可插拔式后台管理系统(Django)
1.实现效果
研究了下django admin的功能,自己实现了一个简单的可插拔式后台管理系统,方便自定义特殊的功能,而且作为一个独立单独的django app,可以整体拷贝到其他项目中作为后台数据管理系统,对数据进行增删改查和自定义操作。下图是拷贝到一个图书管理系统中的后台效果:
2.实现思路
2.1 url的设计和分发
Django自带的admin,对于不同app的不同model表,都会动态的生成类似下面的四条url,分别对应着后台数据的增删改查页面。而为了实现动态路由需要配置两处,一是在项目全局urls.py文件中urlpatterns = [ url(r'^admin/', admin.site.urls),], 二是在每个app的admin.py文件中对model表进行了注册admin.site.register(model),这两处都涉及到了一个admin.site对象,因此我们需要实现自己的site对象即可。
查看:http://127.0.0.1:8008/admin/app01/book/
添加:http://127.0.0.1:8008/admin/app01/book/add/
更新:http://127.0.0.1:8008/admin/app01/book/1/change/
删除:http://127.0.0.1:8008/admin/app01/book/1/delete/
另外,对于django的url多级路由格式需要了解下:url(r" ",([ url(), url()], None, None)), 为 一个三元元祖([],None,None),而元祖中的列表[]又可以嵌套多个相同格式的url([],None,None),如下面的代码实现了三条路由:
url(r'^myAdmin2/', ([ url(r'^book1/',views.index1),
url(r'^book2/',([ url(r'^change/',views.index2), url(r'^add/',views.index3)],None,None ))],
None,None),)
对应的url如下:
http://127.0.0.1:8008/myAdmin2/book1/
http://127.0.0.1:8008/myAdmin2/book2/change/
http://127.0.0.1:8008/myAdmin2/book2/add/
根据上述的思路和多级url路由,可以定义同样的路由设置,一是设置urls.py中全局路由,二是在app的admin.py文件中注册model,三是实现自己的myAdmin.site对象。对应的代码依次如下:
urls.py
from django.conf.urls import urlfrom myAdmin.service.site import site # 引入自定义的site.py 文件中生成的site单例对象 urlpatterns = [ url(r'^myAdmin/', site.urls), ]
app01/admin.py
from myAdmin.service.site import site from app01 import models site.register(models.Book) site.register(models.Author) site.register(models.Publish)
myAdmin/service/site.py
class ModelAdmin(object): def __init__(self, model): self.model = model self.model_name = self.model._meta.model_name self.app_label = self.model._meta.app_label @property def urls(self): return self.get_urls(), None, None def get_urls(self): patterns = [url(r'^$', self.list_view, name='%s_%s_list'%(self.model_name,self.app_label)), url(r'^add/$', self.add_view,name='%s_%s_add'%(self.model_name,self.app_label)), url(r'^(.+)/change/$', self.change_view,name='%s_%s_change'%(self.model_name,self.app_label)), url(r'^(.+)/delete/$', self.delete_view,name='%s_%s_delete'%(self.model_name,self.app_label)), ] return patterns class AdminSite(object): def __init__(self): self._registry = {} def register(self, model, admin_class=None): #对应model表注册时的site.register() if not admin_class: admin_class = ModelAdmin admin_obj = admin_class(model) self._registry[model] = admin_obj @property def urls(self): #对应全局路由中的site.urls return self.get_urls(), None, None def get_urls(self): patterns = [] for model, admin_obj in self._registry.items(): urls = url(r'^{0}/{1}/'.format(model._meta.app_label, model._meta.model_name), admin_obj.urls) patterns.append(urls) return patterns site = AdminSite()
在site.py代码中有三处值得注意,
1. site = AdminSite(), 这里是采用了python模块的天然单例模式,由于每个app中都会采用site对象,因此在整个项目中只能有一个site对象。
2. AdminSite中的get_urls(self)函数
urls = url(r'^{0}/{1}/'.format(model._meta.app_label, model._meta.model_name), admin_obj.urls) 实现了第一级动态路由,即/app01/model/
3. ModelAdmin中的get_urls(self)函数
patterns = [url(r'^$', self.list_view, name='%s_%s_list'%(self.model_name,self.app_label)),
url(r'^add/$', self.add_view,name='%s_%s_add'%(self.model_name,self.app_label)),
url(r'^(.+)/change/$', self.change_view,name='%s_%s_change'%(self.model_name,self.app_label)),
url(r'^(.+)/delete/$', self.delete_view,name='%s_%s_delete'%(self.model_name,self.app_label)), ]
实现了第二级动态路由,即 app01/model, app01/model/add/, app01/model/id/change/, app01/model/id/delete 增删改查四条路径。
2.2 实现增删改查处理函数
在上面url设计中,在ModelAdmin类中定义了相应的处理函数,如下面self.list_view,self.add_view,self.change_view,self.delete_view,需要对其依次实现。
patterns = [url(r'^$', self.list_view, name='%s_%s_list'%(self.model_name,self.app_label)),
url(r'^add/$', self.add_view,name='%s_%s_add'%(self.model_name,self.app_label)),
