如何为多对多关系创建initial_data Json fixture？

I am creating an initializing file for for my django project database. I am doing this using a file called initial_data.json which i have created. For example the following code (when syncdb is run) creates in the model Word a new row where name="apple":

我正在为我的django项目数据库创建一个初始化文件。我正在使用我创建的名为initial_data.json的文件来完成此操作。例如，以下代码（运行syncdb时）在模型Word中创建一个新行，其中name =“apple”：

[ { "model": "sites.word", "pk": 1, "fields": { "name": "apple" } } ]

[{“model”：“sites.word”，“pk”：1，“fields”：{“name”：“apple”}}]

I have managed to this so far for several models, the problem is with models that have a many-to-many field. I've looked around for the correct way to do this and have come up empty.

到目前为止，我已经为几个模型设法了，问题在于具有多对多字段的模型。我四处寻找正确的方法，并且空洞。

So, for example, if a Model mood has many Interests how would I write in the Json file that mood-1's interests are interest-1, interest-2 and interest-3.

那么，例如，如果一个模型情绪有很多兴趣，我怎么写在Json文件中，情绪-1的兴趣是兴趣-1，兴趣-2和兴趣-3。

What is the proper way to write in Json a models many-to-many relation?

在Json中编写多对多模型的正确方法是什么？

EDIT:

编辑：

@pastylegs solution was correct, I was just having trouble because the numbering of my interests was off in the Json file so it couldn't match them with there moods.

@pastylegs解决方案是正确的，我只是遇到了麻烦，因为我的兴趣编号在Json文件中已经关闭，所以它无法与那些情绪相匹配。

2 个解决方案

#1

I'm pretty sure the manytomany field of your model can be written like a simple list:

我很确定你的模型的多个字段可以写成一个简单的列表：

 [
    { 
        "model": "sites.word", 
        "pk": 1,
        "fields": { 
            "name": "apple",
            "my_m2m_field_name": [1,2,3],
        }
     }
]

where 1, 2, 3 are the primary keys for the relations

其中1,2,3是关系的主键

#2

what I like to do is use the dumpdata command. I fire up a test site, use the admin form or the app itself to add just the data that I want to use in my fixture, then I run

我喜欢做的是使用dumpdata命令。我启动一个测试站点，使用管理表单或应用程序本身添加我想要在我的夹具中使用的数据，然后我运行

./manage.py dumpdata appname > appname/fixtures/initial_data.json

You can dump all the apps together if you leave out appname but I like to do it separately for each model.

如果省略appname，可以将所有应用程序一起转储，但我喜欢为每个模型单独执行。

If you're using 1.3 (I'm not yet) then you can use --exclude to not dump some parts. I've aslo seen there is a --indent option to make the output pretty (just found that now while answering your question).

如果您使用的是1.3（我还没有）那么您可以使用--exclude来不转储某些部分。我已经看到有一个--inten选项使输出漂亮（刚刚发现，现在回答你的问题）。

It's one of those things that's easy to miss in the documentation. ;-)

这是文档中很容易遗漏的内容之一。 ;-)

#1