fix(notes): take last numeric segment of href, not the first

The notes listing renders each thread link as
/msg/pms/{folder}/{noteID}/#message. extractIntFromHref returned the
first numeric segment it found, which was always the folder index (1
for the inbox), so every NotePreview.ID came out as 1 and any
follow-up GetNote(np.ID) call failed with "this message has either
been deleted or is not yours".

Surfaced by an end-to-end smoke run against the live site. Limited
to the notes parser; the other extractIntFromHref callers
(/view/{id}/, /journal/{id}/) only ever have a single numeric segment
so they are unaffected.

examples/favorites_page/main.go

@@ -0,0 +1,79 @@
+// favorites_page exercises the per-page favorites listing API
+// ([Client.FavoritesPage]) against the live FA site so a caller can see
+// exactly what fields come back: HasNext, NextPage, len(Items), and a
+// sample of the tag data lifted from each figure's data-tags attribute.
+//
+// Favorites pagination is cursor-based: each page returns an opaque
+// NextPage token that addresses the next page. Pass it back in on the
+// next call; treat empty as end-of-pagination.
+//
+// Required environment variables:
+//
+//	FA_A           — the `a` session cookie
+//	FA_B           — the `b` session cookie
+//	CF_CLEARANCE   — (optional) cf_clearance cookie if Cloudflare challenges
+//	FA_UA          — (optional) User-Agent matching CF_CLEARANCE
+//
+// Usage:
+//
+//	go run ./examples/favorites_page <username> [maxPages]
+package main
+
+import (
+	"context"
+	"fmt"
+	"log"
+	"os"
+	"strconv"
+
+	fa "git.anthrove.art/public/go-fa-api"
+)
+
+func main() {
+	if len(os.Args) < 2 {
+		log.Fatalf("usage: %s <username> [maxPages]", os.Args[0])
+	}
+	user := os.Args[1]
+	maxPages := 0
+	if len(os.Args) >= 3 {
+		if n, err := strconv.Atoi(os.Args[2]); err == nil && n > 0 {
+			maxPages = n
+		}
+	}
+
+	opts := []fa.Option{
+		fa.WithCookies(fa.Cookies{A: os.Getenv("FA_A"), B: os.Getenv("FA_B")}),
+	}
+	if cf := os.Getenv("CF_CLEARANCE"); cf != "" {
+		opts = append(opts, fa.WithCloudflare(fa.CFCookies{Clearance: cf}))
+	}
+	if ua := os.Getenv("FA_UA"); ua != "" {
+		opts = append(opts, fa.WithUserAgent(ua))
+	}
+	client := fa.New(opts...)
+
+	cursor := ""
+	pageNum := 0
+	for {
+		pageNum++
+		lp, err := client.FavoritesPage(context.Background(), user, cursor)
+		if err != nil {
+			log.Fatalf("FavoritesPage(cursor=%q): %v", cursor, err)
+		}
+		fmt.Printf("=== page %d  cursor=%q  items=%d  HasNext=%v  NextPage=%q ===\n",
+			pageNum, cursor, len(lp.Items), lp.HasNext, lp.NextPage)
+		for i, sub := range lp.Items {
+			fmt.Printf("  [%d] id=%d  rating=%s  author=%s  title=%q\n",
+				i, sub.ID, sub.Rating, sub.Author.Name, sub.Title)
+		}
+		if !lp.HasNext {
+			fmt.Printf("\nreached end of pagination after %d page(s)\n", pageNum)
+			return
+		}
+		if maxPages > 0 && pageNum >= maxPages {
+			fmt.Printf("\nstopped at maxPages=%d (HasNext was still true; next cursor=%q)\n", maxPages, lp.NextPage)
+			return
+		}
+		cursor = lp.NextPage
+	}
+}

fixtures_refresh_test.go

@@ -161,7 +161,7 @@ func TestRefreshFixtures(t *testing.T) {
 		},
 		{
 			name:     "favorites_page1.html",
-			url:      urls.Favorites(favoritesUser, 1),
+			url:      urls.Favorites(favoritesUser),
 			requires: []string{favoritesUser},
 			notes:    "favorites per-item Author should be the original artist",
 		},

gallery.go

@@ -3,6 +3,7 @@ package fa
 import (
 	"context"
 	"iter"
+	"strconv"
 
 	"github.com/PuerkitoBio/goquery"
 
