gendry is a Go library that helps you operate database. Based on go-sql-driver/mysql, it provides a series of simple but useful tools to prepare parameters for calling methods in standard library database/sql.
The name gendry comes from the role in the hottest drama The Game of Throne, in which Gendry is not only the bastardy of the late king Robert Baratheon but also a skilled blacksmith. Like the one in drama, this library also forge something which is called SQL.
gendry consists of three isolated parts, and you can use each one of them partially:
manager is used for initializing database connection pool(i.e sql.DB),
you can set almost all parameters for those mysql driver supported.For example, initializing a database connection pool:
var db *sql.DB
var err error
db, err = manager
.New(dbName, user, password, host)
.Set(
manager.SetCharset("utf8"),
manager.SetAllowCleartextPasswords(true),
manager.SetInterpolateParams(true),
manager.SetTimeout(1 * time.Second),
manager.SetReadTimeout(1 * time.Second)
).Port(3302).Open(true)In fact, all things manager does is just for concatting the dataSourceName
the format of a dataSourceName is：
[username[:password]@][protocol[(address)]]/dbname[?param1=value1&...¶mN=valueN]
manager is based on go-mysql-driver/mysql, and if you don't know some of the manager.SetXXX series functions, see it on mysql driver's github home page.And for more details see manager's doc
builder isn't an ORM, in fact one of the most important reasons we create Gendry is we don't like ORM. So Gendry just provides some simple APIs to help you building sqls:
where := map[string]interface{}{
"city": []string{"beijing", "shanghai"},
// The in operator can be omitted by default,
// which is equivalent to:
// "city in": []string{"beijing", "shanghai"},
"score": 5,
"age >": 35,
"address": builder.IsNotNull,
"_or": []map[string]interface{}{
{
"x1": 11,
"x2 >=": 45,
},
{
"x3": "234",
"x4 <>": "tx2",
},
},
"_orderby": "bonus desc",
"_groupby": "department",
}
table := "some_table"
selectFields := []string{"name", "age", "sex"}
cond, values, err := builder.BuildSelect(table, where, selectFields)
//cond = SELECT name,age,sex FROM some_table WHERE (((x1=? AND x2>=?) OR (x3=? AND x4!=?)) AND score=? AND city IN (?,?) AND age>? AND address IS NOT NULL) GROUP BY department ORDER BY bonus DESC
//values = []interface{}{11, 45, "234", "tx2", 5, "beijing", "shanghai", 35}
rows, err := db.Query(cond, values...)In the where param, automatically add 'in' operator by value type(reflect.Slice).
where := map[string]interface{}{
"city": []string{"beijing", "shanghai"},
}the same as
where := map[string]interface{}{
"city in": []string{"beijing", "shanghai"},
}And, the library provide a useful API for executing aggregate queries like count, sum, max, min, avg
where := map[string]interface{}{
"score > ": 100,
"city": []interface{}{"Beijing", "Shijiazhuang", }
}
// AggregateSum, AggregateMax, AggregateMin, AggregateCount, AggregateAvg are supported
result, err := AggregateQuery(ctx, db, "tableName", where, AggregateSum("age"))
sumAge := result.Int64()
result, err = AggregateQuery(ctx, db, "tableName", where, AggregateCount("*"))
numberOfRecords := result.Int64()
result, err = AggregateQuery(ctx, db, "tableName", where, AggregateAvg("score"))
averageScore := result.Float64()If you want to clear the zero value in the where map, you can use builder.OmitEmpty
where := map[string]interface{}{
"score": 0,
"age >": 35,
}
finalWhere := builder.OmitEmpty(where, []string{"score", "age"})
// finalWhere = map[string]interface{}{"age >": 35}
// support: Bool, Array, String, Float32, Float64, Int, Int8, Int16, Int32, Int64, Uint, Uint8, Uint16, Uint32, Uint64, Uintptr, Map, Slice, Interface, StructFor complex queries, NamedQuery may be helpful:
cond, vals, err := builder.NamedQuery("select * from tb where name={{name}} and id in (select uid from anothertable where score in {{m_score}})", map[string]interface{}{
"name": "caibirdme",
"m_score": []float64{3.0, 5.8, 7.9},
})
assert.Equal("select * from tb where name=? and id in (select uid from anothertable where score in (?,?,?))", cond)
assert.Equal([]interface{}{"caibirdme", 3.0, 5.8, 7.9}, vals)slice type can be expanded automatically according to its length, thus these sqls are very convenient for DBA to review.
For critical system, this is recommended
For more detail, see builder's doc or just use godoc
type Person struct {
Name string
Age int
}
rows, err := db.Query("SELECT age as m_age, name from g_xxx where xxx")
defer rows.Close()
var students []Person
for rows.Next() {
var student Person
rows.Scan(student.Age, student.Name)
students = append(students, student)
}type Person struct {
Name string `ddb:"name"`
Age int `ddb:"m_age"`
}
rows, err := db.Query("SELECT age as m_age, name from g_xxx where xxx")
defer rows.Close()
var students []Person
scanner.Scan(rows, &students)Types which implement the interface
type ByteUnmarshaler interface {
UnmarshalByte(data []byte) error
}will take over the corresponding unmarshal work.
type human struct {
Age int `ddb:"ag"`
Extra *extraInfo `ddb:"ext"`
}
type extraInfo struct {
Hobbies []string `json:"hobbies"`
LuckyNumber int `json:"ln"`
}
func (ext *extraInfo) UnmarshalByte(data []byte) error {
return json.Unmarshal(data, ext)
}
//if the type of ext column in a table is varchar(stored legal json string) or json(mysql5.7)
var student human
err := scanner.Scan(rows, &student)
// ...The extra tag of the struct will be used by scanner resolve data from response.The default tag name is ddb:"tagname", but you can specify your own such as:
scanner.SetTagName("json")
type Person struct {
Name string `json:"name"`
Age int `json:"m_age"`
}
// ...
var student Person
scanner.Scan(rows, &student)scanner.SetTagName is a global setting and it can be invoked only once
rows, _ := db.Query("select name, age as m_age from person")
result, err := scanner.ScanMap(rows)
for _, record := range result {
fmt.Println(record["name"], record["m_age"])
}ScanMap scans data from rows and returns a []map[string]interface{}
int, float, string type may be stored as []uint8 by mysql driver, ScanMap just copy those value into the map. If you're sure that there's no binary data type in your mysql table(in most cases, this is true), you can use ScanMapDecode instead which will convert []uint8 to int, float64 or string
For more detail, see scanner's doc
PS：
- Don't forget close rows if you don't use ScanXXXClose
- The second parameter of Scan must be a reference