Merge remote-tracking branch 'origin/feature/293a89c5-bd98-4de4-9afd-5ca371e5e679-recipe-detail-and-kitchen-reading-mode' into task/source-attribution-links_c1a97243-3e41-42fc-a1a0-1ac684ccb463

safeSourceURL accepted hosts that Go's url.Parse takes but the browser's
URL parser refuses, such as "https://[bad]/curry", so the detail API could
send a source.url that sourceUrl() in web/src/dinners/detail.ts rejects,
failing the whole recipe screen. The server now also requires a host the
browser parses as it stands: a bracketed host must be an IPv6 address
without a zone; any other must be ASCII letters, digits and hyphens in
non-empty dot-separated labels, with no punycode label and, when its last
label is a number, a valid dotted quad; and the host must be spelled out,
not percent-encoded. Any other link is withheld as before.

web/src/dinners/testdata/source_url_cases.json pins the source object
the handler sends for each stored link. The Go server test checks the
handler against it, and the recipe screen test renders each source from
it, so the two parsers cannot drift apart on those links.

Task: c1a97243-3e41-42fc-a1a0-1ac684ccb463

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

internal/dinners/detail.go

@@ -58,15 +58,16 @@ type Ingredient struct {
 	RawText string  `json:"raw_text"`
 }
 
-// Source is the site a recipe was fetched from. It has no url: the page's
-// link, pasted or offered as a candidate, may carry a personal token, and
-// no stored link can be shown to be free of one, so like a dinner's
-// recipe_url and a candidate's page it never leaves the server. SiteName
-// is the name the page itself gives its site, and null when it gives none
-// or the stored name could have come from a link (see pageSiteName), so a
-// client shows "Recipe from {site}" only when SiteName is not null.
+// Source is the site a recipe was fetched from. SiteName is the name the
+// page itself gives its site, and null when it gives none or the stored
+// name could have come from a link (see pageSiteName), so a client shows
+// "Recipe from {site}" only when SiteName is not null. URL is the page's
+// link only when it is verifiably free of secrets (see safeSourceURL), and
+// null otherwise: a link pasted or offered as a candidate may carry a
+// personal token, so a client links the credit only when URL is not null.
 type Source struct {
 	SiteName *string `json:"site_name"`
+	URL      *string `json:"url"`
 }
 
 // Get returns dinner id with its recipe and the reason it was suggested,
@@ -149,7 +150,7 @@ func readRecipe(ctx context.Context, q *db.Queries, recipeID string, d *Detail)
 		if err != nil {
 			return err
 		}
-		d.Source = &Source{SiteName: site}
+		d.Source = &Source{SiteName: site, URL: safeSourceURL(rec.SourceUrl)}
 	}
 	ings, err := q.ListRecipeIngredients(ctx, recipeID)
 	if err != nil {
@@ -209,6 +210,69 @@ func pageSiteName(stored sql.NullString, sourceURL string) (*string, error) {
 	return &name, nil
 }
 