@@ -12,28 +13,137 @@ import (
 // Gallery iterates the submissions in a user's main gallery, newest first.
 //
 // Each yielded *Submission carries only the fields visible on the listing
-// page: ID, Title, Author (for favorites), ThumbURL, and Rating. Call
+// page: ID, Title, Author (for favorites), ThumbURL, Rating, and the Tags
+// / CategorizedTags parsed from the figure's data-tags attribute. Call
 // [Client.GetSubmission] with the ID to load the full record.
 func (c *Client) Gallery(ctx context.Context, name string, opts ListOptions, reqOpts ...Option) iter.Seq2[*Submission, error] {
-	return c.listGallerySection(ctx, name, urls.Gallery, opts, reqOpts)
+	return c.listPagedSection(ctx, name, urls.Gallery, opts, reqOpts)
 }
 
 // Scraps iterates the user's scraps folder. Same yield shape as Gallery.
 func (c *Client) Scraps(ctx context.Context, name string, opts ListOptions, reqOpts ...Option) iter.Seq2[*Submission, error] {
-	return c.listGallerySection(ctx, name, urls.Scraps, opts, reqOpts)
+	return c.listPagedSection(ctx, name, urls.Scraps, opts, reqOpts)
 }
 
 // Favorites iterates the user's favorited submissions. The yielded
 // *Submission's Author field reflects the original artist (not the user
 // whose favorites we are walking).
+//
+// Favorites use a fave-ID cursor for pagination, not sequential page
+// numbers, so [ListOptions.StartPage] is ignored — the walk always
+// begins at the newest favorite. [ListOptions.MaxPages] still bounds
+// the crawl.
 func (c *Client) Favorites(ctx context.Context, name string, opts ListOptions, reqOpts ...Option) iter.Seq2[*Submission, error] {
-	return c.listGallerySection(ctx, name, urls.Favorites, opts, reqOpts)
+	return func(yield func(*Submission, error) bool) {
+		cursor := ""
+		pagesFetched := 0
+		for {
+			if opts.reachedLimit(pagesFetched) {
+				return
+			}
+			lp, err := c.FavoritesPage(ctx, name, cursor, reqOpts...)
+			if err != nil {
+				yield(nil, err)
+				return
+			}
+			pagesFetched++
+			if len(lp.Items) == 0 {
+				return
+			}
+			for _, s := range lp.Items {
+				if !yield(s, nil) {
+					return
+				}
+			}
+			if !lp.HasNext {
+				return
+			}
+			cursor = lp.NextPage
+		}
+	}
 }
 
