Merge pull request #7699 from ipfs/fix/dnslink-dir-listing

fix(dnslink-gw): breadcrumbs and CID column when dir listing

assets/bindata_version_hash.go

@@ -2,5 +2,5 @@
 package assets
 
 const (
-	BindataVersionHash = "514e5ae28d8adb84955801b56ef47aca44bf9cc8"
+	BindataVersionHash = "605b5945438e1fe2eaf8a6571cca7ecda12d5599"
 )

core/corehttp/gateway_handler.go

@@ -392,8 +392,9 @@ func (i *gatewayHandler) getOrHeadHandler(w http.ResponseWriter, r *http.Request
 
 	hash := resolvedPath.Cid().String()
 
-	// Storage for gateway URL to be used when linking to other rootIDs. This
-	// will be blank unless subdomain resolution is being used for this request.
+	// Gateway root URL to be used when linking to other rootIDs.
+	// This will be blank unless subdomain or DNSLink resolution is being used
+	// for this request.
 	var gwURL string
 
 	// Get gateway hostname and build gateway URL.
@@ -403,13 +404,16 @@ func (i *gatewayHandler) getOrHeadHandler(w http.ResponseWriter, r *http.Request
 		gwURL = ""
 	}
 
+	dnslink := hasDNSLinkOrigin(gwURL, urlPath)
+
 	// See comment above where originalUrlPath is declared.
 	tplData := listingTemplateData{
 		GatewayURL:  gwURL,
+		DNSLink:     dnslink,
 		Listing:     dirListing,
 		Size:        size,
 		Path:        urlPath,
-		Breadcrumbs: breadcrumbs(urlPath),
+		Breadcrumbs: breadcrumbs(urlPath, dnslink),
 		BackLink:    backLink,
 		Hash:        hash,
 	}

core/corehttp/gateway_indexPage.go

@@ -13,6 +13,7 @@ import (
 // structs for directory listing
 type listingTemplateData struct {
 	GatewayURL  string
+	DNSLink     bool
 	Listing     []directoryItem
 	Size        string
 	Path        string
@@ -34,16 +35,16 @@ type breadcrumb struct {
 	Path string
 }
 
-func breadcrumbs(urlPath string) []breadcrumb {
+func breadcrumbs(urlPath string, dnslinkOrigin bool) []breadcrumb {
 	var ret []breadcrumb
 
 	p, err := ipfspath.ParsePath(urlPath)
 	if err != nil {
 		// No breadcrumbs, fallback to bare Path in template
 		return ret
 	}
-
 	segs := p.Segments()
+	contentRoot := segs[1]
 	for i, seg := range segs {
 		if i == 0 {
 			ret = append(ret, breadcrumb{Name: seg})
@@ -55,13 +56,38 @@ func breadcrumbs(urlPath string) []breadcrumb {
 		}
 	}
 
+	// Drop the /ipns/<fqdn> prefix from breadcrumb Paths when directory
+	// listing on a DNSLink website (loaded due to Host header in HTTP
+	// request).  Necessary because the hostname most likely won't have a
+	// public gateway mounted.
+	if dnslinkOrigin {
+		prefix := "/ipns/" + contentRoot
+		for i, crumb := range ret {
+			if strings.HasPrefix(crumb.Path, prefix) {
+				ret[i].Path = strings.Replace(crumb.Path, prefix, "", 1)
+			}
+		}
+		// Make contentRoot breadcrumb link to the website root
+		ret[1].Path = "/"
+	}
+
 	return ret
 }
 
 func shortHash(hash string) string {
 	return (hash[0:4] + "\u2026" + hash[len(hash)-4:])
 }
 
