From 1b8b5060f1ca8d540ebf6a071617aae3cd725185 Mon Sep 17 00:00:00 2001
From: Evan Reichard <evan@reichard.io>
Date: Thu, 26 Oct 2023 20:20:58 -0400
Subject: [PATCH] [fix] server word count, [add] client word count

---
 assets/reader/index.js    | 23 ++++++++++++++++++++++-
 metadata/epub.go          | 31 +++----------------------------
 metadata/metadata_test.go |  2 +-
 3 files changed, 26 insertions(+), 30 deletions(-)

diff --git a/assets/reader/index.js b/assets/reader/index.js
index 8baf329..d335c38 100644
--- a/assets/reader/index.js
+++ b/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
    **/
diff --git a/metadata/epub.go b/metadata/epub.go
index e5904a1..b518ba8 100644
--- a/metadata/epub.go
+++ b/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
-		}
-	}
-}
diff --git a/metadata/metadata_test.go b/metadata/metadata_test.go
index 322591b..e4a47a4 100644
--- a/metadata/metadata_test.go
+++ b/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 {