[fix] server word count, [add] client word count

api/app-routes.go

@@ -209,7 +209,12 @@ func (api *API) createAppResourcesRoute(routeName string, args ...map[string]any
 				}
 
 				// Search
-				searchResults := search.SearchBook(*sParams.Query, bType)
+				searchResults, err := search.SearchBook(*sParams.Query, bType)
+				if err != nil {
+					errorPage(c, http.StatusInternalServerError, fmt.Sprintf("Search Error: %v", err))
+					return
+				}
+
 				templateVars["Data"] = searchResults
 				templateVars["BookType"] = *sParams.BookType
 			}

assets/reader/index.js

@@ -45,9 +45,15 @@ class EBookReader {
    * Load progress and generate locations
    **/
   async setupReader() {
+    // Get Word Count (If Needed)
+    if (this.bookState.words == 0)
+      this.bookState.words = await this.countWords();
+
     // Load Progress
     let { cfi } = await this.getCFIFromXPath(this.bookState.progress);
-    if (!cfi) this.bookState.currentWord = 0;
+    this.bookState.currentWord = cfi
+      ? this.bookState.percentage * (this.bookState.words / 100)
+      : 0;
 
     let getStats = function () {
       // Start Timer
@@ -1038,6 +1044,21 @@ class EBookReader {
     );
   }
 
+  /**
+   * Count the words of the book. Useful for keeping a more accurate track
+   * of progress percentage. Implementation returns the same number as the
+   * server side implementation.
+   **/
+  countWords() {
+    // Iterate over each item in the spine, render, and count words.
+    return this.book.spine.spineItems.reduce(async (totalCount, item) => {
+      let currentCount = await totalCount;
+      let newDoc = await item.load(this.book.load.bind(this.book));
+      let itemCount = newDoc.innerText.trim().split(/\s+/).length;
+      return currentCount + itemCount;
+    }, 0);
+  }
+
   /**
    * Save settings to localStorage
    **/

metadata/epub.go

@@ -1,11 +1,10 @@
 package metadata
 
 import (
-	"io"
 	"strings"
 
+	"github.com/PuerkitoBio/goquery"
 	"github.com/taylorskalyo/goreader/epub"
-	"golang.org/x/net/html"
 )
 
 func getEPUBMetadata(filepath string) (*MetadataInfo, error) {
@@ -32,33 +31,9 @@ func countEPUBWords(filepath string) (int64, error) {
 	var completeCount int64
 	for _, item := range rf.Spine.Itemrefs {
 		f, _ := item.Open()
-		tokenizer := html.NewTokenizer(f)
-		newCount, err := countTokenizerWords(*tokenizer)
-		if err != nil {
-			return 0, err
-		}
-		completeCount = completeCount + newCount
+		doc, _ := goquery.NewDocumentFromReader(f)
+		completeCount = completeCount + int64(len(strings.Fields(doc.Text())))
 	}
 
 	return completeCount, nil
 }
-
-func countTokenizerWords(tokenizer html.Tokenizer) (int64, error) {
-	var err error
-	var totalWords int64
-	for {
-		tokenType := tokenizer.Next()
-		token := tokenizer.Token()
-		if tokenType == html.TextToken {
-			currStr := string(token.Data)
-			totalWords = totalWords + int64(len(strings.Fields(currStr)))
-		} else if tokenType == html.ErrorToken {
-			err = tokenizer.Err()
-		}
-		if err == io.EOF {
-			return totalWords, nil
-		} else if err != nil {
-			return 0, err
-		}
-	}
-}

metadata/metadata_test.go

@@ -5,7 +5,7 @@ import (
 )
 
 func TestGetWordCount(t *testing.T) {
-	var want int64 = 30477
+	var want int64 = 30080
 	wordCount, err := countEPUBWords("../_test_files/alice.epub")
 
 	if wordCount != want {

search/search.go

@@ -38,29 +38,37 @@ type SearchItem struct {
 	UploadDate string
 }
 
-func SearchBook(query string, bookType BookType) (allEntries []SearchItem) {
-	log.Info(query)
+func SearchBook(query string, bookType BookType) ([]SearchItem, error) {
 	if bookType == BOOK_FICTION {
 		// Search Fiction
 		url := "https://libgen.is/fiction/?q=" + url.QueryEscape(query) + "&language=English&format=epub"
-		body := getPage(url)
-		allEntries = parseLibGenFiction(body)
+		body, err := getPage(url)
+		if err != nil {
+			return nil, err
+		}
+		return parseLibGenFiction(body)
 	} else if bookType == BOOK_NON_FICTION {
 		// Search NonFiction
 		url := "https://libgen.is/search.php?req=" + url.QueryEscape(query)
-		body := getPage(url)
-		allEntries = parseLibGenNonFiction(body)
+		body, err := getPage(url)
+		if err != nil {
+			return nil, err
+		}
+		return parseLibGenNonFiction(body)
+	} else {
+		return nil, errors.New("Invalid Book Type")
+	}
 }
 