+// helper to detect DNSLink website context
+// (when hostname from gwURL is matching /ipns/<fqdn> in path)
+func hasDNSLinkOrigin(gwURL string, path string) bool {
+	if gwURL != "" {
+		fqdn := stripPort(strings.TrimPrefix(gwURL, "//"))
+		return strings.HasPrefix(path, "/ipns/"+fqdn)
+	}
+	return false
+}
+
 var listingTemplate *template.Template
 
 func init() {

core/corehttp/gateway_test.go

@@ -391,6 +391,10 @@ func TestIPNSHostnameRedirect(t *testing.T) {
 	}
 }
 
+// Test directory listing on DNSLink website
+// (scenario when Host header is the same as URL hostname)
+// This is basic regression test: additional end-to-end tests
+// can be found in test/sharness/t0115-gateway-dir-listing.sh
 func TestIPNSHostnameBacklinks(t *testing.T) {
 	ns := mockNamesys{}
 	ts, api, ctx := newTestServerAndNode(t, ns)
@@ -437,15 +441,15 @@ func TestIPNSHostnameBacklinks(t *testing.T) {
 		t.Fatal(err)
 	}
 
-	// expect correct backlinks
+	// expect correct links
 	body, err := ioutil.ReadAll(res.Body)
 	if err != nil {
 		t.Fatalf("error reading response: %s", err)
 	}
 	s := string(body)
 	t.Logf("body: %s\n", string(body))
 
-	if !matchPathOrBreadcrumbs(s, "/ipns/<a href=\"/ipns/example.net\">example.net</a>/<a href=\"/ipns/example.net/foo%3F%20%23%3C%27\">foo? #&lt;&#39;</a>") {
+	if !matchPathOrBreadcrumbs(s, "/ipns/<a href=\"//example.net/\">example.net</a>/<a href=\"//example.net/foo%3F%20%23%3C%27\">foo? #&lt;&#39;</a>") {
 		t.Fatalf("expected a path in directory listing")
 	}
 	if !strings.Contains(s, "<a href=\"/foo%3F%20%23%3C%27/./..\">") {
@@ -454,6 +458,10 @@ func TestIPNSHostnameBacklinks(t *testing.T) {
 	if !strings.Contains(s, "<a href=\"/foo%3F%20%23%3C%27/file.txt\">") {
 		t.Fatalf("expected file in directory listing")
 	}
+	if !strings.Contains(s, "<a class=\"ipfs-hash\" href=\"https://cid.ipfs.io/#") {
+		// /~https://github.com/ipfs/dir-index-html/issues/42
+		t.Fatalf("expected links to cid.ipfs.io in CID column when on DNSLink website")
+	}
 	if !strings.Contains(s, k2.Cid().String()) {
 		t.Fatalf("expected hash in directory listing")
 	}
@@ -487,6 +495,10 @@ func TestIPNSHostnameBacklinks(t *testing.T) {
 	if !strings.Contains(s, "<a href=\"/file.txt\">") {
 		t.Fatalf("expected file in directory listing")
 	}
+	if !strings.Contains(s, "<a class=\"ipfs-hash\" href=\"https://cid.ipfs.io/#") {
+		// /~https://github.com/ipfs/dir-index-html/issues/42
+		t.Fatalf("expected links to cid.ipfs.io in CID column when on DNSLink website")
+	}
 	if !strings.Contains(s, k.Cid().String()) {
 		t.Fatalf("expected hash in directory listing")
 	}
@@ -511,7 +523,7 @@ func TestIPNSHostnameBacklinks(t *testing.T) {
 	s = string(body)
 	t.Logf("body: %s\n", string(body))
 
-	if !matchPathOrBreadcrumbs(s, "/ipns/<a href=\"/ipns/example.net\">example.net</a>/<a href=\"/ipns/example.net/foo%3F%20%23%3C%27\">foo? #&lt;&#39;</a>/<a href=\"/ipns/example.net/foo%3F%20%23%3C%27/bar\">bar</a>") {
+	if !matchPathOrBreadcrumbs(s, "/ipns/<a href=\"//example.net/\">example.net</a>/<a href=\"//example.net/foo%3F%20%23%3C%27\">foo? #&lt;&#39;</a>/<a href=\"//example.net/foo%3F%20%23%3C%27/bar\">bar</a>") {
 		t.Fatalf("expected a path in directory listing")
 	}
 	if !strings.Contains(s, "<a href=\"/foo%3F%20%23%3C%27/bar/./..\">") {
@@ -545,7 +557,7 @@ func TestIPNSHostnameBacklinks(t *testing.T) {
 	s = string(body)
 	t.Logf("body: %s\n", string(body))
 
