For example, there's this snippet in the dgo doc
// If omitempty is not set, then edges with empty values (0 for int/float, "" for string, false
// for bool) would be created for values not specified explicitly.
type Person struct {
Uid string `json:"uid,omitempty"`
Name string `json:"name,omitempty"`
Age int `json:"age,omitempty"`
Dob *time.Time `json:"dob,omitempty"`
Married bool `json:"married,omitempty"`
Raw []byte `json:"raw_bytes,omitempty"`
Friends []Person `json:"friend,omitempty"`
Location loc `json:"loc,omitempty"`
School []School `json:"school,omitempty"`
DType []string `json:"dgraph.type,omitempty"`
}
The comment is stated as if omitting omitempty is a bad thing.
As always, it depends... on your specific use-case and data model.
Having omitempty for a field results in that field missing in the json.Marshal result if its value is the types default zero value. Without omitempty the json will always include that field, and it will be the types default zero value if not set to something else. For example having following struct missing omitempty for the age field:
type Person struct {
Uid string `json:"uid,omitempty"`
Name string `json:"name,omitempty"`
Age int `json:"age"
}
p := Person{
Uid: "0x1",
}
js, _ := json.Marshal(&p)
The resulting json will have no name field, but age with the default zero value for an int, which is 0:
{"uid": "0x1", "age": 0}
If you send this json to a Dgraph mutation, age will be set to 0 for that node and overwrite the nodes current age value, while name will be unchanged. So in this case, having omitempty for the age fields makes sense, otherwise you would have to always set the age field to the correct value before calling a mutation with the json, even if you only want to change the name field of the node.
If you want a field to always be set to its default zero value if not set otherwise in the struct, then you don't use omitempty for that field.
Related
My data structure is
{
_id: ObjectID,
...
fields: [
{ name: "Aryan" },
{ books : [ 1,2,3 ] },
]
}
In our application a user can define his own fields data structure but with a key value structure. So, we had no way of knowing the structure of the data.
So in our document struct we had
type Document struct {
Fields map[string]interface{}
}
as the second parameter returned by mongo was primitive.A ( []interface{} under the hood ), the individual item could have been an array, map, anything.
But we couldn't range over that for it being.
How can I get the individual values like the book ids [1,2,3] or the name value "Aryan" ?
After a couple of attempts to solving this, I was at a state where my current element it self is an interface{} ( [1,2,3] in this case ) and I couldn't get the individual 1,2,3s.
But finally managed to solve it
for k, val := range doc.Fields {
v := reflect.ValueOf(val)
switch reflect.TypeOf(t).Kind() {
case reflect.Slice:
// getting the individual ids
for i := 0; i < v.Len(); i++ {
fmt.Println(v.Index(i))
}
case reflect.String:
default:
// handle
}
}
Note: v.Index(i) returns a data type of Value .
to change the data type
v.Index(i).Interface().(string) // string
v.Index(i).Interface().(float64) // float
Ideally you should avoid working with interface{} type since it's very error prone and the compiler can't help you. The idiomatic way is to define a struct for your model with BSON tags like in this example`
type MyType struct {
ID primitive.ObjectID `bson:"_id,omitempty"`
Fields []Field `bson:"fields,omitempty"`
}
type Field struct {
Name string `bson:"name,omitempty"`
Books []int `bson:"books,omitempty"`
}
Field here is defined as the reunion of all possible fields which again is not ideal but at least the compiler can help you and developers know what to expect from the database document.
I am working on a project that uses combination of Go and MongoDB. I am stuck at a place where I have a struct like:
type Booking struct {
// booking fields
Id int `json:"_id,omitempty" bson:"_id,omitempty"`
Uid int `json:"uid,omitempty" bson:"uid,omitempty"`
IndustryId int `json:"industry_id,omitempty" bson:"industry_id,omitempty"`
LocationId int `json:"location_id,omitempty" bson:"location_id,omitempty"`
BaseLocationId int `json:"base_location_id,omitempty" bson:"base_location_id,omitempty"`
}
In this struct, the field Id is of int type. But as we know MongoDB's default id is of bsonObject type. Some times, the system generates the default MongoDB id in Id field.
