t.bilibili

beclab · Dec 13, 2024 · 61d4c07 · 61d4c07
1 parent 0c11cf2
commit 61d4c07
Show file tree

Hide file tree

Showing 2 changed files with 13 additions and 6 deletions.
diff --git a/backend-server/crawler/entry.go b/backend-server/crawler/entry.go
@@ -19,6 +19,14 @@ import (
 func EntryCrawler(entry *model.Entry, feedUrl, userAgent, cookie string, certificates, fetchViaProxy bool) {
 	//entryID, entryUrl, entryTitle, imageUrl, author string, entryPublishedAt int64, feed *model.Feed) (string, string, int64) {
 	common.Logger.Info("crawler entry start", zap.String("url", entry.URL))
+	if common.Domain(entry.URL) == "t.bilibili.com" {
+		entry.FullContent = entry.Content
+		entry.Language = "zh-cn"
+		if entry.ImageUrl == "" && entry.Content != "" {
+			entry.ImageUrl = common.GetImageUrlFromContent(entry.Content)
+		}
+		return
+	}
 	entry.RawContent = FetchRawContnt(
 		entry.URL,
 		entry.Title,

diff --git a/backend-server/model/entry.go b/backend-server/model/entry.go
@@ -14,12 +14,11 @@ type Entry struct {
 	URL               string  `json:"url"`
 	CommentsURL       string  `json:"comments_url"`
 	PublishedAtParsed time.Time
-	PublishedAt       int64 `json:"published_at"`
-
-	CreatedAt  time.Time `json:"createdAt"`
-	UpdatedAt  time.Time `json:"updatedAt"`
-	Content    string    `json:"content"`
-	RawContent string    `json:"raw_content"`
+	PublishedAt       int64     `json:"published_at"`
+	CreatedAt         time.Time `json:"createdAt"`
+	UpdatedAt         time.Time `json:"updatedAt"`
+	Content           string    `json:"content"`
+	RawContent        string    `json:"raw_content"`
 	//PureContent string             `bson:"pure_content"`
 	FullContent string `json:"full_content"`
 	//DocId       string   `json:"doc_id"`