be done, like setting a specific user as Responsible person, or running a Server Action.
• E-mail actions allows to define e-mail messages to be sent .

Conditions can be activated at scheduled moments. This is the sole purpose of the "Conditions on Time" section. The scheduler action evaluates these conditions to decide what rules to trigger. This allows you to configure actions like "send an e-mail reminder two days after last action date".


The remaining conditions are evaluated immediately when a record is saved. This allows you to configure actions like "send an e-mail when state is changed to In Progress".


While you get the feeling of having a lot of potential to easily customize a customer's very specific business rules, actually the module reveals to fall short on the expectations.
For instance, it's not possible to create conditions for:

• When a new Issue is created, sending an e-mail informing the new Issue;
• When the Responsible is changed, sending an e-mail informing the new responsible person.

Also, the e-mail actions definition is very confusing and limited. The e-mail Subject line is not configurable, and the message body can only use a small list of keyword to include data from the document.

The reis_base_action_rule_triggers module overcomes these limitations and unlocks a lot of the power of the standard module. This is done through two simple additions:

• Conditions can be defined by an "Evaluated expression", that can work on record's old and new values.
• Actions can send an e-mail using v6.1 Email templates, a lot more powerful than the "e-mail actions" options, making them obsolete.

In the condition expressions some variables are available (inspiration came from Oracle's database triggers):

• inserting, creating: boolean indicating it's a new record is being created;
• updating, writing: boolean indicating it's an existing record being changed;
• new: a dictionary with all the values after the record is changed;
• old: a dictionary with all the values before the record was changed;
• changed: a dictionary with only the values that actually changed, even if rewritten (empty if not updating);
• obj: browseable record object, allowing dot notation, with the new/changed record.

Here are some real examples used in a production environment for the Project Issue module:


New issue is created, or is reactivated:
inserting or changed.get('state') == 'draft'


Issue is closed or cancelled:
changed.get('state') in ('done', 'cancel')


Issue's Responsible changed:
old.get('user_id') and new.get('user_id')
and old.get('user_id') != new.get('user_id') 
and not new.get('date_open')  #date_open is written only by Open issue

I hope this turn out useful to more people, and would love to hear about your experience with this module.

How to: allow Project Users to record Time on Tasks for anyone

If you have hr_timesheet installed, users also in the Employee group will only be able to record time on tasks for themselves. If you need Project Users or Managers to be able to record time for other persons, just add the following Record Rule: