EBook

A little app I threw together today is a (very VERY) simple reader for EPUB ebooks.

It works by stripping out the text from the EPUB structure (if it can figure out the filenames, sometimes Calibre creates ridiculous names), and then showing you the result. It doesn’t currently try to parse the formatting.

Make sure SparkFS is running. Then run !EBook.

Drag an EPUB file to the EBook icon (doesn’t matter what the filetype is, it’ll set it to &DDC (Zip) to see if SparkFS can open it; and when done it’ll set it to &AAE as this seems to be what MuView expects EPUB files to be).

The file, if it passes basic checks, will be stripped and combined according to the details in the manifest:

When it is done, a big scrolling window will open containing the text of the entire book. I plan on making something better, but it works for now:

(you can pick from a variety of colour themes, the one shown is PseudoSepia)

Not bad for something I “just decided to have a crack at writing at half two this afternoon”. :-)

hxxp://heyrick.ddns.net/files/ebook001.zip

it’ll set it to &DDC (Zip) to see if SparkFS can open it

Yeah don’t do that Rick. SparkFS will happily allow access to any ZIP format file without changing its type if you use the SparkFS prefix – take your full pathname, change all the colons into #, and then use the filename “SparkFS#yourfilename_with_colons_replaced:$”. ie if your filename is adfs::$.myfile then you can navigate around inside it as SparkFS#adfs##$.myfile:$ – is that not documented somewhere?

So you can detect if it’s a zip by using XOS_File,5 on the resulting filename – if you get an error, it isn’t, otherwise it’ll be a directory.

is that not documented somewhere?

Page 17 of the SparkFS manual.

:D Thanks for the plug, Rick – I think I’ve seen that text somewhere before… :D

Page 17 of the SparkFS manual.

You read a manual? :)

You read a manual? :)

Well, I do have a lot of free time these days…

Stupid thought, but would it make sense to output (or export) to (very simple) HTML so that some formatting could be retained (alignment, bold, italic) and possibly any embedded pictures (if ebooks have such things) could be linked via tags?

And then you can just use the built-in HTML renderer that any decent OS…I’ll get my coat.

Horrible evil thought: Convert it into a Draw Text Area and… (really, really, don’t do that. Uggh)

hxxp://heyrick.ddns.net/files/ebook002.zip

Loads of enhancements today:

• Can now access archives directly without messing with the filetypes (thanks nemo!).
• Sets the reader title to the filename only.
• Renders using Corpus – much better looking. ;-)
  Bumped the text size up to 14pt, makes for easier reading.
• Can now be switched to double-spaced.
• Improved splitter parser to cope with more documents, such as Cory Doctrow’s “Little Brother”.
• Now has basic support for bold and italics (if specified directly, not as a style) and also renders headings in bold.
• Now converts &lt;, &gt;, and &amp; to <, >, and & respectively.
• If you have DADebug running, unhandled entities and UTF-8 will be logged.
• Now supports running a new instance if a file typed &AAE is double clicked upon [filetype is the one used by MuView].

Here’s a picture:

Some books:

• Clive’s “Exile” – was available for free earlier in the year, if you picked up a copy, it’ll work in !EBook. ;-)

• Little Brother – https://craphound.com/littlebrother/download/
  EPUB is under “Fan conversions”, and I’m suprised they haven’t made a movie out of this yet. Maybe it’s a little too close to potential truth?

• Homeland (sequel to the above) – https://craphound.com/homeland/download/

• Pride and Prejudice (and NO zombies) – http://www.gutenberg.org/ebooks/1342

• Alice, you know which one – http://www.gutenberg.org/ebooks/11

• Kafka’s epic pile of mindscrew – http://www.gutenberg.org/ebooks/5200

• Many things of which I cannot speak – http://www.gutenberg.org/ebooks/25525

• An autistic spacecraft pilot – http://heyrick.ddns.net/files/no-explanations.epub
  Hosted on my private server because the original costs $0,99 which is not only a really annoying price for those who don’t use USD (it’ll probably cost me twice as much again just in conversion/handling charges); but more importantly the one I downloaded two years ago has a very clear Creative Commons licence in the front few pages, which permits redistribution.

• And so on…

Yeah don’t do that Rick.

:-) Thanks for the suggestions. Works nicely, so no need to fritz with the filetypes.

Stupid thought, but would it make sense to output (or export) to (very simple) HTML

EPUB is actually a restricted form of XHTML/CSS (most implementations do not tolerate markup errors, that’s why tools like Calibre are important).

And then you can just use the built-in HTML renderer that any decent OS

Sure, if that HTML renderer can cope with rendering page by page, with user-defined text size, spacing, margins etc – and not as a big scrolling window of text.

The underlying technology is very similar, the method of presentation differs somewhat.

Convert it into a Draw Text Area and… (really, really, don’t do that. Uggh)

I ought to add that as an Easter Egg, and call it “nemo mode”. ;-)

I get that the presentation differs. I was just thinking for RISC OS, since your program just concatenates things together, a similar (but more elaborate, for hopefully not much work) could be achieved by feeding the HTML to NetSurf. Note, I’m not suggesting that this would suffice as an ebook reader, but rather that it’d give a prettier view of the content in line with what your program was doing.