+// safeSourceURL is sourceURL when it is verifiably free of secrets — an
+// https link to a host, with no user or password, no query (not even a
+// bare "?") and no fragment (not even a bare "#") — else nil. Those are
+// the places a link carries a token; a link failing any check, or not
+// parsing, is withheld rather than trimmed, since a trimmed link is not the
+// page the recipe came from. The client (sourceUrl in
+// web/src/dinners/detail.ts) parses the link with the browser's URL parser
+// and refuses the whole recipe when it fails, so the host and port must
+// also be ones it parses: see browserHost.
+func safeSourceURL(sourceURL string) *string {
+	u, err := url.Parse(sourceURL)
+	if err != nil || u.Scheme != "https" || u.Opaque != "" || u.Host == "" || u.User != nil ||
+		u.RawQuery != "" || u.ForceQuery || u.Fragment != "" || strings.Contains(sourceURL, "#") {
+		return nil
+	}
+	// url.Parse decodes a percent-encoded host; the link must spell its
+	// host out as it is checked.
+	authority, _, _ := strings.Cut(sourceURL[len(u.Scheme)+len("://"):], "/")
+	if authority != u.Host || !browserHost(u) {
+		return nil
+	}
+	if port := u.Port(); port != "" {
+		if n, err := strconv.ParseUint(port, 10, 16); err != nil || n == 0 {
+			return nil
+		}
+	}
+	return &sourceURL
+}
+
+// browserHost says whether u's host is one the browser's URL parser (the
+// WHATWG URL Standard) accepts as it stands, where url.Parse accepts more,
+// such as "[bad]". The rule is narrower than the browser's: a bracketed
+// host is an IPv6 address without a zone; any other is ASCII letters,
+// digits and hyphens in dot-separated labels, none empty but a trailing
+// one, none punycode ("xn--"), since the browser refuses invalid punycode
+// and checking it is not worth a decoder; and a host whose last label is
+// a number, which the browser reads as an IPv4 address, must be a dotted
+// quad. A link to any other host is withheld.
+func browserHost(u *url.URL) bool {
+	host := u.Hostname()
+	if strings.HasPrefix(u.Host, "[") {
+		ip := net.ParseIP(host)
+		return ip != nil && strings.Contains(host, ":")
+	}
+	labels := strings.Split(strings.TrimSuffix(host, "."), ".")
+	for _, label := range labels {
+		if label == "" || len(label) >= 4 && strings.EqualFold(label[:4], "xn--") {
+			return false
+		}
+		for _, c := range label {
+			if !('a' <= c && c <= 'z' || 'A' <= c && c <= 'Z' || '0' <= c && c <= '9' || c == '-') {
+				return false
+			}
+		}
+	}
+	last := labels[len(labels)-1]
+	if strings.Trim(last, "0123456789") == "" || len(last) >= 2 && strings.EqualFold(last[:2], "0x") {
+		ip := net.ParseIP(host)
+		return ip != nil && ip.To4() != nil && !strings.Contains(host, ":")
+	}
+	return true
+}
+
 // duplicatePosition is the error for two recipe items of kind at position.
 // Position is the only record of recipe order, and the id that breaks the
 // tie says nothing about it, so a recipe with a duplicate is refused rather

internal/dinners/dinners_test.go

@@ -112,6 +112,63 @@ func TestPageSiteNameWithholdsHosts(t *testing.T) {
 	}
 }
 
+// TestSafeSourceURLOnlyWithoutSecrets: a recipe link is shown only when it
+// is https to a host with no user, password, query or fragment, each of
+// which can carry a token; any other link, or one that does not parse, is
+// withheld whole.
+func TestSafeSourceURLOnlyWithoutSecrets(t *testing.T) {
+	t.Parallel()
+	for _, link := range []string{
+		"https://www.example.com/recipes/chicken-curry",
+		"https://example.com",
+		"https://example.com:8443/stew/",
+		"https://[2001:db8::1]/curry",
+		"https://1.2.3.4/curry",
+		"https://Kitchen-2.Example.com./curry",
+	} {
+		if got := dinners.SafeSourceURL(link); got == nil || *got != link {
+			t.Errorf("SafeSourceURL(%q) = %v, want the link", link, deref(got))
+		}
+	}
+	for _, link := range []string{
+		"https://example.com/curry?access_token=query-token",
+		"https://example.com/curry?",
+		"https://user:password@example.com/curry",
+		"https://token@example.com/curry",
+		"https://@example.com/curry",
+		"http://example.com/curry",
+		"https://example.com/curry#token=fragment-token",
+		"https://example.com/curry#",
+		"ftp://example.com/curry",
+		"https:example.com/curry",
+		"https:///curry",
+		"https://example.com:99999/curry",
+		"https://example.com:0/curry",
+		"//example.com/curry",
+		"https://secret-token\x7f.example/%zz",
+		"",
+		// Hosts url.Parse accepts but the browser's URL parser refuses
+		// (or reads as another host), which would fail the whole recipe
+		// screen.
+		"https://[bad]/curry",
+		"https://[fe80::1%25en0]/curry",
+		"https://[1.2.3.4]/curry",
+		"https://1.2.3.999/curry",
+		"https://1.2.3.4.5/curry",
+		"https://999/curry",
+		"https://example.0x/curry",
+		"https://xn--a/curry",
+		"https://ex%61mple.com/curry",
+		"https://ex%2fample.com/curry",
+		"https://a..b/curry",
+		"https://café.example/curry",
+	} {
+		if got := dinners.SafeSourceURL(link); got != nil {
+			t.Errorf("SafeSourceURL(%q) = %q, want nil", link, *got)
+		}
+	}
+}
+
 func deref(s *string) any {
 	if s == nil {
 		return nil

internal/dinners/export_test.go

@@ -74,3 +74,6 @@ func (s *Service) RunGroundJob(ctx context.Context, p GroundJob) (string, error)
 
