logging: add support for hashing data (#4434)

* logging: add support for hashing data

* Update modules/logging/filters.go

Co-authored-by: wiese <wiese@users.noreply.github.com>

* Update modules/logging/filters.go

Co-authored-by: wiese <wiese@users.noreply.github.com>

Co-authored-by: wiese <wiese@users.noreply.github.com>

Author: dunglas

caddytest/integration/caddyfile_adapt/log_filters.txt

@@ -8,18 +8,21 @@ log {
 			uri query {
 				replace foo REDACTED
 				delete bar
+				hash baz
 			}
 			request>headers>Authorization replace REDACTED
 			request>headers>Server delete
 			request>headers>Cookie cookie {
 				replace foo REDACTED
 				delete bar
+				hash baz
 			}
 			request>remote_ip ip_mask {
 				ipv4 24
 				ipv6 32
 			}
 			request>headers>Regexp regexp secret REDACTED
+			request>headers>Hash hash
 		}
 	}
 }
@@ -52,10 +55,17 @@ log {
 								{
 									"name": "bar",
 									"type": "delete"
+								},
+								{
+									"name": "baz",
+									"type": "hash"
 								}
 							],
 							"filter": "cookie"
 						},
+						"request\u003eheaders\u003eHash": {
+							"filter": "hash"
+						},
 						"request\u003eheaders\u003eRegexp": {
 							"filter": "regexp",
 							"regexp": "secret",
@@ -79,6 +89,10 @@ log {
 								{
 									"parameter": "bar",
 									"type": "delete"
+								},
+								{
+									"parameter": "baz",
+									"type": "hash"
 								}
 							],
 							"filter": "query"

modules/logging/filters.go

@@ -15,7 +15,9 @@
 package logging
 
 import (
+	"crypto/sha256"
 	"errors"
+	"fmt"
 	"net"
 	"net/http"
 	"net/url"
@@ -34,6 +36,7 @@ func init() {
 	caddy.RegisterModule(QueryFilter{})
 	caddy.RegisterModule(CookieFilter{})
 	caddy.RegisterModule(RegexpFilter{})
+	caddy.RegisterModule(HashFilter{})
 }
 
 // LogFieldFilter can filter (or manipulate)
@@ -65,6 +68,35 @@ func (DeleteFilter) Filter(in zapcore.Field) zapcore.Field {
 	return in
 }
 
+// hash returns the first 4 bytes of the SHA-256 hash of the given data as hexadecimal
+func hash(s string) string {
+	return fmt.Sprintf("%.4x", sha256.Sum256([]byte(s)))
+}
+
+// HashFilter is a Caddy log field filter that
+// replaces the field with the initial 4 bytes of the SHA-256 hash of the content.
+type HashFilter struct {
+}
+
+// CaddyModule returns the Caddy module information.
+func (HashFilter) CaddyModule() caddy.ModuleInfo {
+	return caddy.ModuleInfo{
+		ID:  "caddy.logging.encoders.filter.hash",
+		New: func() caddy.Module { return new(HashFilter) },
+	}
+}
+
+// UnmarshalCaddyfile sets up the module from Caddyfile tokens.
+func (f *HashFilter) UnmarshalCaddyfile(d *caddyfile.Dispenser) error {
+	return nil
+}
+
+// Filter filters the input field with the replacement value.
+func (f *HashFilter) Filter(in zapcore.Field) zapcore.Field {
+	in.String = hash(in.String)
+	return in
+}
+
 // ReplaceFilter is a Caddy log field filter that
 // replaces the field with the indicated string.
 type ReplaceFilter struct {
@@ -195,23 +227,27 @@ func (m IPMaskFilter) Filter(in zapcore.Field) zapcore.Field {
 type filterAction string
 
 const (
-	// Replace value(s) of query parameter(s).
+	// Replace value(s).
 	replaceAction filterAction = "replace"
-	// Delete query parameter(s).
+
+	// Hash value(s).
+	hashAction filterAction = "hash"
+
+	// Delete.
 	deleteAction filterAction = "delete"
 )
 
 func (a filterAction) IsValid() error {
 	switch a {
-	case replaceAction, deleteAction:
+	case replaceAction, deleteAction, hashAction:
 		return nil
 	}
 
 	return errors.New("invalid action type")
 }
 
 type queryFilterAction struct {
-	// `replace` to replace the value(s) associated with the parameter(s) or `delete` to remove them entirely.
+	// `replace` to replace the value(s) associated with the parameter(s), `hash` to replace them with the 4 initial bytes of the SHA-256 of their content or `delete` to remove them entirely.
 	Type filterAction `json:"type"`
 
