Skip to content

Commit

Permalink
docs: update README and CHANGELOG for v0.10
Browse files Browse the repository at this point in the history
  • Loading branch information
fauu committed Nov 15, 2022
1 parent 0245ae4 commit 748d517
Show file tree
Hide file tree
Showing 2 changed files with 72 additions and 10 deletions.
65 changes: 64 additions & 1 deletion CHANGELOG.md
Original file line number Diff line number Diff line change
Expand Up @@ -8,6 +8,68 @@

## [Unreleased]

## [0.10] – 2022-11-15

### Changed

* User selection within current chunk outside Edit mode is no longer removed
when the cursor leaves the client browser tab.
* If there is a user selection within current chunk outside edit mode when a new
chunk comes in, the current chunk is now replaced entirely, not just the
selected part. (To have just a part of it replaced instead, enter Edit mode by
double-clicking the current chunk area and select the text to be replaced
there.)
* Startup time with OCR enabled has been improved in some cases by initializing
Recognizer in a separate thread.
* Remote OCR request that has failed to complete is now retried twice in some
cases before a failure is reported to the client.
* The current chunk is now faded out instead of hidden while OCR is in progress.
* Characters from unicode block [Specials][unicode-specials] are now stripped
from incoming chunks if `chunk.correct = true` (it is so by default).
* Default furigana font size has been decreased and further downscaling is now
applied to it depending on the length of the annotation relatively to the base
text—this in order to prevent the furigana overflowing and introducing
inconsistencies in spacing between characters. (For how to restore previous
behaviour, see the
[„Current chunk” font sizing][wiki-current-chunk-font-sizing] section in the
Wiki.)
* Furigana font can now be scaled in `custom.css` using the variable
`--chunk-furigana-font-scale`. It is now also scaled with `--chunk-font-size`.
(For details, see the
[„Current chunk” font sizing][wiki-current-chunk-font-sizing] section
in the Wiki.)
* Removed the CSS variable `--chunk-line-height` for now, since due to browser
quirks and furigana interactions it was not fully functional for regulating
the line height in `custom.css`.
* Added CSS variables `--font-ui` and `--font-jp` for easier customization of
font faces. (See the [Font faces][wiki-font-faces] section in the Wiki)
* Added CSS variables `--chunk-font-weight` and `--chunk-furigana-font-weight`
for easier customization of chunk font weight. (See the
[„Current chunk” font weight][wiki-current-chunk-font-weight]
section in the Wiki.)
* Make user selection in current chunk visually stand out a bit more.

[unicode-specials]: https://en.wikipedia.org/wiki/Specials_(Unicode_block)
[wiki-current-chunk-font-sizing]: https://github.com/fauu/Kamite/wiki/Styling-recipes#current-chunk-font-sizing
[wiki-font-faces]: https://github.com/fauu/Kamite/wiki/Styling-recipes#font-faces
[wiki-current-chunk-font-weight]: https://github.com/fauu/Kamite/wiki/Styling-recipes#current-chunk-font-weight

### Fixed

* “Manga OCR” Online OCR engine works again (switched to
[a Hugging Face Space by Detomo][manga-ocr-hf-detomo]).
* Current chunk label now has correct line height on Chrome.
* Status panel fade is now correctly recalculated in a certain situation where
it previously was not.
* Empty incoming chunks are now rejected in some circumstances when they might
not have been previously.
* Likely reduced the incidence of Tesseract executable availability check
failing when it should not.
* *Select highlighted* action no longer appears when it would be a no-op because
of the chunk selection already being equal to the highlight.

[manga-ocr-hf-detomo]: https://huggingface.co/spaces/Detomo/Japanese-OCR

## [0.9] – 2022-10-31

### Fixed
Expand Down Expand Up @@ -243,7 +305,8 @@

Initial release.

[Unreleased]: https://github.com/fauu/Kamite/compare/v0.9...HEAD
[Unreleased]: https://github.com/fauu/Kamite/compare/v0.10...HEAD
[0.10]: https://github.com/fauu/Kamite/releases/tag/v0.10
[0.9]: https://github.com/fauu/Kamite/releases/tag/v0.9
[0.8]: https://github.com/fauu/Kamite/releases/tag/v0.8
[0.7]: https://github.com/fauu/Kamite/releases/tag/v0.7
Expand Down
17 changes: 8 additions & 9 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -378,10 +378,10 @@ Kamite integrates with four alternative OCR (Optical Character Recognition)
providers to enable the extraction of text from manga pages displayed on screen.
The available OCR engines are:

* “Manga OCR” Online (a Hugging Face Space by Gryan Galario)
* “Manga OCR” Online (a Hugging Face Space by Detomo)

<https://github.com/kha-white/manga-ocr>\
<https://huggingface.co/spaces/gryan-galario/manga-ocr-demo>
<https://huggingface.co/spaces/Detomo/Japanese-OCR>

* “Manga OCR” (Local)

Expand Down Expand Up @@ -420,10 +420,10 @@ system resources.
#### Setting up “Manga OCR” Online

> **Warning**
> The “Manga OCR” Online engine depends on a third-party online
> service ([a Hugging Face Space by Gryan Galario][manga-ocr-hf-gg]), so using
> it involves sending screenshots of portions of your screen to a third-party.
> Here is [the stated privacy policy of Hugging Face][huggingface-privacy-policy].
> The “Manga OCR” Online engine depends on a third-party online service
> ([a Hugging Face Space by Detomo][manga-ocr-hf]), so using it involves sending
> screenshots of portions of your screen to a third-party. Here is
> [the stated privacy policy of Hugging Face][huggingface-privacy-policy].
The online API used by the “Manga OCR” Online engine is freely accessible and
consequently *does not* require any setup.
Expand Down Expand Up @@ -1864,8 +1864,7 @@ Kamite never saves your data to disk.
Kamite never sends your data through the network, with the following exceptions:
* When `ocr.engine` is set to `mangaocr_online`, screenshots of portions of your
screen are sent to a [Hugging Face Space][manga-ocr-hf-gg] for text
recognition.
screen are sent to a [Hugging Face Space][manga-ocr-hf] for text recognition.
* When `ocr.engine` is set to `ocrspace`, screenshots of portions of your screen
are sent to [OCR.space] for text recognition.
Expand Down Expand Up @@ -1941,5 +1940,5 @@ the original license notices.
[Gomics-v]: https://github.com/fauu/gomicsv
[Sway]: https://swaywm.org/
[manga-ocr]: https://github.com/kha-white/manga-ocr
[manga-ocr-hf-gg]: https://huggingface.co/spaces/gryan-galario/manga-ocr-demo
[manga-ocr-hf]: https://huggingface.co/spaces/Detomo/Japanese-OCR
[OCR.space]: https://ocr.space/

0 comments on commit 748d517

Please sign in to comment.