From 11833b45cfbec636653d64cec3f8db42d70f051e Mon Sep 17 00:00:00 2001
From: Matt Joiner <anacrolix@gmail.com>
Date: Tue, 26 Sep 2023 22:19:51 +1000
Subject: [PATCH] Support scraping from HTTP trackers

---
 cmd/torrent/scrape.go  | 23 +++++----------------
 tracker/client.go      |  2 ++
 tracker/http/scrape.go | 47 ++++++++++++++++++++++++++++++++++++++++++
 tracker/udp.go         | 12 +++++++++++
 tracker/udp/scrape.go  |  8 ++++---
 5 files changed, 71 insertions(+), 21 deletions(-)
 create mode 100644 tracker/http/scrape.go

diff --git a/cmd/torrent/scrape.go b/cmd/torrent/scrape.go
index ce5fc06c..01a59b7f 100644
--- a/cmd/torrent/scrape.go
+++ b/cmd/torrent/scrape.go
@@ -3,12 +3,11 @@ package main
 import (
 	"context"
 	"fmt"
-	"net/url"
 
 	"github.com/davecgh/go-spew/spew"
 
 	"github.com/anacrolix/torrent"
-	"github.com/anacrolix/torrent/tracker/udp"
+	"github.com/anacrolix/torrent/tracker"
 )
 
 type scrapeCfg struct {
@@ -17,25 +16,13 @@ type scrapeCfg struct {
 }
 
 func scrape(flags scrapeCfg) error {
-	trackerUrl, err := url.Parse(flags.Tracker)
+	cc, err := tracker.NewClient(flags.Tracker, tracker.NewClientOpts{})
 	if err != nil {
-		return fmt.Errorf("parsing tracker url: %w", err)
-	}
-	cc, err := udp.NewConnClient(udp.NewConnClientOpts{
-		Network: trackerUrl.Scheme,
-		Host:    trackerUrl.Host,
-		//Ipv6:    nil,
-		//Logger:  log.Logger{},
-	})
-	if err != nil {
-		return fmt.Errorf("creating new udp tracker conn client: %w", err)
+		err = fmt.Errorf("creating new tracker client: %w", err)
+		return err
 	}
 	defer cc.Close()
-	var ihs []udp.InfoHash
-	for _, ih := range flags.InfoHashes {
-		ihs = append(ihs, ih)
-	}
-	scrapeOut, err := cc.Client.Scrape(context.TODO(), ihs)
+	scrapeOut, err := cc.Scrape(context.TODO(), flags.InfoHashes)
 	if err != nil {
 		return fmt.Errorf("scraping: %w", err)
 	}
diff --git a/tracker/client.go b/tracker/client.go
index 1aaf2561..3b7e2aba 100644
--- a/tracker/client.go
+++ b/tracker/client.go
@@ -9,10 +9,12 @@ import (
 
 	trHttp "github.com/anacrolix/torrent/tracker/http"
 	"github.com/anacrolix/torrent/tracker/udp"
+	"github.com/anacrolix/torrent/types/infohash"
 )
 
 type Client interface {
 	Announce(context.Context, AnnounceRequest, AnnounceOpt) (AnnounceResponse, error)
+	Scrape(ctx context.Context, ihs []infohash.T) (out udp.ScrapeResponse, err error)
 	Close() error
 }
 
diff --git a/tracker/http/scrape.go b/tracker/http/scrape.go
new file mode 100644
index 00000000..6940370f
--- /dev/null
+++ b/tracker/http/scrape.go
@@ -0,0 +1,47 @@
+package httpTracker
+
+import (
+	"context"
+	"log"
+	"net/http"
+	"net/url"
+
+	"github.com/anacrolix/torrent/bencode"
+	"github.com/anacrolix/torrent/tracker/udp"
+	"github.com/anacrolix/torrent/types/infohash"
+)
+
+type scrapeResponse struct {
+	Files files `bencode:"files"`
+}
+
+// Bencode should support bencode.Unmarshalers from a string in the dict key position.
+type files = map[string]udp.ScrapeInfohashResult
+
+func (cl Client) Scrape(ctx context.Context, ihs []infohash.T) (out udp.ScrapeResponse, err error) {
+	_url := cl.url_.JoinPath("..", "scrape")
+	query, err := url.ParseQuery(_url.RawQuery)
+	if err != nil {
+		return
+	}
+	for _, ih := range ihs {
+		query.Add("info_hash", ih.AsString())
+	}
+	_url.RawQuery = query.Encode()
+	log.Printf("%q", _url.String())
+	req, err := http.NewRequestWithContext(ctx, http.MethodGet, _url.String(), nil)
+	if err != nil {
+		return
+	}
+	resp, err := cl.hc.Do(req)
+	if err != nil {
+		return
+	}
+	defer resp.Body.Close()
+	var decodedResp scrapeResponse
+	err = bencode.NewDecoder(resp.Body).Decode(&decodedResp)
+	for _, ih := range ihs {
+		out = append(out, decodedResp.Files[ih.AsString()])
+	}
+	return
+}
diff --git a/tracker/udp.go b/tracker/udp.go
index db486948..cf681887 100644
--- a/tracker/udp.go
+++ b/tracker/udp.go
@@ -4,8 +4,11 @@ import (
 	"context"
 	"encoding/binary"
 
+	"github.com/anacrolix/generics"
+
 	trHttp "github.com/anacrolix/torrent/tracker/http"
 	"github.com/anacrolix/torrent/tracker/udp"
+	"github.com/anacrolix/torrent/types/infohash"
 )
 
 type udpClient struct {
@@ -13,6 +16,15 @@ type udpClient struct {
 	requestUri string
 }
 
+func (c *udpClient) Scrape(ctx context.Context, ihs []infohash.T) (out udp.ScrapeResponse, err error) {
+	return c.cl.Client.Scrape(
+		ctx,
+		generics.SliceMap(ihs, func(from infohash.T) udp.InfoHash {
+			return from
+		}),
+	)
+}
+
 func (c *udpClient) Close() error {
 	return c.cl.Close()
 }
diff --git a/tracker/udp/scrape.go b/tracker/udp/scrape.go
index 331f109e..13a69b99 100644
--- a/tracker/udp/scrape.go
+++ b/tracker/udp/scrape.go
@@ -5,7 +5,9 @@ type ScrapeRequest []InfoHash
 type ScrapeResponse []ScrapeInfohashResult
 
 type ScrapeInfohashResult struct {
-	Seeders   int32
-	Completed int32
-	Leechers  int32
+	// I'm not sure why the fields are named differently for HTTP scrapes.
+	// https://www.bittorrent.org/beps/bep_0048.html
+	Seeders   int32 `bencode:"complete"`
+	Completed int32 `bencode:"downloaded"`
+	Leechers  int32 `bencode:"incomplete"`
 }
-- 
2.44.0