Merge pull request #5998 from thaJeztah/lazy_regexp

use lazyregexp to compile regexes on first use

Author: thaJeztah

.golangci.yml

@@ -7,6 +7,7 @@ linters:
     - dupword       # Detects duplicate words.
     - durationcheck
     - errchkjson
+    - forbidigo
     - gocritic      # Metalinter; detects bugs, performance, and styling issues.
     - gocyclo
     - gofumpt       # Detects whether code was gofumpt-ed.
@@ -66,6 +67,11 @@ linters-settings:
             desc: Use github.com/google/uuid instead.
           - pkg: "io/ioutil"
             desc: The io/ioutil package has been deprecated, see https://go.dev/doc/go1.16#ioutil
+  forbidigo:
+    forbid:
+      - pkg: ^regexp$
+        p: ^regexp\.MustCompile
+        msg: Use internal/lazyregexp.New instead.
   gocyclo:
     min-complexity: 16
   gosec:

cli-plugins/manager/plugin.go

@@ -8,14 +8,14 @@ import (
 	"os"
 	"os/exec"
 	"path/filepath"
-	"regexp"
 	"strings"
 
 	"github.com/docker/cli/cli-plugins/metadata"
+	"github.com/docker/cli/internal/lazyregexp"
 	"github.com/spf13/cobra"
 )
 
-var pluginNameRe = regexp.MustCompile("^[a-z][a-z0-9]*$")
+var pluginNameRe = lazyregexp.New("^[a-z][a-z0-9]*$")
 
 // Plugin represents a potential plugin with all it's metadata.
 type Plugin struct {

cli/command/container/opts.go

@@ -8,12 +8,12 @@ import (
 	"path"
 	"path/filepath"
 	"reflect"
-	"regexp"
 	"strconv"
 	"strings"
 	"time"
 
 	"github.com/docker/cli/cli/compose/loader"
+	"github.com/docker/cli/internal/lazyregexp"
 	"github.com/docker/cli/opts"
 	"github.com/docker/docker/api/types/container"
 	mounttypes "github.com/docker/docker/api/types/mount"
@@ -40,7 +40,7 @@ const (
 	seccompProfileUnconfined = "unconfined"
 )
 
-var deviceCgroupRuleRegexp = regexp.MustCompile(`^[acb] ([0-9]+|\*):([0-9]+|\*) [rwm]{1,3}$`)
+var deviceCgroupRuleRegexp = lazyregexp.New(`^[acb] ([0-9]+|\*):([0-9]+|\*) [rwm]{1,3}$`)
 
 // containerOptions is a data object with all the options for creating a container
 type containerOptions struct {

cli/command/image/build.go

@@ -10,7 +10,6 @@ import (
 	"io"
 	"os"
 	"path/filepath"
-	"regexp"
 	"runtime"
 	"strings"
 
@@ -23,6 +22,7 @@ import (
 	"github.com/docker/cli/cli/internal/jsonstream"
 	"github.com/docker/cli/cli/streams"
 	"github.com/docker/cli/cli/trust"
+	"github.com/docker/cli/internal/lazyregexp"
 	"github.com/docker/cli/opts"
 	"github.com/docker/docker/api"
 	"github.com/docker/docker/api/types"
@@ -432,7 +432,7 @@ func validateTag(rawRepo string) (string, error) {
 	return rawRepo, nil
 }
 
-var dockerfileFromLinePattern = regexp.MustCompile(`(?i)^[\s]*FROM[ \f\r\t\v]+(?P<image>[^ \f\r\t\v\n#]+)`)
+var dockerfileFromLinePattern = lazyregexp.New(`(?i)^[\s]*FROM[ \f\r\t\v]+(?P<image>[^ \f\r\t\v\n#]+)`)
 
 // resolvedTag records the repository, tag, and resolved digest reference
 // from a Dockerfile rewrite.

cli/command/system/info.go

@@ -8,7 +8,6 @@ import (
 	"errors"
 	"fmt"
 	"io"
-	"regexp"
 	"sort"
 	"strings"
 
@@ -19,6 +18,7 @@ import (
 	"github.com/docker/cli/cli/command/formatter"
 	"github.com/docker/cli/cli/debug"
 	flagsHelper "github.com/docker/cli/cli/flags"
+	"github.com/docker/cli/internal/lazyregexp"
 	"github.com/docker/cli/templates"
 	"github.com/docker/docker/api/types/swarm"
 	"github.com/docker/docker/api/types/system"
@@ -142,7 +142,7 @@ func addServerInfo(ctx context.Context, dockerCli command.Cli, format string, in
 
 // placeHolders does a rudimentary match for possible placeholders in a
 // template, matching a '.', followed by an letter (a-z/A-Z).
-var placeHolders = regexp.MustCompile(`\.[a-zA-Z]`)
+var placeHolders = lazyregexp.New(`\.[a-zA-Z]`)
 
