faster translation endpoint

app/service/productDescriptionService/productDescriptionService.go

@@ -5,8 +5,10 @@ import (
 	"encoding/xml"
 	"fmt"
 	"io"
+	"net/http"
 	"slices"
 	"strings"
+	"time"
 
 	"git.ma-al.com/goc_daniel/b2b/app/db"
 	"git.ma-al.com/goc_daniel/b2b/app/model"
@@ -26,27 +28,9 @@ type ProductDescriptionService struct {
 
 func New() *ProductDescriptionService {
 	return &ProductDescriptionService{
-		db:     db.Get(),
-		client: openai.NewClient(option.WithAPIKey("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")),
-	}
-}
-
-func isValidXHTML(s string) bool {
-	decoder := xml.NewDecoder(strings.NewReader(s))
-	hasStartTag := false
-
-	for {
-		tok, err := decoder.Token()
-		if err != nil {
-			if err == io.EOF {
-				return hasStartTag
-			}
-			return false
-		}
-
-		if _, ok := tok.(xml.StartElement); ok {
-			hasStartTag = true
-		}
+		db: db.Get(),
+		client: openai.NewClient(option.WithAPIKey("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"),
+			option.WithHTTPClient(&http.Client{Timeout: 300 * time.Second})),
 	}
 }
 
@@ -80,7 +64,7 @@ func (s *ProductDescriptionService) SaveProductDescription(userID uint, productI
 	for i := 0; i < len(mustBeHTML); i++ {
 		if text, exists := updates[mustBeHTML[i]]; exists {
 			if !isValidXHTML(text) {
-				return responseErrors.ErrInvalidHTML
+				return responseErrors.ErrInvalidXHTML
 			}
 		}
 	}
@@ -136,54 +120,41 @@ func (s *ProductDescriptionService) TranslateProductDescription(userID uint, pro
 		return nil, err
 	}
 
-	request := "Translate to " + lang.ISOCode + " without changing the html structure. You must only translate text visible on website.\n\n"
-	request += "\n"
-	request += "<translation_of_product_description>"
+	request := "Translate to " + lang.ISOCode + " without changing the html structure."
+	request += "\n\n<translation_of_product_description>"
 	request += ProductDescription.Description
-	request += "</translation_of_product_description>"
-	request += "\n"
-	request += "Remember: translate to " + lang.ISOCode + " without changing the html structure. You must only translate text visible on website."
-	request += "\n"
-	request += "<translation_of_product_short_description>"
+	request += "</translation_of_product_description>\n\n"
+	request += "Remember: translate to " + lang.ISOCode + " without changing the html structure."
+	request += "\n\n<translation_of_product_short_description>"
 	request += ProductDescription.DescriptionShort
-	request += "</translation_of_product_short_description>"
-	request += "\n"
-	request += "Remember: translate to " + lang.ISOCode + " without changing the html structure. You must only translate text visible on website."
-	request += "\n"
-	request += "<translation_of_product_meta_description>"
+	request += "</translation_of_product_short_description>\n\n"
+	request += "Remember: translate to " + lang.ISOCode + " without changing the html structure."
+	request += "\n\n<translation_of_product_meta_description>"
 	request += ProductDescription.MetaDescription
-	request += "</translation_of_product_meta_description>"
-	request += "\n"
-	request += "Remember: translate to " + lang.ISOCode + " without changing the html structure. You must only translate text visible on website."
-	request += "\n"
-	request += "<translation_of_product_meta_title>"
+	request += "</translation_of_product_meta_description>\n\n"
+	request += "Remember: translate to " + lang.ISOCode + " without changing the html structure."
+	request += "\n\n<translation_of_product_meta_title>"
 	request += ProductDescription.MetaTitle
-	request += "</translation_of_product_meta_title>"
-	request += "\n"
-	request += "Remember: translate to " + lang.ISOCode + " without changing the html structure. You must only translate text visible on website."
-	request += "\n"
-	request += "<translation_of_product_name>"
+	request += "</translation_of_product_meta_title>\n\n"
+	request += "Remember: translate to " + lang.ISOCode + " without changing the html structure."
+	request += "\n\n<translation_of_product_name>"
 	request += ProductDescription.Name
-	request += "</translation_of_product_name>"
-	request += "\n"
-	request += "Remember: translate to " + lang.ISOCode + " without changing the html structure. You must only translate text visible on website."
-	request += "\n"
-	request += "<translation_of_display_text_available_now>"
+	request += "</translation_of_product_name>\n\n"
+	request += "Remember: translate to " + lang.ISOCode + " without changing the html structure."
+	request += "\n\n<translation_of_display_text_available_now>"
 	request += ProductDescription.AvailableNow
-	request += "</translation_of_display_text_available_now>"
-	request += "\n"
-	request += "Remember: translate to " + lang.ISOCode + " without changing the html structure. You must only translate text visible on website."
-	request += "\n"
-	request += "<translation_of_display_text_available_later>"
+	request += "</translation_of_display_text_available_now>\n\n"
+	request += "Remember: translate to " + lang.ISOCode + " without changing the html structure."
+	request += "\n\n<translation_of_display_text_available_later>"
 	request += ProductDescription.AvailableLater
-	request += "</translation_of_display_text_available_later>"
-	request += "\n"
-	request += "Remember: translate to " + lang.ISOCode + " without changing the html structure. You must only translate text visible on website."
-	request += "\n"
-	request += "<translation_of_product_usage>"
+	request += "</translation_of_display_text_available_later>\n\n"
+	request += "Remember: translate to " + lang.ISOCode + " without changing the html structure."
+	request += "\n\n<translation_of_product_usage>"
 	request += ProductDescription.Usage
 	request += "</translation_of_product_usage>"
 