-	return
+func GoodReadsMostRead(c Cadence) ([]SearchItem, error) {
+	body, err := getPage("https://www.goodreads.com/book/most_read?category=all&country=US&duration=" + string(c))
+	if err != nil {
+		return nil, err
 	}
-
-func GoodReadsMostRead(c Cadence) []SearchItem {
-	body := getPage("https://www.goodreads.com/book/most_read?category=all&country=US&duration=" + string(c))
 	return parseGoodReads(body)
 }
 
-func GetBookURL(id string, bookType BookType) string {
+func GetBookURL(id string, bookType BookType) (string, error) {
 	// Derive Info URL
 	var infoURL string
 	if bookType == BOOK_FICTION {
@@ -70,7 +78,10 @@ func GetBookURL(id string, bookType BookType) string {
 	}
 
 	// Parse & Derive Download URL
-	body := getPage(infoURL)
+	body, err := getPage(infoURL)
+	if err != nil {
+		return "", err
+	}
 
 	// downloadURL := parseLibGenDownloadURL(body)
 	return parseLibGenDownloadURL(body)
@@ -86,8 +97,15 @@ func SaveBook(id string, bookType BookType) (string, error) {
 	}
 
 	// Parse & Derive Download URL
-	body := getPage(infoURL)
-	bookURL := parseLibGenDownloadURL(body)
+	body, err := getPage(infoURL)
+	if err != nil {
+		return "", err
+	}
+	bookURL, err := parseLibGenDownloadURL(body)
+	if err != nil {
+		log.Error("[SaveBook] Parse Download URL Error: ", err)
+		return "", errors.New("Download Failure")
+	}
 
 	// Create File
 	tempFile, err := os.CreateTemp("", "book")
@@ -119,15 +137,29 @@ func SaveBook(id string, bookType BookType) (string, error) {
 	return tempFile.Name(), nil
 }
 
-func getPage(page string) io.ReadCloser {
-	resp, _ := http.Get(page)
-	return resp.Body
+func getPage(page string) (io.ReadCloser, error) {
+	// Set 10s Timeout
+	client := http.Client{
+		Timeout: 10 * time.Second,
 	}
 
-func parseLibGenFiction(body io.ReadCloser) []SearchItem {
+	// Get Page
+	resp, err := client.Get(page)
+	if err != nil {
+		return nil, err
+	}
+
+	// Return Body
+	return resp.Body, err
+}
+
+func parseLibGenFiction(body io.ReadCloser) ([]SearchItem, error) {
 	// Parse
 	defer body.Close()
-	doc, _ := goquery.NewDocumentFromReader(body)
+	doc, err := goquery.NewDocumentFromReader(body)
+	if err != nil {
+		return nil, err
+	}
 
 	// Normalize Results
 	var allEntries []SearchItem
@@ -171,13 +203,16 @@ func parseLibGenFiction(body io.ReadCloser) []SearchItem {
 	})
 
 	// Return Results
-	return allEntries
+	return allEntries, nil
 }
 
-func parseLibGenNonFiction(body io.ReadCloser) []SearchItem {
+func parseLibGenNonFiction(body io.ReadCloser) ([]SearchItem, error) {
 	// Parse
 	defer body.Close()
-	doc, _ := goquery.NewDocumentFromReader(body)
+	doc, err := goquery.NewDocumentFromReader(body)
+	if err != nil {
+		return nil, err
+	}
 
 	// Normalize Results
 	var allEntries []SearchItem
@@ -213,25 +248,31 @@ func parseLibGenNonFiction(body io.ReadCloser) []SearchItem {
 	})
 
 	// Return Results
-	return allEntries
+	return allEntries, nil
 }
 
-func parseLibGenDownloadURL(body io.ReadCloser) string {
+func parseLibGenDownloadURL(body io.ReadCloser) (string, error) {
 	// Parse
 	defer body.Close()
 	doc, _ := goquery.NewDocumentFromReader(body)
 
 	// Return Download URL
 	// downloadURL, _ := doc.Find("#download [href*=cloudflare]").Attr("href")
-	downloadURL, _ := doc.Find("#download h2 a").Attr("href")
-
-	return downloadURL
+	downloadURL, exists := doc.Find("#download h2 a").Attr("href")
+	if exists == false {
+		return "", errors.New("Download URL not found")
 	}
 
-func parseGoodReads(body io.ReadCloser) []SearchItem {
+	return downloadURL, nil
+}
+
+func parseGoodReads(body io.ReadCloser) ([]SearchItem, error) {
 	// Parse
 	defer body.Close()
-	doc, _ := goquery.NewDocumentFromReader(body)
+	doc, err := goquery.NewDocumentFromReader(body)
+	if err != nil {
+		return nil, err
+	}
 
 	// Normalize Results
 	var allEntries []SearchItem
@@ -249,5 +290,5 @@ func parseGoodReads(body io.ReadCloser) []SearchItem {
 	})
 
 	// Return Results
-	return allEntries
+	return allEntries, nil
 }