chore: remove redundant last category if it matches the document title (

#64)

modules/common/document.go

@@ -5,6 +5,7 @@
 package common
 
 import (
+	"strings"
 	"time"
 )
 
@@ -56,6 +57,19 @@ type Document struct {
 
 }
 
+func (document *Document) Cleanup() {
+	document.TrimLastDuplicatedCategory()
+}
+
+func (document *Document) TrimLastDuplicatedCategory() {
+	// Ensure RichCategories is not empty before accessing the last element
+	if len(document.RichCategories) > 0 &&
+		strings.TrimSpace(document.RichCategories[len(document.RichCategories)-1].Label) == strings.TrimSpace(document.Title) {
+		// Remove the last category if it matches the book's title (it's redundant)
+		document.RichCategories = document.RichCategories[:len(document.RichCategories)-1]
+	}
+}
+
 type EditorInfo struct {
 	UserInfo  UserInfo   `json:"user,omitempty" elastic_mapping:"user:{type:object}"` // Information about the last user who updated the document
 	UpdatedAt *time.Time `json:"timestamp,omitempty"`                                 // Timestamp of the last update

plugins/connectors/yuque/collect.go

@@ -239,9 +239,12 @@ func (this *Plugin) collectBooks(connector *common.Connector, datasource *common
 				}
 
 				if !cfg.SkipIndexingBookToc {
+					// Check if the book's Table of Contents (ToC) exists in the map
 					if v, ok := bookTocMap[bookDetail.Book.Slug]; ok {
+						// Assign the ToC to the document's RichCategories
 						document.RichCategories = v
 					} else {
+						// Log a warning if the ToC information is missing
 						log.Debug("missing toc info:", bookDetail.Book.Slug, ",", bookDetail.Book.Name)
 					}
 				}
@@ -268,6 +271,8 @@ func (this *Plugin) collectBooks(connector *common.Connector, datasource *common
 				document.Created = &bookDetail.Book.CreatedAt
 				document.Updated = &bookDetail.Book.UpdatedAt
 
+				document.Cleanup()
+
 				this.save(document)
 			} else {
 				log.Info("skip book:", bookDetail.Book.Name, ",", bookDetail.Book.Public)
@@ -415,6 +420,8 @@ func (this *Plugin) collectDocDetails(connector *common.Connector, datasource *c
 		document.Created = &doc.Doc.CreatedAt
 		document.Updated = &doc.Doc.UpdatedAt
 
+		document.Cleanup()
+
 		this.save(document)
 	}
 }