extract from attrubite

internal/api/http/handler.go

@@ -75,6 +75,14 @@ func (h *Handler) handleRender(c echo.Context) error {
 		return echo.NewHTTPError(400, fmt.Errorf("decode specs: %w", err))
 	}
 
+	extractFrom, ok := map[pb.ExtractFrom]models.ExtractFrom{
+		pb.ExtractFrom_InnerText: models.ExtractFrom_InnerText,
+		pb.ExtractFrom_Attribute: models.ExtractFrom_Attribute,
+	}[specs.CreatedExtractFrom]
+	if !ok {
+		return echo.NewHTTPError(400, "invalid extract from")
+	}
+
 	task := models.Task{
 		TaskType:             models.TaskTypeExtract,
 		URL:                  specs.Url,
@@ -84,6 +92,8 @@ func (h *Handler) handleRender(c echo.Context) error {
 		SelectorDescription:  specs.SelectorDescription,
 		SelectorAuthor:       specs.SelectorAuthor,
 		SelectorCreated:      specs.SelectorCreated,
+		CreatedExtractFrom:   extractFrom,
+		CreatedAttributeName: specs.CreatedAttributeName,
 		SelectorContent:      specs.SelectorContent,
 		SelectorEnclosure:    specs.SelectorEnclosure,
 		Headers:              extractHeaders(c),

internal/extractors/pwextractor/pageparser.go

@@ -114,7 +114,15 @@ func (p *pageParser) extractPost(post playwright.Locator) (models.FeedItem, erro
 
 	item.Enclosure = newLocator(post, p.task.SelectorEnclosure).First().GetAttribute("src")
 
-	createdDateStr := newLocator(post, p.task.SelectorCreated).First().InnerText()
+	var createdDateStr string
+	switch p.task.CreatedExtractFrom {
+	case models.ExtractFrom_InnerText:
+		createdDateStr = newLocator(post, p.task.SelectorCreated).First().InnerText()
+	case models.ExtractFrom_Attribute:
+		createdDateStr = newLocator(post, p.task.SelectorCreated).First().GetAttribute(p.task.CreatedAttributeName)
+	default:
+		return models.FeedItem{}, fmt.Errorf("invalid task.CreatedExtractFrom")
+	}
 	log.Debugf("date=%s", createdDateStr)
 	createdDate, err := p.dateParser.ParseDate(createdDateStr)
 	if err != nil {

internal/models/models.go

@@ -13,6 +13,13 @@ const (
 	TaskTypePageScreenshot = "page_screenshot"
 )
 
+type ExtractFrom int
+
+const (
+	ExtractFrom_InnerText ExtractFrom = 0
+	ExtractFrom_Attribute ExtractFrom = 1
+)
+
 type Task struct {
 	// While adding new fields, dont forget to alter caching func
 	TaskType             TaskType
@@ -23,6 +30,8 @@ type Task struct {
 	SelectorDescription  string
 	SelectorAuthor       string
 	SelectorCreated      string
+	CreatedExtractFrom   ExtractFrom
+	CreatedAttributeName string
 	SelectorContent      string
 	SelectorEnclosure    string
 	Headers              map[string]string