url(r'^(.+)/change/$', self.change_view,name='%s_%s_change'%(self.model_name,self.app_label)),
url(r'^(.+)/delete/$', self.delete_view,name='%s_%s_delete'%(self.model_name,self.app_label)), ]
实现后的site.py代码如下:由于需要用到ModelForm类,定义了两个辅助方法get_modelform_class()和 change_modelform()
def get_modelform_class(self): class Model_form(ModelForm): class Meta: model = self.model fields = '__all__' return Model_form #返回类对象 def change_modelform(self,modelform): for item in modelform: if isinstance(item.field, ModelChoiceField): # ModelChoiceField表示field字段对应的为外键或多对对关系 pop_item_name = item.name item.is_pop=True #为实例动态绑定属性 item_model_name = item.field.queryset.model._meta.model_name item_app_label = item.field.queryset.model._meta.app_label item.pop_url = '/myAdmin/{0}/{1}/add/?pop_item_name={2}'.format(item_app_label, item_model_name,pop_item_name) return modelform #添加数据 def add_view(self, request): modelform_class = self.get_modelform_class() form = modelform_class() form = self.change_modelform(form) if request.method == 'POST': form = modelform_class(request.POST) field_obj = form.save() # url = request.path[:-4] # print url pop_item_name = request.GET.get('pop_item_name') if pop_item_name: result = {'pk':field_obj.pk,'text':str(field_obj),'pop_item_name':pop_item_name} return render(request, 'process_pop.html', {'result':result}) return redirect(self.get_list_url()) return render(request, 'add_view.html', locals()) # 改变数据 def change_view(self, request, number): modelform_class = self.get_modelform_class() model_obj = self.model.objects.filter(id=number).first() form = modelform_class(instance=model_obj) form = self.change_modelform(form) if request.method == 'POST': form = modelform_class(request.POST, instance=model_obj) form.save() return redirect(self.get_list_url()) return render(request, 'change_view.html', locals()) # 删除数据 def delete_view(self, request, number): model_obj = self.model.objects.get(id=number) list_url = self.get_list_url() edit_url = self.get_change_url(model_obj) if request.method=='POST': model_obj.delete() return redirect(list_url) return render(request, 'delete_view.html', locals())
list_view 较为复杂,上面没列出,下面单独列出代码。因为list_view界面中还支持搜索,分类和批量处理三个功能,list_view必须对这三种请求进行捕获和处理。从下面代码中可以看到:定义了两个辅助方法,get_search_condition()和get_filter_condition()来处理搜索和分类过滤(详细见自定义字段的实现)。搜索框和分类过滤器请求通过GET请求提交,get_search_condition()处理搜索框提交的GET请求,将搜索条件封装成一个Q对象返回,get_filter_condition()处理过滤器提交的GET请求,返回Q对象。actions的请求(批量处理)通过POST请求提交,将批处理函数和选定项提交到request.POST,然后list_view进行处理。实现代码如下:
#处理搜索框提交的请求 def get_search_condition(self,request): search_connector = Q() if request.method=='GET': search_content = request.GET.get('search_content','') search_connector.connector = 'or' if search_content and self.search_field: for field in self.search_field: # field_obj = self.model._meta.get_field(field) # if isinstance(field_obj,ManyToManyField) or isinstance(field_obj,ForeignKey): # search_connector.children.append((field + '__name__contains', search_content)) #对于多对多关系,如何实现动态? # else: search_connector.children.append((field + '__contains', search_content)) return search_connector #处理过滤标签的<a>标签提交的请求 def get_filter_condition(self,request): filter_connector = Q() if request.method == 'GET': for filter_field, value in request.GET.items(): if filter_field in self.list_filter: # 设置分页后url会出现page参数,不应做为过滤条件 filter_connector.children.append((filter_field, value)) return filter_connector #查看:显示数据 def list_view(self, request): model = self.model if request.method == 'POST': choice_item = request.POST.get('choice_item') selected_item = request.POST.getlist('selected_item') action_func = getattr(self,choice_item) queryset = model.objects.filter(id__in =selected_item) action_func(queryset) search_condition = self.get_search_condition(request) filter_condition = self.get_filter_condition(request) model_list = model.objects.all().filter(search_condition).filter(filter_condition) showlist= Showlist(self,model_list,model,request) #单独抽象出一个类,用来配置前端数据的显示 return render(request, 'list_view.html', locals())