+	request = cleanForPrompt(request)
+
 	openai_response, err := s.client.Responses.New(context.Background(), responses.ResponseNewParams{
 		Input: responses.ResponseNewParamsInputUnion{OfString: openai.String(request)},
 		Model: openai.ChatModelGPT4_1Mini,
@@ -194,61 +165,136 @@ func (s *ProductDescriptionService) TranslateProductDescription(userID uint, pro
 	}
 	output := openai_response.OutputText()
 
-	// for debugging purposes
-	// fi, err := os.ReadFile("/home/daniel/coding/work/b2b/app/service/productDescriptionService/test.txt") // just pass the file name
+	// for testing purposes
+	// fi, err := os.ReadFile("/home/daniel/coding/work/b2b/app/service/productDescriptionService/test_out.txt") // just pass the file name
 	// output := string(fi)
 
-	success, match := GetStringInBetween(output, "<translation_of_product_description>", "</translation_of_product_description>")
+	success, resolution := resolveResponse(ProductDescription.Description, output, "translation_of_product_description")
 	if !success {
 		return nil, responseErrors.ErrOpenAIBadOutput
 	}
-	ProductDescription.Description = match
+	ProductDescription.Description = resolution
 
-	success, match = GetStringInBetween(output, "<translation_of_product_short_description>", "</translation_of_product_short_description>")
+	success, resolution = resolveResponse(ProductDescription.DescriptionShort, output, "translation_of_product_short_description")
 	if !success {
 		return nil, responseErrors.ErrOpenAIBadOutput
 	}
-	ProductDescription.DescriptionShort = match
+	ProductDescription.DescriptionShort = resolution
 
-	success, match = GetStringInBetween(output, "<translation_of_product_meta_description>", "</translation_of_product_meta_description>")
+	success, resolution = resolveResponse(ProductDescription.MetaDescription, output, "translation_of_product_meta_description")
 	if !success {
 		return nil, responseErrors.ErrOpenAIBadOutput
 	}
-	ProductDescription.MetaDescription = match
+	ProductDescription.MetaDescription = resolution
 
-	success, match = GetStringInBetween(output, "<translation_of_product_meta_title>", "</translation_of_product_meta_title>")
+	success, resolution = resolveResponse(ProductDescription.MetaTitle, output, "translation_of_product_meta_title")
 	if !success {
 		return nil, responseErrors.ErrOpenAIBadOutput
 	}
-	ProductDescription.MetaTitle = match
+	ProductDescription.MetaTitle = resolution
 
-	success, match = GetStringInBetween(output, "<translation_of_product_name>", "</translation_of_product_name>")
+	success, resolution = resolveResponse(ProductDescription.Name, output, "translation_of_product_name")
 	if !success {
 		return nil, responseErrors.ErrOpenAIBadOutput
 	}
-	ProductDescription.Name = match
+	ProductDescription.Name = resolution
 
-	success, match = GetStringInBetween(output, "<translation_of_display_text_available_now>", "</translation_of_display_text_available_now>")
+	success, resolution = resolveResponse(ProductDescription.AvailableNow, output, "translation_of_display_text_available_now")
 	if !success {
 		return nil, responseErrors.ErrOpenAIBadOutput
 	}
-	ProductDescription.AvailableNow = match
+	ProductDescription.AvailableNow = resolution
 
