Create pkg/regexp to better handle init regex

The new Reexp structure, will handle the initialization of regexp
and then will do the MustCompile within a sync.Once field.

Signed-off-by: Daniel J Walsh <dwalsh@redhat.com>

pkg/idtools/usergroupadd_linux.go

@@ -2,11 +2,12 @@ package idtools
 
 import (
 	"fmt"
-	"regexp"
 	"sort"
 	"strconv"
 	"strings"
 	"sync"
+
+	"github.com/containers/storage/pkg/regexp"
 )
 
 // add a user and/or group to Linux /etc/passwd, /etc/group using standard
@@ -24,8 +25,7 @@ var (
 		"usermod": "-%s %d-%d %s",
 	}
 
-	idOutOnce   sync.Once
-	idOutRegexp *regexp.Regexp
+	idOutRegexp = regexp.New(`uid=([0-9]+).*gid=([0-9]+)`)
 	// default length for a UID/GID subordinate range
 	defaultRangeLen   = 65536
 	defaultRangeStart = 100000
@@ -37,10 +37,6 @@ var (
 // /etc/sub{uid,gid} ranges which will be used for user namespace
 // mapping ranges in containers.
 func AddNamespaceRangesUser(name string) (int, int, error) {
-	idOutOnce.Do(func() {
-		idOutRegexp = regexp.MustCompile(`uid=([0-9]+).*gid=([0-9]+)`)
-	})
-
 	if err := addUser(name); err != nil {
 		return -1, -1, fmt.Errorf("adding user %q: %w", name, err)
 	}

pkg/regexp/regexp.go

@@ -0,0 +1,206 @@
+package regexp
+
+import (
+	"io"
+	"regexp"
+	"sync"
+)
+
+// Regexp is a wrapper struct used for wrapping MustCompile regex expressions
+// used init. Using this stucture should help speed up startup time of apps
+// that want to initialize their regex at init time.
+type Regexp struct {
+	once   sync.Once
+	regexp *regexp.Regexp
+	val    string
+}
+
+func New(val string) Regexp {
+	return Regexp{
+		val: val,
+	}
+}
+
+func (re *Regexp) compile() {
+	re.once.Do(func() {
+		re.regexp = regexp.MustCompile(re.val)
+	})
+}
+
+func (re *Regexp) Expand(dst []byte, template []byte, src []byte, match []int) []byte {
+	re.compile()
+	return re.regexp.Expand(dst, template, src, match)
+}
+
+func (re *Regexp) ExpandString(dst []byte, template string, src string, match []int) []byte {
+	re.compile()
+	return re.regexp.ExpandString(dst, template, src, match)
+}
+func (re *Regexp) Find(b []byte) []byte {
+	re.compile()
+	return re.regexp.Find(b)
+}
+
+func (re *Regexp) FindAll(b []byte, n int) [][]byte {
+	re.compile()
+	return re.regexp.FindAll(b, n)
+}
+
+func (re *Regexp) FindAllIndex(b []byte, n int) [][]int {
+	re.compile()
+	return re.regexp.FindAllIndex(b, n)
+}
+
+func (re *Regexp) FindAllString(s string, n int) []string {
+	re.compile()
+	return re.regexp.FindAllString(s, n)
+}
+
+func (re *Regexp) FindAllStringIndex(s string, n int) [][]int {
+	re.compile()
+	return re.regexp.FindAllStringIndex(s, n)
+}
+
+func (re *Regexp) FindAllStringSubmatch(s string, n int) [][]string {
+	re.compile()
+	return re.regexp.FindAllStringSubmatch(s, n)
+}
+
+func (re *Regexp) FindAllStringSubmatchIndex(s string, n int) [][]int {
+	re.compile()
+	return re.regexp.FindAllStringSubmatchIndex(s, n)
+}
+
+func (re *Regexp) FindAllSubmatch(b []byte, n int) [][][]byte {
+	re.compile()
+	return re.regexp.FindAllSubmatch(b, n)
+}
+
+func (re *Regexp) FindAllSubmatchIndex(b []byte, n int) [][]int {
+	re.compile()
+	return re.regexp.FindAllSubmatchIndex(b, n)
+}
+
+func (re *Regexp) FindIndex(b []byte) (loc []int) {
+	re.compile()
+	return re.regexp.FindIndex(b)
+}
+
+func (re *Regexp) FindReaderIndex(r io.RuneReader) (loc []int) {
+	re.compile()
+	return re.regexp.FindReaderIndex(r)
+}
+
+func (re *Regexp) FindReaderSubmatchIndex(r io.RuneReader) []int {
+	re.compile()
+	return re.regexp.FindReaderSubmatchIndex(r)
+}
+
+func (re *Regexp) FindString(s string) string {
+	re.compile()
+	return re.regexp.FindString(s)
+}
+
+func (re *Regexp) FindStringIndex(s string) (loc []int) {
+	re.compile()
+	return re.regexp.FindStringIndex(s)
+}
+
+func (re *Regexp) FindStringSubmatch(s string) []string {
+	re.compile()
+	return re.regexp.FindStringSubmatch(s)
+}
+
+func (re *Regexp) FindStringSubmatchIndex(s string) []int {
+	re.compile()
+	return re.regexp.FindStringSubmatchIndex(s)
+}
+
+func (re *Regexp) FindSubmatch(b []byte) [][]byte {
+	re.compile()
+	return re.regexp.FindSubmatch(b)
+}
+
+func (re *Regexp) FindSubmatchIndex(b []byte) []int {
+	re.compile()
+	return re.regexp.FindSubmatchIndex(b)
+}
+
+func (re *Regexp) LiteralPrefix() (prefix string, complete bool) {
+	re.compile()
+	return re.regexp.LiteralPrefix()
+}
+
+func (re *Regexp) Longest() {
+	re.compile()
+	re.regexp.Longest()
+}
+
+func (re *Regexp) Match(b []byte) bool {
+	re.compile()
+	return re.regexp.Match(b)
+}
+
+func (re *Regexp) MatchReader(r io.RuneReader) bool {
+	re.compile()
+	return re.regexp.MatchReader(r)
+}
+func (re *Regexp) MatchString(s string) bool {
+	re.compile()
+	return re.regexp.MatchString(s)
+}
+
+func (re *Regexp) NumSubexp() int {
+	re.compile()
+	return re.regexp.NumSubexp()
+}
+
+func (re *Regexp) ReplaceAll(src, repl []byte) []byte {
+	re.compile()
+	return re.regexp.ReplaceAll(src, repl)
+}
+
+func (re *Regexp) ReplaceAllFunc(src []byte, repl func([]byte) []byte) []byte {
+	re.compile()
+	return re.regexp.ReplaceAllFunc(src, repl)
+}
+
+func (re *Regexp) ReplaceAllLiteral(src, repl []byte) []byte {
+	re.compile()
+	return re.regexp.ReplaceAllLiteral(src, repl)
+}
+
+func (re *Regexp) ReplaceAllLiteralString(src, repl string) string {
+	re.compile()
+	return re.regexp.ReplaceAllLiteralString(src, repl)
+}
+
+func (re *Regexp) ReplaceAllString(src, repl string) string {
+	re.compile()
+	return re.regexp.ReplaceAllString(src, repl)
+}
+
+func (re *Regexp) ReplaceAllStringFunc(src string, repl func(string) string) string {
+	re.compile()
+	return re.regexp.ReplaceAllStringFunc(src, repl)
+}
+
+func (re *Regexp) Split(s string, n int) []string {
+	re.compile()
+	return re.regexp.Split(s, n)
+}
+
+func (re *Regexp) String() string {
+	re.compile()
+	return re.regexp.String()
+}
+
+func (re *Regexp) SubexpIndex(name string) int {
+	re.compile()
+	return re.regexp.SubexpIndex(name)
+}
+
+func (re *Regexp) SubexpNames() []string {
+	re.compile()
+	return re.regexp.SubexpNames()
+}

