Allow regex to specify databases to dump

conf/mydumper.ini.sample

@@ -17,6 +17,13 @@ chunksize = 128
 # vars= "xx=xx;xx=xx;"
 vars= ""
 
+# Use this to use regexp to control what databases to export. These are optional
+[database]
+# regexp = ^(mysql|sys|information_schema|performance_schema)$
+# As the used regexp lib does not allow for lookarounds, you may use this to invert the whole regexp
+# This option should be refactored as soon as a GPLv3 compliant go-pcre lib is found
+# invert_regexp = on
+
 # Use this to restrict exported data. These are optional
 [where]
 # sample_table1 = created_at >= DATE_SUB(NOW(), INTERVAL 7 DAY)

src/cmd/config.go

@@ -84,10 +84,18 @@ func parseDumperConfig(file string) (*common.Args, error) {
 		}
 	}
 
+	database_regexp, _ := cfg.GetString("database", "regexp")
+	database_invert_regexp, err := cfg.GetBool("database", "invert_regexp")
+	if err != nil {
+		database_invert_regexp = false
+	}
+
 	args.Address = fmt.Sprintf("%s:%d", host, port)
 	args.User = user
 	args.Password = password
 	args.Database = database
+	args.DatabaseRegexp = database_regexp
+	args.DatabaseInvertRegexp = database_invert_regexp
 	args.Table = table
 	args.Outdir = outdir
 	args.ChunksizeInMB = chunksizemb

src/common/common.go

@@ -19,26 +19,28 @@ import (
 
 // Args tuple.
 type Args struct {
-	User            string
-	Password        string
-	Address         string
-	ToUser          string
-	ToPassword      string
-	ToAddress       string
-	ToDatabase      string
-	ToEngine        string
-	Database        string
-	Table           string
-	Outdir          string
-	SessionVars     string
-	Threads         int
-	ChunksizeInMB   int
-	StmtSize        int
-	Allbytes        uint64
-	Allrows         uint64
-	OverwriteTables bool
-	Wheres          map[string]string
-	Selects         map[string]map[string]string
+	User                 string
+	Password             string
+	Address              string
+	ToUser               string
+	ToPassword           string
+	ToAddress            string
+	ToDatabase           string
+	ToEngine             string
+	Database             string
+	DatabaseRegexp       string
+	DatabaseInvertRegexp bool
+	Table                string
+	Outdir               string
+	SessionVars          string
+	Threads              int
+	ChunksizeInMB        int
+	StmtSize             int
+	Allbytes             uint64
+	Allrows              uint64
+	OverwriteTables      bool
+	Wheres               map[string]string
+	Selects              map[string]map[string]string
 
 	// Interval in millisecond.
 	IntervalMs int

src/common/dumper.go

@@ -11,6 +11,7 @@ package common
 
 import (
 	"fmt"
+	"regexp"
 	"strings"
 	"sync"
 	"sync/atomic"
@@ -166,6 +167,19 @@ func allDatabases(log *xlog.Log, conn *Connection) []string {
 	return databases
 }
 
+func filterDatabases(log *xlog.Log, conn *Connection, filter *regexp.Regexp, invert bool) []string {
+	qr, err := conn.Fetch("SHOW DATABASES")
+	AssertNil(err)
+
+	databases := make([]string, 0, 128)
+	for _, t := range qr.Rows {
+		if (!invert && filter.MatchString(t[0].String())) || (invert && !filter.MatchString(t[0].String())) {
+			databases = append(databases, t[0].String())
+		}
+	}
+	return databases
+}
+
 // Dumper used to start the dumper worker.
 func Dumper(log *xlog.Log, args *Args) {
 	pool, err := NewPool(log, args.Threads, args.Address, args.User, args.Password, args.SessionVars)
@@ -180,10 +194,15 @@ func Dumper(log *xlog.Log, args *Args) {
 	conn := pool.Get()
 	var databases []string
 	t := time.Now()
-	if args.Database != "" {
-		databases = strings.Split(args.Database, ",")
+	if args.DatabaseRegexp != "" {
+		r := regexp.MustCompile(args.DatabaseRegexp)
+		databases = filterDatabases(log, conn, r, args.DatabaseInvertRegexp)
 	} else {
-		databases = allDatabases(log, conn)
+		if args.Database != "" {
+			databases = strings.Split(args.Database, ",")
+		} else {
+			databases = allDatabases(log, conn)
+		}
 	}
 	for _, database := range databases {
 		dumpDatabaseSchema(log, conn, args, database)