Plato: A document reader for PDF, ePUB, DjVu, etc.

[baskerville]

Originally Posted by sherman

Just tried this for the first time today. Can't give much feedback due to the afore mentioned touch issues,

Thanks for trying, I won't mind receiving feedback on 0.5.2.

Originally Posted by sherman

I thought that K&P line breaking tries to avoid hyphenation where possible, and especially tries to avoid subsequent hyphenation. And I've just seen three subsequent line breaks ending in a hyphen.

I noticed this when I was testing the renderer on various ePUBs and I misinterpreted the source of the problem. Indeed the value used for the consecutive-hyphens demerits was incorrect (100 instead of 3000)!

[baskerville]

Originally Posted by Symptym

It does seem to have broken the ability to select chapters in books using the menu from the bottom left however and every time I try to choose the chapter in my book again I get sent to the beginning of the last chapter.

This problem is present in the built-in ePUB renderer since its introduction.

Hence you should be able to reproduce with 0.5.2.

[Symptym]

I was able to.

Seems that up to now I've somehow managed never needing to move between chapters.

A question about the adjusted metadata format;

Is there any way to convert the previous format to the new one?

ie. carry over the position in book and any bookmarks?

[baskerville]

Originally Posted by Symptym

A question about the adjusted metadata format;
Is there any way to convert the previous format to the new one?

Yes, there is. I will provide a script to do that when I release 0.5.4.

[sherman]

So, I (hopefully) compiled the latest build, and set the paragraph breaker version to 0.4.0 in the Cargo.toml file before compiling.

First, interface is nice and clean, fast. Booklist is a good start. No series information, or book descriptions, which is a shame. Hoping more metadata in general could be added in the future.

The epub renderer appears to have less chance of subsequent hyphens, but it does still occur. I'm looking at a particularly egregious example right now, which is four consecutive hyphen line endings! Definitely still needs some fine tuning in my opinion.

Automatic ligatures would be a nice feature to include one day, as Plato doesn't appear to do this.

I'm not a fan of the right corner triggering the go-to-page box. It's very easy to hit accidentally. I haven't done any tweaking or configuration beyond default though, so I'm talking about default settings here.

Plato is certainly looking promising. I'm not ready to switch to it yet however, as the current RMSDK renderer is still superior at the moment (for epub). If the renderer can be further refined, and missing features added (I have plenty of books in my library that use tables and/or floats), I would be a very happy person.

[baskerville]

Originally Posted by sherman

Booklist is a good start. No series information, or book descriptions, which is a shame. Hoping more metadata in general could be added in the future.

I might need a little help here: the book description comes from dc:description as far as I know, but how are the series and series index expressed in epub2, epub3 and kepub? I'm only aware of the non-standard meta name="calibre:series{,_index}".

Originally Posted by sherman

The epub renderer appears to have less chance of subsequent hyphens, but it does still occur. I'm looking at a particularly egregious example right now, which is four consecutive hyphen line endings!

I'll look into that. Please note that K&P isn't the only algorithm used.

Originally Posted by sherman

Automatic ligatures would be a nice feature to include one day, as Plato doesn't appear to do this.

Ligatures should be there: the liga OpenType feature is enabled by default in HarfBuzz.

Originally Posted by sherman

I'm not a fan of the right corner triggering the go-to-page box. It's very easy to hit accidentally. I haven't done any tweaking or configuration beyond default though, so I'm talking about default settings here.

I've also been bitten by this, I've therefore shrunk the corner regions in 19caa61.

[davidfor]

Originally Posted by baskerville

I might need a little help here: the book description comes from dc:description as far as I know, but how are the series and series index expressed in epub2, epub3 and kepub? I'm only aware of the non-standard meta name="calibre:series{,_index}".

The DC metadata standard doesn't include series info. There is something in the epub 3 standard. Otherwise, calibre series info is the closest to a standard.

[Namenlos]

Originally Posted by baskerville

mupdf's ePUB engine has several problems[…]I'm therefore in the process of writing an HTML renderer![…]

Have you considered using Webkit (or another html engine) and the Readium SDK? It seems promising, is open source, well documented and backed by companies (eink hardware and french libraries).

[baskerville]

I've released 0.5.4.

The metadata format has changed and manual intervention is required. It might be a good practice to always backup your settings and metadata before upgrading. The attached Python script will perform the conversion:

./fix_metadata.py < .metadata.json > metadata.json
mv metadata.json .metadata.json

Please note that the script assumes that you haven't changed the value of reader.engine in Settings.toml.

List of changes:

• Fix the reported rotation issue.
• Fix chapter navigation in the built-in ePUB renderer.
• Reduce the area of the corner regions to avoid unintentional activations.
• Preload the next and previous pages.
• Load the hyphenation patterns at runtime: you might want to remove the unneeded patterns in the hyphenation-patterns directory.
• Limit the number of consecutive hyphens that the paragraph-breaker library might generate.
• Decrease the size of the book title font.
• Import more metadata from ePUBs.
• Add a setting to trigger the auto import process at startup: import.startup-trigger.
• Add a setting to traverse hidden directories during import: import.traverse-hidden.
• Add a setting to set custom images for the suspend, power off and share intermissions: intermission-images. The setting can be set through the contextual menu of a book strip, in the Set As submenu.

[jcn363]

How can I make my own hyphenate file?
Is there a parameter to control the number of consecutive hyphens that the paragraph-breaker library might generate?
Good luck.

[bn_el]

Thank you !

[baskerville]

Originally Posted by Namenlos

Have you considered using Webkit (or another html engine) and the Readium SDK?

I wrote an ePUB renderer from scratch.

[baskerville]

Originally Posted by jcn363

How can I make my own hyphenate file?

The hyphenation patterns are generated by the hyphenation crate. Within the hyphenation repository, you can build the patterns with:

cargo build --features build_dictionaries

Originally Posted by jcn363

Is there a parameter to control the number of consecutive hyphens that the paragraph-breaker library might generate?

The K&P algorithm doesn't provide such a parameter: it only returns the best solution satisfying the given constraints on the shrinking and stretching of the word spaces. I can only depreciate the subset of solutions that contains consecutive hyphens. But if all the solutions have consecutive hyphens then it won't help. The standard fit algorithm (used when K&P fails) forbids more that two consecutive hyphens.

[sherman]

Originally Posted by baskerville

The hyphenation patterns are generated by the hyphenation crate. Within the hyphenation repository, you can build the patterns with:

cargo build --features build_dictionaries

The K&P algorithm doesn't provide such a parameter: it only returns the best solution satisfying the given constraints on the shrinking and stretching of the word spaces. I can only depreciate the subset of solutions that contains consecutive hyphens. But if all the solutions have consecutive hyphens then it won't help. The standard fit algorithm (used when K&P fails) forbids more that two consecutive hyphens.

The other area to tweak hyphenation is the actual hyphenation rules. For example, the hyphenation dictionary supplied with Kobo ereaders (and probably most other RMSDK readers) sets a minimum of five characters pre and post hyphen (so minimum ten character word). My personal tweak to this is to set it to three pre and post (so minimum six character word).

[baskerville]

Originally Posted by sherman

The other area to tweak hyphenation is the actual hyphenation rules. For example, the hyphenation dictionary supplied with Kobo ereaders (and probably most other RMSDK readers) sets a minimum of five characters pre and post hyphen (so minimum ten character word). My personal tweak to this is to set it to three pre and post (so minimum six character word).

Those parameters are hardcoded (I guess I could fork the library). The current value for en-US is (2, 3).