2.3 自定义字段的实现
在ModelAdmin中可以定义相应的字段,对数据管理显示界面进行设置,从而在model进行注册时能根据需求对这些字段进行更改和扩展,展示出不同的显示效果。下面定义了显示字段,过滤器,搜索和批处理等:
class ModelAdmin(object): list_display = ('__str__',) #自定义显示的字段 list_display_links = () #自定义超链接字段,点击进入编辑页面 list_filter = () #自定义过滤器字段,根据该字段分类 search_field = () #自定义搜索字段,搜索的内容和这些字段进行匹配 actions = () #自定义批处理函数
list_display 的扩展
下面代码为list_display的实现,首先在list_display中加入默认的选择框,编辑和删除超链接,然后对用户配置的list_play进行扩展,得到完整的list_play,再进行渲染,代码如下:
# 定义默认要显式的内容, 编辑,删除操作和选择框,并扩展list_display def edit(self,model_obj=None,isHeader=False): #model_obj: 一个model表对象,isHeader是否是表格的表头字段 if isHeader: return '操作' return mark_safe( '<a href="%s/change/">编辑</a>' % model_obj.pk) # 注意href="%s/change/ 和 href="/%s/change/的区别,前者为当前目录,后者为根目录 def checkbox(self,model_obj=None, isHeader=False): if isHeader: return '选择' return mark_safe('<input type="checkbox" value="%s" name="selected_item"/>'%model_obj.pk) def delete(self,model_obj=None, isHeader=False): if isHeader: return '操作' return mark_safe( '<a href="%s/delete/">删除</a>' % model_obj.pk) def get_list_display(self): new_list_display = [] new_list_display.append(ModelAdmin.checkbox) #加入选择框 new_list_display.extend(self.list_display) #加入用户配置的list_display if not self.list_display_links: new_list_display.append(ModelAdmin.edit) #如果用户未配置超链接字段,加入编辑操作 new_list_display.append(ModelAdmin.delete) #加入删除操作 return new_list_display
上面代码拿到了一个完整的list_display=(checkbox, '', '', exit, delete),而对于list_display的处理和前端显示见下面Showlist 类。
actions 的扩展
和list_diaplay一样,下面代码中,在actions加入默认的批量删除函数,并扩展用户配置的actions批处理函数,拿到了一个完整的actions=(batch_delete, ), 其处理和前端显示见后面Showlist 类
#定义默认的批量删除函数,并扩展用户actions def batch_delete(self,queryset): queryset.delete() batch_delete.short_description = '批量删除' def get_actions(self): new_actions = [] new_actions.append(ModelAdmin.batch_delete) # 加入批量删除操作 new_actions.extend(self.actions) # 扩展用户配置actions return new_actions
在实现list_view()函数时,用到了一个单独的类Showlist, 其中定义了 list_diaplay, list_play_links, list_filter, search_field 和 actions的处理和前端显示逻辑,代码如下:
class Showlist(object): ''' 需要四个参数来初始化实例: model_config: ModelAdmin 的实例对象,决定了其相关配置项 model_list: 发送给前端的表格中要展示的数据对象(Queryset) model:数据表对象 request:视图函数中的request参数 ''' def __init__(self,model_config,model_list,model,request): self.model_config = model_config self.model_list = model_list self.model = model self.request = request # 设置分页 current_page = int(request.GET.get('page',1)) params = self.request.GET base_url = self.request.path all_count = self.model_list.count() #print 'all_count',all_count self.page = page.Pagination(current_page, all_count, base_url, params, per_page_num=4, pager_count=3,) self.page_data = self.model_list[self.page.start:self.page.end]
# 前端actions的显示数据 def get_action_desc(self): # actions list_actions = [] if self.model_config.get_actions(): for action in self.model_config.get_actions(): list_actions.append({ "name": action.__name__, #批处理函数的名字 "desc": action.short_description }) return list_actions
# 前端过滤器的显示数据 def get_filter_dict(self): filter_dict = {} for field in self.model_config.list_filter: params = copy.deepcopy(self.request.GET) selection = self.request.GET.get(field, 0) field_obj = self.model._meta.get_field(field) if isinstance(field_obj, ForeignKey) or isinstance(field_obj, ManyToManyField): #对于多对多或外键字段的处理 data_list = field_obj.rel.to.objects.all() #field_obj.rel.to 能拿到多对多或外键字段对应的另一张model表对象 else: data_list = self.model.objects.all().values('pk', field) temp = [] if params.get(field): #url参数的过滤条件中,如果有该字段的过滤条件,则点击全部时应该删除该字段的过滤条件,从而显示全部数据; del params[field] temp.append("<a href='?