Then you went and added nicer fonts, bold, italic etc anyway, so my idea is probably somewhat redundant, unless there are pictures in ebooks.

Clive’s “Exile” – was available for free earlier in the year,

All Clive’s books, including Birgom’s Diary which isn’t otherwise available yet, are free to friends – and I regard all of you as friends.

http://clive.semmens.org.uk/000-books_r3U.html – that’s a thrown-together page, the links to the books just download a pdf BANG to wherever your downloads go with no warning or anything. But that’s pdfs of the printed version of the books. I don’t have the ebooks, and they don’t exist for some of the titles.

I ought to add that as an Easter Egg, and call it “nemo mode”. ;-)

Is “nemo mode” a bad omen?

No, an edom omen.

That’s roughly what I had in dnim, sey.

‘demo omen’, considering the number of unreleased pearls.

hxxp://heyrick.ddns.net/files/ebook002b.zip

Some more enhancements to finish the day:

• Will try a more intelligent way of starting SparkFS.
• Opening a new book will reposition the scroll to the top of the window.
• Remembers last read position of books that you are reading.
  [note that the data is held in memory, so using multiple copies of !EBook at the same time may cause stale data to overwrite fresh]
• Various parser tweaks to improve output.

‘demo omen’, considering the number of unreleased pearls.

O demon me.

ebook002b seems broken… I drop no-explanations/epub onto the icon and it opens the smallest possible RO window (all the border window buttons) and does nothing else. Sparkfs is running. ebook002 works with the same epub file.

RO 5.24, ARMv7 Pi2, sparkfs 1.45

Thanks for the report. I’ll take a look later on.

Do other books work? Is it only No Explanations?

http://heyrick.ddns.net/files/ebook003.zip

Weird.

So I added a function to look at the input filename, scan backwards from the end looking for a ‘.’ (to get the leafname), and to then copy that to a buffer following the word "Examining ".
This is used in the status window, in place of saying "Examining " followed by the entire path to the file.

Only it didn’t work.
Or, more specifically, it didn’t work for No Explanations. It did work for everything else I tried.

How it failed? Well, it failed in that reading the manifest worked, reading the number of objects to process worked. But it completely skipped over the loop to do the actual reading of the objects.
But only for that one book.

It sounds to me like a stray memory access mucking with stuff, but absolutely nothing stood out. The filename isn’t touched. The counters are correct. The loop counter isn’t relevant as that’s set on entry to the loop…

My ‘fix’ was to create a local string, copy the filename global into that, and then poke around the local copy, not the global one. And, behold! It works.
One day when I have plenty of time and nothing better to do, I’ll try both methods and get the compiler to spit out some assembly. Maybe seeing what’s going on might shed some light? Or, god help me, use DDT. :-/
I cannot use my normal method (DADebug), as inserting a few lines of debug code made it magically work again. Maybe it’s a compiler bug or something? I dunno… I made a fix, it works, that’s all I’m interested in right now.

Tested with:

• The Drowned World, J G Ballard
• Little Brother, Cory Doctrow
• Homeland, Cory Doctrow
• Exile, our very own Clive Semmens
• Alpha, Jus Accardo
• Belinda7, nobody important
• The Snow Spider, Jenny Nimmo
• After the Quake, Haruki Murakami
• Your Name, Shinkai Makoto (lovely film, as expected given the creator!)

Maybe it’s a compiler bug

There ought to be a word for the period of time between that hackneyed claim and the “Ohhhhhhhhh” that will at some point follow. You don’t notice the bugs until they bite.

Here’s an interesting thing I found the other day: Every copy of VirtualAcorn/VirtualRPC that has ever shipped contains a HostFS of course, and its HostFS has an unfortunate little bug: If you dismount a drive all your other drives get unshared. If you don’t use HostFS or Access or never dismount a drive then you wouldn’t notice. You’d correctly guess that it’s emitting a Service_DiscDismounted with too short a filename, and it is – “HostFS::”, but the why is interesting: It does try to append the disc name, but in between it has done a Service_NetFS for dubious reasons, and it treats that service number as the name pointer… which means it actually appends whatever it finds at address &00000055 – normally a zero. This is fortunate, because if it does manage to copy anything as intended it stiffs the machine completely.

Moral: “Working” is not proof of correctness.

There are several, but they’re not repeatable.

Hi Rick,

Firstly thanks for yet another useful application.

I have tried it with one epub document I have and it works but I have tried it with others from Gutenberg, using the epub without images option, and I get issues with errors like “File is either not EPUB or damaged” or “can’t read manifest/ too many manifests”.

This is using some of the example files you have highlighted.

This is with SparkFS 1.45, RISC OS 5.35 (24th Aug 18) on an ARMX6.

RISC OS 5.35 (24th Aug 18)

Really?!! Do you know something we don’t?