I have a requirement in my project where I have to perform a DB operation for getting a particular type of a total number of users. What I am doing is that, filtered all the queries in a slice and passing that Silce to my DB function.
This is the code snippet from where I am calling the DB function
{
filters = []bson.D{
{{Key: "Mykey", Value: myvalue}},
{{Key: "Mykey", Value: myvalue}},
{{Key: "Mykey", Value: myvalue}},
{{Key: "Mykey", Value: myvalue}},
counts, err := dbmain.NoOfDocumentsInfo(MyDBName, myCollectionName, filters...)
}
Below is my called function
func NoOfDocumentsInfo(DB string, col string, filters ...bson.D) ([]int64, error) {
if nil == dbInstance {
if nil == GetDBInstance() {
logger.Error("Not connecting to DB")
err := errors.New("DB connection error")
return nil, err
}
}
logger.Debugf("%s %s", DB, col)
coll := dbInstance.Database(DB).Collection(col)
counts := make([]int64, len(filters))
for i, filter := range filters {
count, err := coll.CountDocuments(context.TODO(), filter)
if err != nil {
logger.Fatal(err)
return nil, err
}
counts[i] = count
}
return counts, nil
}
As you can see I am calling the "coll.CountDocuments" functions multiple times. What I want is to write the code without calling the "coll.CountDocuments" function multiple times by aggregating all the filters into a single query.
I have tried to use the aggregation pipeline but my "cur" and "result" is giving null output. If you run the code you will be able to see it.
func NoOfDocumentsInfo(DB string, col string, filters ...bson.D) ([]int64, error) {
if dbInstance == nil {
if GetDBInstance() == nil {
logger.Error("Not connecting to DB")
err := errors.New("DB connection error")
return nil, err
}
}
logger.Debugf("%s %s", DB, col)
coll := dbInstance.Database(DB).Collection(col)
pipeline := make([]bson.M, 0, len(filters)+2)
pipeline = append(pipeline, bson.M{"$match": bson.M{"$or": filters}})
pipeline = append(pipeline, bson.M{"$group": bson.M{"_id": nil, "count": bson.M{"$sum": 1}}})
pipeline = append(pipeline, bson.M{"$group": bson.M{"_id": nil, "count": bson.M{"$first": "$count"}}})
var result struct {
Count int64 `bson:"count"`
}
cur, err := coll.Aggregate(context.TODO(), pipeline)
if err != nil {
logger.Fatal(err)
return nil, err
}
logger.Debugf("cur: %+v", cur)
err = cur.Decode(&result)
logger.Debugf("result: %+v, err: %v", result, err)
if err != nil {
logger.Fatal(err)
return nil, err
}
return []int64{result.Count}, nil
}
You have to add a field for each filter in $group, you may use $cond to conditionally increment the given counter. But this may very well end up not using indices, and thus being even slower than the separate, original count queries. Also note that using $or may also result in skipping indices. Also note that in $cond you may have to transform filters (e.g. add $ to field names).
You'd better launch concurrent count queries (using go) for each filter, and if they are indexed, they will complete fast. This is how it could look like:
func docCounts(db string, col string, filters ...bson.D) ([]int64, error) {
// ... obtain collection
coll := dbInstance.Database(db).Collection(col)
counts := make([]int64, len(filters))
errs := make([]error, len(filters))
wg := &sync.WaitGroup{}
wg.Add(len(filters))
for i := range filters {
go func(i int) {
defer wg.Done()
counts[i], errs[i] = coll.CountDocuments(context.TODO(), filters[i])
}(i)
}
wg.Wait()
// Produce some kind of error if any of the queries failed.
var err error
for _, e := range errs {
if e != nil {
err = fmt.Errorf("at least one query failed: %w", e)
break
}
}
// Note: starting with Go 1.20, you could simply write:
// err = errors.Join(errs)
return counts, err
}
Related
As you can see I am calling the "coll.CountDocuments" functions multiples times. What I want is to write the code without calling the "coll.CountDocuments" function multiple times by aggregating all the filters into a single query.
