Adding an advanced table block to Wagtail

_advanced-table.scss

// These should match and implement CSS values as per the colour_choices in helpers.py
@import '../abstract/colors';

.advanced-table {
    &__wrapper {
        table {
            background-color: $white;
        }
    }
    &__cell {
        color: $white;
        &--red {
            background-color: $red !important;
        }
        &--orange {
            background-color: $orange !important;
            color: $black;
        }
        &--yellow {
            background-color: $yellow !important;
            color: $black;
        }
        &--green {
            background-color: $green !important;
            color: $black;
        }
        &--aubergine {
            background-color: $aubergine !important;
        }
        &--violet {
            background-color: $violet !important;
        }
        &--purple {
            background-color: $purple !important;
        }
        &--blue {
            background-color: $blue !important;
        }
        &--turquoise {
            background-color: $turquoise !important;
        }
    }
}

.mce-item-table {
    border-collapse: collapse;
    width: 100%;
    border: 1px solid !important;
    th, td {
        border: 1px solid !important;
    }
}

_main.scss

// ...
@import 'advanced-table';

admin.scss

@import 'advanced-table';

advanced_table.html

{% for item in nodes %}
    <div class="advanced-table__wrapper">
        {{ item }}
    </div>
{% endfor %}

base.py

# Add to installed apps

# Note: "richtext" is the name of this app, 
# so you'll need to create that folder with the files referenced in this gist 
# such as `helpers.py`, `constants.py`, `wagtail_hooks.py`, etc.

INSTALLED_APPS = [
    # ...
    "wagtailtinymce",
    "richtext",
]

# Add extra rich text editor 
WAGTAILADMIN_RICH_TEXT_EDITORS = {
    # ...
    "secondary": {
        "WIDGET": "richtext.tinymce.CustomTinyMCERichTextArea",
    },
}

blocks.py

from django.utils.translation import gettext_lazy as _
from wagtail.core.blocks import StructBlock, RichTextBlock
from richtext.helpers import get_nodes_for_block


class AdvancedTableBlock(StructBlock):
    """A rich text field that implements TinyMCE for advanced table editing"""

    class Meta:
        template = "advanced_table.html"
        label = _("Advanced table")
        icon = "table"

    content = RichTextBlock(
        required=True,
        editor="secondary",
    )

    # modify the context to clean the db data and return a list of table nodes
    def get_context(self, value, parent_context=None):
        context = super().get_context(value, parent_context=parent_context)
        content = value.get("content")

        nodes = get_nodes_for_block(content)
        context.update(
            nodes=nodes,
        )

        return context

constants.py

TABLE_ATTRS = {
    "style": True,
}

TR_ATTRS = {
    "class": True,
    "style": True,
}

TH_TD_ATTRS = {
    "class": True,
    "colspan": True,
    "rowspan": True,
    "style": True,
    "scope": True,
}

ALLOWED_TAGS = [
    "table",
    "thead",
    "tbody",
    "tfoot",
    "tr",
    "th",
    "td",
    "b",
    "strong",
    "br",
]

ALLOWED_ROOT_NODES = [
    "table",
]

ALLOWED_STYLE_PROPERTIES_BLOCK = [
    "text-align",
    "vertical-align",
]

ALLOWED_STYLE_PROPERTIES_DB = ALLOWED_STYLE_PROPERTIES_BLOCK + [
    "height",
    "width",
]

ALLOWED_CLASSNAMES = [
    "advanced-table__cell",
]

fields.py

from wagtail.core.blocks import StreamBlock
from wagtail.core.fields import StreamField
from .blocks import AdvancedTableBlock


def AdvancedTableStreamField(blank=False, default="", help_text=""):
    required = not blank

    return StreamField(
        StreamBlock(
            [
                ("advanced_table", AdvancedTableBlock()),
            ],
            required=required,
        ),
        blank=blank,
        default=default,
        help_text=help_text,
    )