-// listGallerySection is the shared engine for Gallery / Scraps / Favorites.
-// urlFn picks the section-specific URL builder; the rest of the pagination
-// machinery is identical across all three sections.
-func (c *Client) listGallerySection(
+// GalleryPage fetches a single page of /gallery/{name}/ and returns the
+// items along with whether more pages exist. Pages are 1-based; pass 0 or
+// 1 for the first page. Use this when driving pagination manually
+// (resuming from a checkpoint, distributing pages across workers); use
+// [Client.Gallery] when you just want every item in order.
+//
+// On a non-final page the returned [ListingPage].NextPage is the next
+// page number as a decimal string ("2", "3", …) — pass it back to the
+// next call after [strconv.Atoi], or treat it as opaque.
+func (c *Client) GalleryPage(ctx context.Context, name string, page int, reqOpts ...Option) (*ListingPage, error) {
+	return c.fetchNumberedPage(ctx, name, page, urls.Gallery, reqOpts)
+}
+
+// ScrapsPage is the single-page counterpart to [Client.Scraps]. See
+// [Client.GalleryPage] for usage notes.
+func (c *Client) ScrapsPage(ctx context.Context, name string, page int, reqOpts ...Option) (*ListingPage, error) {
+	return c.fetchNumberedPage(ctx, name, page, urls.Scraps, reqOpts)
+}
+
+// FavoritesPage fetches a single page of /favorites/{name}/, addressed
+// by the cursor FA emitted on the previous page (empty string for the
+// first page). FA paginates favorites with a fave-ID cursor — not a
+// sequential page number — so the caller must walk forward by passing
+// the returned [ListingPage].NextPage value into the next call. Passing
+// a guessed cursor (e.g. "2") makes FA silently return the first page
+// and the loop will not terminate.
+func (c *Client) FavoritesPage(ctx context.Context, name string, cursor string, reqOpts ...Option) (*ListingPage, error) {
+	out := &ListingPage{}
+	err := c.fetch(ctx, urls.FavoritesCursor(name, cursor), func(doc *goquery.Document) error {
+		items, nextURL, hasNext := parseListingPage(doc, c.cfg.jsonListings)
+		out.Items = items
+		out.HasNext = hasNext
+		if hasNext {
+			out.NextPage = favoritesCursorFromURL(nextURL)
+			// If the markup was unrecognisable, refuse to claim a next
+			// page rather than re-fetching the first one in a loop.
+			if out.NextPage == "" {
+				out.HasNext = false
+			}
+		}
+		return nil
+	}, reqOpts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+// fetchNumberedPage is the shared primitive for page-number-based
+// listings (Gallery / Scraps). urlFn picks the section-specific URL
+// builder; the rest of the pagination machinery is identical.
+func (c *Client) fetchNumberedPage(
+	ctx context.Context,
+	name string,
+	page int,
+	urlFn func(string, int) string,
+	reqOpts []Option,
+) (*ListingPage, error) {
+	if page < 1 {
+		page = 1
+	}
+	out := &ListingPage{}
+	err := c.fetch(ctx, urlFn(name, page), func(doc *goquery.Document) error {
+		items, _, hasNext := parseListingPage(doc, c.cfg.jsonListings)
+		out.Items = items
+		out.HasNext = hasNext
+		if hasNext {
+			out.NextPage = strconv.Itoa(page + 1)
+		}
+		return nil
+	}, reqOpts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+// listPagedSection is the shared engine for the page-number-based
+// listing iterators (Gallery / Scraps). Favorites has its own loop in
+// [Client.Favorites] because its pagination is cursor-based.
+func (c *Client) listPagedSection(
 	ctx context.Context,
 	name string,
 	urlFn func(string, int) string,
@@ -47,28 +157,21 @@ func (c *Client) listGallerySection(
 			if opts.reachedLimit(pagesFetched) {
 				return
 			}
-			var (
-				items   []*Submission
-				hasNext bool
-			)
-			err := c.fetch(ctx, urlFn(name, page), func(doc *goquery.Document) error {
-				items, hasNext = parseGalleryPage(doc, c.cfg.jsonListings)
-				return nil
-			}, reqOpts...)
+			lp, err := c.fetchNumberedPage(ctx, name, page, urlFn, reqOpts)
 			if err != nil {
 				yield(nil, err)
 				return
 			}
 			pagesFetched++
-			if len(items) == 0 {
+			if len(lp.Items) == 0 {
 				return
 			}
-			for _, s := range items {
+			for _, s := range lp.Items {
 				if !yield(s, nil) {
 					return
 				}
 			}
-			if !hasNext {
+			if !lp.HasNext {
 				return
 			}
 			page++