func NoOfDocumentsInfo(DB string, col string, filters ...bson.D) ([]int64, error) {
if nil == dbInstance {
if nil == GetDBInstance() {
logger.Error("Not connecting to DB")
err := errors.New("DB connection error")
return nil, err
}
}
logger.Debugf("%s %s", DB, col)
coll := dbInstance.Database(DB).Collection(col)
counts := make([]int64, len(filters))
for i, filter := range filters {
count, err := coll.CountDocuments(context.TODO(), filter)
if err != nil {
logger.Fatal(err)
return nil, err
}
counts[i] = count
}
return counts, nil
}
I have tried to used aggragation pipeline but "cur" and "result" is giving null output.
`func NoOfDocumentsInfo(DB string, col string, filters ...bson.D) ([]int64, error) {
if dbInstance == nil {
if GetDBInstance() == nil {
logger.Error("Not connecting to DB")
err := errors.New("DB connection error")
return nil, err
}
}
logger.Debugf("%s %s", DB, col)
coll := dbInstance.Database(DB).Collection(col)
pipeline := make([]bson.M, 0, len(filters)+2)
pipeline = append(pipeline, bson.M{"$match": bson.M{"$or": filters}})
pipeline = append(pipeline, bson.M{"$group": bson.M{"_id": nil, "count": bson.M{"$sum": 1}}})
pipeline = append(pipeline, bson.M{"$group": bson.M{"_id": nil, "count": bson.M{"$first": "$count"}}})
var result struct {
Count int64 `bson:"count"`
}
cur, err := coll.Aggregate(context.TODO(), pipeline)
if err != nil {
logger.Fatal(err)
return nil, err
}
logger.Debugf("cur: %+v", cur)
err = cur.Decode(&result)
logger.Debugf("result: %+v, err: %v", result, err)
if err != nil {
logger.Fatal(err)
return nil, err
}
return []int64{result.Count}, nil
}`
A much simpler approach would be the one that I'm going to share here. Let's start with the code:
package main
import (
"context"
"fmt"
"time"
"go.mongodb.org/mongo-driver/bson"
"go.mongodb.org/mongo-driver/mongo"
"go.mongodb.org/mongo-driver/mongo/options"
)
var (
dbInstance *mongo.Client
ctx context.Context
cancel context.CancelFunc
)
func NoOfDocumentsInfo(client *mongo.Client, DB string, col string, filters bson.A) (int64, error) {
coll := client.Database(DB).Collection(col)
myFilters := bson.D{
bson.E{
Key: "$and",
Value: filters,
},
}
counts, err := coll.CountDocuments(ctx, myFilters)
if err != nil {
panic(err)
}
return counts, nil
}
func main() {
ctx, cancel = context.WithTimeout(context.Background(), 20*time.Second)
defer cancel()
// set MongoDB connection
clientOptions := options.Client().ApplyURI("mongodb://root:root#localhost:27017")
mongoClient, err := mongo.Connect(ctx, clientOptions)
if err != nil {
panic(err)
}
defer mongoClient.Disconnect(ctx)
// query with filters
numDocs, err := NoOfDocumentsInfo(mongoClient, "demodb", "myCollection", bson.A{
bson.D{bson.E{Key: "Name", Value: bson.D{bson.E{Key: "$eq", Value: "John Doe"}}}},
bson.D{bson.E{Key: "Song", Value: bson.D{bson.E{Key: "$eq", Value: "White Roses"}}}},
})
if err != nil {
panic(err)
}
fmt.Println("num docs:", numDocs)
}
Let's see the relevant changes applied to the code:
Expect a parameter called filters of type bson.A which is the type for the array in the MongoDB environment.
