dev day #92

internal/services/badwords.go

@@ -1,8 +1,10 @@
 package services
 
 import (
+	"html"
 	"regexp"
 	"strings"
+	"unicode/utf8"
 )
 
 // A compact list of Czech and English bad words with family-friendly replacements.
@@ -138,6 +140,56 @@ func FilterBadWords(s string) (string, bool) {
 	return out, replaced
 }
 
+func MaskBadWords(s string) (string, []string) {
+	if strings.TrimSpace(s) == "" { return s, nil }
+	out := s
+	originals := []string{}
+	for _, cr := range compiledRepls {
+		out = cr.re.ReplaceAllStringFunc(out, func(m string) string {
+			originals = append(originals, m)
+			n := len([]rune(m))
+			if n <= 0 { return m }
+			return strings.Repeat("*", n)
+		})
+	}
+	return out, originals
+}
+
+func MaskBadWordsHTML(s string) string {
+	if strings.TrimSpace(s) == "" { return html.EscapeString(s) }
+	var b strings.Builder
+	i := 0
+	for i < len(s) {
+		sub := s[i:]
+		best := ""
+		bestLen := 0
+		for _, cr := range compiledRepls {
+			if loc := cr.re.FindStringIndex(sub); loc != nil && loc[0] == 0 {
+				m := sub[:loc[1]]
+				if len(m) > bestLen {
+					best = m
+					bestLen = len(m)
+				}
+			}
+		}
+		if bestLen > 0 {
+			n := utf8.RuneCountInString(best)
+			b.WriteString(`<span class="cw" title="`)
+			b.WriteString(html.EscapeString(best))
+			b.WriteString(`">`)
+			b.WriteString(strings.Repeat("*", n))
+			b.WriteString("</span>")
+			i += bestLen
+			continue
+		}
+		r, size := utf8.DecodeRuneInString(sub)
+		if r == utf8.RuneError && size == 0 { break }
+		b.WriteString(html.EscapeString(sub[:size]))
+		i += size
+	}
+	return b.String()
+}
+
 // ContainsSensitiveWords returns true and the matched words if content contains strong/explicit terms.
 func ContainsSensitiveWords(s string) (bool, []string) {
 	if strings.TrimSpace(s) == "" { return false, nil }