dev day #89

internal/services/spam.go

@@ -17,8 +17,7 @@ func EvaluateSpamScore(s string) (float64, []string) {
 		rules = append(rules, "too_short")
 	}
 	// Excessive repeated characters like 'aaaaaa' or '!!!!'
-	repeatRe := regexp.MustCompile(`([a-zA-Z!?.])\1{4,}`)
-	if repeatRe.MatchString(content) {
+	if hasExcessiveRepetition(content, 5) {
 		rules = append(rules, "repeated_chars")
 	}
 	// Low vowel ratio suggests gibberish in Czech/English latin text
@@ -53,3 +52,33 @@ func EvaluateSpamScore(s string) (float64, []string) {
 	if score > 1.0 { score = 1.0 }
 	return score, rules
 }
+
+// hasExcessiveRepetition checks if s contains a run of the same character of length >= minRun
+// Limited to ASCII letters and the punctuation characters ! ? . to mirror the previous intent.
+func hasExcessiveRepetition(s string, minRun int) bool {
+	if minRun < 2 { minRun = 2 }
+	run := 1
+	var prev rune
+	first := true
+	for _, r := range s {
+		if first {
+			prev = r
+			first = false
+			continue
+		}
+		if r == prev && (isAsciiLetter(r) || r == '!' || r == '?' || r == '.') {
+			run++
+			if run >= minRun {
+				return true
+			}
+		} else {
+			prev = r
+			run = 1
+		}
+	}
+	return false
+}
+
+func isAsciiLetter(r rune) bool {
+	return (r >= 'a' && r <= 'z') || (r >= 'A' && r <= 'Z')
+}