Extend filtering to sequences of vowels/consonants (#5)

.golangci.yml

@@ -7,7 +7,7 @@ linters-settings:
     check-type-assertions: true
     check-blank: true
   gocyclo:
-    min-complexity: 20
+    min-complexity: 40
   dupl:
     threshold: 100
   misspell:

filter_test.go

@@ -5,7 +5,7 @@ import (
 	"testing"
 )
 
-const logLine = `foo-bar-16 i2 2022/09/02 09:48:29.199655 baz handler failed to get cox: failed to get cox type: unknown cox type string: {quux} [R method:GET path:/abcd5s8 ra:2022-09-02T09:48:29 form:'map[bar:[11_2022-09-02] lox_id:[] cucumber:[RedCat1509_2022-09-02] cucumber_id:[132072] faux_id:[afExxSDFKHgBJcwxDgIxDETR1vEAWVVHqXo6PcBjfoaDF29f_I8jYTZZVyKeiXzPlP9O9k3SrZtY3IeqA] cox_alarm:[{payout}}] cox_carrot:[OCD] cox_type:[{marks}] creative:[62_203206_123ebd32047fe640] foo_boo_99diks:[https://peebee.jeeass-foo.site/pushforw?lockid=sdd32432dUR1vEAWVVHqXo6PcBjfoaDF29f_Ik3SrZtY3FzXvq0fP1IeqA] goal:[99diks] gps_pos:[1231230D-0BE9-41EF-B146-123123123B9BC7] baz:[123123124-0BE9-41EF-B146-123123123123] ip_address:[1.2.333.4] labelle:[72_206706_125329e2047fe640] poob_id:[62_206706_550eb9e2047fe640] baz_limit:[1234]]' header:'map[User-Agent:[Mozilla/5.0 (iPhone; CPU iPhone OS 14_4_2 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Mobile/15E148] X-Boo-Id:[12312380d-0be9-41ef-b146-3452352323] X-Forwarded-For:[111.332.555.333] X-Forwarded-Proto:[https] X-Forwarded-For:[123.321.123.321]]' foocksorized gees_valeed deeedre_lee:0]`
+const logLine = `foo-bar-16 i2 2022/09/02 09:48:29.199655 baz cdfgzaa aouios handler failed to get cox: failed to get cox type: unknown cox type string: {quux} [R method:GET path:/abcd5s8 ra:2022-09-02T09:48:29 form:'map[bar:[11_2022-09-02] lox_id:[] cucumber:[RedCat1509_2022-09-02] cucumber_id:[132072] faux_id:[afExxSDFKHgBJcwxDgIxDETR1vEAWVVHqXo6PcBjfoaDF29f_I8jYTZZVyKeiXzPlP9O9k3SrZtY3IeqA] cox_alarm:[{payout}}] cox_carrot:[OCD] cox_type:[{marks}] creative:[62_203206_123ebd32047fe640] foo_boo_99diks:[https://peebee.jeeass-foo.site/pushforw?lockid=sdd32432dUR1vEAWVVHqXo6PcBjfoaDF29f_Ik3SrZtY3FzXvq0fP1IeqA] goal:[99diks] gps_pos:[1231230D-0BE9-41EF-B146-123123123B9BC7] baz:[123123124-0BE9-41EF-B146-123123123123] ip_address:[1.2.333.4] labelle:[72_206706_125329e2047fe640] poob_id:[62_206706_550eb9e2047fe640] baz_limit:[1234]]' header:'map[User-Agent:[Mozilla/5.0 (iPhone; CPU iPhone OS 14_4_2 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Mobile/15E148] X-Boo-Id:[12312380d-0be9-41ef-b146-3452352323] X-Forwarded-For:[111.332.555.333] X-Forwarded-Proto:[https] X-Forwarded-For:[123.321.123.321]]' foocksorized gees_valeed deeedre_lee:0]`
 