%s' class='list-group-item is_selected'>全部</a>" % params.urlencode()) else: #不含有该字段的过滤条件,点击时不处理 temp.append("<a href='#' class='list-group-item'>全部</a>") for item in data_list: if isinstance(field_obj, ForeignKey) or isinstance(field_obj, ManyToManyField): #多对多或外键字段,拿到的为对象 id = item.pk text = str(item) params[field] = id #多对多或外键字段,以id做为过滤条件 else: #普通字段拿到的为字典 id = item['pk'] text = item[field] params[field] = text #普通字段以字段名称做为过滤条件 tag_url = params.urlencode() if selection == str(id) or selection == text: #判断此时url过滤字段中选中的条件,为其添加特殊style样式 temp.append("<a href='?%s' class='list-group-item is_selected'>%s</a>" % (tag_url, text)) else: temp.append("<a href='?%s' class='list-group-item'>%s</a>" % (tag_url, text)) filter_dict[field_obj] = temp return filter_dict #前端表格表头的显示数据 def get_head_list(self): head_list = [] for field in self.model_config.get_list_display(): if isinstance(field, str): #判断函数和字符窜 if field == '__str__': #用户未配置时默认的list_play=('__str__',) value = self.model._meta.model_name else: field_obj = self.model._meta.get_field(field) # 拿到字符窜对应的field对象 value = field_obj.verbose_name # 通过拿到verbose_name 来显示中文 else: value = field(self.model_config, isHeader=True) # 获取标题,传入isHeader, 注意此处传入的self.model_config if value: head_list.append(value) return head_list #前端表格内容的显示数据 def get_data_list(self): data_list = [] for model_obj in self.page_data: #分页截取的某一页的数据列表 row_list = [] for field in self.model_config.get_list_display(): if isinstance(field, str): #判断是字符窜或函数 try : field_obj = self.model_config.model._meta.get_field(field) #判断设置的显式列是否为多对多关系,处理相应的多个数据 if isinstance(field_obj, ManyToManyField): temp_list = getattr(model_obj, field).all() #print temp_list ret = [] for temp in temp_list: #print temp ret.append(str(temp)) #转换为字符窜后进行拼接 value = ','.join(ret) #print value else: value = getattr(model_obj, field) # 通过反射拿到字符窜对应的值 if field in self.model_config.list_display_links: # 判断该字段是否设置为超链接,放在此处表明了多对多关系设置在超链接列中无效 value = mark_safe('<a href="%s/change/">%s</a>' % (model_obj.pk, value)) except Exception as e: #print e value = getattr(model_obj, field) else: value = field(self.model_config, model_obj) # 获取内容,传入model_obj,不用传入isHeader if value: row_list.append(value) data_list.append(row_list) # print data_list return data_list
2.4 增加自定义的url 路径
可以为某个model表单独增加一个url接口,来处理特殊的业务;首先需要在site.py 文件中定义接口,然后在app的admin.py注册文件中进行定义处理逻辑。在下面的代码中model表通过覆盖父类的extra_urls()函数来增加了一条url和相应的处理逻辑。
site.py定义的接口如下:
def get_urls(self): patterns = [url(r'^$', self.list_view, name='%s_%s_list'%(self.model_name,self.app_label)), url(r'^add/$', self.add_view,name='%s_%s_add'%(self.model_name,self.app_label)), url(r'^(.+)/change/$', self.change_view,name='%s_%s_change'%(self.model_name,self.app_label)), url(r'^(.+)/delete/$', self.delete_view,name='%s_%s_delete'%(self.model_name,self.app_label)), ] patterns.extend(self.extra_url()) return patterns #定义url接口,modelConfigure通过继承覆盖来配置额外的url def extra_url(self): return []
admin.py 定义处理逻辑如下:
class BookConfig(ModelAdmin):
# 通过下面三个函数,为book添加一条单独的url处理逻辑,实现点击id值,为title添加喜欢或不喜欢 def list_id(self,model_obj=None, isHeader=False): if isHeader: return 'ID' return mark_safe('<a href="like_book/%s">%s</a>'%(model_obj.pk, model_obj.pk)) def like_book(self,request,obj_id): model_obj = models.Book.objects.get(id = obj_id) if '(喜欢)' not in model_obj.title: new_title = '%s (喜欢)'%model_obj.title else: new_title = model_obj.title.replace('(喜欢)','') models.Book.objects.filter(id=obj_id).update(title = new_title) return redirect(self.get_list_url()) def extra_url(self): temp = [url(r'like_book/(\d+)',self.like_book)] return temp list_display = (list_id, 'title','price','author','publish')
site.register(models.Book,BookConfig)
3 总结
通过上述部分,实现了一个完成的后台管理系统,有两个小特色,一是插拔式,方便在其他项目中进行复用;二是代码中保留了扩展字段和自定义url接口,能够根据不同的业务需求扩展特殊的功能。项目源代码及基本使用见下面github。
项目源代码: https://github.com/silence-cho/Myadmin