From b54d454824f964334139a5637663e7e147a8e730 Mon Sep 17 00:00:00 2001
From: Matt Joiner <anacrolix@gmail.com>
Date: Mon, 14 Jul 2025 16:05:02 +1000
Subject: [PATCH] Reduce webseed download waste

Don't close request on cancellation, try to drain it.

Shorten webseed requests to wanted region.
---
 webseed-peer.go       | 11 +++++++++--
 webseed-request.go    |  2 +-
 webseed-requesting.go | 25 ++++++++++++++++++++-----
 webseed/client.go     |  7 ++++++-
 4 files changed, 36 insertions(+), 9 deletions(-)

diff --git a/webseed-peer.go b/webseed-peer.go
index eb6320ce..cba9d3e6 100644
--- a/webseed-peer.go
+++ b/webseed-peer.go
@@ -317,6 +317,9 @@ func (ws *webseedPeer) readChunks(wr *webseedRequest) (err error) {
 		var n int
 		n, err = io.ReadFull(wr.request.Body, buf)
 		ws.peer.readBytes(int64(n))
+		if webseed.PrintDebug && wr.cancelled.Load() {
+			fmt.Printf("webseed read error after cancellation: %v\n", err)
+		}
 		if err != nil {
 			err = fmt.Errorf("reading chunk: %w", err)
 			return
@@ -332,14 +335,18 @@ func (ws *webseedPeer) readChunks(wr *webseedRequest) (err error) {
 		// webseed requests are triggered, we want to ensure our existing request is up to date.
 		wr.next++
 		err = ws.peer.receiveChunk(&msg)
-		stop := err != nil || !ws.keepReading(wr)
+		stop := err != nil || wr.next >= wr.end
+		if !stop {
+			if !ws.keepReading(wr) {
+				wr.Cancel()
+			}
+		}
 		ws.peer.locker().Unlock()
 
 		if err != nil {
 			err = fmt.Errorf("processing chunk: %w", err)
 		}
 		if stop {
-			// TODO: Keep reading until the buffer is drained.
 			return
 		}
 	}
diff --git a/webseed-request.go b/webseed-request.go
index 43f36883..62e2f1ea 100644
--- a/webseed-request.go
+++ b/webseed-request.go
@@ -21,7 +21,7 @@ type webseedRequest struct {
 }
 
 func (me *webseedRequest) Close() {
-	me.request.Cancel()
+	me.request.Close()
 }
 
 // Record that it was exceptionally cancelled.
diff --git a/webseed-requesting.go b/webseed-requesting.go
index 23616163..91e3f75f 100644
--- a/webseed-requesting.go
+++ b/webseed-requesting.go
@@ -159,15 +159,30 @@ func (cl *Client) globalUpdateWebSeedRequests() {
 				continue
 			}
 			t := requestKey.t
-			// Run the request to the end of the file for now. TODO: Set a reasonable end so the
-			// remote doesn't oversend.
 			peer := t.webSeeds[requestKey.url]
 			panicif.NotEq(peer.hostKey, costKey)
 			printPlan()
 			begin := t.getRequestIndexContainingOffset(requestKey.startOffset)
-			// TODO: Find an actual end, so we don't lose lots of data when requests are cancelled.
-			end := t.endRequestIndexForFileIndex(requestKey.fileIndex)
-			panicif.Eq(begin, end)
+			fileEnd := t.endRequestIndexForFileIndex(requestKey.fileIndex)
+			last := begin
+			for {
+				if !t.wantReceiveChunk(last) {
+					break
+				}
+				if last >= fileEnd-1 {
+					break
+				}
+				last++
+			}
+			// Request shouldn't exist if this occurs.
+			panicif.LessThan(last, begin)
+			// Hello C++ my old friend.
+			end := last + 1
+			if webseed.PrintDebug {
+				fmt.Printf("shortened webseed request for %v: [%v-%v) to [%v-%v)\n",
+					requestKey.filePath(), begin, last+1, begin, end)
+			}
+			panicif.GreaterThan(end, fileEnd)
 			peer.spawnRequest(begin, end)
 		}
 	}
diff --git a/webseed/client.go b/webseed/client.go
index a3a1a2d6..9cd59c1f 100644
--- a/webseed/client.go
+++ b/webseed/client.go
@@ -47,7 +47,12 @@ type Request struct {
 
 func (r Request) Cancel() {
 	r.cancel()
-	r.bodyPipe.CloseWithError(context.Canceled)
+}
+
+func (r Request) Close() {
+	// We aren't cancelling because we want to know if we can keep receiving buffered data after
+	// cancellation. PipeReader.Close always returns nil.
+	_ = r.bodyPipe.Close()
 }
 
 type Client struct {
-- 
2.51.0