gallery_page_test.go

@@ -0,0 +1,198 @@
+package fa
+
+import (
+	"context"
+	"fmt"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"sync"
+	"sync/atomic"
+	"testing"
+)
+
+// fakeGalleryPage builds a minimal gallery-page response with two figures.
+// nextHref is the next-page URL emitted in the Next form; empty means no
+// Next button (last page).
+func fakeGalleryPage(startID int, nextHref string) string {
+	var b strings.Builder
+	b.WriteString(`<html><body>`)
+	for i := 0; i < 2; i++ {
+		id := startID + i
+		fmt.Fprintf(&b, `
+<figure id="sid-%d" class="t-image r-general">
+  <a href="/view/%d/" title="Sub %d">
+    <img data-tags="u_someartist c_artwork_digital t_all s_wolf wolf" src="//d.example/t/%d.png"/>
+  </a>
+  <figcaption>
+    <p>Sub %d</p>
+    <a href="/user/someartist/">someartist</a>
+  </figcaption>
+</figure>`, id, id, id, id, id)
+	}
+	if nextHref != "" {
+		fmt.Fprintf(&b, `<form action=%q method="get"><button class="button standard" type="submit">Next</button></form>`, nextHref)
+	}
+	b.WriteString(`</body></html>`)
+	return b.String()
+}
+
+func TestGalleryPage_HasNextPropagates(t *testing.T) {
+	var requests atomic.Int32
+	mux := http.NewServeMux()
+	mux.HandleFunc("/gallery/u/", func(w http.ResponseWriter, _ *http.Request) {
+		requests.Add(1)
+		_, _ = w.Write([]byte(fakeGalleryPage(1000, "/gallery/u/2/")))
+	})
+	mux.HandleFunc("/gallery/u/2/", func(w http.ResponseWriter, _ *http.Request) {
+		requests.Add(1)
+		_, _ = w.Write([]byte(fakeGalleryPage(2000, "")))
+	})
+	srv := httptest.NewServer(mux)
+	defer srv.Close()
+	client := newE2EClient(t, srv)
+
+	first, err := client.GalleryPage(context.Background(), "u", 1)
+	if err != nil {
+		t.Fatalf("GalleryPage(1): %v", err)
+	}
+	if !first.HasNext {
+		t.Error("first.HasNext = false; want true")
+	}
+	if first.NextPage != "2" {
+		t.Errorf("first.NextPage = %q; want \"2\"", first.NextPage)
+	}
+	if len(first.Items) != 2 {
+		t.Fatalf("first.Items len = %d; want 2", len(first.Items))
+	}
+	if first.Items[0].ID != 1000 {
+		t.Errorf("first.Items[0].ID = %d; want 1000", first.Items[0].ID)
+	}
+	if len(first.Items[0].Tags) == 0 || len(first.Items[0].CategorizedTags.Species) == 0 {
+		t.Errorf("first.Items[0]: tags not populated from data-tags: %+v", first.Items[0])
+	}
+
+	last, err := client.GalleryPage(context.Background(), "u", 2)
+	if err != nil {
+		t.Fatalf("GalleryPage(2): %v", err)
+	}
+	if last.HasNext {
+		t.Error("last.HasNext = true; want false (last page)")
+	}
+	if last.NextPage != "" {
+		t.Errorf("last.NextPage = %q; want empty", last.NextPage)
+	}
+
+	if requests.Load() != 2 {
+		t.Errorf("requests = %d; want 2", requests.Load())
+	}
+}
+
+func TestScrapsPage_HitsScrapsRoute(t *testing.T) {
+	var gotPath string
+	mux := http.NewServeMux()
+	mux.HandleFunc("/scraps/u/", func(w http.ResponseWriter, r *http.Request) {
+		gotPath = r.URL.Path
+		_, _ = w.Write([]byte(fakeGalleryPage(1, "")))
+	})
+	srv := httptest.NewServer(mux)
+	defer srv.Close()
+	client := newE2EClient(t, srv)
+
+	if _, err := client.ScrapsPage(context.Background(), "u", 1); err != nil {
+		t.Fatalf("ScrapsPage: %v", err)
+	}
+	if gotPath != "/scraps/u/" {
+		t.Errorf("gotPath = %q; want /scraps/u/", gotPath)
+	}
+}
+
+func TestFavoritesPage_CursorChain(t *testing.T) {
+	var requests []string
+	var mu sync.Mutex
+	record := func(p string) {
+		mu.Lock()
+		requests = append(requests, p)
+		mu.Unlock()
+	}
+	mux := http.NewServeMux()
+	mux.HandleFunc("/favorites/u/", func(w http.ResponseWriter, r *http.Request) {
+		record(r.URL.Path)
+		_, _ = w.Write([]byte(fakeGalleryPage(1000, "/favorites/u/9999/next")))
+	})
+	mux.HandleFunc("/favorites/u/9999/next", func(w http.ResponseWriter, r *http.Request) {
+		record(r.URL.Path)
+		_, _ = w.Write([]byte(fakeGalleryPage(2000, "")))
+	})
+	srv := httptest.NewServer(mux)
+	defer srv.Close()
+	client := newE2EClient(t, srv)
+
+	first, err := client.FavoritesPage(context.Background(), "u", "")
+	if err != nil {
+		t.Fatalf("FavoritesPage(first): %v", err)
+	}
+	if !first.HasNext {
+		t.Fatal("first.HasNext = false; want true")
+	}
+	if first.NextPage != "9999" {
+		t.Errorf("first.NextPage = %q; want \"9999\" (cursor)", first.NextPage)
+	}
+
+	last, err := client.FavoritesPage(context.Background(), "u", first.NextPage)
+	if err != nil {
+		t.Fatalf("FavoritesPage(cursor): %v", err)
+	}
+	if last.HasNext {
+		t.Error("last.HasNext = true; want false")
+	}
+	if last.NextPage != "" {
+		t.Errorf("last.NextPage = %q; want empty", last.NextPage)
+	}
+
+	want := []string{"/favorites/u/", "/favorites/u/9999/next"}
+	mu.Lock()
+	defer mu.Unlock()
+	if len(requests) != len(want) {
+		t.Fatalf("requests = %v; want %v", requests, want)
+	}
+	for i, w := range want {
+		if requests[i] != w {
+			t.Errorf("requests[%d] = %q; want %q", i, requests[i], w)
+		}
+	}
+}
+
+// TestFavorites_IteratorTerminates guards against the cursor-loop
+// regression that brought us here: with sequential page numbers, the
+// Favorites iterator never terminated because FA fell back to page 1
+// for every fake-numbered cursor.
+func TestFavorites_IteratorTerminates(t *testing.T) {
+	mux := http.NewServeMux()
+	mux.HandleFunc("/favorites/u/", func(w http.ResponseWriter, _ *http.Request) {
+		_, _ = w.Write([]byte(fakeGalleryPage(1, "/favorites/u/42/next")))
+	})
+	mux.HandleFunc("/favorites/u/42/next", func(w http.ResponseWriter, _ *http.Request) {
+		_, _ = w.Write([]byte(fakeGalleryPage(3, "")))
+	})
+	srv := httptest.NewServer(mux)
+	defer srv.Close()
+	client := newE2EClient(t, srv)
+
+	count := 0
+	for sub, err := range client.Favorites(context.Background(), "u", ListOptions{}) {
+		if err != nil {
+			t.Fatalf("Favorites: %v", err)
+		}
+		if sub == nil {
+			t.Fatal("nil sub")
+		}
+		count++
+		if count > 10 {
+			t.Fatalf("iterator did not terminate; count > 10")
+		}
+	}
+	if count != 4 {
+		t.Errorf("count = %d; want 4 (2 per page * 2 pages)", count)
+	}
+}

