chore: upgrade GitHub Actions to latest versions

.github/workflows/docs.yml

@@ -26,14 +26,14 @@ jobs:
   build:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v6.0.2
+      - uses: actions/checkout@v7.0.0
         with:
           submodules: true
           # Sitemap lastmod comes from the latest content commit.
           fetch-depth: 0
 
       - name: Checkout tago
-        uses: actions/checkout@v6.0.2
+        uses: actions/checkout@v7.0.0
         with:
           repository: tamnd/tago
           path: .tago-src
@@ -107,7 +107,7 @@ jobs:
       group: cloudflare-pages-chemrxiv-cli
       cancel-in-progress: true
     steps:
-      - uses: actions/checkout@v6.0.2
+      - uses: actions/checkout@v7.0.0
         with:
           fetch-depth: 1
           sparse-checkout: scripts/

.github/workflows/release.yml

@@ -68,7 +68,7 @@ jobs:
 
       # Tools GoReleaser shells out to for signing and SBOMs.
       - uses: sigstore/cosign-installer@v3
-      - uses: anchore/sbom-action/download-syft@v0
+      - uses: anchore/sbom-action/download-syft@v0.24.0
 
       - uses: goreleaser/goreleaser-action@v6
         with:

chemrxiv/chemrxiv.go

@@ -85,14 +85,20 @@ func (c *Client) graphqlEndpoint() string {
 // Search fetches preprints matching term. An empty term returns the most
 // recent preprints. Returns up to limit results and the server total count.
 func (c *Client) Search(ctx context.Context, term string, limit int) ([]Preprint, int, error) {
+	return c.SearchFrom(ctx, term, 0, limit)
+}
+
+// SearchFrom fetches preprints matching term starting from the given skip
+// offset. Returns up to limit results and the server total count.
+func (c *Client) SearchFrom(ctx context.Context, term string, skip, limit int) ([]Preprint, int, error) {
 	if limit <= 0 {
 		limit = 20
 	}
 
 	const pageSize = 25
 	var out []Preprint
 	total := 0
-	skip := 0
+	cursor := skip
 
 	for {
 		fetch := pageSize
@@ -102,7 +108,7 @@ func (c *Client) Search(ctx context.Context, term string, limit int) ([]Preprint
 
 		q := fmt.Sprintf(
 			`{ itemsByKeyword(skip:%d, limit:%d, term:%s) { totalCount itemHits { item { id doi title statusDate publishedDate version authors { firstName lastName institutions { name } } categories { name } keywords license metrics { viewCount downloadCount } asset { original { url } } } } } }`,
-			skip, fetch, jsonString(term),
+			cursor, fetch, jsonString(term),
 		)
 
 		var resp graphqlResponse[searchData]
@@ -117,7 +123,7 @@ func (c *Client) Search(ctx context.Context, term string, limit int) ([]Preprint
 		total = result.TotalCount
 
 		for _, hit := range result.ItemHits {
-			out = append(out, wireToPreprint(hit.Item, len(out)+1))
+			out = append(out, wireToPreprint(hit.Item, skip+len(out)+1))
 			if len(out) >= limit {
 				return out, total, nil
 			}
@@ -126,14 +132,19 @@ func (c *Client) Search(ctx context.Context, term string, limit int) ([]Preprint
 		if len(result.ItemHits) < fetch {
 			break
 		}
-		skip += fetch
+		cursor += fetch
 	}
 	return out, total, nil
 }
 
 // Recent returns the most recent preprints (empty term search).
 func (c *Client) Recent(ctx context.Context, limit int) ([]Preprint, int, error) {
-	return c.Search(ctx, "", limit)
+	return c.SearchFrom(ctx, "", 0, limit)
+}
+
+// RecentFrom returns recent preprints starting from the given skip offset.
+func (c *Client) RecentFrom(ctx context.Context, skip, limit int) ([]Preprint, int, error) {
+	return c.SearchFrom(ctx, "", skip, limit)
 }
 
 // GetPreprint fetches a single preprint by its ChemRxiv ID.