pkg/regexp/regexp_test.go

@@ -0,0 +1,29 @@
+package regexp
+
+import (
+	"testing"
+)
+
+func TestMatchString(t *testing.T) {
+	r := New(`[0-9]+`)
+
+	if !r.MatchString("100") {
+		t.Fatalf("Should have matched")
+	}
+
+	if r.MatchString("test") {
+		t.Fatalf("Should not have matched")
+	}
+}
+
+func TestFindStringSubmatch(t *testing.T) {
+	r := New(`a=([0-9]+).*b=([0-9]+)`)
+
+	if len(r.FindStringSubmatch("a=1,b=2")) != 3 {
+		t.Fatalf("Should have matched 3")
+	}
+
+	if len(r.FindStringSubmatch("a=1")) != 0 {
+		t.Fatalf("Should not have matched 0")
+	}
+}

pkg/stringid/stringid.go

@@ -9,35 +9,26 @@ import (
 	"math"
 	"math/big"
 	"math/rand"
-	"regexp"
 	"strconv"
 	"strings"
 	"sync"
 	"time"
+
+	"github.com/containers/storage/pkg/regexp"
 )
 
 const shortLen = 12
 
 var (
-	validShortID *regexp.Regexp
-	validHex     *regexp.Regexp
-	onceRegex    sync.Once
+	validShortID = regexp.New("^[a-f0-9]{12}$")
+	validHex     = regexp.New(`^[a-f0-9]{64}$`)
 
 	rngLock sync.Mutex
 	rng     *rand.Rand // A RNG with seeding properties we control. It can only be accessed with randLock held.
 )
 
-func initRegex() {
-	onceRegex.Do(func() {
-		validShortID = regexp.MustCompile("^[a-f0-9]{12}$")
-		validHex = regexp.MustCompile(`^[a-f0-9]{64}$`)
-	})
-}
-
 // IsShortID determines if an arbitrary string *looks like* a short ID.
 func IsShortID(id string) bool {
-	initRegex()
-
 	return validShortID.MatchString(id)
 }
 
@@ -88,8 +79,6 @@ func GenerateNonCryptoID() string {
 
 // ValidateID checks whether an ID string is a valid image ID.
 func ValidateID(id string) error {
-	initRegex()
-
 	if ok := validHex.MatchString(id); !ok {
 		return fmt.Errorf("image ID %q is invalid", id)
 	}