helpers.py

import bleach
from bleach.css_sanitizer import CSSSanitizer
from bs4 import BeautifulSoup
from django.utils.safestring import mark_safe
from richtext.constants import (
    TABLE_ATTRS,
    TR_ATTRS,
    TH_TD_ATTRS,
    ALLOWED_TAGS,
    ALLOWED_ROOT_NODES,
    ALLOWED_STYLE_PROPERTIES_BLOCK,
    ALLOWED_STYLE_PROPERTIES_DB,
    ALLOWED_CLASSNAMES,
)


block_css_sanitizer = CSSSanitizer(
    allowed_css_properties=ALLOWED_STYLE_PROPERTIES_BLOCK,
)


def colour_choices():
    return (
        ("red", _("Red")),
        ("orange", _("Orange")),
        ("yellow", _("Yellow")),
        ("green", _("Green")),
        ("aubergine", _("Aubergine")),
        ("violet", _("Violet")),
        ("purple", _("Purple")),
        ("blue", _("Blue")),
        ("turquoise", _("Turquoise")),
    )


db_css_sanitizer = CSSSanitizer(
    allowed_css_properties=ALLOWED_STYLE_PROPERTIES_DB,
)


def get_style_formats():
    colours = colour_choices()
    formats = []

    for item in colours:
        element_classname = "advanced-table__cell"
        modifier_classname = f"advanced-table__cell--{item[0]}"
        title = str(item[1])
        style = {
            "title": title,
            "selector": "tr,th,td",
            "classes": [element_classname, modifier_classname],
        }
        formats.append(style)

    return formats


def get_allowed_attrs():
    attrs = set(
        list(TABLE_ATTRS.keys()) + list(TR_ATTRS.keys()) + list(TH_TD_ATTRS.keys())
    )
    return list(attrs)


def clean_data_for_db(data):

    # convert data to bs object
    data = BeautifulSoup(data, "html.parser")

    # remove any disallowed root nodes
    for child in data.contents:
        if child.name and child.name.lower() not in ALLOWED_ROOT_NODES:
            child.decompose()

    # remove any disallowed classes
    for child in data.find_all():
        try:
            classes = child["class"]
            child["class"] = [
                x for x in classes if x.startswith(tuple(ALLOWED_CLASSNAMES))
            ]
        except KeyError:
            pass

    # remove any disallowed style properties
    data = mark_safe(
        bleach.clean(
            str(data),
            tags=ALLOWED_TAGS,
            attributes=get_allowed_attrs(),
            css_sanitizer=db_css_sanitizer,
            strip_comments=True,
            strip=True,
        )
    )

    return data


def get_nodes_for_block(data):

    # leave only classes on elements
    data = mark_safe(
        bleach.clean(
            str(data),
            tags=ALLOWED_TAGS,
            attributes=get_allowed_attrs(),
            css_sanitizer=block_css_sanitizer,
            strip_comments=True,
            strip=True,
        )
    )

    # convert data to bs object
    data = BeautifulSoup(data, "html.parser")

    # split nodes into list
    nodes = [
        mark_safe(x.prettify())
        for x in data.contents
        if x.name and x.name.lower() in ALLOWED_ROOT_NODES
    ]

    return nodes

Pipfile

...
wagtailtinymce = {git = "https://github.com/Junatum/wagtailtinymce.git"}
bleach = {extras = ["css"], version = "*"}

tinymce.py

from wagtail.core.rich_text import features
from wagtailtinymce.rich_text import TinyMCERichTextArea
from richtext.helpers import get_style_formats, clean_data_for_db


class CustomTinyMCERichTextArea(TinyMCERichTextArea):
    @classmethod
    def getDefaultArgs(cls):
        args = super().getDefaultArgs()

        # config for tinymce button bar
        args["buttons"] = [
            [
                ["undo", "redo"],
                ["styleselect"],
                ["bold"],
                ["table"],
                ["alignleft", "aligncenter", "alignright"],
                ["pastetext"],
            ]
        ]

        # config for tinymce init items
        args["passthru_init_keys"] = {
            "selector": "textarea",
            "style_formats": get_style_formats(),
            "style_formats_merge": False,
            "style_formats_autohide": True,
            "content_css": "/static/css/admin.css",
            "table_toolbar": """
                tabledelete |
                tableinsertrowbefore tableinsertrowafter tabledeleterow |
                tableinsertcolbefore tableinsertcolafter tabledeletecol
            """,
            "table_appearance_options": False,
            "table_advtab": False,
            "table_cell_advtab": False,
        }

        return args

    def __init__(self, attrs=None, **kwargs):
        super().__init__(attrs, **kwargs)

