Add docs for DRF-based choosers

pull/8921/head
Matt Westcott 2022-07-14 15:11:13 +01:00 zatwierdzone przez LB (Ben Johnston)
rodzic 19a7708c7c
commit c4d6c7d6c5
1 zmienionych plików z 171 dodań i 0 usunięć

Wyświetl plik

@ -77,3 +77,174 @@ def register_viewset():
```
Registering a chooser viewset will also set up a chooser widget to be used whenever a ForeignKey field to that model appears in a `WagtailAdminModelForm` - see [](./forms). In particular, this means that a panel definition such as `FieldPanel("author")`, where `author` is a foreign key to the `Person` model, will automatically use this chooser interface.
## Chooser viewsets for non-model datasources
While the generic chooser views are primarily designed to use Django models as the data source, choosers based on other sources such as REST API endpoints can be implemented by overriding the individual methods that deal with data retrieval.
Within `wagtail.admin.views.generic.chooser`:
- `BaseChooseView.get_object_list()` - returns a list of records to be displayed in the chooser. (In the default implementation, this is a Django QuerySet, and the records are model instances.)
- `BaseChooseView.columns` - a list of `wagtail.admin.ui.tables.Column` objects specifying the fields of the record to display in the final table
- `BaseChooseView.apply_object_list_ordering(objects)` - given a list of records as returned from `get_object_list`, returns the list with the desired ordering applied
- `ChosenViewMixin.get_object(pk)` - returns the record identified by the given primary key
- `ChosenResponseMixin.get_chosen_response_data(item)` - given a record, returns the dictionary of data that will be passed back to the chooser widget to populate it (consisting of items `id` and `title`, unless the chooser widget's JavaScript has been customised)
Within `wagtail.admin.widgets`:
- `BaseChooser.get_instance(value)` - given a value that may be a record, a primary key or None, returns the corresponding record or None
- `BaseChooser.get_value_data_from_instance(item)` - given a record, returns the dictionary of data that will populate the chooser widget (consisting of items `id` and `title`, unless the widget's JavaScript has been customised)
For example, the following code will implement a chooser that runs against a JSON endpoint for the User model at `http://localhost:8000/api/users/`, set up with Django REST Framework using the default configuration and no pagination:
```python
from django.views.generic.base import View
import requests
from wagtail.admin.ui.tables import Column, TitleColumn
from wagtail.admin.views.generic.chooser import (
BaseChooseView, ChooseViewMixin, ChooseResultsViewMixin, ChosenResponseMixin, ChosenViewMixin, CreationFormMixin
)
from wagtail.admin.viewsets.chooser import ChooserViewSet
from wagtail.admin.widgets import BaseChooser
class BaseUserChooseView(BaseChooseView):
@property
def columns(self):
return [
TitleColumn(
"title",
label="Title",
accessor='username',
id_accessor='id',
url_name=self.chosen_url_name,
link_attrs={"data-chooser-modal-choice": True},
),
Column(
"email", label="Email", accessor="email"
)
]
def get_object_list(self):
r = requests.get("http://localhost:8000/api/users/")
r.raise_for_status()
results = r.json()
return results
def apply_object_list_ordering(self, objects):
return objects
class UserChooseView(ChooseViewMixin, CreationFormMixin, BaseUserChooseView):
pass
class UserChooseResultsView(ChooseResultsViewMixin, CreationFormMixin, BaseUserChooseView):
pass
class UserChosenViewMixin(ChosenViewMixin):
def get_object(self, pk):
r = requests.get("http://localhost:8000/api/users/%d/" % int(pk))
r.raise_for_status()
return r.json()
class UserChosenResponseMixin(ChosenResponseMixin):
def get_chosen_response_data(self, item):
return {
"id": item["id"],
"title": item["username"],
}
class UserChosenView(UserChosenViewMixin, UserChosenResponseMixin, View):
pass
class BaseUserChooserWidget(BaseChooser):
def get_instance(self, value):
if value is None:
return None
elif isinstance(value, dict):
return value
else:
r = requests.get("http://localhost:8000/api/users/%d/" % int(value))
r.raise_for_status()
return r.json()
def get_value_data_from_instance(self, instance):
return {
"id": instance["id"],
"title": instance["username"],
}
class UserChooserViewSet(ChooserViewSet):
icon = "user"
choose_one_text = "Choose a user"
choose_another_text = "Choose another user"
edit_item_text = "Edit this user"
choose_view_class = UserChooseView
choose_results_view_class = UserChooseResultsView
chosen_view_class = UserChosenView
base_widget_class = BaseUserChooserWidget
user_chooser_viewset = UserChooserViewSet("user_chooser", url_prefix="user-chooser")
```
If the data source implements its own pagination - meaning that the pagination mechanism built into the chooser should be bypassed - the `BaseChooseView.get_results_page(request)` method can be overridden instead of `get_object_list`. This should return an instance of `django.core.paginator.Page`. For example, if the API in the above example followed the conventions of the Wagtail API, implementing pagination with `offset` and `limit` URL parameters and returning a dict consisting of `meta` and `results`, the `BaseUserChooseView` implementation could be modified as follows:
```python
from django.core.paginator import Page, Paginator
class APIPaginator(Paginator):
"""
Customisation of Django's Paginator class for use when we don't want it to handle
slicing on the result set, but still want it to generate the page numbering based
on a known result count.
"""
def __init__(self, count, per_page, **kwargs):
self._count = int(count)
super().__init__([], per_page, **kwargs)
@property
def count(self):
return self._count
class BaseUserChooseView(BaseChooseView):
@property
def columns(self):
return [
TitleColumn(
"title",
label="Title",
accessor='username',
id_accessor='id',
url_name=self.chosen_url_name,
link_attrs={"data-chooser-modal-choice": True},
),
Column(
"email", label="Email", accessor="email"
)
]
def get_results_page(self, request):
try:
page_number = int(request.GET.get('p', 1))
except ValueError:
page_number = 1
r = requests.get("http://localhost:8000/api/users/", params={
'offset': (page_number - 1) * self.per_page,
'limit': self.per_page,
})
r.raise_for_status()
result = r.json()
paginator = APIPaginator(result['meta']['total_count'], self.per_page)
page = Page(result['items'], page_number, paginator)
return page
```