Ability to override reading mode for every single bookshelf item

Keeping the bookshelf page alive would make it feels better.

No reloading, no waiting, and no need to scroll back to where you were.

Ability to limit the max width of images to have better experience.

Trim the blank areas at the edge of images.

Only white area will be trimmed.

• White border detection processor
• Toggle and hotkey
• Help dialog

Maybe you have seen this concept in the Lab:

Actually this was ported from a web project that I made several months ago, which was built by using THREE, an awesome 3D rendering lib for web browsers.

So it's not quite like a native solution, as I have to do a lot of tedious work to get .NET and WebView to communicate with each other, and the performance is poor, because I can only transfer images from .NET into WebView in base64 format, which is not ideal.

I am attempting to implement this using pure .NET technology, such as MonoGame. Although this will take a considerable amount of time, I believe it's worth the effort.

Left: After
Right: Current

Currently RuliaReader can only list the files under main folder, all the sub folders and their contents will be ignored.
Is it possible to add sub folder support, so that we can move into the sub folder in the main folder view and see what's inside.
Many thanks.

目前为止RuliaReader貌似只能扫描主目录下的可阅读内容，而主文件夹下的所有子文件夹及其内容会被忽略。请问能不能增加对子文件夹的阅读支持呢，这样就能在主文件夹的视图下点击进入子文件夹，并查看其中的可阅读内容。十分感谢。

FileMeta

FileMeta is the abstraction of a single file. It should make file loading faster.

• Can analyze all files under a bookshelf folder. There should be a button to do this in the folder page.
• Start automatically when a new folder is being added
• Get the pixel size of every single image in archive and store them
• Use FileMeta when loading a file
• Create the FileMeta when a file has been loaded into the reader
• Files in small size or have few images should not be cached
• There should be one or more indicators to tell users that files are being analyzed
• Make a helping dialog for the first time usage

Thumbnail cache

• Create thumbnails and save them to cache folder
• Thumbnail should be clearable

• Reduce semaphore initial count from 6 to 2
• Fix a condition where semaphore may not be released

This is due to the incorrect implementation of the archive reading function.

别的阅读器可以添加路径作为书籍，直接读取路径下的图片文件，这边希望能加上，只支持压缩包还是不太方便。

Suggest aliases when setting alias in bookshelf.

Example:

• D:\MyManga\SLAM DUNK => SLAM DUNK
• Z:\动画\俺马子与幼驯染的惨烈修罗场（伪）\漫画\裕时悠示×まりも => 俺马子与幼驯染的惨烈修罗场（伪）

It needs a little bit refactoring.

Rulia employs a certain mechanism to swiftly extract image sizes, but it struggles with some unique JPEG formats like this one.

Provide a fallback to ensure that the image sizes for these variants can be read without any issues.

Off:

C:\Users\Rulia\Desktop\The story of Rulia

On:

The story of Rulia

• epub: Supported since 0.8.0.
• mobi: Supported since 0.9.0.

For more information please read: https://github.com/RuliaReader/Rulia/wiki/EPUB-and-Mobi-support

I have no clue why it happens. But there is no value safey check from the start. Anyway I should add this check to ensure noise reduction processor won't blow up.

Replace MetadataExtractor with a brand new module would increase the loading speed.

Only images should be scaled:

This is a known issue. I limited the rendering size of a single image to save memory, but in some conditions it would make it a bad reading experience, like when using vertical scrolling mode.

So I have two options:

• Increase the rendering size
• Split the image into multiple relatively small images in vertical scrolling mode

It should execute in sequence.

This thing shouldn't be there:

And add two entries:

• Bookshelf
• Last read

Well it is not a big deal as Rulia doesn't display entry keys (filenames in archive) on user interface, but after #39 was add, this could be a flaw.

Have to detect the encoding before unpacking archives, which is not an easy job. For now I found this solution:

https://github.com/kuyur/unicue
https://kuyur.net/blog/archives/635

Build a encoding guesser in that way.

• Loading module refactor
• File that is <= 768MB would be loaded into memory completely. This makes loading progress faster, but would takes more memory (up to 768MB).

Make a component to make it easier for one-hand operation.

I have implemented a neural network for determining whether two images are the left and right parts of the same original image. This is very helpful for the image layout process, as it can automatically stitch together two suitable images, eliminating the need for manual switching by the user.

Convert it to onnx format and embed it into Rulia by using built-in Windows AI. (Microsoft.ML.OnnxRuntime is not capable with UWP anymore)