gallery_parser.go

@@ -19,6 +19,15 @@ import (
 // pure HTML the same behaviour as before [WithExperimentalJSONListings]
 // existed.
 func parseGalleryPage(doc *goquery.Document, useJSON bool) (items []*Submission, hasNext bool) {
+	items, _, hasNext = parseListingPage(doc, useJSON)
+	return items, hasNext
+}
+
+// parseListingPage parses one page of a listing endpoint and also returns
+// the raw next-page URL FA emits in its "Next" pagination form. Callers
+// that need to chain across cursor-based pages (Favorites) consume the
+// URL; callers that don't (Gallery / Scraps) can ignore it.
+func parseListingPage(doc *goquery.Document, useJSON bool) (items []*Submission, nextURL string, hasNext bool) {
 	var jsonData listingJSONMap
 	if useJSON {
 		jsonData = readListingJSON(doc)
@@ -28,8 +37,8 @@ func parseGalleryPage(doc *goquery.Document, useJSON bool) (items []*Submission,
 			items = append(items, s)
 		}
 	})
-	hasNext = detectNextPage(doc)
-	return items, hasNext
+	nextURL, hasNext = nextPageURL(doc)
+	return items, nextURL, hasNext
 }
 
 // parseGalleryFigure lifts a single submission preview from a
@@ -85,6 +94,15 @@ func parseGalleryFigure(sel *goquery.Selection, jsonData listingJSONMap) *Submis
 		}
 	}
 
+	// data-tags on the figure's <img> carries both the unprefixed keyword
+	// list and the prefixed system tags (s_/c_/a_/u_/t_). Splitting it lets
+	// callers classify listing items without an extra /view/ fetch.
+	if img := sel.Find("img[data-tags]").First(); img.Length() > 0 {
+		if raw, ok := img.Attr("data-tags"); ok {
+			applyListingDataTags(s, raw)
+		}
+	}
+
 	// JSON enrichment preferred sources for the fields it carries.
 	if jsonData != nil {
 		if entry, ok := jsonData[id]; ok {
@@ -105,3 +123,35 @@ func parseGalleryFigure(sel *goquery.Selection, jsonData listingJSONMap) *Submis
 