To overcome this I have modified the struct like this:
type Booking struct {
// booking fields
Id int `json:"_id,omitempty" bson:"_id,omitempty"`
BsonId bson.ObjectId `json:"bson_id" bson:"_id,omitempty"`
Uid int `json:"uid,omitempty" bson:"uid,omitempty"`
IndustryId int `json:"industry_id,omitempty" bson:"industry_id,omitempty"`
LocationId int `json:"location_id,omitempty" bson:"location_id,omitempty"`
BaseLocationId int `json:"base_location_id,omitempty" bson:"base_location_id,omitempty"`
}
In above struct, I have mapped the same _id field in two different struct fields Id (type int) and BsonId (type bson.ObjectId). I want if id of integer type comes, it maps under Id otherwise under BsonId.
But this thing is giving following error:
Duplicated key '_id' in struct models.Booking
How can I implement this type of thing with Go Structs ??
Update:
Here is the code I have written for custom marshaling/unmarshaling:
func (booking *Booking) SetBSON(raw bson.Raw) (err error) {
type bsonBooking Booking
if err = raw.Unmarshal((*bsonBooking)(booking)); err != nil {
return
}
booking.BsonId, err = booking.Id
return
}
func (booking *Booking) GetBSON() (interface{}, error) {
booking.Id = Booking.BsonId
type bsonBooking *Booking
return bsonBooking(booking), nil
}
But this is giving type mismatch error of Id and BsonId fields. What should I do now ?
In your original struct you used this tag for the Id field:
bson:"_id,omitempty"
This means if the value of the Id field is 0 (zero value for the int type), then that field will not be sent to MongoDB. But the _id property is mandatory in MongoDB, so in this case the MongoDB server will generate an ObjectId for it.
To overcome this, the easiest is to ensure the Id will is always non-zero; or if the 0 is a valid id, remove the omitempty option from the tag.
If your collection must allow ids of mixed type (int and ObjectId), then the easiest would be to define the Id field with type of interface{} so it can accomodate both (actually all) types of key values:
Id interface{} `json:"_id,omitempty" bson:"_id,omitempty"`
Yes, working with this may be a little more cumbersome (e.g. if you explicitly need the id as int, you need to use type assertion), but this will solve your issue.
If you do need 2 id fields, one with int type and another with ObjectId type, then your only option will be to implement custom BSON marshaling and unmarshaling. This basically means to implement the bson.Getter and / or bson.Setter interfaces (one method for each) on your struct type, in which you may do anything you like to populate your struct or assemble the data to be actually saved / inserted. For details and example, see Accessing MongoDB from Go.
Here's an example how using custom marshaling it may look like:
Leave out the Id and BsonId fields from marshaling (using the bson:"-" tag), and add a 3rd, "temporary" id field:
type Booking struct {
Id int `bson:"-"`
BsonId bson.ObjectId `bson:"-"`
TempId interface{} `bson:"_id"`
// rest of your fields...
}
So whatever id you have in your MongoDB, it will end up in TempId, and only this id field will be sent and saved in MongoDB's _id property.
Use the GetBSON() method to set TempId from the other id fields (whichever is set) before your struct value gets saved / inserted, and use SetBSON() method to "copy" TempId's value to one of the other id fields based on its dynamic type after the document is retrieved from MongoDB:
func (b *Booking) GetBSON() (interface{}, error) {
if b.Id != 0 {
b.TempId = b.Id
} else {
b.TempId = b.BsonId
}
return b, nil
}
func (b *Booking) SetBSON(raw bson.Raw) (err error) {
if err = raw.Unmarshal(b); err != nil {
return
}
if intId, ok := b.TempId.(int); ok {
b.Id = intId
} else bsonId, ok := b.TempId.(bson.ObjectId); ok {
b.BsonId = bsonId
} else {
err = errors.New("invalid or missing id")
}
return
}
Note: if you dislike the TempId field in your Booking struct, you may create a copy of Booking (e.g. tempBooking), and only add TempId into that, and use tempBooking for marshaling / unmarshaling. You may use embedding (tempBooking may embed Booking) so you can even avoid repetition.
