Hooked in character counts per page for estimation, needs some cleanup.

API/Controllers/BookController.cs

@@ -2,6 +2,7 @@
 using System.IO;
 using System.Linq;
 using System.Threading.Tasks;
+using API.Constants;
 using API.Data;
 using API.DTOs.Reader;
 using API.Entities.Enums;
@@ -40,11 +41,14 @@ public class BookController : BaseApiController
     /// <param name="chapterId"></param>
     /// <returns></returns>
     [HttpGet("{chapterId}/book-info")]
-    public async Task<ActionResult<BookInfoDto>> GetBookInfo(int chapterId)
+    [ResponseCache(CacheProfileName = ResponseCacheProfiles.Hour, VaryByQueryKeys = ["chapterId", "includeWordCounts"])]
+    public async Task<ActionResult<BookInfoDto>> GetBookInfo(int chapterId, bool includeWordCounts = false)
     {
         var dto = await _unitOfWork.ChapterRepository.GetChapterInfoDtoAsync(chapterId);
         if (dto == null) return BadRequest(await _localizationService.Translate(User.GetUserId(), "chapter-doesnt-exist"));
         var bookTitle = string.Empty;
+        IDictionary<int, int>? pageWordCounts = null;
+
         switch (dto.SeriesFormat)
         {
             case MangaFormat.Epub:
@@ -52,6 +56,12 @@ public class BookController : BaseApiController
                 var mangaFile = (await _unitOfWork.ChapterRepository.GetFilesForChapterAsync(chapterId))[0];
                 using var book = await EpubReader.OpenBookAsync(mangaFile.FilePath, BookService.LenientBookReaderOptions);
                 bookTitle = book.Title;
+
+                if (includeWordCounts)
+                {
+                    // TODO: Cache this in temp/chapterId folder to avoid having to process file each time
+                    pageWordCounts = await _bookService.GetWordCountsPerPage(mangaFile.FilePath);
+                }
                 break;
             }
             case MangaFormat.Pdf:
@@ -72,7 +82,7 @@ public class BookController : BaseApiController
                 break;
         }
 
-        return Ok(new BookInfoDto()
+        var info = new BookInfoDto()
         {
             ChapterNumber = dto.ChapterNumber,
             VolumeNumber = dto.VolumeNumber,
@@ -84,7 +94,14 @@ public class BookController : BaseApiController
             LibraryId = dto.LibraryId,
             IsSpecial = dto.IsSpecial,
             Pages = dto.Pages,
-        });
+            PageWordCounts = pageWordCounts
+        };
+
+
+
+
+
+        return Ok(info);
     }
 
     /// <summary>

API/Controllers/ReaderController.cs

@@ -41,6 +41,7 @@ public class ReaderController : BaseApiController
     private readonly IEventHub _eventHub;
     private readonly IScrobblingService _scrobblingService;
     private readonly ILocalizationService _localizationService;
+    private readonly IBookService _bookService;
 
     /// <inheritdoc />
     public ReaderController(ICacheService cacheService,
@@ -48,7 +49,8 @@ public class ReaderController : BaseApiController
         IReaderService readerService, IBookmarkService bookmarkService,
         IAccountService accountService, IEventHub eventHub,
         IScrobblingService scrobblingService,
-        ILocalizationService localizationService)
+        ILocalizationService localizationService,
+        IBookService bookService)
     {
         _cacheService = cacheService;
         _unitOfWork = unitOfWork;
@@ -59,6 +61,7 @@ public class ReaderController : BaseApiController
         _eventHub = eventHub;
         _scrobblingService = scrobblingService;
         _localizationService = localizationService;
+        _bookService = bookService;
     }
 
     /// <summary>
@@ -218,11 +221,11 @@ public class ReaderController : BaseApiController
     /// <remarks>This is generally the first call when attempting to read to allow pre-generation of assets needed for reading</remarks>
     /// <param name="chapterId"></param>
     /// <param name="extractPdf">Should Kavita extract pdf into images. Defaults to false.</param>
-    /// <param name="includeDimensions">Include file dimensions. Only useful for image based reading</param>
+    /// <param name="includeDimensions">Include file dimensions. Only useful for image-based reading</param>
+    /// <param name="includeWordCounts">Include epub word counts per page. Only useful for epub-based reading</param>
     /// <returns></returns>
     [HttpGet("chapter-info")]
-    [ResponseCache(CacheProfileName = ResponseCacheProfiles.Hour, VaryByQueryKeys = ["chapterId", "extractPdf", "includeDimensions"
+    [ResponseCache(CacheProfileName = ResponseCacheProfiles.Hour, VaryByQueryKeys = ["chapterId", "extractPdf", "includeDimensions"])]
-    ])]
     public async Task<ActionResult<ChapterInfoDto>> GetChapterInfo(int chapterId, bool extractPdf = false, bool includeDimensions = false)
     {
         if (chapterId <= 0) return Ok(null); // This can happen occasionally from UI, we should just ignore
@@ -846,6 +849,7 @@ public class ReaderController : BaseApiController
         // Patch in the reading progress
         await _unitOfWork.ChapterRepository.AddChapterModifiers(User.GetUserId(), chapter);
 
