Again, span­ish on­ly be­cause it's a video... in span­ish.

Re­sul­ta que me olvidé que sí habían graba­do mi char­la de do­cu­tils y com­pañi­a. Gra­cias a Ger­mán por hac­erme acor­dar y mostrarme adonde es­taba!

Y ... acá es­tá:

Fi­nal­ly, a new re­lease of rst2pdf!

You can get it at its site: http://rst2pdf.­google­code.­com

rst2pdf is a tool to con­vert re­struc­tured text (a light, cool markup lan­guage) to PDF us­ing re­port­lab in­stead of La­TeX.

It has been used for many things, from book­s, to mag­a­zi­nes, to brochures, to man­u­al­s, to web­sites and has lots of fea­tures:

• Font em­bed­d­ing (T­TF or Type1 fonts)

• Cas­­cad­ing Stylesheets

• Ex­treme­­ly flex­i­ble plug­in ar­chi­tec­­ture (y­ou can do things like ren­der the head­­ings from ar­bi­­trary SVG files!)

• Sphinx in­­te­­gra­­tion.

• Con­­fig­urable page lay­outs

• Cus­­tom cov­­er pages via tem­­plates

• And much, much more...

The big­gest change in 0.16 is sure­ly the im­proved sup­port for Sphinx 1.0.x so if you are us­ing Sphinx, you re­al­ly want this ver­sion.

Al­so, it has a ton of bug­fix­es, and a few mi­nor but use­ful new fea­tures.

Here's the whole changel­og if you don't be­lieve me:

• Fixed Is­­sue 343: Plugged mem­o­ry leak in the RSON pars­er.

• Fix for Is­­sue 287: there is still a cor­n­er case if you have two sec­­tions with the same ti­tle, at the same lev­­el, in the same page, in dif­fer­­ent files where the links will break.

• Fixed Is­­sue 367: ger­­man-lo­­cal­ized dates are MM. DD. YYYY so when used in sphinx's tem­­plate cov­­er they ap­­peared weird, like a list item. Fixed with a mi­nor work­around in the tem­­plate.

• Fixed Is­­sue 366: links to "#" make no sense on a PDF file

• Made de­f­i­ni­­tions from de­f­i­ni­­tion lists more sty­lable.

• Moved de­f­i­ni­­tion lists to Split­­Ta­bles, so you can have very long de­f­i­ni­­tion­s.

• Fixed Is­­sue 318: Im­­ple­­men­t­ed Do­­main spe­­cif­ic in­­dex­es for Sphinx 1.0.x

• Fixed In­­dex links when us­ing Sphinx/pdf­builder.

• Fixed Is­­sue 360: Set lit­er­al.­­word­Wrap to None by de­­fault so it does­n't in­­her­it word­Wrap CJK when you use the oth­­er­­wise cor­rect ja­­pa­­nese set­t­ings. In any case, lit­er­al blocks are not sup­­posed to wrap at al­l.

• Switched pdf­builder to use Split­­Ta­bles by de­­fault (it made no sense not to do it)

• Fixed Is­­sue 365: some TTF fonts don't val­i­­date but they work any­way.

• Set a valid de­­fault baseurl for Sphinx (makes it much faster!)

• New fea­­ture: --use-num­bered-links to show sec­­tion num­bers in links to sec­­tion­s, like "See sec­­tion 2.3 Ter­mi­­na­­tion"

• Added stylesheets for land­s­cape pa­per sizes (i.e: a4-­­land­s­cape.style)

• Fixed Is­­sue 364: Some op­­tions not re­spec­t­ed when passed in per-­­doc op­­tions in sphinx.

• Fixed Is­­sue 361: mul­ti­­ple line­breaks in line blocks were col­lapsed.

• Fixed Is­­sue 363: strange char­ac­ters in some cas­es in math di­rec­­tive.

• Fixed Is­­sue 362: Smarter au­­to-en­­clos­ing of equa­­tions in $...$

• Fixed Is­­sue 358: --re­al--­­foot­notes de­­faults to False, but help text in­­di­­cates de­­fault is True

• Fixed Is­­sue 359: Wrong --­­fit-back­­­ground-­­mode help string

• Fixed Is­­sue 356: mis­s­ing cells if a cell spawns rows and col­umn­s.

• Fixed Is­­sue 349: Work cor­rec­t­­ly with lan­guages that are avail­able in form aa_bb and not aa (ex­am­­ple: zh_c­n)

• Fixed Is­­sue 345: give file/­­line in­­­fo when there is an er­ror in a raw PDF di­rec­­tive.

• Fixed Is­­sue 336: JPEG im­ages should work even with­­out PIL (but give a warn­ing be­­cause sizes will prob­a­bly be wrong)

• Fixed Is­­sue 351: foot­note/c­i­­ta­­tion re­f­er­ences were gen­er­at­ed in­­­cor­rec­t­­ly, which caused prob­lems if there was a ci­­ta­­tion with the same text as a head­­ing.

