recursive fetching of reply parents + thread deduplication

1 files changed, 40 insertions, 0 deletions

diff --git a/backend/notes/note.go b/backend/notes/note.go
index 3b32aa1..591caff 100644
--- a/backend/notes/note.go
+++ b/backend/notes/note.go
@@ -1,6 +1,7 @@
 package notes
 
 import (
+	"sort"
 	"time"
 )
 
@@ -24,3 +25,42 @@ type Facet struct {
 	Type      string `json:"type"`
 	Value     string `json:"value"` // the link destination, whether url, profile or tag
 }
+
+func isChainContained(smallChain, bigChain []string) bool {
+	if len(smallChain) > len(bigChain) {
+		return false
+	}
+
+	// check if smallChain is a suffix of bigChain
+	offset := len(bigChain) - len(smallChain)
+	for i, uri := range smallChain {
+		if bigChain[offset+i] != uri {
+			return false
+		}
+	}
+	return true
+}
+
+func DeduplicateThreads(posts []Note) []Note {
+	// sort by parent chain length (longest first)
+	sort.Slice(posts, func(i, j int) bool {
+		return len(posts[i].Parents) > len(posts[j].Parents)
+	})
+
+	filtered := []Note{}
+	for _, post := range posts {
+		isRedundant := false
+		for _, kept := range filtered {
+			// check if this post's chain is contained in any kept post
+			if isChainContained(post.Parents, kept.Parents) && post.URI == kept.URI {
+				isRedundant = true
+				break
+			}
+		}
+		if !isRedundant {
+			filtered = append(filtered, post)
+		}
+	}
+
+	return filtered
+}