 // PageSiteName is pageSiteName.
 var PageSiteName = pageSiteName
+
+// SafeSourceURL is safeSourceURL.
+var SafeSourceURL = safeSourceURL

internal/server/dinnerdetail_test.go

@@ -5,6 +5,7 @@ import (
 	"encoding/json"
 	"net/http"
 	"net/http/httptest"
+	"os"
 	"reflect"
 	"strings"
 	"testing"
@@ -111,7 +112,7 @@ func TestGetDinnerGroundedSideQuest(t *testing.T) {
 			{Name: str("onion"), Qty: str("1"), RawText: "1 onion"},
 		},
 		Steps:           []string{"Fry with the garlic.", "Chop the onion."},
-		Source:          &dinners.Source{SiteName: str("Curry House")},
+		Source:          &dinners.Source{SiteName: str("Curry House"), URL: str(curryPage)},
 		GroundingStatus: "grounded",
 		WhyPicked:       str(reason),
 	}
@@ -216,51 +217,113 @@ func TestGetDinnerNotFound(t *testing.T) {
 	}
 }
 
-// TestGetDinnerSourceHidesTokenURL: a recipe grounded in a pasted link that
-// carries a token in its path or query shows its site name, but never the
-// link or any part of it.
-func TestGetDinnerSourceHidesTokenURL(t *testing.T) {
+// TestGetDinnerSourceURLOnlyWhenSafe: a recipe grounded in a clean https
+// link shows it as source.url; one whose stored link carries a token in its
+// query or userinfo, is http, or has a fragment shows its site name but
+// source.url null and nothing of the link.
+func TestGetDinnerSourceURLOnlyWhenSafe(t *testing.T) {
 	t.Parallel()
 	e := newDinnersEnv(t)
 	const (
-		pathToken  = "path-token-abc123"
+		clean      = "https://kitchen.example/recipes/chicken-curry"
 		queryToken = "query-token-def456"
-		pathPage   = "https://secret-path.example/u/" + pathToken + "/chicken-curry"
 		queryPage  = "https://secret-query.example/curry?access_token=" + queryToken
 	)
-	e.pages.Bodies[pathPage] = recipestest.RecipePage("Chicken Curry", "Curry House", "https://a.example/curry.jpg", 40)
+	e.pages.Bodies[clean] = recipestest.RecipePage("Chicken Curry", "Curry House", "https://a.example/curry.jpg", 40)
 	e.pages.Bodies[queryPage] = recipestest.RecipePage("Weeknight Curry", "B Kitchen", "https://b.example/curry.jpg", 25)
 	e.images.JPEGs["https://a.example/curry.jpg"] = recipestest.JPEG(t, 400, 320, 1)
 	e.images.JPEGs["https://b.example/curry.jpg"] = recipestest.JPEG(t, 400, 320, 2)
-	for _, tc := range []struct{ dish, page, site string }{
-		{"Chicken curry", pathPage, "Curry House"},
-		{"Weeknight curry", queryPage, "B Kitchen"},
-	} {
-		id := createDinner(t, e.h, `{"name":"`+tc.dish+`"}`) // runner not started: stays pending
+	ground := func(dish, page string) (id string) {
+		t.Helper()
+		id = createDinner(t, e.h, `{"name":"`+dish+`"}`) // runner not started: stays pending
 		e.model.QueueGenerate(recipestest.NormalizedReply("stovetop", 25))
-		if rec, _ := callDinners(t, e.h, http.MethodPost, dinnersPath+"/"+id+"/ground", `{"recipe_url":"`+tc.page+`"}`); rec.Code != http.StatusOK {
-			t.Fatalf("POST ground %s = %d: %s", tc.page, rec.Code, rec.Body)
+		if rec, _ := callDinners(t, e.h, http.MethodPost, dinnersPath+"/"+id+"/ground", `{"recipe_url":"`+page+`"}`); rec.Code != http.StatusOK {
+			t.Fatalf("POST ground %s = %d: %s", page, rec.Code, rec.Body)
 		}
 		var stored string
 		if err := e.st.Reader.QueryRow(`SELECT source_url FROM recipes WHERE id = (SELECT recipe_id FROM dinners WHERE id = ?)`, id).Scan(&stored); err != nil {
 			t.Fatal(err)
 		}
-		if stored != tc.page {
-			t.Fatalf("stored source_url = %q, want %q", stored, tc.page)
+		if stored != page {
+			t.Fatalf("stored source_url = %q, want %q", stored, page)
 		}
+		return id
+	}
+	source := func(what, id string) map[string]any {
+		t.Helper()
+		raw, _ := getDetail(t, e.h, id)
+		src, ok := raw["source"].(map[string]any)
+		if !ok {
+			t.Fatalf("source of %s = %v, want an object", what, raw["source"])
+		}
+		return src
+	}
 
