DOM manipulation on process plugins

Problem/Motivation

Manipulating HTML during migration is something that migration process plugins now handle via simple text manipulation, using regular expressions. It is more reliable to manipulate HTML using a parser, for example using the PHP DOMDocument and related classes.

Some manipulation examples can be found on #2958281: Allow manipulating html from process plugins. More reasons can be found on this issue.

Proposed resolution

One way to do it is to have a plugin to import the HTML string into an object representing the HTML, then several plugins that manipulate that object, and finally export the object back into a HTML string.

Remaining tasks

Current children tickets:

• #2958281: Allow manipulating html from process plugins: Contains the dom migrate process plugin, which both imports and exports HTML/DOMObject.
• #2958285: Allow replacing based on a xpath expression: Contains an analogue of str_replace, but using a DOMObject and xpath to find the target.
• #2958672: Use migration lookup on text fields: Contains a child of dom_str_replace, that uses migration_lookup process plugin for replacing ids on a DOMObject
• #3042539: Apply styles configured for CKEditor: Contains the dom_apply_styles migrate process plugin, which applies styles configured for the Editor module based on xpath expressions in the plugin configuration.

User interface changes

N.A.

API changes

None for now, unless we want to generalize this and make migrate process plugins that can accept a subset of values, e.g. an interface, and handle incompatibilities.
If that is the case, I guess this would become more a Drupal core ticket than a migrate_plus one; since it is a major change in how process plugins work.
I would suggest to avoid this for now if possible, we can always try this when we have enough experience with the approach.

Data model changes

N.A.