-	if !matchPathOrBreadcrumbs(s, "/ipns/<a href=\"/ipns/example.net\">example.net</a>") {
+	if !matchPathOrBreadcrumbs(s, "/ipns/<a href=\"//example.net/\">example.net</a>") {
 		t.Fatalf("expected a path in directory listing")
 	}
 	if !strings.Contains(s, "<a href=\"/good-prefix/\">") {

core/corehttp/hostname.go

@@ -129,7 +129,7 @@ func HostnameOption() ServeOption {
 				if !gw.NoDNSLink && isDNSLinkRequest(r.Context(), coreAPI, host) {
 					// rewrite path and handle as DNSLink
 					r.URL.Path = "/ipns/" + stripPort(host) + r.URL.Path
-					childMux.ServeHTTP(w, r)
+					childMux.ServeHTTP(w, withHostnameContext(r, host))
 					return
 				}
 
@@ -143,10 +143,6 @@ func HostnameOption() ServeOption {
 			if gw, hostname, ns, rootID, ok := knownSubdomainDetails(host, knownGateways); ok {
 				// Looks like we're using a known gateway in subdomain mode.
 
-				// Add gateway hostname context for linking to other root ids.
-				// Example: localhost/ipfs/{cid}
-				ctx := context.WithValue(r.Context(), "gw-hostname", hostname)
-
 				// Assemble original path prefix.
 				pathPrefix := "/" + ns + "/" + rootID
 
@@ -201,7 +197,7 @@ func HostnameOption() ServeOption {
 				r.URL.Path = pathPrefix + r.URL.Path
 
 				// Serve path request
-				childMux.ServeHTTP(w, r.WithContext(ctx))
+				childMux.ServeHTTP(w, withHostnameContext(r, hostname))
 				return
 			}
 			// We don't have a known gateway. Fallback on DNSLink lookup
@@ -213,7 +209,7 @@ func HostnameOption() ServeOption {
 			if !cfg.Gateway.NoDNSLink && isDNSLinkRequest(r.Context(), coreAPI, host) {
 				// rewrite path and handle as DNSLink
 				r.URL.Path = "/ipns/" + stripPort(host) + r.URL.Path
-				childMux.ServeHTTP(w, r)
+				childMux.ServeHTTP(w, withHostnameContext(r, host))
 				return
 			}
 
@@ -234,6 +230,17 @@ type wildcardHost struct {
 	spec *config.GatewaySpec
 }
 
+// Extends request context to include hostname of a canonical gateway root
+// (subdomain root or dnslink fqdn)
+func withHostnameContext(r *http.Request, hostname string) *http.Request {
+	// This is required for links on directory listing pages to work correctly
+	// on subdomain and dnslink gateways. While DNSlink could read value from
+	// Host header, subdomain gateways have more comples rules (knownSubdomainDetails)
+	// More: /~https://github.com/ipfs/dir-index-html/issues/42
+	ctx := context.WithValue(r.Context(), "gw-hostname", hostname)
+	return r.WithContext(ctx)
+}
+
 func prepareKnownGateways(publicGateways map[string]*config.GatewaySpec) gatewayHosts {
 	var hosts gatewayHosts
 

test/sharness/t0114-gateway-subdomains.sh

@@ -432,7 +432,8 @@ test_expect_success "valid parent directory path in directory listing at {cid}.i
   test_should_contain "<a href=\"/ipfs/ipns/bar\">bar</a>" list_response
 '
 
-# Note we test for sneaky subdir names  {cid}.ipfs.example.com/ipfs/ipns/ :^)
+# Note 1: we test for sneaky subdir names  {cid}.ipfs.example.com/ipfs/ipns/ :^)
+# Note 2: example.com/ipfs/.. present in HTML will be redirected to subdomain, so this is expected behavior
 test_expect_success "valid breadcrumb links in the header of directory listing at {cid}.ipfs.example.com/sub/dir" '
   curl -s -H "Host: $DIR_FQDN" http://127.0.0.1:$GWAY_PORT/ipfs/ipns/ > list_response &&
   test_should_contain "Index of" list_response &&

