Source:
https://github.com/akeneo/pim-api-docs/blob/master/content/rest-api/update.md

Markdown Parser to try
https://github.com/thephpleague/commonmark/blob/main/src/Parser/MarkdownParser.php

Given there are 2 attributes:

1. description (text)
2. ingredients (text)

When transform executed
Then ingredients from the description identified by a regex are copied to the ingredients field.

• find better API for extractor construction
• implement Query to unify access to conditions
• implement common methods - work with Query

• Inroduce an interface like ParentAwareReppsitory or SubitemRepository or ComplexRepository or ParentChild or
  with the method getParentFields(): array like family_cdoe for family options

• replace PersistRepository::$groupFieldNames with that

• Consider renaming x_code to x because it is how already used in Akeneo, see attribute field in the option API

• [] RepositoryFactory::createWriteRepository - wrap API automatically base on its supported interfaces

The goal of the commans(s) is to generate empty profiles

profile:etl:create
profile:connection:create

https://api.akeneo.com/api-reference.html#Referenceentitymediafile

Problem

Duplicating ref.entity colours must happen after duplicating main colours otherwise it fails on finding main colours.

Possible solution

Try to order entities by their deps first which might be a hard task.

Workaround

Duplicate entities one-by-one: first main colours, then colours.

See: https://docs.akeneo.com/5.0/manipulate_pim_data/rule/general_information_on_rule_format.html#remove

To achieve it, it needs to connect to the Categories API and fetch all children.

1. Clean up transformation output: print one-by-one
2. Output stats:

• transformation errors
• load errors

See Akeneo Rules add.
This action adds new items to arrays, like categories or multi-selects.

Create a doc example for add attribute to a family, link it to a main page if needed.

Save stats to yaml and load errors to CSV

• ask what is the appropriate license in similar cases
• add a file LICENSE

Actions will be more powerful if we add additional condition

actions:
    -
        type: set
        field: name
        locale: en_GB
        scope: null
        expression: 'trim(value("name", null, "en_GB"))'
        apply_if: 'hasValue("name", null, "en_GB"))'

apply_if is an expression.
Action is applied only if apply_if is evaluated as true.

• consider examples
• test how to substitute the clean action from Akeneo rules, add to the documentation page

Add description of setting/adding associations from here
https://docs.akeneo.com/5.0/manipulate_pim_data/rule/general_information_on_rule_format.html#enrichment-rule-structure

1. Identify listable/upsertable
2. create new extractors/loaders

Implement comparison in Resource::set to avoid diff if nothing changed

test for windows

Assumption:
again by creation the main one didn't exist and theey had to be ordered first

Idea
new type of writer - collect all first without batching, order and save in batches (using same old?).

Try with clean data w/o ref. entites

Problem

At the moment it returns an array that contain a data patch.

Using arrays - not good approach. Value array, patch - not clear terminology.

Ideas

Option 1. Return patch resource

public function  execute(Resource $resource): ?Resource

As same concept as now, but returns an object.

Pro:

1. Similar approach.
2. Can be used in loader as Resource - so it allows to get rid of arrays also there.

Con:

1. Patch - the concept is still not 100% clear.

Option 2. Return void, all changes inside the $resource.

public function  execute(Resource $resource): void

Pro:

1. Resource changed by ref.
2. CopyAll (migration) doesn't need any actions - it is just same object as in the beginning, so patch will be an empty array.

Con:

1. It needs another approach for creating patches for loader - perhaps as array_diff between initial and result of transformations.

How to make a decision

1. Find examples

Improve transform by outputting the before/after info as a list.

• Extract common part of all loaders
• Return Response/LoadResult list directly from implementations
• Implement ResourceCollection to replace loader buffer and add meaningful methods like getResourceType(), get('reference_entity_code')

For add and remove actions it is important to see what is an actual patch that is send to PIM

Akeneo users are familiar with the concepts of the Rules Engine.
In order to make the tool more understandable for them, it is better to use same terms and same config.

Additionally, change --etl-profile to --rules-profile (or profile or rules)?

Rename main command from transform to "run-rules" or "run"?

The command should show a question In interactive mode:

You're connecting to http://akeneo-host to transform data. Continue?
[y,n]

where http://akeneo-host is a destination connection host.

See details
https://api.akeneo.com/php-client/exception.html#

E.g.

• No user found for the given username and password (see https://api.akeneo.com/php-client/exception.html#unprocessable-entity-exception)
  must fail immediately

Events

• onProgress (incl. BeforeAfter)
• onBeforeTransformation - to set current resource to Holder
• onLoadErrors