Add a substr options to DedupeBase process plugin

+++ b/core/modules/migrate/lib/Drupal/migrate/Plugin/migrate/process/MachineName.php
@@ -37,7 +38,20 @@ public function transform($value, MigrateExecutable $migrate_executable, Row $ro
+      $new_value = substr($new_value, $start, $length);

Lets use Unicode::substr()

We also need to discuss how the truncating will work with dedupe. This process plugin is nearly always used with dedupe and that currently has no context of the maximum length of the machine name.

We discussed this on IRC and decided that the best approach would be to:

1. Add a default length of 32 to the machine name plugin
2. Add a length configuration option to DedupeBase
3. Update any migrations using the machine name and dedupe plugins to specify a length of 32 to dedupe.
4. Update DedupeBase so that it can handle creating a non duplicate value at the correct length.

Thanks for working on this.

1. +++ b/core/modules/migrate/lib/Drupal/migrate/Plugin/migrate/process/DedupeBase.php
   @@ -31,6 +33,16 @@ public function transform($value, MigrateExecutable $migrate_executable, Row $ro
   +      throw new MigrateException('Input character length should be an integer.');
   

   Should we expand on this to say that "null" or no value will capture the entire string?

2. +++ b/core/modules/migrate/tests/Drupal/migrate/Tests/process/DedupeEntityTest.php
   @@ -82,10 +83,44 @@ public function testDedupe($count, $postfix = '') {
   +    if (isset($start)) {
   +      $configuration['start'] = $start;
   +    }
   +    if (isset($length)) {
   +      $configuration['length'] = $length;
   +    }
   

   I can't see where we use configuration where we couldn't simply use $start and $length?

Other than above, It's looking good so far.

@bdone, ignore my last comment I missed this line in dreditor as it was greyed out.

plugin = new DedupeEntity($configuration, 'dedupe_entity', array(), $this->getMigration(), $this->entityQueryFactory);

The error messages ready better, thanks.

Looks good to me - the one open issue here is the idea of defaulting machine names to 32 characters, but I also don't see how that would work without introducing complection between the machine_name and dedupe_entity plugins.

Committed df301e7 and pushed to 8.x. Thanks!