test/sharness/t0115-gateway-dir-listing.sh

@@ -0,0 +1,146 @@
+#!/usr/bin/env bash
+#
+# Copyright (c) Protocol Labs
+
+test_description="Test directory listing (dir-index-html) on the HTTP gateway"
+
+
+. lib/test-lib.sh
+
+## ============================================================================
+## Start IPFS Node and prepare test CIDs
+## ============================================================================
+
+test_expect_success "ipfs init" '
+  export IPFS_PATH="$(pwd)/.ipfs" &&
+  ipfs init --profile=test > /dev/null
+'
+
+test_launch_ipfs_daemon --offline
+
+test_expect_success "Add the test directory" '
+  mkdir -p rootDir/ipfs &&
+  mkdir -p rootDir/ipns &&
+  mkdir -p rootDir/api &&
+  mkdir -p rootDir/ą/ę &&
+  echo "I am a txt file on path with utf8" > rootDir/ą/ę/file-źł.txt &&
+  echo "I am a txt file in confusing /api dir" > rootDir/api/file.txt &&
+  echo "I am a txt file in confusing /ipfs dir" > rootDir/ipfs/file.txt &&
+  echo "I am a txt file in confusing /ipns dir" > rootDir/ipns/file.txt &&
+  DIR_CID=$(ipfs add -Qr --cid-version 1 rootDir) &&
+  FILE_CID=$(ipfs files stat /ipfs/$DIR_CID/ą/ę/file-źł.txt | head -1)
+'
+
+## ============================================================================
+## Test dir listing on path gateway (eg. 127.0.0.1:8080/ipfs/)
+## ============================================================================
+
+test_expect_success "path gw: backlink on root CID should point at self" '
+  curl -sD - http://127.0.0.1:$GWAY_PORT/ipfs/${DIR_CID}/ > list_response &&
+  test_should_contain "Index of" list_response &&
+  test_should_contain "<a href=\"/ipfs/$DIR_CID/\">..</a>" list_response
+'
+
+test_expect_success "path gw: Etag should be present" '
+  curl -sD - http://127.0.0.1:$GWAY_PORT/ipfs/${DIR_CID}/ą/ę > list_response &&
+  test_should_contain "Index of" list_response &&
+  test_should_contain "Etag: \"DirIndex-" list_response
+'
+
+test_expect_success "path gw: breadcrumbs should point at /ipfs namespace mounted at Origin root" '
+  test_should_contain "/ipfs/<a href=\"/ipfs/$DIR_CID\">$DIR_CID</a>/<a href=\"/ipfs/$DIR_CID/%C4%85\">ą</a>/<a href=\"/ipfs/$DIR_CID/%C4%85/%C4%99\">ę</a>" list_response
+'
+
+test_expect_success "path gw: backlink should point at parent directory" '
+  test_should_contain "<a href=\"/ipfs/$DIR_CID/%C4%85/%C4%99/..\">..</a>" list_response
+'
+
+test_expect_success "path gw: name column should be a link to its content path" '
+  test_should_contain "<a href=\"/ipfs/$DIR_CID/%C4%85/%C4%99/file-%C5%BA%C5%82.txt\">file-źł.txt</a>" list_response
+'
+
+test_expect_success "path gw: hash column should be a CID link with filename param" '
+  test_should_contain "<a class=\"ipfs-hash\" href=\"/ipfs/$FILE_CID?filename=file-%25C5%25BA%25C5%2582.txt\">" list_response
+'
+
+## ============================================================================
+## Test dir listing on subdomain gateway (eg. <cid>.ipfs.localhost:8080)
+## ============================================================================
+
+DIR_HOSTNAME="${DIR_CID}.ipfs.localhost"
+# note: we skip DNS lookup by running curl with --resolve $DIR_HOSTNAME:127.0.0.1
+
+test_expect_success "path gw: backlink on root CID should point origin root" '
+  curl -sD - --resolve $DIR_HOSTNAME:$GWAY_PORT:127.0.0.1 http://$DIR_HOSTNAME:$GWAY_PORT/ > list_response &&
+  test_should_contain "Index of" list_response &&
+  test_should_contain "<a href=\"/\">..</a>" list_response
+'
+
+test_expect_success "path gw: Etag should be present" '