Build the myFilters variable which is of type bson.D (slice) with the following single item (bson.E) in this way:
The Key is the logical operator
The Value is the array passed into the function
Build the array to pass to the function with all of the needed filters (e.g. two equal conditions: one on the Name key and the other on the Song).
Finally, I also did some improvements on how you've opened the MongoDB connection and how you've released the allocated resources.
Let me know if this solves your issue, thanks!
I have two columns in a CSV file. I am accessing only the first column using the SearchData() function.
The problem is that I want to access the data as an array but when I return an array string in the AccessData() function and write the products[0] in the SearchData(), it gives me all the data by removing the bracket sign [] only and when I write products[1], it gives me runtime error: index out of range [1] with length 1.
Required result
products[0] = First Item
products[1] = Second Item
...
so on
Code
func AccessData(number int) string {
content, err := ioutil.ReadFile("products/data1.csv")
if err != nil {
log.Fatal(err)
}
Data := string(content)
sliceData := strings.Split(Data, ",")
return sliceData[number]
}
func SearchData(){
for i := 0; i <= 34; i = i + 2 {
products := AccessData(i)
fmt.Println(products)
}
}
This should do the trick:
func firstColumns(filename string) []string {
f, err := os.Open(filename)
if err != nil {
log.Fatal(err)
}
defer f.Close()
r := csv.NewReader(f)
var result []string
for {
row, err := r.Read()
if err != nil {
if err == io.EOF {
break
}
log.Fatal(err)
}
if len(row) > 0 {
result = append(result, row[0])
}
}
return result
}
func main() {
data := firstColumns("products/data1.csv")
fmt.Println(data)
fmt.Println(data[1])
}
This turns the the first column of every row into a []string which can be access index.
The output is:
[First item Second item]
Second item
How can I insert an array of documents into MongoDB with mgo library using only a single DB call as in db.collection.insert()?
I have the following Transaction structure:
type Transaction struct {
Brand string `json:"brand"`
Name string `json:"name"`
Plu string `json:"plu"`
Price string `json:"price"`
}
From a POST request I will recieve an array of these structures. I want to insert them into MongoDB as individual documents but using a single DB call as explained in db.collection.insert()
I tried using c.Insert of mgo
The following is the code snippet:
func insertTransaction(c *gin.Context) {
var transactions []Transaction
err := c.BindJSON(&transactions)
if err != nil {
c.AbortWithStatusJSON(http.StatusBadRequest, map[string]string{"error":"invalid JSON"})
return
}
err = InsertTransactons(transactions)
if err != nil {
c.AbortWithStatusJSON(http.StatusInternalServerError, &map[string](interface{}){
"status": "error",
"code": "500",
"message": "Internal server error",
"error": err,
})
return
}
c.JSON(http.StatusCreated, &map[string](interface{}){
"status": "success",
"code": "0",
"message": "created",
})
}
func InsertTransactons(u []Transaction) error {
s := GetSession()
defer s.Close()
c := s.DB(DB).C(TransactionColl)
err := c.Insert(u...)
if err != nil {
return err
}
return nil
}
But as I compile and run the code, I get the following error:
go/database.go:34:17: cannot use u (type *[]Transaction) as type
[]interface {} in argument to c.Insert
You cannot pass []*Transaction as []interface{}. You need to convert each Transaction to inferface{} to change its memory layout.
var ui []interface{}
for _, t := range u{
ui = append(ui, t)
}
Pass ui to c.Insert instead
Create slice of interface for document structs by appending and then inserting data using Bulk insert which takes variable arguments.
type Bulk struct {
// contains filtered or unexported fields
}
func (b *Bulk) Insert(docs ...interface{})
For inserting documents in Bulk
const INSERT_COUNT int = 10000
type User struct {
Id bson.ObjectId `bson:"_id,omitempty" json:"_id"`
Email string `bson:"email" json:"email"`
}
func (self *User) Init() {
self.Id = bson.NewObjectId()
}
Call Bulk() function on collection returned from db connection. Bulk() function returns pointer to *Bulk.
bulk := dbs.Clone().DB("").C("users").Bulk()
bulk.Insert(users...)