 // BenchmarkFilterAlphanumeric checks bespoke implementation.
 // BenchmarkFilterAlphanumeric-12    	 1912300	       614.7 ns/op	       0 B/op	       0 allocs/op.
@@ -14,14 +14,14 @@ func BenchmarkFilterAlphanumeric(b *testing.B) {
 
 	for i := 0; i < b.N; i++ {
 		s := []byte(logLine)
-		a := filterAlphanumeric(s, 150)
+		a := filterDynamic(s, 150)
 		_ = a
 	}
 }
 
 func TestAlphanum(t *testing.T) {
-	filtered := string(filterAlphanumeric([]byte(logLine), 150))
-	expected := `X X X/X/X X:X:X.X baz handler failed to get cox: failed to get cox type: unknown cox type string: {quux} [R method:GET path:/X ra:X:X:X form:'map[bar:`
+	filtered := string(filterDynamic([]byte(logLine), 150))
+	expected := `X X X/X/X X:X:X.X baz X X handler failed to get cox: failed to get cox type: unknown cox type string: {quux} [R method:GET path:/X ra:X:X:X form:'map[`
 
 	if expected != filtered {
 		t.Fatalf("unexpected filtered: %s", filtered)
@@ -31,7 +31,7 @@ func TestAlphanum(t *testing.T) {
 func TestShortLine(t *testing.T) {
 	line := "foo-bar-12 i3 2022/09/15 11:24:10.689412 baz 0-275 foo bar"
 	expected := "X X X/X/X X:X:X.X baz X foo bar"
-	filtered := string(filterAlphanumeric([]byte(line), 120))
+	filtered := string(filterDynamic([]byte(line), 120))
 
 	if expected != filtered {
 		t.Fatalf("unexpected filtered: %s", filtered)

fllter.go

@@ -1,24 +1,42 @@
 package main
 
-// filterAlphanumeric replaces a-zA-Z_-% sequences that have at least one digit with X.
+// filterDynamic replaces a-zA-Z_-% sequences that have at least one digit or 5+ consecutive consolants/vowels with X.
 // Does not allocate, uses original slice.
-func filterAlphanumeric(data []byte, l int) []byte {
+func filterDynamic(data []byte, l int) []byte {
 	hasDigit := false
 	wordStart := -1
+	maxConsecutive := 0
+	consecutive := 0
 
 	res := data[:0]
 
-	var i int
+	var (
+		i            int
+		prevCharType byte
+	)
 
 	for i = 0; i < len(data); i++ {
 		c := data[i]
 
 		isAlpha := false
+		charType := byte(0)
 
 		switch {
 		case c >= 'a' && c <= 'z':
+			if c == 'a' || c == 'e' || c == 'i' || c == 'o' || c == 'u' || c == 'y' || c == 'w' {
+				charType = 'v' // Vowel.
+			} else {
+				charType = 'c' // Consonant.
+			}
+
 			isAlpha = true
 		case c >= 'A' && c <= 'Z':
+			if c == 'A' || c == 'E' || c == 'I' || c == 'O' || c == 'U' || c == 'Y' || c == 'W' {
+				charType = 'v'
+			} else {
+				charType = 'c'
+			}
+
 			isAlpha = true
 		case c >= '0' && c <= '9':
 			isAlpha = true
@@ -27,16 +45,29 @@ func filterAlphanumeric(data []byte, l int) []byte {
 			isAlpha = true
 		}
 
+		if charType == prevCharType {
+			consecutive++
+		} else {
+			if consecutive > maxConsecutive {
+				maxConsecutive = consecutive
+			}
+
+			prevCharType = charType
+			consecutive = 1
+		}
+
 		// Finish current word.
 		if wordStart >= 0 && !isAlpha {
-			if hasDigit {
+			if hasDigit || maxConsecutive > 4 {
 				res = append(res, 'X')
 			} else {
 				res = append(res, data[wordStart:i]...)
 			}
 
 			wordStart = -1
 			hasDigit = false
+			maxConsecutive = 0
+			consecutive = 0
 		}
 
 		if wordStart == -1 {

main.go

@@ -134,7 +134,7 @@ func main() {
 		}
 
 		if top > 0 {
-			filtered := filterAlphanumeric(line, length)
+			filtered := filterDynamic(line, length)
 
 			d.Reset()