Patch REST API to Partial Update MongoDB in .NET

2020-06-06 02:29发布

问题:

I have an object

{
  "_id": "testobject",
  "A": "First line",
  "B": "Second line",
  "C": "Third line"
}

I want to send a REST PATCH request to my API to only update one of these properties

{
  "_id": "testobject",
  "C": "Forth line"
}

This gets parsed into a class

public class SomeObject {
  public string A { get; set; }
  public string B { get; set; }
  public string C { get; set; }
}

I now need to update the existing document in MongoDB but only updating the property C.

I could create an update definition just for this one record

UpdateDefinition<SomeObject> update = Builders<SomeObject>.Update.Set(x => x.C, <value of property C>)

Or I could hard code a check on each property to see if it is empty

IList<UpdateDefinition<SomeObject>> updates = new List<UpdateDefinition<SomeObject>>();
if (!string.IsNullOrEmpty(C)) {
  updates.Add(UpdateDefinition<SomeObject> update = Builders<SomeObject>.Update.Set(x => x.C, <value of property C>));
}
if (!string.IsNullOrEmpty(C)) {
  updates.Add(UpdateDefinition<SomeObject> update = Builders<SomeObject>.Update.Set(x => x.C, <value of property C>));
}

However, if I have many properties and many sub properties this could get very large very fast. The other issue is that if I set the value of the property to be intentionally empty then it would not update the record at all due to it looking for the field to non-empty.

How can I dynamically do partial updates to MongoDB documents in .NET so that I have a generic PATCH API call that can take any of the parameters the document has and only update the properties specified?

回答1:

I suggest that you avoid relying on 1.x legacy API, as it's perfectly supported in 2.x as well, as shown in the sample code below.

var client = new MongoClient();
var database = client.GetDatabase("test");
var collection = database.GetCollection<BsonDocument>("test");

var changesJson = "{ a : 1, b : 2 }";
var changesDocument = BsonDocument.Parse(changesJson);

var filter = Builders<BsonDocument>.Filter.Eq("_id", 1);

UpdateDefinition<BsonDocument> update = null;
foreach (var change in changesDocument)
{
    if (update == null)
    {
        var builder = Builders<BsonDocument>.Update;
        update = builder.Set(change.Name, change.Value);
    }
    else
    {
        update = update.Set(change.Name, change.Value);
    }
}

//following 3 lines are for debugging purposes only
//var registry = BsonSerializer.SerializerRegistry;
//var serializer = registry.GetSerializer<BsonDocument>();
//var rendered = update.Render(serializer, registry).ToJson();

//you can also use the simpler form below if you're OK with bypassing the UpdateDefinitionBuilder (and trust the JSON string to be fully correct)
update = new BsonDocumentUpdateDefinition<BsonDocument>(new BsonDocument("$set", changesDocument));

var result = collection.UpdateOne(filter, update);

Credits go to Robert Stam for providing the code sample.



回答2:

You can use

IMongoUpdate updateDoc = new UpdateDocument("$set", doc);
collection.Update(Query.EQ("_id",id), updateDoc);

However, you should be careful.

If you first deserialize your document into SomeObject, all of the fields will get their default value (null for strings, 0 for ints etc). And if you use that object for the update, the fields that didn't exist in your json string would be updated to their default value.

If you use

var bsonDoc = BsonSerializer.Deserialize<BsonDocument>(jsonString);  
IMongoUpdate updateDoc = new UpdateDocument("$set", bsonDoc);
collection.Update(Query.EQ("_id",id), updateDoc);

your document on the database will be updated only for the fields that are present in your jsonString



回答3:

Not sure anyone is here >= June '20 however I did the following. I'm using NewtonSoft JObject/JArray and I wanted to create a mongo update parser/function that wouldn't know the incoming schema and would build out nested documents as well. Another thing I had to get used to (I'm new to Mongo) was the syntax of the keys in the Bson Update document i.e.

{ "key.full.path.into.nested.document" : "valueToSet" }

So, after trying a few ways to manually/recursively account for the nesting/containing path of incoming JSON doc, I finally found and can just use JToken.Path property perfectly for this.

Anyway, hopefully this is something someone will find useful. It's just an example and makes a few assumptions about the document structure but is pretty useful in its current form. And, like me, I think it might help a few people that are learning Mongo and their C# driver while also using JSON.Net to wrap the incoming REST requests.

    public BsonDocument ParseUpdateRequest(JObject req)
    {
        BsonDocument bson = new BsonDocument();
        Parse(req, ref bson);

        BsonDocument parsedBson = new BsonDocument();
        parsedBson["$set"] = bson;
        return parsedBson;            
    }
    private void Parse(JObject req, ref BsonDocument bson)
    {
        /**
         * Could use a parent key/node in each recursion call or just use the JToken path
         * string.IsNullOrEmpty(parentNode) ? field.Key : parentNode + "." + field.Key;
         **/ 
        string key;
        JToken val;
        foreach (var field in req)
        {
            key = field.Value.Path;                
            val = field.Value;                
            switch (val.Type)
            {
                case JTokenType.String:
                    bson.Add(key, (string)val);
                    break;
                case JTokenType.Integer:
                    bson.Add(key, (int)val);
                    break;
                case JTokenType.Float:
                    bson.Add(key, (float)val);
                    break;
                case JTokenType.Date:
                    DateTime dt = (DateTime)val;
                    bson.Add(key, dt.ToUniversalTime());                        
                    break;
                case JTokenType.Array:
                    BsonArray bsonArray = ParseArray((JArray)val);
                    bson.Add(key, bsonArray);
                    break;
                case JTokenType.Object:
                    Parse((JObject)val, ref bson);
                    break;
            }
        }
        return;
    }

    private BsonArray ParseArray(JArray source)
    {
        BsonArray bson = new BsonArray();            
        foreach (JToken field in source)
        {   
            switch (field.Type)
            {
                case JTokenType.String:
                    bson.Add((string)field);
                    break;                    
                case JTokenType.Date:
                    DateTime dt = (DateTime)field;
                    bson.Add(dt.ToUniversalTime());
                    break;
                case JTokenType.Integer:
                    bson.Add((int)field);
                    break;
                case JTokenType.Float:
                    bson.Add((float)field);
                    break;
                case JTokenType.Object:
                    BsonDocument nestedDoc = new BsonDocument();
                    Parse((JObject)field, ref nestedDoc);
                    bson.Add(nestedDoc);
                    break;
            }
        }
        return bson;
    }

And here's some simple test code I wrote:

            ModelUser user = new ModelUser();
        ControllerApp app = new ControllerApp();
        ControllerApp.Instance.User = user;
        JObject req = new JObject();
        req["first"] = "First";
        req["last"] = "Last";
        req["usertype"] = "parent";
        req["pw"] = "q345n3452345n2345";
        req["array"] = JArray.Parse("[ '1', '2', '3' ]");
        req["dateTest"] = DateTime.UtcNow;
        req["profile"] = new JObject();
        req["profile"]["name"] = new JObject();
        req["profile"]["name"]["first"] = "testUpdateFirst";

        BsonDocument bd;
        bd = user.ParseUpdateRequest(req);
        string s = bd.ToJson();