        # add rules added by wagtail hooks, as they aren't picked up otherwise
        rules = []
        for rule in features.converter_rules_by_converter["editorhtml"].values():
            for item in rule:
                rules.append(item)
        self.converter.converter_rules = rules

    def value_from_datadict(self, data, files, name):
        original_value = super(TinyMCERichTextArea, self).value_from_datadict(
            data, files, name
        )
        if original_value is None:
            return None

        # convert to database format
        db_format_data = self.converter.to_database_format(original_value)

        # clean html to remove unwanted attrs and styles
        cleaned_data = clean_data_for_db(db_format_data)

        return cleaned_data

wagtail_hooks.py

from wagtail.admin.rich_text import HalloPlugin
from wagtail.admin.rich_text.converters.editor_html import WhitelistRule
from wagtail.core import hooks
from wagtail.core.whitelist import allow_without_attributes, attribute_rule
from richtext.constants import TABLE_ATTRS, TR_ATTRS, TH_TD_ATTRS


@hooks.register("register_rich_text_features")
def register_embed_feature(features):
    features.register_editor_plugin(
        "editorhtml",
        "table",
        HalloPlugin(
            name="tinymcetable",
        ),
    )
    features.register_editor_plugin(
        "editorhtml",
        "thead",
        HalloPlugin(
            name="tinymcethead",
        ),
    )
    features.register_editor_plugin(
        "editorhtml",
        "tbody",
        HalloPlugin(
            name="tinymcetbody",
        ),
    )
    features.register_editor_plugin(
        "editorhtml",
        "tfoot",
        HalloPlugin(
            name="tinymcetfoot",
        ),
    )
    features.register_editor_plugin(
        "editorhtml",
        "tr",
        HalloPlugin(
            name="tinymcetr",
        ),
    )
    features.register_editor_plugin(
        "editorhtml",
        "th",
        HalloPlugin(
            name="tinymceth",
        ),
    )
    features.register_editor_plugin(
        "editorhtml",
        "td",
        HalloPlugin(
            name="tinymcetd",
        ),
    )

    features.register_converter_rule(
        "editorhtml",
        "table",
        [
            WhitelistRule("table", attribute_rule(TABLE_ATTRS)),
        ],
    )
    features.register_converter_rule(
        "editorhtml",
        "thead",
        [
            WhitelistRule("thead", allow_without_attributes),
        ],
    )
    features.register_converter_rule(
        "editorhtml",
        "tbody",
        [
            WhitelistRule("tbody", allow_without_attributes),
        ],
    )
    features.register_converter_rule(
        "editorhtml",
        "tfoot",
        [
            WhitelistRule("tfoot", allow_without_attributes),
        ],
    )
    features.register_converter_rule(
        "editorhtml",
        "tr",
        [
            WhitelistRule("tr", attribute_rule(TR_ATTRS)),
        ],
    )
    features.register_converter_rule(
        "editorhtml",
        "th",
        [
            WhitelistRule("th", attribute_rule(TH_TD_ATTRS)),
        ],
    )
    features.register_converter_rule(
        "editorhtml",
        "td",
        [
            WhitelistRule("td", attribute_rule(TH_TD_ATTRS)),
        ],
    )

A couple notes for future viewers of this code:

richtext is the name of this app. So you'll want to update all such references to the name of the app that you actually put this code into.

If you're using Wagtail 3, you'll need to use the wagtail-hallo app, since Wagtail has removed Hallo.js from its main codebase, and this code relies on that functionality.

If you're using Django 4.x (and possibly also 3.x, though I'm not sure), you'll need to use a newer fork of the wagtailtinymce package, as it is not compatible with Django 4. I'm personally using this line in my requirements.txt to bring in a working fork:

git+https://github.com/Junatum/wagtailtinymce.git@custom-1.2#egg=wagtailtinymce