 	return s
 }
+
+// applyListingDataTags splits the whitespace-separated data-tags attribute
+// FA emits on listing-page <img> elements and routes each token to either
+// CategorizedTags (when the token has a known single-letter prefix
+// s_/c_/a_/u_/t_) or Tags (everything else).
+//
+// The prefix mapping mirrors the /view/ parser in submission_parser.go so a
+// listing-path Submission carries the same categorisation a /view/-path one
+// would, modulo tokens FA can't represent in this flat attribute (multi-word
+// tags, the a_ vs u_ distinction).
+func applyListingDataTags(s *Submission, raw string) {
+	for _, tok := range strings.Fields(raw) {
+		if len(tok) >= 3 && tok[1] == '_' {
+			name := tok[2:]
+			switch tok[0] {
+			case 's':
+				s.CategorizedTags.Species = append(s.CategorizedTags.Species, name)
+				continue
+			case 'c':
+				s.CategorizedTags.Characters = append(s.CategorizedTags.Characters, name)
+				continue
+			case 'a', 'u':
+				s.CategorizedTags.Artists = append(s.CategorizedTags.Artists, name)
+				continue
+			case 't':
+				s.CategorizedTags.Types = append(s.CategorizedTags.Types, name)
+				continue
+			}
+		}
+		s.Tags = append(s.Tags, tok)
+	}
+}

gallery_parser_test.go

@@ -62,6 +62,99 @@ func TestParseGalleryPage_Synthetic(t *testing.T) {
 	}
 }
 
+func TestParseGalleryFigure_DataTags(t *testing.T) {
+	const html = `<html><body>
+<figure id="sid-2001" class="t-image r-general">
+  <a href="/view/2001/" title="Mixed Tags">
+    <img data-tags="u_someartist c_artwork_digital t_all s_wolf wolf solo digital landscape" src="//d.example/thumb/2001.png"/>
+  </a>
+</figure>
+<figure id="sid-2002" class="t-image r-general">
+  <a href="/view/2002/" title="No Tags">
+    <img src="//d.example/thumb/2002.png"/>
+  </a>
+</figure>
+<figure id="sid-2003" class="t-image r-general">
+  <a href="/view/2003/" title="Only Keywords">
+    <img data-tags="wolf solo" src="//d.example/thumb/2003.png"/>
+  </a>
+</figure>
+</body></html>`
+	doc, err := goquery.NewDocumentFromReader(strings.NewReader(html))
+	if err != nil {
+		t.Fatalf("setup: %v", err)
+	}
+	items, _ := parseGalleryPage(doc, false)
+	if len(items) != 3 {
+		t.Fatalf("items = %d; want 3", len(items))
+	}
+
+	// Mixed prefixed + unprefixed.
+	mixed := items[0]
+	wantTags := []string{"wolf", "solo", "digital", "landscape"}
+	if !equalStrings(mixed.Tags, wantTags) {
+		t.Errorf("items[0].Tags = %v; want %v", mixed.Tags, wantTags)
+	}
+	if !equalStrings(mixed.CategorizedTags.Species, []string{"wolf"}) {
+		t.Errorf("items[0].Species = %v", mixed.CategorizedTags.Species)
+	}
+	if !equalStrings(mixed.CategorizedTags.Characters, []string{"artwork_digital"}) {
+		t.Errorf("items[0].Characters = %v", mixed.CategorizedTags.Characters)
+	}
+	if !equalStrings(mixed.CategorizedTags.Types, []string{"all"}) {
+		t.Errorf("items[0].Types = %v", mixed.CategorizedTags.Types)
+	}
+	if !equalStrings(mixed.CategorizedTags.Artists, []string{"someartist"}) {
+		t.Errorf("items[0].Artists = %v", mixed.CategorizedTags.Artists)
+	}
+
+	// Missing data-tags: both slices stay nil.
+	if items[1].Tags != nil {
+		t.Errorf("items[1].Tags = %v; want nil", items[1].Tags)
+	}
+	if items[1].CategorizedTags.Species != nil ||
+		items[1].CategorizedTags.Characters != nil ||
+		items[1].CategorizedTags.Artists != nil ||
+		items[1].CategorizedTags.Types != nil {
+		t.Errorf("items[1].CategorizedTags = %+v; want zero", items[1].CategorizedTags)
+	}
+
+	// Unprefixed-only: everything lands in Tags.
+	if !equalStrings(items[2].Tags, []string{"wolf", "solo"}) {
+		t.Errorf("items[2].Tags = %v", items[2].Tags)
+	}
+	if items[2].CategorizedTags.Species != nil {
+		t.Errorf("items[2].Species = %v; want nil", items[2].CategorizedTags.Species)
+	}
+}
+
+func TestParseGalleryPage_RealFixtureTags(t *testing.T) {
+	raw := loadFixture(t, "gallery_page1.html")
+	doc, err := goquery.NewDocumentFromReader(bytes.NewReader(raw))
+	if err != nil {
+		t.Fatalf("read doc: %v", err)
+	}
+	items, _ := parseGalleryPage(doc, false)
+	if len(items) == 0 {
+		t.Fatal("real fixture: no items parsed")
+	}
+	var withTags, withSpecies int
+	for _, it := range items {
+		if len(it.Tags) > 0 {
+			withTags++
+		}
+		if len(it.CategorizedTags.Species) > 0 {
+			withSpecies++
+		}
+	}
+	if withTags == 0 {
+		t.Error("no items got Tags populated from data-tags")
+	}
+	if withSpecies == 0 {
+		t.Error("no items got CategorizedTags.Species populated from data-tags")
+	}
+}
+
 func TestParseGalleryPage_RealFixture(t *testing.T) {
 	raw := loadFixture(t, "gallery_page1.html")
 	doc, err := goquery.NewDocumentFromReader(bytes.NewReader(raw))