+        // TODO: We need to actually use word count from the pages
         if (series.Format == MangaFormat.Epub)
         {
             var progressCount = chapter.WordCount;

API/DTOs/Reader/BookInfoDto.cs

@@ -1,4 +1,5 @@
-using API.Entities.Enums;
+using System.Collections.Generic;
+using API.Entities.Enums;
 
 namespace API.DTOs.Reader;
 
@@ -15,4 +16,9 @@ public sealed record BookInfoDto : IChapterInfoDto
     public int Pages { get; set; }
     public bool IsSpecial { get; set; }
     public string ChapterTitle { get; set; } = default! ;
+    /// <summary>
+    /// For Epub reader, this will contain Page number -> word count. All other times will be null.
+    /// </summary>
+    /// <remarks>This is optionally returned by includeWordCounts</remarks>
+    public IDictionary<int, int>? PageWordCounts { get; set; }
 }

API/Services/BookService.cs

@@ -14,6 +14,7 @@ using API.Entities.Enums;
 using API.Extensions;
 using API.Services.Tasks.Scanner.Parser;
 using API.Helpers;
+using API.Services.Tasks.Metadata;
 using Docnet.Core;
 using Docnet.Core.Converters;
 using Docnet.Core.Models;
@@ -59,6 +60,7 @@ public interface IBookService
     Task<ICollection<BookChapterItem>> GenerateTableOfContents(Chapter chapter);
     Task<string> GetBookPage(int page, int chapterId, string cachedEpubPath, string baseUrl, List<PersonalToCDto> ptocBookmarks, List<AnnotationDto> annotations);
     Task<Dictionary<string, int>> CreateKeyToPageMappingAsync(EpubBookRef book);
+    Task<IDictionary<int, int>?> GetWordCountsPerPage(string bookFilePath);
 }
 
 public class BookService : IBookService
@@ -955,6 +957,50 @@ public class BookService : IBookService
         return dict;
     }
 
+    public async Task<IDictionary<int, int>?> GetWordCountsPerPage(string bookFilePath)
+    {
+        var ret = new Dictionary<int, int>();
+        try
+        {
+            using var book = await EpubReader.OpenBookAsync(bookFilePath, LenientBookReaderOptions);
+            var mappings = await CreateKeyToPageMappingAsync(book);
+
+            var doc = new HtmlDocument {OptionFixNestedTags = true};
+
+
+            var bookPages = await book.GetReadingOrderAsync();
+            foreach (var contentFileRef in bookPages)
+            {
+                var page = mappings[contentFileRef.Key];
+                var content = await contentFileRef.ReadContentAsync();
+                doc.LoadHtml(content);
+
+                var body = doc.DocumentNode.SelectSingleNode("//body");
+
+                if (body == null)
+                {
+                    _logger.LogError("{FilePath} has no body tag! Generating one for support. Book may be skewed", book.FilePath);
+                    doc.DocumentNode.SelectSingleNode("/html").AppendChild(HtmlNode.CreateNode("<body></body>"));
+                    body = doc.DocumentNode.SelectSingleNode("//html/body");
+                }
+
+                // Find all words in the html body
+                // TEMP: REfactor this to use WordCountAnalyzerService
+                var textNodes = body!.SelectNodes("//text()[not(parent::script)]");
+                ret.Add(page, textNodes?.Sum(node => node.InnerText.Count(char.IsLetter)) ?? 0);
+
+            }
+
+        }
+        catch (Exception ex)
+        {
+            _logger.LogError(ex, "There was an issue calculating word counts per page");
+            return null;
+        }
+
+        return ret;
+    }
+
     /// <summary>
     /// Parses out Title from book. Chapters and Volumes will always be "0". If there is any exception reading book (malformed books)
     /// then null is returned. This expects only an epub file

