Make test - pytest pg_tests or mysql_tests

[reywang18]

I have Ubuntu 18.10 version, python 3.7 .
Can we have a mysql / PGsql back end, sample test files ?

Is there IRC channel for this ?

[mouadhkaabachi]

I added postgres dump for sales demo cube, you can use it
https://github.com/abilian/olapy/blob/master/cubes_templates/postgres_sales_db_dump.sql

[mouadhkaabachi]

to run db tests by your own (in OlaPy Not OlaPy-web repo) , you need to export string connection before running make test, something like :
export SQLALCHEMY_DATABASE_URI=postgresql://postgres@localhost/olapy_test
you can check tests in travisCI
https://travis-ci.org/abilian/olapy/builds/475224654

[reywang18]

Finally, I got py_test working.
Few things, I noticed. In tests, we need DB owner privileges/drop tables etc.
I prefer to leave tables around and run test with lowest DB privileges if possible.

[mouadhkaabachi]

thank you reywang18 for your feedback, we will take consideration your suggestion in the next olapy-web update/release ;)

[reywang18]

Just noticed about this,
SET
CREATE EXTENSION
ERROR: must be owner of extension plpgsql
SET

And the test suite is heavy at pytest, can I have a comment on it?

assert result["result"]["supply_time"][0] == 24

Can you refer it back to which table and aggregate (full sql here)?
How we get back 24 from DB ?

[mouadhkaabachi]

assert result["result"]["supply_time"][0] == 24 is used with a custom cube with some configuration, which is the foodmart cube,
By default olapy construct the cube if it follows certain pattern (star schema model with 'Facts' name for facts table, and the same key column name betwe facts table and another table 'dimension' {exple: Geography.Continent => Facts.Continent} ) but if you have tables (dataset) that does't follow this pattern,
example: facts table name is Foo and there is no common column name betwen dimension and facts table {Geography.MyContinent => Foo.Cont} you need to specify those relations in a config object like the cube_config object in
https://github.com/abilian/olapy/blob/3d305ef79ebf669cdc7323eedc76a6ee1e9605df/tests/conftest.py#L19

so assert result["result"]["supply_time"][0] == 24 correspond to a custom test cube where you can find column supply_time in the facts table
the custom cube is foodmart
I attached the dump file for this dataset
I hope it's clear
dump_foodart.sql.tar.gz

[mouadhkaabachi]

here is the MDX (Sql) query to execute:
https://github.com/abilian/olapy/blob/3d305ef79ebf669cdc7323eedc76a6ee1e9605df/tests/queries.py#L257

[reywang18]

Hi, Thanks for the info. Here is my issue I guess, I am more SQL guy not MDX savvy yet. So the syntax is a bit weird to me. """SELECT FROM [main] --- this is not a DB table ? WHERE ([Measures].[supply_time]) So I think if it has some comments to it, new guy like me; might easy to follow .

…

On Thu, Jan 24, 2019 at 6:34 AM Mouadh Kaabachi ***@***.***> wrote: here is the MDX (Sql) query to execute: https://github.com/abilian/olapy/blob/3d305ef79ebf669cdc7323eedc76a6ee1e9605df/tests/queries.py#L257 — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#6 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AEg7oqia02qmf2Z7zuUbmp7K8MsDPi3Oks5vGZougaJpZM4ZnGTk> .

[mouadhkaabachi]

mdx -> extension of sql used for olap engines
in reality FROM [main] does not matter, the important thing is when we call load_cube it is where we will specify the cube (database with many tables not one table only) to launch, and in the mdx request you can put 'FROM [anything]'
https://github.com/abilian/olapy/blob/3d305ef79ebf669cdc7323eedc76a6ee1e9605df/tests/conftest.py#L91
on the other hand I put 'main' in load_cube because I use by default sqlite (in memory) which is very fast for testing and sqlite (in memory) create a db called 'main'

[reywang18]

mdx -> extension of sql used for olap engines Any command line tool from Linux , or we need to use Python and import function to run mdx SQL here...

…

On Thu, Jan 24, 2019 at 12:17 PM Mouadh Kaabachi ***@***.***> wrote: mdx -> extension of sql used for olap engines in reality FROM [main] does not matter, the important thing is when we call load_cube it is where we will specify the cube (database with many tables not one table only) to launch, and in the mdx request you can put 'FROM [anything]' https://github.com/abilian/olapy/blob/3d305ef79ebf669cdc7323eedc76a6ee1e9605df/tests/conftest.py#L91 on the other hand I put 'main' in load_cube because I use by default sqlite (in memory) which is very fast for testing and sqlite (in memory) create a db called 'main' — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#6 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AEg7olTW533wTyyD5CCCQBwSrY789_foks5vGeqTgaJpZM4ZnGTk> .

[mouadhkaabachi]

executing mdx from linux cli, I never tried this, I did not have the need to do it
for olapy you need to import MdxEngine, load the cube and execute the mdx query
like this demo
https://github.com/abilian/olapy/blob/master/demos/olapy-core/demo.py
But if you are looking other tools to execute mdx query you can try xmondrian (based on mondrian) which a good tool (java based)
https://rpbouman.blogspot.com/2016/03/need-mondrian-war-checkout-xmondrian.html
or you can test iccube which i personally used so many times to get inspired