XBMC SageTv Plugin

[mikwilli]

Brent,
For me I have a several reasons. Probably the biggest is that I enjoy coding and this is a challenge. Second is that I have SageTV on a computer connected to a TV, but the other TVs in my house which are not directly connected to it have Xboxes. While its true you can watch any of the recorded shows easily from the Xboxes you don't get the episode information, video format (So you know if its possible to watch through Xbox), etc. Also with this I'll have the ability to browse through the TV listings, set up recordings, etc.

Also since my other TVs are all STD Def using my existing xbox hardware is much cheaper than a media extender or small computer attached to them.

Hope this helps.

[Brent]

Originally Posted by mikwilli

Hope this helps.

It does and it makes sense too. Thanks!

[kricker]

Originally Posted by Brent

Ok. Sorry to pollute this thread buy I have to ask you guys in the know.

Why would you go through this trouble? What do you like with XBMC that you don't about SageTV. And what do you need from SageTV that you aren't getting from XBMC...

I'm a curious one - especially about HTPC software.

It's a matter of taste, just like anything. Best for you to try it out yourself and see how it fits into your HTPC wants/needs.

3rd party win32 builds
Grab Jester's latest win32 build from his link. It has native SMB support for windows.

[mikwilli]

Well I just can't figure out what the problem is. The code is a cakewalk but its not returning what it should. For the node in question the dump is;
Type: ELEMENT_NODE
Name: objectRef
Value: None
Type: ATTRIBUTE_NODE
Name: ref
Value: 27805663
Type: TEXT_NODE
Name: #text
Value: 27805663

but running this which works elsewhere doesn't work on this node
value = node.getAttribute('ref')

value stays as an empty string. From the dump we can see that an attribute called ref exists and the extraction command can't get any simpler so I am not sure what the issue is. Google has failed me

For reference I have no trouble using the same command and getting a result with this node:
Type: ELEMENT_NODE
Name: arrayRef
Value: None
Type: ATTRIBUTE_NODE
Name: ref
Value: 9246537
Type: TEXT_NODE
Name: #text
Value: 9246537
Type: ATTRIBUTE_NODE
Name: size
Value: 2
Type: TEXT_NODE
Name: #text
Value: 2

[mikwilli]

Argh, I knew it was something simple and boneheaded. I was calling the unlink command for the dom before processing the command. The nodelist I had was getting smashed by the garbage cleanup. All I had to do was wait to do gc until after the call.

[stuckless]

mikwilli, Glad you figured it out

And Brent, much like mikwilli response to your xbmc, I have 1 main TV where I have the HD extender, but I also have 2 other TVs in the house that get used less frequently. On one I have a MVP, and on the other I have a Xbox. So having a script like this for the rare times I do use the xbox is great.

[mikwilli]

I really like this API. It is able to provide point pieces of information really quickly. I did find though that once I had it ramped up to grab all the needed pieces for the search window though that it was doing 16 http calls per item and so a search of my recorded shows made over 200 HTTP calls which turned out to be too slow. What I am doing is a hybrid approach. I will use this API to get the information quickly like AiringID or MediaFileID, then use the webserver XML for broader information. It should be the most optimal solution. I was going to look into JSON, but it looks like it is supported in Python 2.6 and the latest Atlantis build of XBMC is still using 2.4

I will update later next week with (hopefully) a working copy to try out.

[stuckless]

I'll put some thought into this as well. I had a couple of ideas around the making request calls that would return more information. For example, once you have a media file ref, you could make a call that would return all/most information about media item in a single call. Unfortunately, been busy working on some other stuff lately

[jhh]

Originally Posted by stuckless

This script looks pretty cool. I'm a long time XMBC fan, but I've recently had to let it go, since it does not support mpeg4 HD (at least not on the xbox).

The webserver supports re-encoding/streaming on the fly - if you have a powerhorse as a server you could consider doing so and the xbmc could then show the re-encoded content.

As an alternative you could build a windows/linux box with XBMC on as the limitations are XBOX related and most likely not XBMC.

Just me trying to keep SageTV interested developers on the platform

[jreichen]

Originally Posted by stuckless

I'll put some thought into this as well. I had a couple of ideas around the making request calls that would return more information. For example, once you have a media file ref, you could make a call that would return all/most information about media item in a single call. Unfortunately, been busy working on some other stuff lately

You probably already know this, but nielm created a DTD for the XML documents that the web server creates. That's one format you could use to return larger data sets.