+	cleanID := ground("Chicken curry", clean)
+	if src := source("a clean link", cleanID); !reflect.DeepEqual(src, map[string]any{"site_name": "Curry House", "url": clean}) {
+		t.Errorf("source of a clean link = %v, want site_name Curry House and url %s", src, clean)
+	}
+
+	queryID := ground("Weeknight curry", queryPage)
+	withheld := func(what, id, site string, secrets ...string) {
+		t.Helper()
 		rec, _ := callDinners(t, e.h, http.MethodGet, dinnersPath+"/"+id, "")
 		if rec.Code != http.StatusOK {
-			t.Fatalf("GET dinner = %d: %s", rec.Code, rec.Body)
+			t.Fatalf("GET %s = %d: %s", what, rec.Code, rec.Body)
 		}
-		for _, secret := range []string{tc.page, "secret-path.example", "secret-query.example", pathToken, queryToken, "access_token", `"url"`} {
+		for _, secret := range secrets {
 			if strings.Contains(rec.Body.String(), secret) {
-				t.Errorf("detail of %s contains %q: %s", tc.dish, secret, rec.Body)
+				t.Errorf("detail of %s contains %q: %s", what, secret, rec.Body)
 			}
 		}
-		raw, _ := getDetail(t, e.h, id)
-		if src, _ := raw["source"].(map[string]any); !reflect.DeepEqual(src, map[string]any{"site_name": tc.site}) {
-			t.Errorf("source of %s = %v, want only site_name %q", tc.dish, raw["source"], tc.site)
+		if src := source(what, id); !reflect.DeepEqual(src, map[string]any{"site_name": site, "url": nil}) {
+			t.Errorf("source of %s = %v, want site_name %q and url null", what, src, site)
+		}
+	}
+	withheld("a link with a query token", queryID, "B Kitchen", queryPage, "secret-query.example", queryToken, "access_token")
+
+	// The other unsafe shapes, stored as a link grounded elsewhere would
+	// leave them.
+	for _, tc := range []struct{ what, link string }{
+		{"a link with a userinfo token", "https://user-token-ghi789@kitchen.example/recipes/chicken-curry"},
+		{"a link with a userinfo password", "https://cook:password-jkl012@kitchen.example/recipes/chicken-curry"},
+		{"an http link", "http://kitchen.example/recipes/chicken-curry"},
+		{"a link with a fragment", "https://kitchen.example/recipes/chicken-curry#token=fragment-mno345"},
+	} {
+		if _, err := e.st.Writer.Exec(`UPDATE recipes SET source_url = ? WHERE id = (SELECT recipe_id FROM dinners WHERE id = ?)`, tc.link, cleanID); err != nil {
+			t.Fatal(err)
+		}
+		withheld(tc.what, cleanID, "Curry House", tc.link, "kitchen.example", "-token-", "password", "fragment")
+	}
+
+	// web/src/dinners/testdata/source_url_cases.json pins source as this
+	// handler sends it for each stored link; the recipe screen's tests