internal/urls/routes.go

@@ -36,10 +36,25 @@ func Scraps(name string, page int) string {
 	return Host + "/scraps/" + safeName(name) + "/" + pageSegment(page)
 }
 
-// Favorites returns the URL for a user's favorites page. FA uses a numeric
-// page parameter; the first page is 1.
-func Favorites(name string, page int) string {
-	return Host + "/favorites/" + safeName(name) + "/" + pageSegment(page)
+// Favorites returns the URL for the first page of a user's favorites.
+// FA paginates favorites with a fave-ID cursor (see [FavoritesCursor]),
+// not sequential page numbers — passing /favorites/{user}/{N}/ with a
+// small integer N silently falls back to the first page. Use this for
+// the first page only; follow the cursor returned in [ListingPage].NextPage
+// for subsequent pages.
+func Favorites(name string) string {
+	return Host + "/favorites/" + safeName(name) + "/"
+}
+
+// FavoritesCursor returns the URL for a follow-up favorites page,
+// addressed by the fave-ID cursor FA emits on the previous page's "Next"
+// form (e.g. /favorites/{user}/1951234825/next). The cursor is opaque
+// to the SDK — pass through whatever [ListingPage].NextPage gave you.
+func FavoritesCursor(name, cursor string) string {
+	if cursor == "" {
+		return Favorites(name)
+	}
+	return Host + "/favorites/" + safeName(name) + "/" + cursor + "/next"
 }
 
 // Journal returns the URL for a single journal entry.

notes_parser.go

@@ -52,11 +52,17 @@ func parseNoteListItem(item *goquery.Selection) *NotePreview {
 	}
 
 	// Note ID lives in the href: /msg/pms/{folder}/{id}/#message. Strip the
-	// fragment first so extractIntFromHref picks the trailing numeric path.
+	// fragment first, then take the *last* numeric segment — the folder
+	// number (e.g. 1) appears before the note ID and would otherwise win
+	// the "first numeric segment" race in extractIntFromHref.
 	if i := strings.Index(href, "#"); i != -1 {
 		href = href[:i]
 	}
-	np.ID = NoteID(extractIntFromHref(href))
+	for _, seg := range strings.Split(href, "/") {
+		if n, err := parseID[NoteID](seg); err == nil && n != 0 {
+			np.ID = n
+		}
+	}
 
 	// Read/unread: classes on the subject link.
 	if class, _ := subjectLink.Attr("class"); strings.Contains(class, "note-unread") || strings.Contains(class, "unread") && !strings.Contains(class, "note-read") {

pagination.go

@@ -6,6 +6,30 @@ import (
 	"github.com/PuerkitoBio/goquery"
 )
 
