From 17c4df85a9306bbba53c4c7549a0fa09ca31874a Mon Sep 17 00:00:00 2001
From: Matt Joiner <anacrolix@gmail.com>
Date: Wed, 4 Jun 2025 10:39:35 +1000
Subject: [PATCH] Limit webseeds by host and reduce default to 5 per Client

---
 client.go             |  8 +++++---
 torrent.go            | 11 +++++++++++
 webseed-peer.go       | 23 +++++++++++------------
 webseed-requesting.go |  7 +++++++
 4 files changed, 34 insertions(+), 15 deletions(-)

diff --git a/client.go b/client.go
index f7ee73e7..dbe0b31e 100644
--- a/client.go
+++ b/client.go
@@ -29,6 +29,7 @@ import (
 	"github.com/anacrolix/log"
 	"github.com/anacrolix/missinggo/v2"
 	"github.com/anacrolix/missinggo/v2/bitmap"
+	"github.com/anacrolix/missinggo/v2/panicif"
 	"github.com/anacrolix/missinggo/v2/pproffd"
 	"github.com/anacrolix/sync"
 	"github.com/cespare/xxhash"
@@ -94,7 +95,7 @@ type Client struct {
 	numHalfOpen   int
 
 	websocketTrackers  websocketTrackers
-	numWebSeedRequests int
+	numWebSeedRequests map[webseedHostKeyHandle]int
 
 	activeAnnounceLimiter limiter.Instance
 	httpClient            *http.Client
@@ -1933,8 +1934,9 @@ func (cl *Client) Stats() ClientStats {
 	return cl.statsLocked()
 }
 
-func (cl *Client) underWebSeedHttpRequestLimit() bool {
-	return cl.numWebSeedRequests < 10
+func (cl *Client) underWebSeedHttpRequestLimit(key webseedHostKeyHandle) bool {
+	panicif.Zero(key)
+	return cl.numWebSeedRequests[key] < 5
 }
 
 func (cl *Client) countWebSeedHttpRequests() (num int) {
diff --git a/torrent.go b/torrent.go
index dc9571ba..1df51972 100644
--- a/torrent.go
+++ b/torrent.go
@@ -19,6 +19,7 @@ import (
 	"strings"
 	"text/tabwriter"
 	"time"
+	"unique"
 	"unsafe"
 
 	"github.com/RoaringBitmap/roaring"
@@ -3049,6 +3050,7 @@ func (t *Torrent) addWebSeed(url string, opts ...AddWebSeedsOpt) bool {
 				}
 			},
 		},
+		hostKey: t.deriveWebSeedHostKey(url),
 	}
 	ws.peer.initRequestState()
 	for _, opt := range opts {
@@ -3076,6 +3078,15 @@ func (t *Torrent) addWebSeed(url string, opts ...AddWebSeedsOpt) bool {
 	return true
 }
 
+func (t *Torrent) deriveWebSeedHostKey(urlStr string) (ret webseedHostKeyHandle) {
+	u, err := url.Parse(urlStr)
+	if err != nil {
+		t.slogger().Warn("error parsing webseed URL", "url", urlStr, "err", err)
+		return unique.Make(webseedHostKey(urlStr))
+	}
+	return unique.Make(webseedHostKey(u.Hostname()))
+}
+
 func (t *Torrent) peerIsActive(p *Peer) (active bool) {
 	t.iterPeers(func(p1 *Peer) {
 		if p1 == p {
diff --git a/webseed-peer.go b/webseed-peer.go
index bedbb927..f02845a4 100644
--- a/webseed-peer.go
+++ b/webseed-peer.go
@@ -2,7 +2,6 @@ package torrent
 
 import (
 	"context"
-	"errors"
 	"fmt"
 	"io"
 	"iter"
@@ -13,7 +12,7 @@ import (
 
 	"github.com/RoaringBitmap/roaring"
 	g "github.com/anacrolix/generics"
-
+	"github.com/anacrolix/missinggo/v2/panicif"
 	"github.com/anacrolix/torrent/metainfo"
 	pp "github.com/anacrolix/torrent/peer_protocol"
 	"github.com/anacrolix/torrent/webseed"
@@ -26,6 +25,7 @@ type webseedPeer struct {
 	activeRequests   map[*webseedRequest]struct{}
 	locker           sync.Locker
 	lastUnhandledErr time.Time
+	hostKey          webseedHostKeyHandle
 }
 
 func (me *webseedPeer) nominalMaxRequests() maxRequests {
@@ -44,7 +44,11 @@ func (me *webseedPeer) numRequests() int {
 }
 
 func (me *webseedPeer) shouldUpdateRequests() bool {
-	return me.numRequests() < me.client.MaxRequests && me.peer.t.cl.underWebSeedHttpRequestLimit()
+	return me.moreRequestsAllowed()
+}
+
+func (me *webseedPeer) moreRequestsAllowed() bool {
+	return me.numRequests() < me.client.MaxRequests && me.peer.t.cl.underWebSeedHttpRequestLimit(me.hostKey)
 }
 
 func (me *webseedPeer) updateRequests() {
@@ -133,7 +137,8 @@ func (ws *webseedPeer) spawnRequest(begin, end RequestIndex) {
 		end:     end,
 	}
 	ws.activeRequests[&wsReq] = struct{}{}
-	ws.peer.t.cl.numWebSeedRequests++
+	panicif.Zero(ws.hostKey)
+	ws.peer.t.cl.numWebSeedRequests[ws.hostKey]++
 	ws.slogger().Debug(
 		"starting webseed request",
 		"begin", begin,
@@ -175,19 +180,13 @@ func (ws *webseedPeer) runRequest(webseedRequest *webseedRequest) {
 
 func (ws *webseedPeer) deleteActiveRequest(wr *webseedRequest) {
 	g.MustDelete(ws.activeRequests, wr)
-	ws.peer.t.cl.numWebSeedRequests--
+	ws.peer.t.cl.numWebSeedRequests[ws.hostKey]--
 }
 
 func (ws *webseedPeer) spawnRequests() {
 	next, stop := iter.Pull(ws.inactiveRequests())
 	defer stop()
-	for {
-		if !ws.peer.t.cl.underWebSeedHttpRequestLimit() {
-			break
-		}
-		if ws.numRequests() >= ws.client.MaxRequests {
-			break
-		}
+	for ws.moreRequestsAllowed() {
 		req, ok := next()
 		if !ok {
 			break
diff --git a/webseed-requesting.go b/webseed-requesting.go
index 66893c11..607cba40 100644
--- a/webseed-requesting.go
+++ b/webseed-requesting.go
@@ -1,10 +1,17 @@
 package torrent
 
 import (
+	"unique"
+
 	"github.com/anacrolix/torrent/metainfo"
 	requestStrategy "github.com/anacrolix/torrent/request-strategy"
 )
 
+type (
+	webseedHostKey       string
+	webseedHostKeyHandle = unique.Handle[webseedHostKey]
+)
+
 /*
 - Go through all the requestable pieces in order of priority, availability, whether there are peer requests, partial, infohash.
 - For each piece calculate files involved. Record each file not seen before and the piece index.
-- 
2.51.0