===== Mapping Service =====

This service integrates in the feed import pipeline. The purpose is to transform customer supplied data into picalike supplied data. Currently we consider:


  * categories (shop categories are turned into our internal category structure)
  * gender
  * brand
  * attributes

git: https://git.picalike.corpex-kunden.de/v5/mapping-service


===== High level Description =====

The service iterates over the meta_db_collection and performs the mapping for each product. The results will be stored in the document for the product under the key “mapping_service”.


===== Data Types =====


  * category: string [exactly one category per product]
  * brand: string [exactly one brand per product]
  * gender: list [multiple genders per product are possible]
  * attributes: list [multiple attributes per product are possible]


===== Operations =====


  * http://pci01.picalike.corpex-kunden.de:8053/ (backend)
  * http://pci01.picalike.corpex-kunden.de:8052/ (frontend: mapping-tool)


=== Deploy this service ===

bash ice_tea_build.sh && bash ice_tea_push.sh && bash ice_tea_deploy.sh

if shops are currently running, they will be restarted after deployment, so you should make sure, that no big shops are running, otherwise you are potentially completely mapping them twice


=== Start this service ===

docker restart\\
/home/picalike/docker_bin/mapping-service/ice_tea_run.sh


=== Use this service ===

Shops are usually triggered through the shop-conveyor-belt. You need to keep in mind that both the shop-conveyor-belt as well as the mapping-service keep track of an internal state. If the shop is busy in the mapping-service the start will fail silently.

{{/dokuwiki/lib/images/smileys/icon_exclaim.gif|:!:}} The state for the mapping-service is kept in “osa-mongo.mapping_service.status” (search for busy true)… if the shop is not really running, you can just delete the entry from the status-collection or set busy: false


===== Super Special Knowledge =====


  * the service writes some statistics in the mongo collection: mapping_service.status
  * the service should at some point take over as the brand mapping service (currently in solr_updater), but the collections were never moved from osa_db.brand_* to mapping_service.brand_* . {{/dokuwiki/lib/images/smileys/icon_exclaim.gif|:!:}} although the service is configured to use the currently correct collections in osa_db.brand_*, it is not recommended to switch to this service (not tested)


===== Special Solutions =====

To rename a category, use the script ''%%one_time_mapping_changes.py%%'' from the git repository. It is not fully parametrized, so it needs some adaption before it can be used. But the rename is just a series of update_one in two(!) mongo collections.


===== (Posible) future functionalities =====

A visual mapping service skeleton is implemented. Check the [[https://git.picalike.corpex-kunden.de/incubator/visual-mapping|repository]] and the [[http://dokuwiki.picalike.corpex-kunden.de/visual-mapping|wiki page]]. For now, it only adds color information, but it's not deployed anywhere. It's a work in progress that has been put on hold together with OSA and it's referenced here for knowledge purposes only.


===== Dependencies =====

The **category** mapper relies on the Mongo Collection osa_db.category_relation

The Mongo Collections involved are:


  * mapping_service.picalike_categories
  * mapping_service.shop_categories
  * mapping_service.category_mapping

The **brand** mapper depends on the following collections:


  * osa_db.brands_picalike
  * osa_db.brands_shop
  * osa_db.brands_mapping

The **gender** mapping uses:


  * osa_db.gender_relation
  * osa_db.gender_category_text_relation