From 789efa5deef53071b57479d37e4022bf372c4eef Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?K=C3=A9vin=20Dunglas?= <dunglas@gmail.com>
Date: Tue, 23 Nov 2021 18:00:20 +0100
Subject: [PATCH] logging: add a regexp filter (#4426)

---
 .../caddyfile_adapt/log_filters.txt           |  6 ++
 modules/logging/filters.go                    | 57 +++++++++++++++++++
 modules/logging/filters_test.go               | 11 ++++
 3 files changed, 74 insertions(+)

diff --git a/caddytest/integration/caddyfile_adapt/log_filters.txt b/caddytest/integration/caddyfile_adapt/log_filters.txt
index 2693e9c6..1bef4c21 100644
--- a/caddytest/integration/caddyfile_adapt/log_filters.txt
+++ b/caddytest/integration/caddyfile_adapt/log_filters.txt
@@ -19,6 +19,7 @@ log {
 				ipv4 24
 				ipv6 32
 			}
+			request>headers>Regexp regexp secret REDACTED
 		}
 	}
 }
@@ -55,6 +56,11 @@ log {
 							],
 							"filter": "cookie"
 						},
+						"request\u003eheaders\u003eRegexp": {
+							"filter": "regexp",
+							"regexp": "secret",
+							"value": "REDACTED"
+						},
 						"request\u003eheaders\u003eServer": {
 							"filter": "delete"
 						},
diff --git a/modules/logging/filters.go b/modules/logging/filters.go
index cf3b5cc1..ded08acb 100644
--- a/modules/logging/filters.go
+++ b/modules/logging/filters.go
@@ -19,6 +19,7 @@ import (
 	"net"
 	"net/http"
 	"net/url"
+	"regexp"
 	"strconv"
 
 	"github.com/caddyserver/caddy/v2"
@@ -32,6 +33,7 @@ func init() {
 	caddy.RegisterModule(IPMaskFilter{})
 	caddy.RegisterModule(QueryFilter{})
 	caddy.RegisterModule(CookieFilter{})
+	caddy.RegisterModule(RegexpFilter{})
 }
 
 // LogFieldFilter can filter (or manipulate)
@@ -426,6 +428,58 @@ OUTER:
 	return in
 }
 
+// RegexpFilter is a Caddy log field filter that
+// replaces the field matching the provided regexp with the indicated string.
+type RegexpFilter struct {
+	// The regular expression pattern defining what to replace.
+	RawRegexp string `json:"regexp,omitempty"`
+
+	// The value to use as replacement
+	Value string `json:"value,omitempty"`
+
+	regexp *regexp.Regexp
+}
+
+// CaddyModule returns the Caddy module information.
+func (RegexpFilter) CaddyModule() caddy.ModuleInfo {
+	return caddy.ModuleInfo{
+		ID:  "caddy.logging.encoders.filter.regexp",
+		New: func() caddy.Module { return new(RegexpFilter) },
+	}
+}
+
+// UnmarshalCaddyfile sets up the module from Caddyfile tokens.
+func (f *RegexpFilter) UnmarshalCaddyfile(d *caddyfile.Dispenser) error {
+	for d.Next() {
+		if d.NextArg() {
+			f.RawRegexp = d.Val()
+		}
+		if d.NextArg() {
+			f.Value = d.Val()
+		}
+	}
+	return nil
+}
+
+// Provision compiles m's regexp.
+func (m *RegexpFilter) Provision(ctx caddy.Context) error {
+	r, err := regexp.Compile(m.RawRegexp)
+	if err != nil {
+		return err
+	}
+
+	m.regexp = r
+
+	return nil
+}
+
+// Filter filters the input field with the replacement value if it matches the regexp.
+func (f *RegexpFilter) Filter(in zapcore.Field) zapcore.Field {
+	in.String = f.regexp.ReplaceAllString(in.String, f.Value)
+
+	return in
+}
+
 // Interface guards
 var (
 	_ LogFieldFilter = (*DeleteFilter)(nil)
@@ -433,14 +487,17 @@ var (
 	_ LogFieldFilter = (*IPMaskFilter)(nil)
 	_ LogFieldFilter = (*QueryFilter)(nil)
 	_ LogFieldFilter = (*CookieFilter)(nil)
+	_ LogFieldFilter = (*RegexpFilter)(nil)
 
 	_ caddyfile.Unmarshaler = (*DeleteFilter)(nil)
 	_ caddyfile.Unmarshaler = (*ReplaceFilter)(nil)
 	_ caddyfile.Unmarshaler = (*IPMaskFilter)(nil)
 	_ caddyfile.Unmarshaler = (*QueryFilter)(nil)
 	_ caddyfile.Unmarshaler = (*CookieFilter)(nil)
+	_ caddyfile.Unmarshaler = (*RegexpFilter)(nil)
 
 	_ caddy.Provisioner = (*IPMaskFilter)(nil)
+	_ caddy.Provisioner = (*RegexpFilter)(nil)
 
 	_ caddy.Validator = (*QueryFilter)(nil)
 )
diff --git a/modules/logging/filters_test.go b/modules/logging/filters_test.go
index 6871bea0..99289c3e 100644
--- a/modules/logging/filters_test.go
+++ b/modules/logging/filters_test.go
@@ -3,6 +3,7 @@ package logging
 import (
 	"testing"
 
+	"github.com/caddyserver/caddy/v2"
 	"go.uber.org/zap/zapcore"
 )
 
@@ -67,3 +68,13 @@ func TestValidateCookieFilter(t *testing.T) {
 		t.Fatalf("unknown action type must be invalid")
 	}
 }
+
+func TestRegexpFilter(t *testing.T) {
+	f := RegexpFilter{RawRegexp: `secret`, Value: "REDACTED"}
+	f.Provision(caddy.Context{})
+
+	out := f.Filter(zapcore.Field{String: "foo-secret-bar"})
+	if out.String != "foo-REDACTED-bar" {
+		t.Fatalf("field has not been filtered: %s", out.String)
+	}
+}