+  curl -sD - --resolve $DIR_HOSTNAME:$GWAY_PORT:127.0.0.1 http://$DIR_HOSTNAME:$GWAY_PORT/ą/ę > list_response &&
+  test_should_contain "Index of" list_response &&
+  test_should_contain "Etag: \"DirIndex-" list_response
+'
+
+test_expect_success "subdomain gw: breadcrumbs should leverage path-based router mounted on the parent domain" '
+  test_should_contain "/ipfs/<a href=\"//localhost:$GWAY_PORT/ipfs/$DIR_CID\">$DIR_CID</a>/<a href=\"//localhost:$GWAY_PORT/ipfs/$DIR_CID/%C4%85\">ą</a>/<a href=\"//localhost:$GWAY_PORT/ipfs/$DIR_CID/%C4%85/%C4%99\">ę</a>" list_response
+'
+
+test_expect_success "path gw: name column should be a link to content root mounted at subdomain origin" '
+  test_should_contain "<a href=\"/%C4%85/%C4%99/file-%C5%BA%C5%82.txt\">file-źł.txt</a>" list_response
+'
+
+test_expect_success "path gw: hash column should be a CID link to path router with filename param" '
+  test_should_contain "<a class=\"ipfs-hash\" href=\"//localhost:$GWAY_PORT/ipfs/$FILE_CID?filename=file-%25C5%25BA%25C5%2582.txt\">" list_response
+'
+
+## ============================================================================
+## Test dir listing on DNSLink gateway (eg. example.com)
+## ============================================================================
+
+# DNSLink test requires a daemon in online mode with precached /ipns/ mapping
+test_kill_ipfs_daemon
+DNSLINK_HOSTNAME="website.example.com"
+export IPFS_NS_MAP="$DNSLINK_HOSTNAME:/ipfs/$DIR_CID"
+test_launch_ipfs_daemon
+
+# Note that:
+# - this type of gateway is also tested in gateway_test.go#TestIPNSHostnameBacklinks
+#   (go tests and sharness tests should be kept in sync)
+# - we skip DNS lookup by running curl with --resolve $DNSLINK_HOSTNAME:127.0.0.1
+
+test_expect_success "dnslink gw: backlink on root CID should point origin root" '
+  curl -v -sD - --resolve $DNSLINK_HOSTNAME:$GWAY_PORT:127.0.0.1 http://$DNSLINK_HOSTNAME:$GWAY_PORT/ > list_response &&
+  test_should_contain "Index of" list_response &&
+  test_should_contain "<a href=\"/\">..</a>" list_response
+'
+
+test_expect_success "dnslink gw: Etag should be present" '
+  curl -sD - --resolve $DNSLINK_HOSTNAME:$GWAY_PORT:127.0.0.1 http://$DNSLINK_HOSTNAME:$GWAY_PORT/ą/ę > list_response &&
+  test_should_contain "Index of" list_response &&
+  test_should_contain "Etag: \"DirIndex-" list_response
+'
+
+test_expect_success "dnslink gw: breadcrumbs should point at content root mounted at dnslink origin" '
+  test_should_contain "/ipns/<a href=\"//$DNSLINK_HOSTNAME:$GWAY_PORT/\">website.example.com</a>/<a href=\"//$DNSLINK_HOSTNAME:$GWAY_PORT/%C4%85\">ą</a>/<a href=\"//$DNSLINK_HOSTNAME:$GWAY_PORT/%C4%85/%C4%99\">ę</a>" list_response
+'
+
+test_expect_success "dnslink gw: name column should be a link to content root mounted at dnslink origin" '
+  test_should_contain "<a href=\"/%C4%85/%C4%99/file-%C5%BA%C5%82.txt\">file-źł.txt</a>" list_response
+'
+
+# DNSLink websites don't have public gateway mounted by default
+# See: /~https://github.com/ipfs/dir-index-html/issues/42
+test_expect_success "dnslink gw: hash column should be a CID link to cid.ipfs.io" '
+  test_should_contain "<a class=\"ipfs-hash\" href=\"https://cid.ipfs.io/#$FILE_CID\" target=\"_blank\" rel=\"noreferrer noopener\">" list_response
+'
+
+## ============================================================================
+## End of tests, cleanup
+## ============================================================================
+
+test_kill_ipfs_daemon
+test_expect_success "clean up ipfs dir" '
+  rm -rf "$IPFS_PATH"
+'
+test_done