Assign it to variable which will be used to call Insert() method using Bulk pointer receiver.
func main(){
// Database
dbs, err := mgo.Dial("mongodb://localhost/")
if err != nil {
panic(err)
}
// Collections
uc := dbs.Clone().DB("").C("users")
defer dbs.Clone().DB("").Session.Close()
for n := 0; n < b.N; n++ {
count := INSERT_COUNT
users := make([]interface{}, count)
for i := 0; i < count; i++ {
loop_user := User{}
loop_user.Init()
loop_user.Email = fmt.Sprintf("report-%d#example.com", i)
users[i] = loop_user
}
bulk := uc.Bulk()
bulk.Unordered()
bulk.Insert(users...)
_, bulkErr := bulk.Run()
if bulkErr != nil {
panic(err)
}
}
}
On app engine I have a large number of entities of a particular kind.
I want to run a function on each entity (e.g. edit the entity or copy it)
I would do this in a taskqueue but a taskqueue is limited to 10 minutes runtime and each function call is prone to many kinds of errors. What is the best way to do this?
Here's my solution although I'm hoping someone out there has a better solution. I also wonder if this is prone to fork bombs e.g. if the task runs twice, it will set off two chains of iteration.. ! I'm only using it to iterate a few hundred thousand entities, although the operation on each entity is expensive.
First I create a taskqueue for running each individual function call on an entity one at a time:
queue:
- name: entity-iter
rate: 100/s
max_concurrent_requests: 1
retry_parameters:
task_retry_limit: 3
task_age_limit: 30m
min_backoff_seconds: 200
and then I have an iterate entity method which, given the kind, will call your delay func on each entity with the key.
package sysadmin
import (
"google.golang.org/appengine/datastore"
"golang.org/x/net/context"
"google.golang.org/appengine/log"
"google.golang.org/appengine/delay"
"google.golang.org/appengine/taskqueue"
)
func ForEachEntity(kind string, f *delay.Function) *delay.Function {
var callWithNextKey *delay.Function // func(c context.Context, depth int, cursorString string) error
callWithNextKey = delay.Func("something", func(c context.Context, depth int, cursorString string) error {
q := datastore.NewQuery(kind).KeysOnly()
if cursorString != "" {
if curs, err := datastore.DecodeCursor(cursorString); err != nil {
log.Errorf(c, "error decoding cursor %v", err)
return err
} else {
q = q.Start(curs)
}
}
it := q.Run(c)
if key, err := it.Next(nil); err != nil {
if err == datastore.Done {
log.Infof(c, "Done %v", err)
return nil
}
log.Errorf(c, "datastore error %v", err)
return err
} else {
curs, _ := it.Cursor()
if t, err := f.Task(key); err != nil {
return err
} else if _, err = taskqueue.Add(c, t, "entity-iter"); err != nil {
log.Errorf(c, "error %v", err)
return err
}
if depth - 1 > 0 {
if err := callWithNextKey.Call(c, depth - 1, curs.String()); err != nil {
log.Errorf(c, "error2 %v", err)
return err
}
}
}
return nil
})
return callWithNextKey
}
example usage:
var DoCopyCourse = delay.Func("something2", CopyCourse)
var DoCopyCourses = ForEachEntity("Course", DoCopyCourse)
func CopyCourses(c context.Context) {
//sharedmodels.MakeMockCourses(c)
DoCopyCourses.Call(c, 9999999, "")
}
Basically after doing a query I'd like to take the resulting rows and produce a []map[string]interface{}, but I do not see how to do this with the API since the Rows.Scan() function needs a specific number of parameters matching the requested number of columns (and possibly the types as well) to correctly obtain the data.