 	// The name of the query parameter.
@@ -224,9 +260,9 @@ type queryFilterAction struct {
 // QueryFilter is a Caddy log field filter that filters
 // query parameters from a URL.
 //
-// This filter updates the logged URL string to remove or replace query
-// parameters containing sensitive data. For instance, it can be used
-// to redact any kind of secrets which were passed as query parameters,
+// This filter updates the logged URL string to remove, replace or hash
+// query parameters containing sensitive data. For instance, it can be
+// used to redact any kind of secrets which were passed as query parameters,
 // such as OAuth access tokens, session IDs, magic link tokens, etc.
 type QueryFilter struct {
 	// A list of actions to apply to the query parameters of the URL.
@@ -271,6 +307,14 @@ func (m *QueryFilter) UnmarshalCaddyfile(d *caddyfile.Dispenser) error {
 				}
 				qfa.Value = d.Val()
 
+			case "hash":
+				if !d.NextArg() {
+					return d.ArgErr()
+				}
+
+				qfa.Type = hashAction
+				qfa.Parameter = d.Val()
+
 			case "delete":
 				if !d.NextArg() {
 					return d.ArgErr()
@@ -304,6 +348,11 @@ func (m QueryFilter) Filter(in zapcore.Field) zapcore.Field {
 				q[a.Parameter][i] = a.Value
 			}
 
+		case hashAction:
+			for i := range q[a.Parameter] {
+				q[a.Parameter][i] = hash(a.Value)
+			}
+
 		case deleteAction:
 			q.Del(a.Parameter)
 		}
@@ -316,7 +365,7 @@ func (m QueryFilter) Filter(in zapcore.Field) zapcore.Field {
 }
 
 type cookieFilterAction struct {
-	// `replace` to replace the value of the cookie or `delete` to remove it entirely.
+	// `replace` to replace the value of the cookie, `hash` to replace it with the 4 initial bytes of the SHA-256 of its content or `delete` to remove it entirely.
 	Type filterAction `json:"type"`
 
 	// The name of the cookie.
@@ -330,7 +379,7 @@ type cookieFilterAction struct {
 // cookies.
 //
 // This filter updates the logged HTTP header string
-// to remove or replace cookies containing sensitive data. For instance,
+// to remove, replace or hash cookies containing sensitive data. For instance,
 // it can be used to redact any kind of secrets, such as session IDs.
 //
 // If several actions are configured for the same cookie name, only the first
@@ -378,6 +427,14 @@ func (m *CookieFilter) UnmarshalCaddyfile(d *caddyfile.Dispenser) error {
 				}
 				cfa.Value = d.Val()
 
+			case "hash":
+				if !d.NextArg() {
+					return d.ArgErr()
+				}
+
+				cfa.Type = hashAction
+				cfa.Name = d.Val()
+
 			case "delete":
 				if !d.NextArg() {
 					return d.ArgErr()
@@ -415,6 +472,11 @@ OUTER:
 				transformedRequest.AddCookie(c)
 				continue OUTER
 
+			case hashAction:
+				c.Value = hash(c.Value)
+				transformedRequest.AddCookie(c)
+				continue OUTER
+
 			case deleteAction:
 				continue OUTER
 			}

modules/logging/filters_test.go

@@ -13,14 +13,15 @@ func TestQueryFilter(t *testing.T) {
 		{replaceAction, "notexist", "REDACTED"},
 		{deleteAction, "bar", ""},
 		{deleteAction, "notexist", ""},
+		{hashAction, "hash", ""},
 	}}
 
 	if f.Validate() != nil {
 		t.Fatalf("the filter must be valid")
 	}
 
-	out := f.Filter(zapcore.Field{String: "/path?foo=a&foo=b&bar=c&bar=d&baz=e"})
-	if out.String != "/path?baz=e&foo=REDACTED&foo=REDACTED" {
+	out := f.Filter(zapcore.Field{String: "/path?foo=a&foo=b&bar=c&bar=d&baz=e&hash=hashed"})
+	if out.String != "/path?baz=e&foo=REDACTED&foo=REDACTED&hash=e3b0c442" {
 		t.Fatalf("query parameters have not been filtered: %s", out.String)
 	}
 }
@@ -45,10 +46,11 @@ func TestCookieFilter(t *testing.T) {
 	f := CookieFilter{[]cookieFilterAction{
 		{replaceAction, "foo", "REDACTED"},
 		{deleteAction, "bar", ""},
+		{hashAction, "hash", ""},
 	}}
 
-	out := f.Filter(zapcore.Field{String: "foo=a; foo=b; bar=c; bar=d; baz=e"})
-	if out.String != "foo=REDACTED; foo=REDACTED; baz=e" {
+	out := f.Filter(zapcore.Field{String: "foo=a; foo=b; bar=c; bar=d; baz=e; hash=hashed"})
+	if out.String != "foo=REDACTED; foo=REDACTED; baz=e; hash=1a06df82" {
 		t.Fatalf("cookies have not been filtered: %s", out.String)
 	}
 }
@@ -78,3 +80,12 @@ func TestRegexpFilter(t *testing.T) {
 		t.Fatalf("field has not been filtered: %s", out.String)
 	}
 }
+
+func TestHashFilter(t *testing.T) {
+	f := HashFilter{}
+
+	out := f.Filter(zapcore.Field{String: "foo"})
+	if out.String != "2c26b46b" {
+		t.Fatalf("field has not been filtered: %s", out.String)
+	}
+}