 // needsServerInfo detects if the given template uses any server information.
 // If only client-side information is used in the template, we can skip

cli/command/trust/key_generate.go

@@ -5,12 +5,12 @@ import (
 	"fmt"
 	"os"
 	"path/filepath"
-	"regexp"
 	"strings"
 
 	"github.com/docker/cli/cli"
 	"github.com/docker/cli/cli/command"
 	"github.com/docker/cli/cli/trust"
+	"github.com/docker/cli/internal/lazyregexp"
 	"github.com/pkg/errors"
 	"github.com/spf13/cobra"
 	"github.com/theupdateframework/notary"
@@ -41,7 +41,7 @@ func newKeyGenerateCommand(dockerCli command.Streams) *cobra.Command {
 }
 
 // key names can use lowercase alphanumeric + _ + - characters
-var validKeyName = regexp.MustCompile(`^[a-z0-9][a-z0-9\_\-]*$`).MatchString
+var validKeyName = lazyregexp.New(`^[a-z0-9][a-z0-9\_\-]*$`).MatchString
 
 // validate that all of the key names are unique and are alphanumeric + _ + -
 // and that we do not already have public key files in the target dir on disk

cli/command/trust/signer_add.go

@@ -6,13 +6,13 @@ import (
 	"io"
 	"os"
 	"path"
-	"regexp"
 	"strings"
 
 	"github.com/docker/cli/cli"
 	"github.com/docker/cli/cli/command"
 	"github.com/docker/cli/cli/command/image"
 	"github.com/docker/cli/cli/trust"
+	"github.com/docker/cli/internal/lazyregexp"
 	"github.com/docker/cli/opts"
 	"github.com/pkg/errors"
 	"github.com/spf13/cobra"
@@ -45,7 +45,7 @@ func newSignerAddCommand(dockerCLI command.Cli) *cobra.Command {
 	return cmd
 }
 
-var validSignerName = regexp.MustCompile(`^[a-z0-9][a-z0-9\_\-]*$`).MatchString
+var validSignerName = lazyregexp.New(`^[a-z0-9][a-z0-9\_\-]*$`).MatchString
 
 func addSigner(ctx context.Context, dockerCLI command.Cli, options signerAddOptions) error {
 	signerName := options.signer

cli/compose/template/template.go

@@ -7,18 +7,30 @@ import (
 	"fmt"
 	"regexp"
 	"strings"
+
+	"github.com/docker/cli/internal/lazyregexp"
 )
 
 const (
 	delimiter = "\\$"
 	subst     = "[_a-z][_a-z0-9]*(?::?[-?][^}]*)?"
 )
 
-var defaultPattern = regexp.MustCompile(fmt.Sprintf(
+var defaultPattern = lazyregexp.New(fmt.Sprintf(
 	"%s(?i:(?P<escaped>%s)|(?P<named>%s)|{(?P<braced>%s)}|(?P<invalid>))",
 	delimiter, delimiter, subst, subst,
 ))
 
+// regexper is an internal interface to allow passing a [lazyregexp.Regexp]
+// in places where a custom ("regular") [regexp.Regexp] is accepted. It defines
+// only the methods we currently use.
+type regexper interface {
+	FindAllStringSubmatch(s string, n int) [][]string
+	FindStringSubmatch(s string) []string
+	ReplaceAllStringFunc(src string, repl func(string) string) string
+	SubexpNames() []string
+}
+
 // DefaultSubstituteFuncs contains the default SubstituteFunc used by the docker cli
 var DefaultSubstituteFuncs = []SubstituteFunc{
 	softDefault,
@@ -51,10 +63,16 @@ type SubstituteFunc func(string, Mapping) (string, bool, error)
 // SubstituteWith substitutes variables in the string with their values.
 // It accepts additional substitute function.
 func SubstituteWith(template string, mapping Mapping, pattern *regexp.Regexp, subsFuncs ...SubstituteFunc) (string, error) {
+	return substituteWith(template, mapping, pattern, subsFuncs...)
+}
+
+// SubstituteWith substitutes variables in the string with their values.
+// It accepts additional substitute function.
+func substituteWith(template string, mapping Mapping, pattern regexper, subsFuncs ...SubstituteFunc) (string, error) {
 	var err error
 	result := pattern.ReplaceAllStringFunc(template, func(substring string) string {
 		matches := pattern.FindStringSubmatch(substring)
-		groups := matchGroups(matches, pattern)
+		groups := matchGroups(matches, defaultPattern)
 		if escaped := groups["escaped"]; escaped != "" {
 			return escaped
 		}
@@ -93,38 +111,42 @@ func SubstituteWith(template string, mapping Mapping, pattern *regexp.Regexp, su
 
 // Substitute variables in the string with their values
 func Substitute(template string, mapping Mapping) (string, error) {
-	return SubstituteWith(template, mapping, defaultPattern, DefaultSubstituteFuncs...)
+	return substituteWith(template, mapping, defaultPattern, DefaultSubstituteFuncs...)
 }
 
 // ExtractVariables returns a map of all the variables defined in the specified
 // composefile (dict representation) and their default value if any.
 func ExtractVariables(configDict map[string]any, pattern *regexp.Regexp) map[string]string {
+	return extractVariables(configDict, pattern)
+}
+
+func extractVariables(configDict map[string]any, pattern regexper) map[string]string {
 	if pattern == nil {
 		pattern = defaultPattern
 	}
 	return recurseExtract(configDict, pattern)
 }
 
-func recurseExtract(value any, pattern *regexp.Regexp) map[string]string {
+func recurseExtract(value any, pattern regexper) map[string]string {
 	m := map[string]string{}
 
-	switch value := value.(type) {
+	switch val := value.(type) {
 	case string:
-		if values, is := extractVariable(value, pattern); is {
+		if values, is := extractVariable(val, pattern); is {
 			for _, v := range values {
 				m[v.name] = v.value
 			}
 		}
 	case map[string]any:
-		for _, elem := range value {
+		for _, elem := range val {
 			submap := recurseExtract(elem, pattern)
-			for key, value := range submap {
-				m[key] = value
+			for k, v := range submap {
+				m[k] = v
 			}
 		}
 
 	case []any:
-		for _, elem := range value {
+		for _, elem := range val {
 			if values, is := extractVariable(elem, pattern); is {
 				for _, v := range values {
 					m[v.name] = v.value
@@ -141,7 +163,7 @@ type extractedValue struct {
 	value string
 }
 
-func extractVariable(value any, pattern *regexp.Regexp) ([]extractedValue, bool) {
+func extractVariable(value any, pattern regexper) ([]extractedValue, bool) {
 	sValue, ok := value.(string)
 	if !ok {
 		return []extractedValue{}, false
@@ -227,7 +249,7 @@ func withRequired(substitution string, mapping Mapping, sep string, valid func(s
 	return value, true, nil
 }
 
-func matchGroups(matches []string, pattern *regexp.Regexp) map[string]string {
+func matchGroups(matches []string, pattern regexper) map[string]string {
 	groups := make(map[string]string)
 	for i, name := range pattern.SubexpNames()[1:] {
 		groups[name] = matches[i+1]

cli/compose/template/template_test.go

@@ -169,15 +169,15 @@ func TestSubstituteWithCustomFunc(t *testing.T) {
 		return value, true, nil
 	}
 
-	result, err := SubstituteWith("ok ${FOO}", defaultMapping, defaultPattern, errIsMissing)
+	result, err := substituteWith("ok ${FOO}", defaultMapping, defaultPattern, errIsMissing)
 	assert.NilError(t, err)
 	assert.Check(t, is.Equal("ok first", result))
 
-	result, err = SubstituteWith("ok ${BAR}", defaultMapping, defaultPattern, errIsMissing)
+	result, err = substituteWith("ok ${BAR}", defaultMapping, defaultPattern, errIsMissing)
 	assert.NilError(t, err)
 	assert.Check(t, is.Equal("ok ", result))
 
-	_, err = SubstituteWith("ok ${NOTHERE}", defaultMapping, defaultPattern, errIsMissing)
+	_, err = substituteWith("ok ${NOTHERE}", defaultMapping, defaultPattern, errIsMissing)
 	assert.Check(t, is.ErrorContains(err, "required variable"))
 }
 
@@ -278,7 +278,7 @@ func TestExtractVariables(t *testing.T) {
 	}
 	for _, tc := range testCases {
 		t.Run(tc.name, func(t *testing.T) {
-			actual := ExtractVariables(tc.dict, defaultPattern)
+			actual := extractVariables(tc.dict, defaultPattern)
 			assert.Check(t, is.DeepEqual(actual, tc.expected))
 		})
 	}

cli/context/store/store.go

@@ -14,17 +14,17 @@ import (
 	"net/http"
 	"path"
 	"path/filepath"
-	"regexp"
 	"strings"
 
+	"github.com/docker/cli/internal/lazyregexp"
 	"github.com/docker/docker/errdefs"
 	"github.com/opencontainers/go-digest"
 	"github.com/pkg/errors"
 )
 
 const restrictedNamePattern = "^[a-zA-Z0-9][a-zA-Z0-9_.+-]+$"
 
-var restrictedNameRegEx = regexp.MustCompile(restrictedNamePattern)
+var restrictedNameRegEx = lazyregexp.New(restrictedNamePattern)
 
 // Store provides a context store for easily remembering endpoints configuration
 type Store interface {