+// ListingPage is one page of a listing endpoint (Gallery / Scraps /
+// Favorites). It carries everything an external caller needs to drive
+// pagination by hand: the items, whether FA exposed a "next page" link,
+// and an opaque NextPage token to pass back into the next per-page call.
+//
+// External scrapers that want to manage their own loop (resume from a
+// checkpoint, run pages in parallel, throttle differently) should call
+// the per-page methods ([Client.GalleryPage], [Client.ScrapsPage],
+// [Client.FavoritesPage]) and stop when HasNext is false. Callers that
+// just want every item in order should keep using the iter.Seq2-shaped
+// methods ([Client.Gallery] et al.), which walk pages internally.
+//
+// NextPage's contents differ by endpoint — for Gallery / Scraps it is
+// the next 1-based page number as a decimal string ("2", "3", …); for
+// Favorites it is the fave-ID cursor FA emits on the "Next" form
+// (because favorites pagination is cursor-based, not page-number-based).
+// Treat the value as opaque: pass whatever you got back to the next
+// call without parsing.
+type ListingPage struct {
+	Items    []*Submission
+	HasNext  bool
+	NextPage string // "" when !HasNext; otherwise the opaque token to pass back
+}
+
 // ListOptions configures the pagination of a simple iterator method like
 // [Client.Gallery] or [Client.Notes]. Filtered iterators ([Client.Search],
 // [Client.Browse]) use their own option structs that fold the same fields
@@ -42,17 +66,57 @@ func (o ListOptions) reachedLimit(pagesFetched int) bool {
 // FA's beta theme renders pagination as either a Next form button or a
 // hyperlink with a recognisable label.
 func detectNextPage(doc *goquery.Document) bool {
-	if doc.Find("form button.button.standard:contains('Next')").Length() > 0 {
-		return true
+	url, _ := nextPageURL(doc)
+	return url != ""
+}
+
+// nextPageURL returns the action/href that the "Next" pagination control
+// would navigate to, along with a flag indicating whether one was found.
+// Returns ("", false) on the last page (FA emits no Next form/anchor, or
+// emits it inside an HTML comment that doesn't parse as an element).
+func nextPageURL(doc *goquery.Document) (string, bool) {
+	var action string
+	doc.Find("form").EachWithBreak(func(_ int, f *goquery.Selection) bool {
+		if f.Find("button.button.standard:contains('Next')").Length() == 0 {
+			return true
+		}
+		action, _ = f.Attr("action")
+		return false
+	})
+	if action != "" {
+		return action, true
 	}
-	hit := false
+	var href string
 	doc.Find("a.button.standard, a.button-link, a.pagination-next").EachWithBreak(func(_ int, sel *goquery.Selection) bool {
 		text := strings.ToLower(trimText(sel))
 		if strings.Contains(text, "next") || strings.Contains(text, "older") {
-			hit = true
+			href, _ = sel.Attr("href")
 			return false
 		}
 		return true
 	})
-	return hit
+	if href == "" {
+		return "", false
+	}
+	return href, true
+}
+
+// favoritesCursorFromURL extracts the fave-ID cursor segment from a
+// /favorites/{user}/{cursor}/next URL. Returns "" if the URL does not
+// match that shape (in which case the caller treats the listing as
+// exhausted rather than chasing a malformed cursor).
+func favoritesCursorFromURL(rawURL string) string {
+	// Strip query / fragment, then split. Favorites paths can be relative
+	// ("/favorites/u/123/next") or absolute — handle both.
+	rawURL = strings.TrimPrefix(rawURL, "https://www.furaffinity.net")
+	rawURL = strings.TrimPrefix(rawURL, "http://www.furaffinity.net")
+	if i := strings.IndexAny(rawURL, "?#"); i >= 0 {
+		rawURL = rawURL[:i]
+	}
+	parts := strings.Split(strings.Trim(rawURL, "/"), "/")
+	// Expect ["favorites", "{user}", "{cursor}", "next"].
+	if len(parts) != 4 || parts[0] != "favorites" || parts[3] != "next" {
+		return ""
+	}
+	return parts[2]
 }

submission.go

@@ -35,10 +35,19 @@ type Submission struct {
 	Gender          Gender
 	Description     string // raw HTML; sanitise before rendering to a browser
 	DescriptionText string // plaintext convenience
-	Tags            []string
+	// Tags holds the user-supplied keyword tags. On /view/-path Submissions
+	// these come from div.submission-tags anchors. On listing-path
+	// Submissions (Gallery/Scraps/Favorites/Browse/Search/SubmissionInbox)
+	// they come from the figure's data-tags attribute, which carries the
+	// same keywords FA renders on /view/ for that submission.
+	Tags []string
 	// CategorizedTags groups FA's prefixed system tags by category.
-	// FA emits these as tag-block entries inside div.submission-tags with
-	// prefixes s_ (species), c_ (character), a_/u_ (artist), and t_ (type).
+	// On /view/-path Submissions FA emits these as tag-block entries inside
+	// div.submission-tags with prefixes s_ (species), c_ (character),
+	// a_/u_ (artist), and t_ (type). On listing-path Submissions the same
+	// prefixed tokens are parsed out of the figure's data-tags attribute;
+	// the a_ vs u_ distinction is lost there because FA collapses both into
+	// u_ in that flat list.
 	CategorizedTags CategorizedTags
 	FileURL         string // absolute CDN URL; pass to Download
 	ThumbURL        string