-	success, match = GetStringInBetween(output, "<translation_of_display_text_available_later>", "</translation_of_display_text_available_later>")
+	success, resolution = resolveResponse(ProductDescription.AvailableLater, output, "translation_of_display_text_available_later")
 	if !success {
 		return nil, responseErrors.ErrOpenAIBadOutput
 	}
-	ProductDescription.AvailableLater = match
+	ProductDescription.AvailableLater = resolution
 
-	success, match = GetStringInBetween(output, "<translation_of_product_usage>", "</translation_of_product_usage>")
+	success, resolution = resolveResponse(ProductDescription.Usage, output, "translation_of_product_usage")
 	if !success {
 		return nil, responseErrors.ErrOpenAIBadOutput
 	}
-	ProductDescription.Usage = match
+	ProductDescription.Usage = resolution
 
 	return &ProductDescription, nil
 }
 
+// isValidXHTML checks if the string obeys the XHTML format
+func isValidXHTML(s string) bool {
+	r := strings.NewReader(s)
+	d := xml.NewDecoder(r)
+
+	// Configure the decoder for HTML; leave off strict and autoclose for XHTML
+	d.Strict = true
+	d.AutoClose = xml.HTMLAutoClose
+	d.Entity = xml.HTMLEntity
+	for {
+		_, err := d.Token()
+		switch err {
+		case io.EOF:
+			return true // We're done, it's valid!
+		case nil:
+		default:
+			return false // Oops, something wasn't right
+		}
+	}
+}
+
+func cleanForPrompt(s string) string {
+	r := strings.NewReader(s)
+	d := xml.NewDecoder(r)
+
+	prompt := ""
+
+	// Configure the decoder for HTML; leave off strict and autoclose for XHTML
+	d.Strict = true
+	d.AutoClose = xml.HTMLAutoClose
+	d.Entity = xml.HTMLEntity
+	for {
+		token, err := d.Token()
+		if err == io.EOF {
+			break
+		}
+
+		switch v := token.(type) {
+		case xml.StartElement:
+			prompt += "<" + AttrName(v.Name)
+
+			for _, attr := range v.Attr {
+				if v.Name.Local == "img" && attr.Name.Local == "alt" {
+					prompt += fmt.Sprintf(` %s="%s"`, AttrName(attr.Name), attr.Value)
+				}
+			}
+
+			prompt += ">"
+		case xml.EndElement:
+			prompt += "</" + AttrName(v.Name) + ">"
+		case xml.CharData:
+			prompt += string(v)
+		case xml.Comment:
+		case xml.ProcInst:
+		case xml.Directive:
+		}
+	}
+
+	return prompt
+}
+
+func resolveResponse(original string, response string, key string) (bool, string) {
+	success, match := GetStringInBetween(response, "<"+key+">", "</"+key+">")
+	if !success || !isValidXHTML(match) {
+		return false, ""
+	}
+
+	success, resolution := RebuildFromResponse("<"+key+">"+original+"</"+key+">", "<"+key+">"+match+"</"+key+">")
+	if !success {
+		return false, ""
+	}
+
+	return true, resolution[2+len(key) : len(resolution)-3-len(key)]
+}
+
 // GetStringInBetween returns empty string if no start or end string found
 func GetStringInBetween(str string, start string, end string) (success bool, result string) {
 	s := strings.Index(str, start)
@@ -263,3 +309,140 @@ func GetStringInBetween(str string, start string, end string) (success bool, res
 
