/reviewed-on nexedi/erp5!859

Previously, this could only be node_uid.
In future we should find a more generic approach to make this configurable.

Also, add two tests to testInventoryAPI to check queries with selection_domain
for getInventoryList and getMovementHistoryList.

Currently, the domain selection is implictly on node_category

Without this change, "source/a/b" in non-strict mode indexes:
  (document.uid, source.uid, b.uid, 1)
  (document.uid, source.uid, a.uid, 0)
  (document.uid, source.uid, source.uid, 0)
This last line does not contain anything which cannot be found by looking
at the base category uid column, so it is wasting disk (and index) space,
costing performance.
But keep indexing a Base Category document if is it not the base category
for considered relation. It is not clear whether stopping indexation at the
first encountered Base Category document is intentional, or if recursion
should only stop when reaching the base category of considered relation.
With this change, "source/a/b" in non-strict mode indexes:
  (document.uid, source.uid, b.uid, 1)
  (document.uid, source.uid, a.uid, 0)
removing the redundancy.

Verify there is a single line per use.
Test summary lines on per-category level.
Always compare use by uid available on lines, not by fetching a random
document on each aggregate line.

A new column ledger in the stock table and some related keys
(stored in erp5_mysql_innodb_catalog) have also been created
Adds tests for inventory queries based on ledger in testInventoryAPI

Mark the test as expected failure for now.

Update callers which could not be fixed so they disable the exception.
Reviewed-by: Jerôme Perrin <jerome@nexedi.com>

This avoid differences like 07:55:27 != 07:55:26 in tests

Avoid pass next_item_simulation_state to _generateSQLKeywordDict because it will always ignore this parameter and display messages in logs

Add test to parameter with output=1 in getTrackingList method

see http://docs.python.org/2/library/unittest.html#deprecated-aliases

This is not a new feature, this behavior is compatible with the old caching
implementation that was using inventory_stock table.
In other words, this brings backs the backward compatibility in inventory
caching.

(This commit fixes the incomplete tests that was reverted just before.)

Needs more tests.

This reverts commit 74db4d39.

This is not a new feature, this behavior is compatible with the old caching
implementation that was using inventory_stock table.
In other words, this brings backs the backword compatibility in invetory
caching.

We were joining catalog for section, node and resource, mainly in order to
expose _title and _relative_url in brains. We just need to select the
corresponding uids and let brains do another catalog search to get the object
when accessed.
Note that erp5_banking_core is not updated

Create a sql cache for getInventory call
Remove full-inventory optimisation as it now useless
Implement full-inventory feature directly into stock
Make erp5_pdm depends on this optimisation
Install optimisation for all unit tests

Also, movement with 0 price where ignored by omit_asset_increase /
omit_asset_decrease
Signed-off-by: Aurélien Calonne <aurel@nexedi.com>

To migrate, run:
alter table stock add column payment_request_uid bigint unsigned after funding_uid, add index payment_request_uid (payment_request_uid);
alter table inventory_stock add column payment_request_uid bigint unsigned after funding_uid;

To migrate, run:
alter table stock add column funding_uid bigint unsigned after payment_uid, add index funding_uid (funding_uid);
alter table inventory_stock add column funding_uid bigint unsigned after payment_uid;

Mark three tests as expected failure. To filter by category with full inventory document does not work.

This test will make sure that getInventoryList works in the situation which two sections use the same node and one section has full inventory for the node.