+	// render each one, so a link the server sends and the client refuses
+	// fails one side or the other.
+	fixture, err := os.ReadFile("../../web/src/dinners/testdata/source_url_cases.json")
+	if err != nil {
+		t.Fatal(err)
+	}
+	var shared struct {
+		Cases []struct {
+			Stored string
+			Source map[string]any
+		}
+	}
+	if err := json.Unmarshal(fixture, &shared); err != nil {
+		t.Fatal(err)
+	}
+	if len(shared.Cases) == 0 {
+		t.Fatal("source_url_cases.json has no cases")
+	}
+	for _, c := range shared.Cases {
+		if _, err := e.st.Writer.Exec(`UPDATE recipes SET source_url = ? WHERE id = (SELECT recipe_id FROM dinners WHERE id = ?)`, c.Stored, cleanID); err != nil {
+			t.Fatal(err)
+		}
+		if src := source(c.Stored, cleanID); !reflect.DeepEqual(src, c.Source) {
+			got, _ := json.Marshal(src)
+			want, _ := json.Marshal(c.Source)
+			t.Errorf("source of %s = %s, want %s", c.Stored, got, want)
 		}
 	}
 }
@@ -270,14 +333,16 @@ func TestGetDinnerSourceHidesTokenURL(t *testing.T) {
 // token has site_name null and nothing from the link in its detail; a page
 // that names its site keeps that name, even on a token-bearing host; and a
 // stored name shaped like another host, as a redirect leaves, is withheld.
+// The links carry a query, so source.url is withheld too and nothing of the
+// link may appear.
 func TestGetDinnerSiteNameNeverFromLink(t *testing.T) {
 	t.Parallel()
 	e := newDinnersEnv(t)
 	const (
 		bareToken  = "private-token-abc123"
 		namedToken = "named-token-def456"
-		bare       = "https://" + bareToken + ".example.com/recipe"
-		named      = "https://www." + namedToken + ".example.com/recipe"
+		bare       = "https://" + bareToken + ".example.com/recipe?print=1"
+		named      = "https://www." + namedToken + ".example.com/recipe?print=1"
 	)
 	e.pages.Bodies[bare] = recipestest.RecipePage("Chicken Curry", "", "https://a.example/curry.jpg", 40)
 	e.pages.Bodies[named] = recipestest.RecipePage("Weeknight Curry", "Curry House", "https://b.example/curry.jpg", 25)
@@ -315,8 +380,8 @@ func TestGetDinnerSiteNameNeverFromLink(t *testing.T) {
 			}
 		}
 		raw, _ := getDetail(t, e.h, id)
-		if src, _ := raw["source"].(map[string]any); !reflect.DeepEqual(src, map[string]any{"site_name": want}) {
-			t.Errorf("source of %s = %v, want only site_name %v", what, raw["source"], want)
+		if src, _ := raw["source"].(map[string]any); !reflect.DeepEqual(src, map[string]any{"site_name": want, "url": nil}) {
+			t.Errorf("source of %s = %v, want site_name %v and url null", what, raw["source"], want)
 		}
 	}
 