chemrxiv/chemrxiv_test.go

@@ -413,3 +413,51 @@ func TestRecent(t *testing.T) {
 		t.Error("Recent returned no preprints")
 	}
 }
+
+func TestSearchFromSkip(t *testing.T) {
+	// Verify that SearchFrom sends the skip value in the query and adjusts rank.
+	var capturedQuery string
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		b, _ := io.ReadAll(r.Body)
+		var body map[string]string
+		_ = json.Unmarshal(b, &body)
+		capturedQuery = body["query"]
+		_, _ = w.Write([]byte(sampleSearchResponse))
+	}))
+	defer srv.Close()
+
+	c := newTestClient(srv)
+	preprints, _, err := c.SearchFrom(context.Background(), "oxygen", 10, 5)
+	if err != nil {
+		t.Fatal(err)
+	}
+	// Query should contain skip:10
+	if !strings.Contains(capturedQuery, "skip:10") {
+		t.Errorf("query does not contain skip:10; got: %s", capturedQuery)
+	}
+	// Ranks should start from skip+1 = 11
+	if len(preprints) > 0 && preprints[0].Rank != 11 {
+		t.Errorf("first rank = %d, want 11 (skip=10)", preprints[0].Rank)
+	}
+}
+
+func TestRateLimitedOn429(t *testing.T) {
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		w.WriteHeader(http.StatusTooManyRequests)
+	}))
+	defer srv.Close()
+
+	cfg := DefaultConfig()
+	cfg.BaseURL = srv.URL
+	cfg.Rate = 0
+	cfg.Retries = 2
+	c := NewClient(cfg)
+
+	_, _, err := c.Search(context.Background(), "test", 5)
+	if err == nil {
+		t.Fatal("expected error after 429 retries exhausted, got nil")
+	}
+	if !strings.Contains(err.Error(), "429") {
+		t.Errorf("error should mention 429, got: %v", err)
+	}
+}

cli/cmd_recent.go

@@ -5,18 +5,21 @@ import (
 )
 
 func (a *App) recentCmd() *cobra.Command {
-	return &cobra.Command{
+	var skip int
+	cmd := &cobra.Command{
 		Use:   "recent",
 		Short: "List the most recent ChemRxiv preprints",
 		Long:  "Fetch the most recently submitted ChemRxiv preprints. Default: up to 20 results.",
 		RunE: func(cmd *cobra.Command, _ []string) error {
 			n := a.effectiveLimit(20)
-			a.progressf("fetching %d recent preprints...", n)
-			preprints, _, err := a.client.Recent(cmd.Context(), n)
+			a.progressf("fetching %d recent preprints (skip %d)...", n, skip)
+			preprints, _, err := a.client.RecentFrom(cmd.Context(), skip, n)
 			if err != nil {
 				return mapFetchErr(err)
 			}
 			return a.renderOrEmpty(preprints, len(preprints))
 		},
 	}
+	cmd.Flags().IntVar(&skip, "skip", 0, "number of results to skip before returning")
+	return cmd
 }

cli/cmd_search.go

@@ -5,7 +5,8 @@ import (
 )
 
 func (a *App) searchCmd() *cobra.Command {
-	return &cobra.Command{
+	var skip int
+	cmd := &cobra.Command{
 		Use:   "search <query>",
 		Short: "Search ChemRxiv preprints by keyword",
 		Long: `Search ChemRxiv preprints by keyword using the public GraphQL API.
@@ -14,13 +15,15 @@ Paginates automatically until the requested limit is reached.`,
 		RunE: func(cmd *cobra.Command, args []string) error {
 			query := args[0]
 			n := a.effectiveLimit(20)
-			a.progressf("searching for %q (limit %d)...", query, n)
-			preprints, total, err := a.client.Search(cmd.Context(), query, n)
+			a.progressf("searching for %q (skip %d, limit %d)...", query, skip, n)
+			preprints, total, err := a.client.SearchFrom(cmd.Context(), query, skip, n)
 			if err != nil {
 				return mapFetchErr(err)
 			}
 			a.progressf("found %d total; showing %d", total, len(preprints))
 			return a.renderOrEmpty(preprints, len(preprints))
 		},
 	}
+	cmd.Flags().IntVar(&skip, "skip", 0, "number of results to skip before returning")
+	return cmd
 }