Again, I'd like to generalize this call and take any query and turn it into a []map[string]interface{}, where the map contains column names mapped to the values for that row.
This is likely very inefficient, and I plan on changing the structure later so that interface{} is a struct for a single data point.
How would I do this using just the database/sql package, or if necessary the database/sql/driver package?
Look at using sqlx, which can do this a little more easily than the standard database/sql library:
places := []Place{}
err := db.Select(&places, "SELECT * FROM place ORDER BY telcode ASC")
if err != nil {
fmt.Printf(err)
return
}
You could obviously replace []Place{} with a []map[string]interface{}, but where possible it is better to use a struct if you know the structure of your database. You won't need to undertake any type assertions as you might on an interface{}.
I haven't used it (yet), but I believe the "common" way to do what you are asking (more or less) is to use gorp.
You can create a struct that maintains the map key to the position of the []interface{} slice. By doing this, you do not need to create a predefined struct. For example:
IDOrder: 0
IsClose: 1
IsConfirm: 2
IDUser: 3
Then, you can use it like this:
// create a fieldbinding object.
var fArr []string
fb := fieldbinding.NewFieldBinding()
if fArr, err = rs.Columns(); err != nil {
return nil, err
}
fb.PutFields(fArr)
//
outArr := []interface{}{}
for rs.Next() {
if err := rs.Scan(fb.GetFieldPtrArr()...); err != nil {
return nil, err
}
fmt.Printf("Row: %v, %v, %v, %s\n", fb.Get("IDOrder"), fb.Get("IsConfirm"), fb.Get("IDUser"), fb.Get("Created"))
outArr = append(outArr, fb.GetFieldArr())
}
Sample output:
Row: 1, 1, 1, 2016-07-15 10:39:37 +0000 UTC
Row: 2, 1, 11, 2016-07-15 10:42:04 +0000 UTC
Row: 3, 1, 10, 2016-07-15 10:46:20 +0000 UTC
SampleQuery: [{"Created":"2016-07-15T10:39:37Z","IDOrder":1,"IDUser":1,"IsClose":0,"IsConfirm":1},{"Created":"2016-07-15T10:42:04Z","IDOrder":2,"IDUser":11,"IsClose":0,"IsConfirm":1},{"Created":"2016-07-15T10:46:20Z","IDOrder":3,"IDUser":10,"IsClose":0,"IsConfirm":1}]
Please see the full example below or at fieldbinding:
main.go
package main
import (
"bytes"
"database/sql"
"encoding/json"
"fmt"
)
import (
_ "github.com/go-sql-driver/mysql"
"github.com/junhsieh/goexamples/fieldbinding/fieldbinding"
)
var (
db *sql.DB
)
// Table definition
// CREATE TABLE `salorder` (
// `IDOrder` int(10) unsigned NOT NULL AUTO_INCREMENT,
// `IsClose` tinyint(4) NOT NULL,
// `IsConfirm` tinyint(4) NOT NULL,
// `IDUser` int(11) NOT NULL,
// `Created` datetime NOT NULL,
// `Changed` datetime NOT NULL,
// PRIMARY KEY (`IDOrder`),
// KEY `IsClose` (`IsClose`)
// ) ENGINE=InnoDB DEFAULT CHARSET=utf8;
func main() {
var err error
// starting database server
db, err = sql.Open("mysql", "Username:Password#tcp(Host:Port)/DBName?parseTime=true")
if err != nil {
panic(err.Error()) // Just for example purpose. You should use proper error handling instead of panic
}
defer db.Close()
// SampleQuery
if v, err := SampleQuery(); err != nil {
fmt.Printf("%s\n", err.Error())
} else {
var b bytes.Buffer
if err := json.NewEncoder(&b).Encode(v); err != nil {
fmt.Printf("SampleQuery: %v\n", err.Error())
}
fmt.Printf("SampleQuery: %v\n", b.String())
}
}
func SampleQuery() ([]interface{}, error) {
param := []interface{}{}
param = append(param, 1)
sql := "SELECT "
sql += " SalOrder.IDOrder "
sql += ", SalOrder.IsClose "
sql += ", SalOrder.IsConfirm "
sql += ", SalOrder.IDUser "
sql += ", SalOrder.Created "
sql += "FROM SalOrder "
sql += "WHERE "
sql += "IsConfirm = ? "
sql += "ORDER BY SalOrder.IDOrder ASC "
rs, err := db.Query(sql, param...)
