nick.recoil.org

I’ve pushed my beta code for mythRecommend to Github. It’s a system which allows you to subscribe to a source of regular TV recommendations, and have those recommendations automatically get translated into concrete recording schedule for a MythTV system.

Once a recording has been made, the recording schedule is made dormant, so it will not clutter your existing schedule list as time goes on. You can decide if you would like to turn that recommended program into a permanent schedule or not, and reinstate it if desired.

I hope to add more feed parsers as the code matures, but as things stand there is only one written, specifically for The Guardian here in the UK.

At the end of June, I gave a short talk to the London Ruby Users Group on some code I’ve been working on to enable me to interface with MythTV on a programatic level. MythTV is Open Source software which allows the recording and playback of TV on a Linux box, turning it into a PVR. I use this at home to record TV from Freeview.

MythTV has a web interface written for it in the form of MythWeb, and it allows decent remote interaction with the system. It is also my main mode of interaction when trawling for interesting programmes to record. It’s much more convenient to use a keyboard and mouse to navigate, rather than a remote control. This is especially true here in the UK, where there are many Freeview channels that could contain content of interest.

Creating a pleasant API

However, if you’re looking to play around with integration, and prototype some ideas you have to help you navigate and interact in different ways, you’re not going to have an easy ride. There’s no single clean implementation of the functionality you can access in any one language. PHP drives the main MythWeb code, but it also mixes in some Perl to access some fairly major functionality (downloading/streaming the recorded programmes). There are also Python bindings within the MythTV codebase, but I’m not aware of anything substantial built on top of them, and what is implemented is mainly concerned with metadata.

The main functions I wanted a MythTV integration library to offer were the ability to list what had been recorded, obtain a thumbnail for that recording, and to be able to stream the data to you. This was achieved in the 0.1.0 release. Now we’re up to version 0.2.0, it now supports listing of EPG data, and creation/editing of recording schedules. This paves the way for experiments in automatic scheduling of recordings. There are many potentially interesting sources of TV reviews to scrape. For instance, if you like what The Guardian has to say about Television in its Watch this section, then extracting the titles from that page, and feeding them to ruby-mythtv is easy.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27

  require 'ruby-mythtv'
 
  # Connect to the server
  mythbackend, mythdb = MythTV.connect(:host => 'mythtv.localdomain',
                                       :database_password => 'password')
 
  # Find matches on our search term, and limit the results to 5 matches
  programs = mythdb.list_programs(:conditions => ['title LIKE ?', "%Bruce Parry%"],
                                  :limit => 5)
 
  # Take the first program match, and convert it to a recording schedule
  new_schedule = MythTV::RecordingSchedule.new(programs[0], mythdb)
  new_schedule.save
 
  # Signal the backend of recording changes for our recording schedule entry
  mythbackend.reschedule_recordings(new_schedule.recordid)
 
  # Let the backend resolve matches
  sleep(1)
 
  # Enumerate the list of pending recordings, find ours, and check for any conflicts
  pending_recordings = mythbackend.query_pending
  conflicts = pending_recordings.find { |p| p.recordid == new_schedule.recordid &&
                                            p.recstatus_sym == :rsConflict }
 
  # If conflicts is empty, then all is good. If it is populated, then action needs
  # to be taken, such as bumping the priority, or removing the clashes....

Ruby-mythtv is present both on Rubyforge and Github. Major release versions will be present on Rubyforge, and development versions will be present on Github. This method allows me to push out code which I consider functional, but not necessarily finalised. Release early, release often, as the saying goes, and this is all the more relevant when forking projects and sharing ideas is central to the way Github works. There’s nothing worse than code which is “almost there”, languishing in your “I really should finish it” folder, destined never to see the light of day.

A chap at Smoke and Mirrors, where I’m doing some Freelance work, is creating a system to bridge between video routers and a web-based status page. Most of the suites within the building are plumbed in via fibre-optic KVM connections, and depending on the schedule, different video processing systems can get presented in different rooms at different times. All it takes is a change on the routers.

Although complex, this enables the most cost-effective use of each system, marrying the functionality needed to the number of physical people required to attend the session for any given day. There is a large variation in price and functionality with certain high-end systems, so you always want to try and get maximal use from the systems you have chosen to invest in.

He’s looking at using one of the boards from the wonderfully named Acme Systems from Rome. The idea is to communicate with the video router via its RS-232 serial interface, and retrieve an ASCII representation of the current router configuration. It’s a fascinating area of software AND hardware as glue. A physical device which will run some bespoke code to bridge information from one system into another.

In related news, I’ve also received my Artigo Pico-ITX kit and am now a proud owner of the tiniest system running Linux that I’ve ever seen. It also seems to play happily with the DVB sticks that I had stability issues with on the older Mini-ITX M10k board.

The fan is noticable in a very quiet room, but as soon as there’s anything else to be heard, it’s drowned out, so it’s not quite suitable for your bedroom. They do a fanless system, but it’s half the clock speed. Installing it was done over PXE from my Mac, using the Ubuntu netboot downloads.

I came across a curious issue with EPG data the other weekend. Here’s a screenshot of what was rendered by the TV:

0×92 is the ANSI hex value for a right single quotation mark. In HTML entity terms, it is ’ and should look like ’. Unfortunately, EPG data in the UK is encoded as ISO-8859-1, and does not contain an equivalent character, so the nearest is a simple apostrophe, ‘, ASCII 0×27. The system behind converting source EPG data into 8859-1 obviously suffers from the occasional translation error.

More on quotation marks and character sets can be found in Markus Kuhn’s ASCII and Unicode quotation marks page, should readers be so inclined.

Dominic and I are busy preparing for our little on-stage session at Open Tech 2005 where we’re talking at 4.30 about our fridge-sized PVR. Not actually fridge-sized physically, but certainly in terms of data capacity.

We’re also working on an accompanying website and possibly a mailing list to go with the system we’ll be demonstrating on stage.