@@ -324,13 +389,13 @@ func TestGetDinnerSiteNameNeverFromLink(t *testing.T) {
 	if got := storedSite(bareRecipe); got != bareToken+".example.com" {
 		t.Fatalf("stored site_name = %q, want the link's host", got)
 	}
-	check("a page naming no site", bareID, []string{bare, bareToken, "example.com", `"url"`}, nil)
+	check("a page naming no site", bareID, []string{bare, bareToken, "example.com"}, nil)
 
 	namedID, namedRecipe := ground("Weeknight curry", named)
 	if got := storedSite(namedRecipe); got != "Curry House" {
 		t.Fatalf("stored site_name = %q, want the page's", got)
 	}
-	check("a page naming its site", namedID, []string{named, namedToken, "example.com", `"url"`}, "Curry House")
+	check("a page naming its site", namedID, []string{named, namedToken, "example.com"}, "Curry House")
 
 	const redirectHost = "redirect-token-ghi789.example.net"
 	if _, err := e.st.Writer.Exec(`UPDATE recipes SET site_name = ? WHERE id = ?`, redirectHost, namedRecipe); err != nil {

web/src/dinners/DinnerDetailPage.test.tsx

@@ -9,6 +9,7 @@ import { errorResponse, jsonResponse, stubFetchByPath } from '../test/fetchStub.
 import { mockViewport } from '../test/matchMedia.ts'
 import { GROUNDING_POLL_MS } from './api.ts'
 import { DinnerDetailPage, REGULAR_DINNERS } from './DinnerDetailPage.tsx'
+import sourceCases from './testdata/source_url_cases.json'
 
 const DINNER = '/api/v1/dinners/d1'
 
@@ -186,6 +187,23 @@ describe('DinnerDetailPage', () => {
       expect(screen.queryByRole('link', { name: /Recipe from/ })).not.toBeInTheDocument()
     })
 
+    // Each source exactly as the Go handler sends it for a stored link
+    // (internal/server's TestGetDinnerSourceURLOnlyWhenSafe checks the same
+    // file), so every url the server sends must parse here.
+    it.each(sourceCases.cases)('renders the source the server sends for $stored', async ({ source }) => {
+      serve(detail({ source }))
+      renderAt(['/dinner/d1'])
+
+      const credit = await screen.findByText('Recipe from Curry House')
+      expect(screen.queryByRole('alert')).not.toBeInTheDocument()
+      if (source.url === null) {
+        expect(credit.closest('a')).toBeNull()
+        expect(screen.queryByRole('link', { name: /Recipe from/ })).not.toBeInTheDocument()
+      } else {
+        expect(screen.getByRole('link', { name: 'Recipe from Curry House' })).toHaveAttribute('href', new URL(source.url).href)
+      }
+    })
+
     it('leaves the credit out when the server names no site', async () => {
       serve(detail({ source: { site_name: null } }))
       renderAt(['/dinner/d1'])

web/src/dinners/testdata/source_url_cases.json

@@ -0,0 +1,52 @@
+{
+  "cases": [
+    {
+      "stored": "https://kitchen.example/recipes/chicken-curry",
+      "source": { "site_name": "Curry House", "url": "https://kitchen.example/recipes/chicken-curry" }
+    },
+    {
+      "stored": "https://Kitchen-2.Example.com.:8443/recipes/chicken-curry",
+      "source": { "site_name": "Curry House", "url": "https://Kitchen-2.Example.com.:8443/recipes/chicken-curry" }
+    },
+    {
+      "stored": "https://[2001:db8::1]/recipes/chicken-curry",
+      "source": { "site_name": "Curry House", "url": "https://[2001:db8::1]/recipes/chicken-curry" }
+    },
+    {
+      "stored": "https://192.0.2.7/recipes/chicken-curry",
+      "source": { "site_name": "Curry House", "url": "https://192.0.2.7/recipes/chicken-curry" }
+    },
+    {
+      "stored": "https://[bad]/recipes/chicken-curry",
+      "source": { "site_name": "Curry House", "url": null }
+    },
+    {
+      "stored": "https://[fe80::1%25en0]/recipes/chicken-curry",
+      "source": { "site_name": "Curry House", "url": null }
+    },
+    {
+      "stored": "https://192.0.2.999/recipes/chicken-curry",
+      "source": { "site_name": "Curry House", "url": null }
+    },
+    {
+      "stored": "https://xn--a.example/recipes/chicken-curry",
+      "source": { "site_name": "Curry House", "url": null }
+    },
+    {
+      "stored": "https://kitchen.example/recipes/chicken-curry?access_token=query-token",
+      "source": { "site_name": "Curry House", "url": null }
+    },
+    {
+      "stored": "https://user-token@kitchen.example/recipes/chicken-curry",
+      "source": { "site_name": "Curry House", "url": null }
+    },
+    {
+      "stored": "http://kitchen.example/recipes/chicken-curry",
+      "source": { "site_name": "Curry House", "url": null }
+    },
+    {
+      "stored": "https://kitchen.example/recipes/chicken-curry#token=fragment-token",
+      "source": { "site_name": "Curry House", "url": null }
+    }
+  ]
+}