From a4eb3b52cf3c10ef3d540f8d6d6b63eab4ff0126 Mon Sep 17 00:00:00 2001
From: Daniel Goc <daniel_goc@ma-al.com>
Date: Thu, 2 Apr 2026 15:24:34 +0200
Subject: [PATCH] change incoming HTML to HTML4

---
 .../productTranslationService.go              | 64 +++++++++++++++++--
 1 file changed, 58 insertions(+), 6 deletions(-)

diff --git a/app/service/productTranslationService/productTranslationService.go b/app/service/productTranslationService/productTranslationService.go
index 1b0a747..55f4f66 100644
--- a/app/service/productTranslationService/productTranslationService.go
+++ b/app/service/productTranslationService/productTranslationService.go
@@ -8,6 +8,7 @@ import (
 	"log"
 	"net/http"
 	"os"
+	"regexp"
 	"slices"
 	"strings"
 	"time"
@@ -99,8 +100,9 @@ func (s *ProductTranslationService) SaveProductDescription(userID uint, productI
 	// check that fields description, description_short and usage, if they exist, have a valid html format
 	mustBeHTML := []string{"description", "description_short", "usage"}
 	for i := 0; i < len(mustBeHTML); i++ {
-		if text, exists := updates[mustBeHTML[i]]; exists {
-			if !isValidXHTML(text) {
+		if _, exists := updates[mustBeHTML[i]]; exists {
+			updates[mustBeHTML[i]] = parseAutoCloseTags(updates[mustBeHTML[i]])
+			if !isValidXHTML(updates[mustBeHTML[i]]) {
 				return responseErrors.ErrInvalidXHTML
 			}
 		}
@@ -245,9 +247,17 @@ func cleanForPrompt(s string) string {
 				}
 			}
 
-			prompt += ">"
+			if slices.Contains(xml.HTMLAutoClose, v.Name.Local) {
+				prompt += "/>"
+			} else {
+				prompt += ">"
+			}
+
 		case xml.EndElement:
-			prompt += "</" + attrName(v.Name) + ">"
+			if !slices.Contains(xml.HTMLAutoClose, v.Name.Local) {
+				prompt += "</" + attrName(v.Name) + ">"
+			}
+
 		case xml.CharData:
 			prompt += string(v)
 		case xml.Comment:
@@ -288,6 +298,43 @@ func getStringInBetween(str string, start string, end string) (success bool, res
 	return true, str[s : s+e]
 }
 
+// this converts input into HTML4 format.
+// this really is ad-hoc solution, but it works.
+func parseAutoCloseTags(s string) string {
+	alts := ""
+	for i, name := range xml.HTMLAutoClose {
+		if i > 0 {
+			alts += "|"
+		}
+		alts += name
+	}
+
+	// remove closing </img> tags
+	reClose := regexp.MustCompile(`(?i)<\s*\/\s*(?:` + alts + `)\s*>`)
+	s = reClose.ReplaceAllString(s, "")
+
+	// convert <img ...> → <img ... />
+	// matches <img ...> that do NOT already end with />
+	reOpen := regexp.MustCompile(`(?i)<\s*(` + alts + `)\b([^>]*?)>`)
+	s = reOpen.ReplaceAllStringFunc(s, func(tag string) string {
+		trimmed := strings.TrimSpace(tag)
+
+		// Already self-closed: <img ... />, <br/>
+		if strings.HasSuffix(trimmed, "/>") {
+			return tag
+		}
+
+		// Replace final > with />
+		i := strings.LastIndex(tag, ">")
+		if i < 0 {
+			return tag
+		}
+		return tag[:i] + " />"
+	})
+
+	return s
+}
+
 // isValidXHTML checks if the string obeys the XHTML format
 func isValidXHTML(s string) bool {
 	r := strings.NewReader(s)
@@ -363,7 +410,12 @@ func rebuildFromResponse(s_original string, s_response string) (bool, string) {
 						result += fmt.Sprintf(` %s="%s"`, attrName(attr.Name), attr.Value)
 					}
 				}
-				result += ">"
+
+				if slices.Contains(xml.HTMLAutoClose, v_original.Name.Local) {
+					result += "/>"
+				} else {
+					result += ">"
+				}
 
 			case xml.CharData:
 				result += string(v_response)
@@ -381,7 +433,7 @@ func rebuildFromResponse(s_original string, s_response string) (bool, string) {
 					return false, ""
 				}
 
-				if v_original.Name.Local != "img" {
+				if !slices.Contains(xml.HTMLAutoClose, v_original.Name.Local) {
 					result += "</" + attrName(v_original.Name) + ">"
 				}