From 0e32cc3f176baa3314e08b9eaf6a565e0b61f0fe Mon Sep 17 00:00:00 2001
From: Egor Aristov <ef@efprojects.com>
Date: Tue, 6 May 2025 20:47:10 +0300
Subject: [PATCH] extract from attrubite

---
 internal/api/http/handler.go                  | 32 ++++++++++++-------
 internal/extractors/pwextractor/pageparser.go | 10 +++++-
 internal/models/models.go                     | 31 +++++++++++-------
 3 files changed, 50 insertions(+), 23 deletions(-)

diff --git a/internal/api/http/handler.go b/internal/api/http/handler.go
index 9b34509..13e34aa 100644
--- a/internal/api/http/handler.go
+++ b/internal/api/http/handler.go
@@ -75,18 +75,28 @@ 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,
-		SelectorPost:        specs.SelectorPost,
-		SelectorTitle:       specs.SelectorTitle,
-		SelectorLink:        specs.SelectorLink,
-		SelectorDescription: specs.SelectorDescription,
-		SelectorAuthor:      specs.SelectorAuthor,
-		SelectorCreated:     specs.SelectorCreated,
-		SelectorContent:     specs.SelectorContent,
-		SelectorEnclosure:   specs.SelectorEnclosure,
-		Headers:             extractHeaders(c),
+		TaskType:             models.TaskTypeExtract,
+		URL:                  specs.Url,
+		SelectorPost:         specs.SelectorPost,
+		SelectorTitle:        specs.SelectorTitle,
+		SelectorLink:         specs.SelectorLink,
+		SelectorDescription:  specs.SelectorDescription,
+		SelectorAuthor:       specs.SelectorAuthor,
+		SelectorCreated:      specs.SelectorCreated,
+		CreatedExtractFrom:   extractFrom,
+		CreatedAttributeName: specs.CreatedAttributeName,
+		SelectorContent:      specs.SelectorContent,
+		SelectorEnclosure:    specs.SelectorEnclosure,
+		Headers:              extractHeaders(c),
 	}
 
 	cacheLifetime, err := time.ParseDuration(specs.CacheLifetime)
diff --git a/internal/extractors/pwextractor/pageparser.go b/internal/extractors/pwextractor/pageparser.go
index ef1a99e..b2c9a53 100644
--- a/internal/extractors/pwextractor/pageparser.go
+++ b/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 {
diff --git a/internal/models/models.go b/internal/models/models.go
index 7099d92..d5cf818 100644
--- a/internal/models/models.go
+++ b/internal/models/models.go
@@ -13,19 +13,28 @@ 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
-	URL                 string
-	SelectorPost        string
-	SelectorTitle       string
-	SelectorLink        string
-	SelectorDescription string
-	SelectorAuthor      string
-	SelectorCreated     string
-	SelectorContent     string
-	SelectorEnclosure   string
-	Headers             map[string]string
+	TaskType             TaskType
+	URL                  string
+	SelectorPost         string
+	SelectorTitle        string
+	SelectorLink         string
+	SelectorDescription  string
+	SelectorAuthor       string
+	SelectorCreated      string
+	CreatedExtractFrom   ExtractFrom
+	CreatedAttributeName string
+	SelectorContent      string
+	SelectorEnclosure    string
+	Headers              map[string]string
 }
 
 func (t Task) CacheKey() string {