• Fixed Is­­sue 353: bet­ter han­dling of graphviz, so that it works with­­out vec­­tor­pdf but gives a warn­ing about it.

• Fixed Is­­sue 354: make todo_n­ode from sphinx cus­­tom­iz­a­ble.

• Fixed bug where nest­ed lists broke page lay­out if the page was smal­l.

• Smarter --in­­line-links op­­tion

• New ex­ten­­sion: fan­­cyti­tles, see //ral­si­­na.me/we­blog/­­post­s/B­B906.html

• New fea­­ture: tab-width op­­tion in code-block di­rec­­tive (de­­faults to 8).

• Fixed Is­­sue 340: end­notes/­­foot­notes were not styled.

• Fixed Is­­sue 339: class names us­ing _ were not us­able.

• Fixed Is­­sue 335: ug­­ly crash when us­ing im­ages in some spe­­cif­ic places (looks like a re­­port­lab bug)

• Fixed Is­­sue 329: make the fig­ure align­­men­t/­­class at­tributes work more like La­­TeX than HT­M­L.

• Fixed Is­­sue 328: list item styles were be­ing ig­nored.

• Fixed Is­­sue 186: new --use-float­ing-im­ages makes im­ages with :align: set work like in HT­M­L, with the next flow­able flow­ing be­­side it.

• Fixed Is­­sue 307: head­­er/­­foot­er from stylesheet now sup­­ports in­­­line rest markup and sub­­sti­­tu­­tions de­fined in the main doc­u­­men­t.

• New pdf_­­toc_depth op­­tion for Sphinx/pdf­builder

• New pdf_use_­­toc op­­tion for Sphinx/pdf­builder

• Fixed Is­­sue 308: com­­pat­i­­bil­i­­ty with re­­port­lab from SVN

• Fixed Is­­sue 323: er­rors in the con­­fig.sam­­ple made it work weird.

• Fixed Is­­sue 322: Im­age sub­­sti­­tu­­tions did­n't work in doc­u­­ment ti­­tle.

• Im­­ple­­men­t­ed Is­­sue 321: un­der­­line and strikethrough avail­able in stylesheet.

• Fixed Is­­sue 317: Ug­­ly er­ror mes­sage when file does not ex­ist

That a plug­in ar­chi­tec­ture for a com­plex app is a good idea is one of those things that most peo­ple kin­da agree on. One thing we don't quite agree is how the heck are we go­ing to make out app mod­u­lar?

One way to do it (if you are cod­ing python) is us­ing Yap­sy.

Yap­sy is awe­some. Al­so, yap­sy is a bit un­der­doc­u­ment­ed. Let's see if this post fix­es that a bit and leaves just the awe­some.

Up­date: I had not seen the new Yap­sy doc­s, re­leased a few days ago. They are much bet­ter than what was there be­fore :-)

Here's the gen­er­al idea be­hind yap­sy:

• You cre­ate a Plug­in Man­ag­er that can find and load plug­ins from a list of places (for ex­am­­ple, from ["/us­r/share/ap­p­­name/­­plu­g­in­s", "~/.ap­p­­name/­­plu­g­in­s"]).

• A plug­in cat­e­­go­ry is a class.

• There is a map­ping be­tween cat­e­­go­ry names and cat­e­­go­ry class­es.

• A plug­in is a mod­­ule and a meta­­da­­ta file. The mod­­ule de­fines a class that in­­her­its from a cat­e­­go­ry class, and be­­longs to that cat­e­­go­ry.

  The meta­­da­­ta file has stuff like the plug­in's name, de­scrip­­tion, URL, ver­­sion, etc.

One of the great things about Yap­sy is that it does­n't spec­i­fy too much. A plug­in will be just a python ob­jec­t, you can put what­ev­er you want there, or you can nar­row it down by spec­i­fy­ing the cat­e­go­ry class.

In fac­t, the way I have been do­ing the cat­e­go­ry class­es is:

• Start with an em­p­­ty class

• Im­­ple­­ment two plug­ins of that cat­e­­go­ry

• If there is a chunk that's much alike in both, move it in­­­to the cat­e­­go­ry class.

But trust me, this will all be clear­er with an ex­am­ple :-)

I will be do­ing it with a graph­i­cal PyQt ap­p, but Yap­sy works just as well for head­less of CLI app­s.

Let's start with a sim­ple ap­p: an HTML ed­i­tor with a pre­view wid­get.

Here's the code for the ap­p, which is re­al­ly sim­ple (it does­n't save or do any­thing, re­al­ly, it's just an ex­am­ple):

But this ap­pli­ca­tion has an ob­vi­ous lim­it: you have to type HTML in it. Why not type python code in it and have it con­vert to HTML for dis­play? Or Wi­ki markup, or re­struc­tured tex­t?