Some JSON data I am getting have spaces in the key names. I am using standard encoding/json library to unmarshal the data. However it is unable to understand the keys with spaces in the schema. For e.g. following code:
package main
import (
"encoding/json"
"fmt"
)
func main() {
var jsonBlob = []byte(`[
{"Na me": "Platypus", "Order": "Monotremata"},
{"Na me": "Quoll", "Order": "Dasyuromorphia"}
]`)
type Animal struct {
Name string `json: "Na me"`
Order string `json: "Order,omitempty"`
}
var animals []Animal
err := json.Unmarshal(jsonBlob, &animals)
if err != nil {
fmt.Println("error:", err)
}
fmt.Printf("%+v", animals)
}
Gives the output as:
[{Name: Order:Monotremata} {Name: Order:Dasyuromorphia}]
So in the schema the library removes the space(from Na me) and try to find the key (Name), which is obviously not present. Any suggestion what can I do here?
Your json tag specification is incorrect, that's why the encoding/json library defaults to the field name which is Name. But since there is no JSON field with "Name" key, Animal.Name will remain its zero value (which is the empty string "").
Unmarshaling Order will still work, because the json package will use the field name if json tag specification is missing (tries with both lower and upper-case). Since the field name is identical to the JSON key, it works without extra JSON tag mapping.
You can't have a space in the tag specification after the colon and before the quotation mark:
type Animal struct {
Name string `json:"Na me"`
Order string `json:"Order,omitempty"`
}
With this simple change, it works (try it on the Go Playground):
[{Name:Platypus Order:Monotremata} {Name:Quoll Order:Dasyuromorphia}]
I tried to retrieve a document from my collection with unique id.
I have a collection with fields: name, age, city, and rank. I want to get 'city' results from mongodb using golang.
My struct code
type exp struct {
name string `bson:"name"`
age int `bson:"age"`
city string `bson:"city"`
rank int `bson:"rank"`
}
With the following code to retrieve results from mongodb:
var result []exp //my struct type
err = coll.Find(bson.M{"City":bson.M{}}).Distinct("City",&result)
fmt.Println(result)
With this code I get an empty array as the result. How would I get all the cities?
Try this code
var result []string
err = c.Find(nil).Distinct("city", &result)
if err != nil {
log.Fatal(err)
}
fmt.Println(result)
Due to restrictions in reflection, mgo (as well as encoding/json and other similar packages) is unable to use unexported fields to marshal or unmarshal data. What you need to do is to export your fields by capitalize the first letter:
type exp struct {
Name string `bson:"name"`
Age int `bson:"age"`
City string `bson:"city"`
Rank int `bson:"rank"`
}
A side note: you do not need to specify the bson tags if the desired name is the same as the lowercase field name. The documentation for bson states:
The lowercased field name is used as the key for each exported field,
but this behavior may be changed using the respective field tag.
Edit:
I just realized you did get an empty slice and not a slice with empty struct fields. My answer is then not an actual answer to the question, but it is still an issue that you need to consider.
Using mgo I'm unable to store any meaningful data. Only the _id gets stored
type Person struct {
name string
age int
}
func main() {
session, err := mgo.Dial("localhost")
if err != nil {
log.Fatal(err)
}
defer session.Close()
p := Person{"Joe", 50}
ppl := session.DB("rest").C("people")
ppl.Insert(p)
}
The result in Mongo is just the _id field - no sign of "Joe".
Using go 1.1.2 on Arch linux, MongoDB 2.4.6.
type Person struct {
name string
age int
}
The mgo package can't access unexported (lowercase) fields of your struct (i.e. no other package than the one the struct is defined in can). You need to export them (first letter must be upper case), like this:
type Person struct {
Name string
Age int
}
If you wish to have the field names in lower case in the DB you must provide a struct tag for them, like this:
type Person struct {
Name string `bson:"name"`
Age int `bson:"age"`
}
See the documentation on names:
Names are as important in Go as in any other language. They even have
semantic effect: the visibility of a name outside a package is
determined by whether its first character is upper case. [...]
EDIT:
Gustavo Niemeyer (author of the mgo and bson packages) noted in the comments that unlike the json package, the bson marshaller will lowercase all struct field names when commiting to the database, effectively making the last step in this answer superfluous.