From 140522df81e3940b6dc7233c9771ba45a44ba899 Mon Sep 17 00:00:00 2001
From: Sivachandran Paramasivam <sivachandran.p@gmail.com>
Date: Thu, 21 Sep 2023 19:34:08 +0530
Subject: [PATCH] error_map: canonicalise error

---
 error_map.go      | 31 ++++++++++++++++++++++++++++++-
 error_map_test.go | 19 +++++++++++++++++++
 2 files changed, 49 insertions(+), 1 deletion(-)

diff --git a/error_map.go b/error_map.go
index 6bd178d..6f68781 100644
--- a/error_map.go
+++ b/error_map.go
@@ -1,12 +1,41 @@
 package main
 
 import (
+	"regexp"
 	"sort"
 	"strconv"
 	"sync"
 	"sync/atomic"
 )
 
+var (
+	// capture 1 should capture the canonicalised error message
+	errMsgPatterns = []string{
+		// read tcp 10.10.0.62:60682->63.35.24.107:443: read: connection reset by peer
+		`read tcp [\d.]+:\d+->[\d.]+:\d+: read: (.*)`,
+	}
+
+	errMsgRegexes []*regexp.Regexp
+)
+
+func init() {
+	for _, pattern := range errMsgPatterns {
+		errMsgRegexes = append(errMsgRegexes, regexp.MustCompile(pattern))
+	}
+}
+
+func canonicalizeError(err error) string {
+	msg := err.Error()
+	for _, errMsgRegex := range errMsgRegexes {
+		matches := errMsgRegex.FindStringSubmatch(msg)
+		if matches != nil {
+			return matches[1]
+		}
+	}
+
+	return msg
+}
+
 type errorMap struct {
 	mu sync.RWMutex
 	m  map[string]*uint64
@@ -19,7 +48,7 @@ func newErrorMap() *errorMap {
 }
 
 func (e *errorMap) add(err error) {
-	s := err.Error()
+	s := canonicalizeError(err)
 	e.mu.RLock()
 	c, ok := e.m[s]
 	e.mu.RUnlock()
diff --git a/error_map_test.go b/error_map_test.go
index cad6fb2..0b37157 100644
--- a/error_map_test.go
+++ b/error_map_test.go
@@ -2,6 +2,7 @@ package main
 
 import (
 	"errors"
+	"fmt"
 	"reflect"
 	"testing"
 )
@@ -23,6 +24,24 @@ func TestErrorMapGet(t *testing.T) {
 	}
 }
 
+func TestCanonicalisedError(t *testing.T) {
+	m := newErrorMap()
+	for i := 0; i < 10; i++ {
+		m.add(fmt.Errorf("read tcp 10.10.0.62:%d->63.35.24.107:%d: read: connection reset by peer", 1000+i, 2000+i))
+	}
+	m.add(errors.New("tls timeout"))
+
+	e := errorsByFrequency{
+		{"connection reset by peer", 10},
+		{"tls timeout", 1},
+	}
+	if a := m.byFrequency(); !reflect.DeepEqual(a, e) {
+		t.Logf("Expected: %+v", e)
+		t.Logf("Got: %+v", a)
+		t.Fail()
+	}
+}
+
 func TestByFrequency(t *testing.T) {
 	m := newErrorMap()
 	a := errors.New("A")