 	return true, str[s : s+e]
 }
+
+// Rebuilds HTML using the original HTML as a template and the response as a source
+// Assumes that both original and response have the exact same XML structure
+func RebuildFromResponse(s_original string, s_response string) (bool, string) {
+
+	r_original := strings.NewReader(s_original)
+	d_original := xml.NewDecoder(r_original)
+
+	r_response := strings.NewReader(s_response)
+	d_response := xml.NewDecoder(r_response)
+
+	result := ""
+
+	// Configure the decoder for HTML; leave off strict and autoclose for XHTML
+	d_original.Strict = true
+	d_original.AutoClose = xml.HTMLAutoClose
+	d_original.Entity = xml.HTMLEntity
+
+	d_response.Strict = true
+	d_response.AutoClose = xml.HTMLAutoClose
+	d_response.Entity = xml.HTMLEntity
+
+	token_original, err_original := d_original.Token()
+	token_response, err_response := d_response.Token()
+
+	for {
+		// err_original can only be EOF or nil.
+		if err_original != nil || err_response != nil {
+			if err_original != err_response {
+				return false, ""
+			}
+			return true, result
+		}
+
+		switch v_original := token_original.(type) {
+		case xml.StartElement:
+			switch v_response := token_response.(type) {
+			case xml.StartElement:
+				if v_original.Name.Space != v_response.Name.Space || v_original.Name.Local != v_response.Name.Local {
+					return false, ""
+				}
+
+				result += "<" + AttrName(v_original.Name)
+
+				for _, attr := range v_original.Attr {
+					if v_original.Name.Local != "img" || attr.Name.Local != "alt" {
+						result += fmt.Sprintf(` %s="%s"`, AttrName(attr.Name), attr.Value)
+					}
+				}
+
+				for _, attr := range v_response.Attr {
+					if v_response.Name.Local == "img" && attr.Name.Local == "alt" {
+						result += fmt.Sprintf(` %s="%s"`, AttrName(attr.Name), attr.Value)
+					}
+				}
+				result += ">"
+
+			case xml.CharData:
+				result += string(v_response)
+				token_response, err_response = d_response.Token()
+				continue
+
+			default:
+				return false, ""
+			}
+
+		case xml.EndElement:
+			switch v_response := token_response.(type) {
+			case xml.EndElement:
+				if v_original.Name.Space != v_response.Name.Space || v_original.Name.Local != v_response.Name.Local {
+					return false, ""
+				}
+
+				if v_original.Name.Local != "img" {
+					result += "</" + AttrName(v_original.Name) + ">"
+				}
+
+			case xml.CharData:
+				result += string(v_response)
+				token_response, err_response = d_response.Token()
+				continue
+
+			default:
+				return false, ""
+			}
+
+		case xml.CharData:
+			switch v_response := token_response.(type) {
+			case xml.CharData:
+				result += string(v_response)
+
+			case xml.StartElement:
+				result += string(v_original)
+				token_original, err_original = d_original.Token()
+				continue
+
+			case xml.EndElement:
+				result += string(v_original)
+				token_original, err_original = d_original.Token()
+				continue
+
+			default:
+				return false, ""
+			}
+
+		case xml.Comment:
+			result += "<!--" + string(v_original) + "-->"
+			token_original, err_original = d_original.Token()
+			continue
+
+		case xml.ProcInst:
+			if len(v_original.Inst) == 0 {
+				result += "<?" + v_original.Target + "?>"
+			} else {
+				result += "<?" + v_original.Target + " " + string(v_original.Inst) + "?>"
+			}
+			token_original, err_original = d_original.Token()
+			continue
+
+		case xml.Directive:
+			result += "<!" + string(v_original) + ">"
+			token_original, err_original = d_original.Token()
+			continue
+		}
+
+		token_original, err_original = d_original.Token()
+		token_response, err_response = d_response.Token()
+	}
+}
+
+func AttrName(name xml.Name) string {
+	if name.Space == "" {
+		return name.Local
+	} else {
+		return name.Space + ":" + name.Local
+	}
+}

app/utils/responseErrors/response_errors.go

@@ -40,7 +40,7 @@ var (
 	// Typed errors for product description handler
 	ErrBadAttribute       = errors.New("bad attribute")
 	ErrBadField           = errors.New("this field can not be updated")
-	ErrInvalidHTML        = errors.New("text is not in html format")
+	ErrInvalidXHTML       = errors.New("text is not in xhtml format")
 	ErrOpenAIResponseFail = errors.New("OpenAI responded with failure")
 	ErrOpenAIBadOutput    = errors.New("OpenAI response does not obey the format")
 )
@@ -116,7 +116,7 @@ func GetErrorCode(c fiber.Ctx, err error) string {
 		return i18n.T_(c, "error.err_bad_attribute")
 	case errors.Is(err, ErrBadField):
 		return i18n.T_(c, "error.err_bad_field")
-	case errors.Is(err, ErrInvalidHTML):
+	case errors.Is(err, ErrInvalidXHTML):
 		return i18n.T_(c, "error.err_invalid_html")
 	case errors.Is(err, ErrOpenAIResponseFail):
 		return i18n.T_(c, "error.err_openai_response_fail")
@@ -154,7 +154,7 @@ func GetErrorStatus(err error) int {
 		errors.Is(err, ErrInvalidPassword),
 		errors.Is(err, ErrBadAttribute),
 		errors.Is(err, ErrBadField),
-		errors.Is(err, ErrInvalidHTML):
+		errors.Is(err, ErrInvalidXHTML):
 		return fiber.StatusBadRequest
 	case errors.Is(err, ErrEmailExists):
 		return fiber.StatusConflict