optimize timestamp detection

timestamp.go

@@ -1,35 +1,31 @@
 package logparser
 
-import (
- "regexp"
-)
-
 const (
  lookForTimestampLimit = 100
 )
 
-var (
- timestampRegexes = []*regexp.Regexp{
- regexp.MustCompile(`(^|\s)\d{2}:\d{2}(:\d{2}[^\s"']*)?`),
- regexp.MustCompile(`\d{2} [A-Z][a-z]{2} \d{4}`),
- regexp.MustCompile(`\d{4}-\d{2}-\d{2}`),
- regexp.MustCompile(`\d{4}/\d{2}/\d{2}`),
- regexp.MustCompile(`\d{4}\.\d{2}\.\d{2}`),
- regexp.MustCompile(`[A-Z][a-z]{2} \d{2}`),
- regexp.MustCompile(`\d{2}-\d{2}-\d{4}`),
- regexp.MustCompile(`\d{2}/\d{2}/\d{4}`),
- regexp.MustCompile(`\d{2}\.\d{2}\.\d{4}`),
- regexp.MustCompile(`\d{2}/[A-Z][a-z]{2}/\d{4}`),
- }
-)
-
 func containsTimestamp(line string) bool {
  if len(line) > lookForTimestampLimit {
  line = line[:lookForTimestampLimit]
  }
- for _, re := range timestampRegexes {
- if re.MatchString(line) {
- return true
+ var digits, colons int
+ for _, r := range line {
+ switch {
+ case r >= '0' && r <= '9':
+ digits++
+ if digits > 2 {
+ digits = 0
+ }
+ if digits == 2 && colons == 2 {
+ return true
+ }
+ case r == ':':
+ if digits == 2 {
+ colons++
+ }
+ digits = 0
+ default:
+ digits, colons = 0, 0
  }
  }
  return false

timestamp_test.go

@@ -3,16 +3,10 @@ package logparser
 import (
  "github.com/stretchr/testify/assert"
  "testing"
+ "time"
 )
 
 func Test_containsTimestamp(t *testing.T) {
- assert.True(t, containsTimestamp("2005-08-09"))
- assert.True(t, containsTimestamp("2020/06/26"))
- assert.True(t, containsTimestamp("02/17/2009"))
- assert.True(t, containsTimestamp("25.02.2013"))
- assert.True(t, containsTimestamp("2013.25.02"))
- assert.True(t, containsTimestamp("18:31"))
- assert.True(t, containsTimestamp("18:31:42"))
  assert.True(t, containsTimestamp("18:31:42+03"))
  assert.True(t, containsTimestamp("18:31:42-03"))
  assert.True(t, containsTimestamp("18:31:42+03:30"))
@@ -25,4 +19,29 @@ func Test_containsTimestamp(t *testing.T) {
  assert.True(t, containsTimestamp("2005-08-09T18:31:42"))
  assert.True(t, containsTimestamp("2005-08-09T18:31:42.201"))
  assert.True(t, containsTimestamp(`10/Oct/2000:13:55:36 -0700`))
+ assert.True(t, containsTimestamp(time.ANSIC))
+ assert.True(t, containsTimestamp(time.UnixDate))
+ assert.True(t, containsTimestamp(time.RubyDate))
+ assert.True(t, containsTimestamp(time.RFC850))
+ assert.True(t, containsTimestamp(time.RFC1123))
+ assert.True(t, containsTimestamp(time.RFC1123Z))
+ assert.True(t, containsTimestamp(time.RFC3339))
+ assert.True(t, containsTimestamp(time.RFC3339Nano))
+ assert.True(t, containsTimestamp(time.Stamp))
+ assert.True(t, containsTimestamp(time.StampMilli))
+ assert.True(t, containsTimestamp(time.StampMicro))
+
+ assert.False(t, containsTimestamp("13/32"))
+ assert.False(t, containsTimestamp("13:32"))
+ assert.False(t, containsTimestamp("100/5/100"))
+ assert.False(t, containsTimestamp("1:12:123"))
+ assert.False(t, containsTimestamp("12:aa:12:32"))
+
+}
+
+func Benchmark_containsTimestamp(b *testing.B) {
+ l := `10.42.0.21 - - [30/Oct/2023:11:55:47 +0000] "GET / HTTP/1.1" 200 612 "-" "-" "-"`
+ for n := 0; n < b.N; n++ {
+ containsTimestamp(l)
+ }
 }