API/Services/Tasks/Metadata/WordCountAnalyzerService.cs

@@ -35,7 +35,7 @@ public class WordCountAnalyzerService : IWordCountAnalyzerService
     private readonly IReaderService _readerService;
     private readonly IMediaErrorService _mediaErrorService;
 
-    private const int AverageCharactersPerWord = 5;
+    public const int AverageCharactersPerWord = 5;
 
     public WordCountAnalyzerService(ILogger<WordCountAnalyzerService> logger, IUnitOfWork unitOfWork, IEventHub eventHub,
         ICacheHelper cacheHelper, IReaderService readerService, IMediaErrorService mediaErrorService)

UI/Web/src/app/book-reader/_components/book-reader/book-reader.component.ts

@@ -662,7 +662,7 @@ export class BookReaderComponent implements OnInit, AfterViewInit, OnDestroy {
     this.cdRef.markForCheck();
 
 
-    this.bookService.getBookInfo(this.chapterId).subscribe(async (info) => {
+    this.bookService.getBookInfo(this.chapterId, true).subscribe(async (info) => {
       if (this.readingListMode && info.seriesFormat !== MangaFormat.EPUB) {
         // Redirect to the manga reader.
         const params = this.readerService.getQueryParamsObject(this.incognitoMode, this.readingListMode, this.readingListId);

UI/Web/src/app/book-reader/_models/book-info.ts

@@ -1,4 +1,4 @@
-import { MangaFormat } from "src/app/_models/manga-format";
+import {MangaFormat} from "src/app/_models/manga-format";
 
 export interface BookInfo {
   bookTitle: string;
@@ -6,4 +6,8 @@ export interface BookInfo {
   seriesId: number;
   libraryId: number;
   volumeId: number;
+  /**
+   * Maps the page number to character count. Only available on epub reader.
+   */
+  pageWordCounts: {[key: number]: number};
 }

UI/Web/src/app/book-reader/_services/book.service.ts

@@ -1,9 +1,9 @@
-import { HttpClient } from '@angular/common/http';
+import {HttpClient} from '@angular/common/http';
-import { Injectable } from '@angular/core';
+import {Injectable} from '@angular/core';
-import { TextResonse } from 'src/app/_types/text-response';
+import {TextResonse} from 'src/app/_types/text-response';
-import { environment } from 'src/environments/environment';
+import {environment} from 'src/environments/environment';
-import { BookChapterItem } from '../_models/book-chapter-item';
+import {BookChapterItem} from '../_models/book-chapter-item';
-import { BookInfo } from '../_models/book-info';
+import {BookInfo} from '../_models/book-info';
 
 export interface FontFamily {
   /**
@@ -28,7 +28,8 @@ export class BookService {
   getFontFamilies(): Array<FontFamily> {
     return [{title: 'default', family: 'default'}, {title: 'EBGaramond', family: 'EBGaramond'}, {title: 'Fira Sans', family: 'Fira_Sans'},
     {title: 'Lato', family: 'Lato'}, {title: 'Libre Baskerville', family: 'Libre_Baskerville'}, {title: 'Merriweather', family: 'Merriweather'},
-    {title: 'Nanum Gothic', family: 'Nanum_Gothic'}, {title: 'Open Dyslexic', family: 'OpenDyslexic2'}, {title: 'RocknRoll One', family: 'RocknRoll_One'}, {title: 'Fast Font Serif (Bionic)', family: 'FastFontSerif'}, {title: 'Fast Font Sans (Bionic)', family: 'FastFontSans'}];
+    {title: 'Nanum Gothic', family: 'Nanum_Gothic'}, {title: 'Open Dyslexic', family: 'OpenDyslexic2'}, {title: 'RocknRoll One', family: 'RocknRoll_One'},
+    {title: 'Fast Font Serif (Bionic)', family: 'FastFontSerif'}, {title: 'Fast Font Sans (Bionic)', family: 'FastFontSans'}];
   }
 
   getBookChapters(chapterId: number) {
@@ -39,8 +40,8 @@ export class BookService {
     return this.http.get<string>(this.baseUrl + 'book/' + chapterId + '/book-page?page=' + page, TextResonse);
   }
 
-  getBookInfo(chapterId: number) {
+  getBookInfo(chapterId: number, includeWordCounts: boolean = false) {
-    return this.http.get<BookInfo>(this.baseUrl + 'book/' + chapterId + '/book-info');
+    return this.http.get<BookInfo>(this.baseUrl + `book/${chapterId}/book-info?includeWordCounts=${includeWordCounts}`);
   }
 
   getBookPageUrl(chapterId: number, page: number) {