if err != nil {
return nil, err
}
defer rs.Close()
// create a fieldbinding object.
var fArr []string
fb := fieldbinding.NewFieldBinding()
if fArr, err = rs.Columns(); err != nil {
return nil, err
}
fb.PutFields(fArr)
//
outArr := []interface{}{}
for rs.Next() {
if err := rs.Scan(fb.GetFieldPtrArr()...); err != nil {
return nil, err
}
fmt.Printf("Row: %v, %v, %v, %s\n", fb.Get("IDOrder"), fb.Get("IsConfirm"), fb.Get("IDUser"), fb.Get("Created"))
outArr = append(outArr, fb.GetFieldArr())
}
if err := rs.Err(); err != nil {
return nil, err
}
return outArr, nil
}
fieldbinding package:
package fieldbinding
import (
"sync"
)
// NewFieldBinding ...
func NewFieldBinding() *FieldBinding {
return &FieldBinding{}
}
// FieldBinding is deisgned for SQL rows.Scan() query.
type FieldBinding struct {
sync.RWMutex // embedded. see http://golang.org/ref/spec#Struct_types
FieldArr []interface{}
FieldPtrArr []interface{}
FieldCount int64
MapFieldToID map[string]int64
}
func (fb *FieldBinding) put(k string, v int64) {
fb.Lock()
defer fb.Unlock()
fb.MapFieldToID[k] = v
}
// Get ...
func (fb *FieldBinding) Get(k string) interface{} {
fb.RLock()
defer fb.RUnlock()
// TODO: check map key exist and fb.FieldArr boundary.
return fb.FieldArr[fb.MapFieldToID[k]]
}
// PutFields ...
func (fb *FieldBinding) PutFields(fArr []string) {
fCount := len(fArr)
fb.FieldArr = make([]interface{}, fCount)
fb.FieldPtrArr = make([]interface{}, fCount)
fb.MapFieldToID = make(map[string]int64, fCount)
for k, v := range fArr {
fb.FieldPtrArr[k] = &fb.FieldArr[k]
fb.put(v, int64(k))
}
}
// GetFieldPtrArr ...
func (fb *FieldBinding) GetFieldPtrArr() []interface{} {
return fb.FieldPtrArr
}
// GetFieldArr ...
func (fb *FieldBinding) GetFieldArr() map[string]interface{} {
m := make(map[string]interface{}, fb.FieldCount)
for k, v := range fb.MapFieldToID {
m[k] = fb.FieldArr[v]
}
return m
}
If you really want a map, which is needed in some cases, have a look at dbr, but you need to use the fork (since the pr got rejected in the original repo). The fork seems more up to date anyway:
https://github.com/mailru/dbr
For info on how to use it:
https://github.com/gocraft/dbr/issues/83
package main
import (
"fmt"
"github.com/bobby96333/goSqlHelper"
)
func main(){
fmt.Println("hello")
conn,err :=goSqlHelper.MysqlOpen("user:password#tcp(127.0.0.1:3306)/dbname")
checkErr(err)
row,err := conn.QueryRow("select * from table where col1 = ? and col2 = ?","123","abc")
checkErr(err)
if *row==nil {
fmt.Println("no found row")
}else{
fmt.Printf("%+v",row)
}
}
func checkErr(err error){
if err!=nil {
panic(err)
}
}
output:
&map[col1:abc col2:123]