Kindle: Go back to using digital-text, but improve code handling

kindle/book.go

@@ -7,54 +7,61 @@ import (
 	"github.com/ahobsonsayers/abs-tract/utils"
 	"github.com/antchfx/htmlquery"
 	"github.com/antchfx/xpath"
+	mapset "github.com/deckarep/golang-set/v2"
 	"golang.org/x/net/html"
 )
 
 const publishDateLayout = "Jan 2, 2006"
 
 var (
 	bookCoverSetExpr  = xpath.MustCompile(`.//img/@srcset`)
-	bookFormatExpr    = xpath.MustCompile(`.//a[contains(text(), "Kindle Edition")]//text()`)
+	bookFormatExpr    = xpath.MustCompile(`.//a[matches(., "Kindle|Hardcover|Paperback")]//text()`)
 	bookInfoExpr      = xpath.MustCompile(`.//div[contains(@class, "a-color-secondary")]`)
 	bookTitleExpr     = xpath.MustCompile(`.//h2`)
 	searchResultsExpr = xpath.MustCompile(`//div[contains(@class, "s-result-list")]//div[@data-index and @data-asin]`)
 )
 
 type Book struct {
 	ASIN        string
+	Format      string
 	Title       string
 	Author      string
 	Cover       string
 	PublishDate *time.Time
 }
 
-// BooksFromHTML parses and returns the books from the html of a search results page
+// BooksFromHTML parses the books from the html of a search results page.
 func BooksFromHTML(searchNode *html.Node) ([]Book, error) {
 	resultNodes := htmlquery.QuerySelectorAll(searchNode, searchResultsExpr)
 
 	books := make([]Book, 0, len(resultNodes))
+	seenAsins := mapset.NewSet[string]()
 	for _, resultNode := range resultNodes {
-		if !isKindleBook(resultNode) {
-			continue
-		}
-
+		// Attempt to parse result as a book, recording it if it
+		// could be parsed and the asin has not already been seen
 		book := BookFromHTML(resultNode)
-		if book != nil {
+		if book != nil && !seenAsins.Contains(book.ASIN) {
 			books = append(books, *book)
+			seenAsins.Add(book.ASIN)
 		}
 	}
 
 	return books, nil
 }
 
-// BookFromHTML parses and returns a book from the html
-// of a book result on the search results page
+// BookFromHTML parses a book from the html of a result on the search results page.
+// If a result is not for a book, nil will is returned.
 func BookFromHTML(bookNode *html.Node) *Book {
 	asin := bookAsin(bookNode)
 	if asin == "" {
 		return nil
 	}
 
+	format := bookFormat(bookNode)
+	if format == "" {
+		return nil
+	}
+
 	title := bookTitle(bookNode)
 	if title == "" {
 		return nil
@@ -65,27 +72,28 @@ func BookFromHTML(bookNode *html.Node) *Book {
 
 	return &Book{
 		ASIN:        asin,
+		Format:      format,
 		Title:       title,
 		Author:      author,
 		Cover:       cover,
 		PublishDate: publishDate,
 	}
 }
 
-func isKindleBook(bookNode *html.Node) bool {
+// bookAsin gets the book asim.
+func bookAsin(bookNode *html.Node) string {
+	return htmlquery.SelectAttr(bookNode, "data-asin")
+}
+
+// bookFormat gets the book format
+func bookFormat(bookNode *html.Node) string {
 	bookFormatNode := htmlquery.QuerySelector(bookNode, bookFormatExpr)
 	if bookFormatNode == nil {
-		return false
+		return ""
 	}
-	bookFormatNodeValue := htmlquery.InnerText(bookFormatNode)
-
-	bookFormat := strings.ToLower(bookFormatNodeValue)
-	return strings.Contains(bookFormat, "kindle")
-}
 
-// bookAsin gets the book asim.
-func bookAsin(bookNode *html.Node) string {
-	return htmlquery.SelectAttr(bookNode, "data-asin")
+	bookFormatNodeValue := htmlquery.InnerText(bookFormatNode)
+	return strings.ToLower(bookFormatNodeValue)
 }
 
 // bookTitle gets the book title.

kindle/kindle.go

@@ -71,7 +71,8 @@ func (c *Client) get(
 
 func (c *Client) Search(ctx context.Context, title string, author *string) ([]Book, error) {
 	parameters := map[string]string{
-		"i": "stripbooks",
+		"i": "digital-text",
+		// "i": "stripbooks",
 		"k": title,
 	}
 	if author != nil && *author != "" {

kindle/kindle_test.go

@@ -15,9 +15,13 @@ const (
 )
 
 func TestSearchBook(t *testing.T) {
-	t.Skip("Fails in CI")
+	// t.Skip("Fails in CI")
 	// Should return https://www.amazon.com/dp/B007978NU6
-	books, err := kindle.DefaultClient.Search(context.Background(), TheHobbitTitle, lo.ToPtr(TheHobbitAuthor))
+	books, err := kindle.DefaultClient.Search(
+		context.Background(),
+		TheHobbitTitle,
+		lo.ToPtr(TheHobbitAuthor),
+	)
 	require.NoError(t, err)
 	require.NotEmpty(t, books)