Add an iterate process plugin

This is pretty interesting to me. I'm interested in seeing better examples though. The first example:

+++ b/src/Plugin/migrate/process/Iterate.php
@@ -0,0 +1,157 @@
+ * @code
+ * process:
+ *   field_dates:
+ *     plugin: iterate
+ *     source: field_timestamps
+ *     process:
+ *       plugin: format_date
+ *       from_format: 'U'
+ *       to_format: 'Y-m-d\TH:i:s'
+ * @endcode

As far as I can tell, this is identical to the simpler:

field_dates:
  plugin: format_date
  source: field_timestamps
  from_format: 'U'
  to_format: 'Y-m-d\TH:i:s'

since format_date does not "handle_multiples".

The example using extract is more interesting because extract does "handle_multiples". Still, I think that gets us the same thing as this:

field_node_references:
  -
    plugin: explode
    delimiter: '/'
    source: field_links
  -
    plugin: sub_process
    process:
      '0': '2'
  -
    plugin: migration_lookup
    migration: node

I think more compelling examples would be good. This plugin strikes me as something I would potentially reach for, but I don't think I'd use it for either of these cases.

I just put in a new issue with an alternative approach that manages to pass the complexity onto sub_process. #3314502: Add wrapper process plugin to wrap/unwrap values in arrays

This works like a charm! Thank you!

I think that for the specific use case of having an array with values this is the simpler and better solution than the approach of #3314502: Add wrapper process plugin to wrap/unwrap values in arrays.

Sorry, probably that needs a better review ;)

I think you can get the same effect using the build_array process plugin from #3440904: Process plugin: build an array from source, destination, pipeline.

Instead of this example from the doc block for iterate

process:
  field_dates:
    plugin: iterate
    source: field_timestamps
    process:
      plugin: format_date
      from_format: 'U'
      to_format: 'Y-m-d\TH:i:s'

you can use build_array to prepare the array of arrays that sub_process expects:

process:
  field_dates:
    - plugin: build_array
      source: field_timestamps
      template:
        - 'pipeline:'
    - plugin: sub_process
      process:
        plugin: format_date
        from_format: 'U'
        to_format: 'Y-m-d\TH:i:s'
        source: '0'

As @danflanagan8 pointed out in #3, it is much simpler in this case to let the Migrate API loop through the array for you, but I think this example shows that any time you want to use iterate, you can get the same effect with build_array and sub_process.

Many simple uses of sub_process can also be handled by build_array. The second example from #3 becomes

field_node_references:
  -
    plugin: explode
    delimiter: '/'
    source: field_links
  -
    plugin: build_array
    template:
      - pipeline:2
  -
    plugin: migration_lookup
    migration: node

The build_array plugin uses array_walk_recursive(), which is a lot simpler than sub_process and iterate, both of which create a separate process pipeline. So I prefer to use build_array when it can replace one of the others.

I'll take comment #7 as a good way to keep things a bit more DRY. If you still feel strongly about the addition of this feature, feel free to re-open and propose some alternatives.