Splitting multi-paragraph StreamField blocks into individual blocks

models.py

from __future__ import absolute_import, unicode_literals

from bs4 import BeautifulSoup, Tag
from django.db import models

from wagtail.wagtailadmin.edit_handlers import StreamFieldPanel
from wagtail.wagtailcore import blocks
from wagtail.wagtailcore.fields import StreamField
from wagtail.wagtailcore.models import Page
from wagtail.wagtailcore.rich_text import RichText


class HomePage(Page):
    body = StreamField([
        ('heading', blocks.CharBlock()),
        ('paragraph', blocks.RichTextBlock()),
        ('multi_paragraph_import', blocks.RichTextBlock()),
    ])

    content_panels = Page.content_panels + [
        StreamFieldPanel('body'),
    ]

    def clean(self):
        super().clean()

        is_replacing_body = False  # becomes true if there are multi_paragraph_import blocks to rewrite
        new_body = []  # a list of (block_type, value, id) tuples

        for block in self.body:
            if block.block_type == 'multi_paragraph_import':
                is_replacing_body = True
                soup = BeautifulSoup(block.value.source, 'html5lib')
                # create a paragraph block for each top-level element in the rich text
                for element in soup.body.children:

                    # ignore whitespace between elements
                    if isinstance(element, str) and not element.strip():
                        continue

                    # ignore empty paragraphs
                    if isinstance(element, Tag) and element.name == 'p' and not element.text:
                        continue

                    new_body.append(
                        ('paragraph', RichText(str(element)), None)
                    )
            else:
                # keep all other block types intact
                new_body.append((block.block_type, block.value, block.id))

        if is_replacing_body:
            # need to construct a StreamValue to write back to body; writing the list back
            # will fail because StreamField only recognises (block_type, value) tuples,
            # not (block_type, value, id)
            body_block = self._meta.get_field('body').stream_block
            self.body = blocks.StreamValue(body_block, new_body)