Add Example of Serialization/Deserialization for CSV

[LeviButcher]

Short summary (3-5 sentences) describing the issue.

Currently, there is no up to date example of how to implement a Custom Formatter to return both CSV Response and Read a CSV request (Using text/csv with Content-Type and Accept headers). My use case for needing this example is for allowing batch updates with a CSV file across multiple entity types. This could be elegantly solved if it was possible to accept a request of text/csv and the server automatically de-serialized it and applied the data to a patch endpoint.

Any guidelines would be appreciated and I'm more then willing to write up the example and place over here: https://github.com/OData/ODataSamples

Assemblies affected

*Which assemblies and versions are known to be affected e.g. OData WebApi lib 6.1.0

Most recent Odata Versions.

[xuzhg]

@LeviButcher I am working on a demo.
I have a question about CVS for you or the community. What's your expectation output about the complex property or collection property?

[xuzhg]

@LeviButcher Please let me know your thought about: https://devblogs.microsoft.com/odata/customize-odata-payload-serialization-format-within-asp-net-core-odata/

[sharshi]

@xuzhg great work on the article, It's the only working implementation I was able to find that got this far in solving this.

I'm exploring options for dealing with the navigation property, I see you parse it into one 'cell'. I think there are two general approaches to take for navigation properties:

{
  "name": "John Doe",
  "age": 30,
  "address": {
    "street": "123 Main St",
    "city": "Anytown",
    "zipCode": "12345"
  },
  "hobbies": ["reading", "traveling", "swimming"],
  "contacts": [
    {
      "type": "email",
      "value": "[email protected]"
    },
    {
      "type": "phone",
      "value": "555-1234"
    }
  ]
}

1. columns could be root level: name, age, address, hobbies, contacts and the arrays/objects are parsed as json and take of one cell
   • PRO: one row per result
   • CON: json in a csv is not standard
   • CON: good luck using the data
2. columns could be dot notation: name, age, address.street, address.city, address.zipCode, hobbies, contacts.type, contacts.value:
   • PRO: Data is denormalized
   • CON: The number of rows will grow with every array, and exponentially with the combination of multiple arrays.

name	age	address.street	address.city	address.zipCode	hobbies	contacts.type	contacts.value
John Doe	30	123 Main St	Anytown	12345	reading	email	[email protected]
John Doe	30	123 Main St	Anytown	12345	reading	phone	555-1234
John Doe	30	123 Main St	Anytown	12345	traveling	email	[email protected]
John Doe	30	123 Main St	Anytown	12345	traveling	phone	555-1234
John Doe	30	123 Main St	Anytown	12345	swimming	email	[email protected]
John Doe	30	123 Main St	Anytown	12345	swimming	phone	555-1234