Refactor ColumnType: register classes, return instances with config

- register_column_types() now returns classes instead of instances
- ColumnType.__init__ takes optional config=, baking it into the instance
- get_column_type() returns a ColumnType instance (or None) instead of a
  (name, config) tuple
- get_column_types() returns {col: ColumnType instance} instead of tuples
- Remove get_column_type_class() - no longer needed
- render_cell/validate/transform_value methods no longer take config arg;
  use self.config instead
- render_cell hook takes column_type (ColumnType or None) instead of
  column_type + column_type_config

https://claude.ai/code/session_01SvPEPqHgURTWESRp28pTC3
This commit is contained in:
Claude 2026-03-17 05:18:14 +00:00
commit dd9b83301c
No known key found for this signature in database
11 changed files with 227 additions and 243 deletions

View file

@ -922,13 +922,16 @@ await .get_column_type(database, resource, column)
``column`` - string
The name of the column.
Returns a ``(column_type_name, config)`` tuple for the specified column. ``column_type_name`` is a string like ``"email"`` or ``"url"``, and ``config`` is a dict or ``None``. If no column type is assigned, returns ``(None, None)``.
Returns a :ref:`ColumnType <column_types>` instance with ``.config`` populated for the specified column, or ``None`` if no column type is assigned.
.. code-block:: python
ct_name, config = await datasette.get_column_type(
ct = await datasette.get_column_type(
"mydb", "mytable", "email_col"
)
if ct:
print(ct.name) # "email"
print(ct.config) # None or {...}
.. _datasette_get_column_types:
@ -940,12 +943,13 @@ await .get_column_types(database, resource)
``resource`` - string
The name of the table or view.
Returns a dictionary mapping column names to ``(column_type_name, config)`` tuples for all columns that have assigned types on the given resource.
Returns a dictionary mapping column names to :ref:`ColumnType <column_types>` instances (with ``.config`` populated) for all columns that have assigned types on the given resource.
.. code-block:: python
ct_map = await datasette.get_column_types("mydb", "mytable")
# {"email_col": ("email", None), "site": ("url", None)}
for col_name, ct in ct_map.items():
print(col_name, ct.name, ct.config)
.. _datasette_set_column_type:
@ -995,22 +999,6 @@ Removes the column type assignment for the specified column.
"mydb", "mytable", "location"
)
.. _datasette_get_column_type_class:
.get_column_type_class(column_type_name)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
``column_type_name`` - string
The name of the column type, e.g. ``"email"``.
Returns the registered ``ColumnType`` instance for the given name, or ``None`` if no plugin has registered a column type with that name. This is a synchronous method.
.. code-block:: python
ct = datasette.get_column_type_class("email")
if ct:
print(ct.description) # "Email address"
.. _datasette_add_database:
.add_database(db, name=None, route=None)
@ -2049,6 +2037,14 @@ The internal database schema is as follows:
value text,
unique(database_name, resource_name, column_name, key)
);
CREATE TABLE column_types (
database_name TEXT NOT NULL,
resource_name TEXT NOT NULL,
column_name TEXT NOT NULL,
column_type TEXT NOT NULL,
config TEXT,
PRIMARY KEY (database_name, resource_name, column_name)
);
.. [[[end]]]

View file

@ -474,8 +474,8 @@ Examples: `datasette-publish-fly <https://datasette.io/plugins/datasette-publish
.. _plugin_hook_render_cell:
render_cell(row, value, column, table, pks, database, datasette, request, column_type, column_type_config)
----------------------------------------------------------------------------------------------------------
render_cell(row, value, column, table, pks, database, datasette, request, column_type)
--------------------------------------------------------------------------------------
Lets you customize the display of values within table cells in the HTML table view.
@ -503,11 +503,8 @@ Lets you customize the display of values within table cells in the HTML table vi
``request`` - :ref:`internals_request`
The current request object
``column_type`` - string or None
The name of the :ref:`column type <column_types>` assigned to this column, or ``None`` if no column type is assigned.
``column_type_config`` - dict or None
The configuration dict for the assigned column type, or ``None``.
``column_type`` - :ref:`ColumnType <column_types>` instance or None
The :ref:`ColumnType <column_types>` instance assigned to this column (with ``.config`` populated), or ``None`` if no column type is assigned. You can access ``column_type.name``, ``column_type.config``, etc.
If a column has a :ref:`column type <column_types>` assigned and that column type's ``render_cell`` method returns a non-``None`` value, it will take priority over this plugin hook.
@ -1002,7 +999,7 @@ The permission system then uses this query along with rules from plugins to dete
register_column_types(datasette)
--------------------------------
Return a list of :ref:`ColumnType <column_types>` instances to register custom column types. Column types define how values in specific columns are rendered, validated, and transformed.
Return a list of :ref:`ColumnType <column_types>` **classes** (not instances) to register custom column types. Column types define how values in specific columns are rendered, validated, and transformed.
.. code-block:: python
@ -1023,7 +1020,6 @@ Return a list of :ref:`ColumnType <column_types>` instances to register custom c
database,
datasette,
request,
config,
):
if value:
return markupsafe.Markup(
@ -1032,14 +1028,12 @@ Return a list of :ref:`ColumnType <column_types>` instances to register custom c
).format(color=markupsafe.escape(value))
return None
async def validate(self, value, config, datasette):
async def validate(self, value, datasette):
if value and not value.startswith("#"):
return "Color must start with #"
return None
async def transform_value(
self, value, config, datasette
):
async def transform_value(self, value, datasette):
# Normalize to uppercase
if isinstance(value, str):
return value.upper()
@ -1048,7 +1042,7 @@ Return a list of :ref:`ColumnType <column_types>` instances to register custom c
@hookimpl
def register_column_types(datasette):
return [ColorColumnType()]
return [ColorColumnType]
Each ``ColumnType`` subclass must define the following class attributes:
@ -1060,16 +1054,16 @@ Each ``ColumnType`` subclass must define the following class attributes:
And the following methods, all optional:
``render_cell(self, value, column, table, database, datasette, request, config)``
``render_cell(self, value, column, table, database, datasette, request)``
Return an HTML string to render this cell value, or ``None`` to fall through to the default ``render_cell`` plugin hook chain. When a column type provides rendering, it takes priority over the ``render_cell`` plugin hook.
``validate(self, value, config, datasette)``
``validate(self, value, datasette)``
Validate a value before it is written via the insert, update, or upsert API endpoints. Return ``None`` if valid, or a string error message if invalid. Null values and empty strings skip validation.
``transform_value(self, value, config, datasette)``
``transform_value(self, value, datasette)``
Transform a value before it appears in JSON API output. Return the transformed value. The default implementation returns the value unchanged.
The ``config`` argument passed to these methods is the parsed JSON config dict for the specific column assignment, or ``None`` if no config was provided.
Per-column configuration is available via ``self.config`` in all methods. When a column type is looked up for a specific column (via :ref:`get_column_type <datasette_get_column_type>` or :ref:`get_column_types <datasette_get_column_types>`), the returned instance has ``config`` set to the parsed JSON config dict for that column assignment, or ``None`` if no config was provided.
Column types are assigned to columns via the ``column_types`` key in :ref:`table configuration <metadata_tables>`: