chore: lazy load regexes to save memory.

collection/collection.go

@@ -4,8 +4,7 @@
 package collection
 
 import (
-	"regexp"
-
+	"github.com/corazawaf/coraza/v3/internal/regexp"
 	"github.com/corazawaf/coraza/v3/types"
 )
 
@@ -37,7 +36,7 @@ type Keyed interface {
 	Get(key string) []string
 
 	// FindRegex returns a slice of MatchData for the regex
-	FindRegex(key *regexp.Regexp) []types.MatchData
+	FindRegex(key regexp.Regexp) []types.MatchData
 
 	// FindString returns a slice of MatchData for the string
 	FindString(key string) []types.MatchData

experimental/regexp/regexp.go

@@ -0,0 +1,28 @@
+// Copyright 2023 Juan Pablo Tosso and the OWASP Coraza contributors
+// SPDX-License-Identifier: Apache-2.0
+
+package experimental
+
+import (
+	"fmt"
+
+	"github.com/corazawaf/coraza/v3/experimental/regexp/regexptypes"
+	"github.com/corazawaf/coraza/v3/internal/regexp"
+)
+
+// SetRegexpCompiler sets the regex compiler used by the WAF. This is specially
+// useful when we want to lazily compile regexes in a mono thread environment as
+// we don't need to synchronize the regex compilation.
+func SetRegexpCompiler(fn func(expr string) (regexptypes.Regexp, error)) {
+	if fn == nil {
+		fmt.Println("invalid regex compiler")
+		return
+	}
+
+	if regexp.RegexCompiler != nil {
+		fmt.Println("regex compiler already set")
+		return
+	}
+
+	regexp.RegexCompiler = fn
+}

experimental/regexp/regexptypes/types.go

@@ -0,0 +1,19 @@
+// Copyright 2023 Juan Pablo Tosso and the OWASP Coraza contributors
+// SPDX-License-Identifier: Apache-2.0
+
+package regexptypes
+
+import "regexp"
+
+// Regexp is the interface that wraps the basic MatchString, FindStringSubmatch,
+// FindAllStringSubmatch, SubexpNames, Match and String methods.
+type Regexp interface {
+	MatchString(s string) bool
+	FindStringSubmatch(s string) []string
+	FindAllStringSubmatch(s string, n int) [][]string
+	SubexpNames() []string
+	Match(s []byte) bool
+	String() string
+}
+
+var _ Regexp = (*regexp.Regexp)(nil)

internal/collections/concat.go

@@ -4,11 +4,11 @@
 package collections
 
 import (
-	"regexp"
 	"strings"
 
 	"github.com/corazawaf/coraza/v3/collection"
 	"github.com/corazawaf/coraza/v3/internal/corazarules"
+	"github.com/corazawaf/coraza/v3/internal/regexp"
 	"github.com/corazawaf/coraza/v3/types"
 	"github.com/corazawaf/coraza/v3/types/variables"
 )
@@ -67,7 +67,7 @@ func (c *ConcatKeyed) Get(key string) []string {
 }
 
 // FindRegex returns a slice of MatchData for the regex
-func (c *ConcatKeyed) FindRegex(key *regexp.Regexp) []types.MatchData {
+func (c *ConcatKeyed) FindRegex(key regexp.Regexp) []types.MatchData {
 	var res []types.MatchData
 	for _, d := range c.data {
 		res = append(res, replaceVariable(c.variable, d.FindRegex(key))...)

internal/collections/concat_test.go

@@ -4,10 +4,10 @@
 package collections
 
 import (
-	"regexp"
 	"strings"
 	"testing"
 
+	"github.com/corazawaf/coraza/v3/internal/regexp"
 	"github.com/corazawaf/coraza/v3/types"
 	"github.com/corazawaf/coraza/v3/types/variables"
 )

internal/collections/map.go

@@ -4,11 +4,11 @@
 package collections
 
 import (
-	"regexp"
 	"strings"
 
 	"github.com/corazawaf/coraza/v3/collection"
 	"github.com/corazawaf/coraza/v3/internal/corazarules"
+	"github.com/corazawaf/coraza/v3/internal/regexp"
 	"github.com/corazawaf/coraza/v3/types"
 	"github.com/corazawaf/coraza/v3/types/variables"
 )
@@ -40,7 +40,7 @@ func (c *Map) Get(key string) []string {
 	return values
 }
 
-func (c *Map) FindRegex(key *regexp.Regexp) []types.MatchData {
+func (c *Map) FindRegex(key regexp.Regexp) []types.MatchData {
 	var result []types.MatchData
 	for k, data := range c.data {
 		if key.MatchString(k) {

internal/collections/map_test.go

@@ -15,9 +15,9 @@ package collections
 
 import (
 	"fmt"
-	"regexp"
 	"testing"
 
+	"github.com/corazawaf/coraza/v3/internal/regexp"
 	"github.com/corazawaf/coraza/v3/types/variables"
 )
 

internal/collections/named.go

@@ -5,11 +5,11 @@
 
 import (
 	"fmt"
-	"regexp"
 	"strings"
 
 	"github.com/corazawaf/coraza/v3/collection"
 	"github.com/corazawaf/coraza/v3/internal/corazarules"
+	"github.com/corazawaf/coraza/v3/internal/regexp"
 	"github.com/corazawaf/coraza/v3/types"
 	"github.com/corazawaf/coraza/v3/types/variables"
 )
@@ -94,7 +94,7 @@
 	collection *NamedCollection
 }
 
-func (c *NamedCollectionNames) FindRegex(key *regexp.Regexp) []types.MatchData {
+func (c *NamedCollectionNames) FindRegex(key regexp.Regexp) []types.MatchData {
 	panic("selection operator not supported")
 }
 

internal/collections/named_test.go

@@ -5,9 +5,9 @@ package collections
 
 import (
 	"fmt"
-	"regexp"
 	"testing"
 
+	"github.com/corazawaf/coraza/v3/internal/regexp"
 	"github.com/corazawaf/coraza/v3/types/variables"
 )
 

internal/collections/sized.go

@@ -5,12 +5,12 @@
 
 import (
 	"fmt"
-	"regexp"
 	"strconv"
 	"strings"
 
 	"github.com/corazawaf/coraza/v3/collection"
 	"github.com/corazawaf/coraza/v3/internal/corazarules"
+	"github.com/corazawaf/coraza/v3/internal/regexp"
 	"github.com/corazawaf/coraza/v3/types"
 	"github.com/corazawaf/coraza/v3/types/variables"
 )
@@ -32,7 +32,7 @@
 }
 
 // FindRegex returns a slice of MatchData for the regex
-func (c *SizeCollection) FindRegex(*regexp.Regexp) []types.MatchData {
+func (c *SizeCollection) FindRegex(regexp.Regexp) []types.MatchData {
 	return c.FindAll()
 }
 

internal/corazawaf/rule.go

@@ -6,7 +6,6 @@ package corazawaf
 import (
 	"fmt"
 	"reflect"
-	"regexp"
 	"strconv"
 	"strings"
 	"sync"
@@ -16,6 +15,7 @@ import (
 	"github.com/corazawaf/coraza/v3/experimental/plugins/plugintypes"
 	"github.com/corazawaf/coraza/v3/internal/corazarules"
 	"github.com/corazawaf/coraza/v3/internal/memoize"
+	"github.com/corazawaf/coraza/v3/internal/regexp"
 	"github.com/corazawaf/coraza/v3/types"
 	"github.com/corazawaf/coraza/v3/types/variables"
 )
@@ -50,7 +50,7 @@ type ruleVariableException struct {
 
 	// The key for the variable that is going to be requested
 	// If nil, KeyStr is going to be used
-	KeyRx *regexp.Regexp
+	KeyRx regexp.Regexp
 }
 
 // RuleVariable is compiled during runtime by transactions
@@ -65,7 +65,7 @@ type ruleVariableParams struct {
 
 	// The key for the variable that is going to be requested
 	// If nil, KeyStr is going to be used
-	KeyRx *regexp.Regexp
+	KeyRx regexp.Regexp
 
 	// The string key for the variable that is going to be requested
 	// If KeyRx is not nil, KeyStr is ignored
@@ -454,14 +454,14 @@ func (r *Rule) AddAction(name string, action plugintypes.Action) error {
 // it will be used to match the variable, in case of string it will
 // be a fixed match, in case of nil it will match everything
 func (r *Rule) AddVariable(v variables.RuleVariable, key string, iscount bool) error {
-	var re *regexp.Regexp
+	var re regexp.Regexp
 	if len(key) > 2 && key[0] == '/' && key[len(key)-1] == '/' {
 		key = key[1 : len(key)-1]
 
 		if vare, err := memoize.Do(key, func() (interface{}, error) { return regexp.Compile(key) }); err != nil {
 			return err
 		} else {
-			re = vare.(*regexp.Regexp)
+			re = vare.(regexp.Regexp)
 		}
 	}
 
@@ -524,13 +524,13 @@ func (r *Rule) AddVariable(v variables.RuleVariable, key string, iscount bool) e
 // OK: SecRule !ARGS:id "..."
 // ERROR: SecRule !ARGS: "..."
 func (r *Rule) AddVariableNegation(v variables.RuleVariable, key string) error {
-	var re *regexp.Regexp
+	var re regexp.Regexp
 	if len(key) > 2 && key[0] == '/' && key[len(key)-1] == '/' {
 		key = key[1 : len(key)-1]
 		if vare, err := memoize.Do(key, func() (interface{}, error) { return regexp.Compile(key) }); err != nil {
 			return err
 		} else {
-			re = vare.(*regexp.Regexp)
+			re = vare.(regexp.Regexp)
 		}
 	}
 	// Prevent sigsev

internal/corazawaf/transaction_test.go

@@ -7,7 +7,6 @@ import (
 	"bytes"
 	"fmt"
 	"io"
-	"regexp"
 	"runtime/debug"
 	"strconv"
 	"strings"
@@ -19,6 +18,7 @@ import (
 	"github.com/corazawaf/coraza/v3/experimental/plugins/plugintypes"
 	"github.com/corazawaf/coraza/v3/internal/collections"
 	"github.com/corazawaf/coraza/v3/internal/corazarules"
+	"github.com/corazawaf/coraza/v3/internal/regexp"
 	utils "github.com/corazawaf/coraza/v3/internal/strings"
 	"github.com/corazawaf/coraza/v3/types"
 	"github.com/corazawaf/coraza/v3/types/variables"

internal/corazawaf/waf.go

@@ -9,7 +9,6 @@ import (
 	"io"
 	"io/fs"
 	"os"
-	"regexp"
 	"strconv"
 	"strings"
 	"time"
@@ -18,6 +17,7 @@ import (
 	"github.com/corazawaf/coraza/v3/experimental/plugins/plugintypes"
 	"github.com/corazawaf/coraza/v3/internal/auditlog"
 	"github.com/corazawaf/coraza/v3/internal/environment"
+	"github.com/corazawaf/coraza/v3/internal/regexp"
 	stringutils "github.com/corazawaf/coraza/v3/internal/strings"
 	"github.com/corazawaf/coraza/v3/internal/sync"
 	"github.com/corazawaf/coraza/v3/types"
@@ -119,7 +119,7 @@ type WAF struct {
 	AuditLogParts types.AuditLogParts
 
 	// Contains the regular expression for relevant status audit logging
-	AuditLogRelevantStatus *regexp.Regexp
+	AuditLogRelevantStatus regexp.Regexp
 
 	auditLogWriter plugintypes.AuditLogWriter
 

internal/operators/restpath.go

@@ -7,11 +7,11 @@ package operators
 
 import (
 	"fmt"
-	"regexp"
 	"strings"
 
 	"github.com/corazawaf/coraza/v3/experimental/plugins/plugintypes"
 	"github.com/corazawaf/coraza/v3/internal/memoize"
+	"github.com/corazawaf/coraza/v3/internal/regexp"
 )
 
 var rePathTokenRe = regexp.MustCompile(`\{([^\}]+)\}`)
@@ -21,7 +21,7 @@ var rePathTokenRe = regexp.MustCompile(`\{([^\}]+)\}`)
 // It will later transform the path to a regex and assign the variables to
 // ARGS_PATH
 type restpath struct {
-	re *regexp.Regexp
+	re regexp.Regexp
 }
 
 var _ plugintypes.Operator = (*restpath)(nil)
@@ -36,7 +36,7 @@ func newRESTPath(options plugintypes.OperatorOptions) (plugintypes.Operator, err
 	if err != nil {
 		return nil, err
 	}
-	return &restpath{re: re.(*regexp.Regexp)}, nil
+	return &restpath{re: re.(regexp.Regexp)}, nil
 }
 
 func (o *restpath) Evaluate(tx plugintypes.TransactionState, value string) bool {

internal/operators/rx.go

@@ -7,18 +7,18 @@ package operators
 
 import (
 	"fmt"
-	"regexp"
 	"strconv"
 	"unicode/utf8"
 
 	"rsc.io/binaryregexp"
 
 	"github.com/corazawaf/coraza/v3/experimental/plugins/plugintypes"
 	"github.com/corazawaf/coraza/v3/internal/memoize"
+	"github.com/corazawaf/coraza/v3/internal/regexp"
 )
 
 type rx struct {
-	re *regexp.Regexp
+	re regexp.Regexp
 }
 
 var _ plugintypes.Operator = (*rx)(nil)
@@ -40,7 +40,7 @@ func newRX(options plugintypes.OperatorOptions) (plugintypes.Operator, error) {
 	if err != nil {
 		return nil, err
 	}
-	return &rx{re: re.(*regexp.Regexp)}, nil
+	return &rx{re: re.(regexp.Regexp)}, nil
 }
 
 func (o *rx) Evaluate(tx plugintypes.TransactionState, value string) bool {

internal/operators/rx_test.go

@@ -5,11 +5,11 @@ package operators
 
 import (
 	"fmt"
-	"regexp"
 	"testing"
 
 	"github.com/corazawaf/coraza/v3/experimental/plugins/plugintypes"
 	"github.com/corazawaf/coraza/v3/internal/corazawaf"
+	"github.com/corazawaf/coraza/v3/internal/regexp"
 )
 
 func TestRx(t *testing.T) {