You could, in prin­ci­ple, just im­ple­ment all those mod­es, but then you are as­sum­ing the re­spon­s­abil­i­ty of sup­port­ing ev­ery thing-that-­can-be-­turned-in­to-HTM­L. Your app would be a mono­lith. That's where yap­sy en­ters the scene.

So, let's cre­ate a plug­in cat­e­go­ry, called "For­mat­ter" which takes plain text and re­turns HTM­L. Then we add stuff in the UI so the us­er can choose what for­mat­ter he wants, and im­ple­ment two of those.

Here's our plug­in cat­e­go­ry class:

Of course what good is a plug­in ar­chi­tec­ture with­out any plug­ins for it? So, let's cre­ate two plug­ins.

First: a plug­in that takes python code and re­turns HTM­L, thanks to pyg­ments.

See how it goes in­to a plug­ins fold­er? Lat­er on we will tell yap­sy to search there for plug­ins.

To be rec­og­nized as a plug­in, it needs a meta­da­ta file, too:

And re­al­ly, that's all there is to mak­ing a plug­in. Here's an­oth­er one for com­par­ison, which us­es do­cu­tils to for­mat re­Struc­tured Tex­t:

And here they are in ac­tion:

Of course us­ing cat­e­gories you can do things like a "Tool­s" cat­e­go­ry, where the plug­ins get added to a Tools menu, too.

And here's the ap­pli­ca­tion code:

In short: this is easy to do, and it leads to fix­ing your ap­pli­ca­tion's in­ter­nal struc­ture, so it helps you write bet­ter code.

Full source code for ev­ery­thing.

One of my favourite things about Aran­du­ka as a project is that it's an end­less source of smal­l, lim­it­ed side project­s.

For ex­am­ple, Aran­du­ka is now close to be­ing able to sync my book col­lec­tion to my phone. But... what if what I want to read on the train is not a book but, say, a blog?

Well, blogs pro­vide their con­tent via a feed. And A feed is a col­lec­tion of HTML pieces glued in­to a struc­ture plus some da­ta like au­thor and such.

And there's a great mod­ule for pars­ing them, called feed­pars­er. And I have writ­ten not one, not two, not three, but four RSS ag­gre­ga­tors in the past.

So, how about con­vert­ing the feed in­to some­thing my phone can han­dle? [#] Would it be hard to do?

Well... not re­al­ly hard. It's most­ly a mat­ter of tak­ing a smal­l, sam­ple ePub doc­u­ment (cre­at­ed by Cal­i­bre) writ­ing a few tem­plates, feed­ing it the da­ta from feed­pars­er and zip­ping it up.

For ex­am­ple, this is this blog, as an ePub and here's FBRead­er read­ing it:

As usu­al, the code is open, and it's here in aran­duka's mer­cu­ri­al.

It's not re­al­ly in­ter­est­ing code, and re­quires tem­plite feed­pars­er and who knows what else.

The pro­duced ePub does­n't val­i­date, and it prob­a­bly nev­er will, be­cause it has chunks of the orig­i­nal feed in it, so stan­dard com­pli­ance does­n't de­pend on rss2epub.

Al­so, you get no im­ages. That would im­ply pars­ing and fix­ing all img el­e­ments, I sup­pose, and I am not go­ing to do it right now.

[#] I first saw this fea­ture in pluck­er a long time ago, and I know Cal­i­bre has it too.

Yes, it's yet an­oth­er pro­gram I am work­ing on. But hey, the last few I start­ed are ac­tu­al­ly pret­ty func­tion­al al­ready!

And... I am not do­ing this one alone, which should make it more fun.

It's an eBook (or just any book?) man­ager, that helps you keep your PDF/­Mo­bi/F­B2/what­ev­er or­ga­nized, and should even­tu­al­ly sync them to the de­vice you want to use to read them.

What works now? See the video!

In case that makes no sense to you:

• You can get books from Feed­­Book­s. Those books will get down­load­­ed, added to your database, tagged, the cov­­er fetched, etc. etc.

• You can im­­port your cur­rent fold­er of books in bulk.

  Aran­­du­­ka will use google and oth­­er sources to try to guess (from the file­­name) what book that is and fill in the ex­­tra da­­ta about it.

• You can "guess" the ex­­tra da­­ta.

  By mark­ing cer­­tain da­­ta (say, the ti­tle) as re­li­able, Aran­­du­­ka will try to find some pos­si­ble books that match then you can choose if it's right.

  Of course you can al­­so ed­it that da­­ta man­u­al­­ly.

And that's about it. Planned fea­tures:

• Way too many to list.

The goals are clear:

• It should be beau­ti­­ful (I know it is­n't!)

• It should be pow­er­­ful (not yet!)

• It should be bet­ter than the "com­pe­ti­­tion"

If those three goals are not achieved, it's fail­ure. It may be a fun fail­ure, but it would still be a fail­ure.