[mikwilli]

One thing I found interesting though was that the DTD did not seem to contain all the pieces you find on the web pages. For instance I was not able to find isArchived. Since its not in the XML and not in the sage.api that I can find (It may well be there and I am missing it) I am unsure if this is something that is no longer applicable.

stuckless,
I loaded up your SVN trunk into eclipse and was poking around. Its got a lot of good stuff in it. I was thinking that really all that may be needed would be to cast the objects back into the Sage pieces like sage.Airing, etc. I wasn't able to find these object definitions in the Sage.jar though. Any idea where they are housed?

[jreichen]

Originally Posted by mikwilli

One thing I found interesting though was that the DTD did not seem to contain all the pieces you find on the web pages. For instance I was not able to find isArchived. Since its not in the XML and not in the sage.api that I can find (It may well be there and I am missing it) I am unsure if this is something that is no longer applicable.

There isn't an isArchived function in the Sage api. Instead, it's

isTVFile && isLibraryFile

[mikwilli]

Ah, thanks for the info. Is there a doc that I should be reading instead of posting? I know there is a Studio doc is this info there?

[jreichen]

Andy created a sticky for Studio Information Resources. That has several links to helpful documentation.

I usually open an existing STV in Studio to see how to do things. I'm no Studio expert, but I can follow the code well enough to find out how to do something.

Other Java plugins are also good places to look. The first thing I worked on was nielm's webserver and learned a lot from looking through its code.

If your intent is to use this thread for user support, it's probably a good idea to ask development questions elsewhere so users don't have to filter through them to find answers to their questions. The SageTV Studio forum is a good place to post questions about development.

[stuckless]

You probably already know this, but nielm created a DTD for the XML documents that the web server creates. That's one format you could use to return larger data sets.

Actually I didn't know this , but I would think that if niem's server already supports sending back large chunks, then there wouldn't really be a need for me to do it as well... at least not in the same format. Anything that I would do, would either be done via a generator, or custom templated on the fly. I'll explain...

I'd either generate replies that represent object items, much like greg does in his API, and then serialize them back as XML or JSON. The other approach would be to allow a user to pass reply object template during the request, and then just have the reply code, fill in the blanks so to speak.

mikwilli, regarding the APIs... The APIs that I created mirrors the ficticous APIs that Sage documents. You can't get a concrete sage object that represents a media file, etc, but you can get an object that you then call services against. It's actually because of the services nature of the Sage API, that making a remote API was quite easy. If you want to work in Java, then you can use my java wrappers, which work like how Sage documents their API, or you can use Greg's wrappers (from the studio tools project) which works more like traditional objects.

[mikwilli]

Ok cool. This was great information. Really the webserver would be perfect for this XBMC python script except for 2 things. First is that the xml does not respond to the pagelen argument. So that means that when I request the recorded progs, scheduled progs, EPG info, etc. It always passes the full list. Since Xboxes are just P3s with limited memory this makes the script too slow. The second issue is that the order by/sort by commands are not honored by the xml.

So what ends up happening is that to grab say the last 5 recorded programs I end up processing everything. I'll check the code that was pointed out which should give me some hints as to what may be possible that I am currently unaware of. Either way I think it wouldn't be too hard to add some extra pieces which would do the bulk of the sorting/processing on the bigger server side instead of at the Xbox.

[stuckless]

It always passes the full list. Since Xboxes are just P3s with limited memory this makes the script too slow

Have you looked in using a SAX parser instead of a DOM parser? I tend to use SAX for xml files that are quite large. It's a little hard to work with at first, but once you get used to SAX, it's actually quite good. I used to DOM parsing for everything, but lately I've been using SAX, and it is quite fast and uses very little memory.

[mikwilli]

stuckless,
I didn't realize that the SAX parser was faster. I will check into using it. Also I added a new Handler to your Sage API. It doesn't touch any existing code aside from sagex.remote.SagexServlet which allows it to be called. The handler basically just doesn't make remote references. If its a primitive or array of primatives is prints them out in the value tags. Same with strings. If its a sage object then it calls the SageXMLWriter that Nielm wrote in sagexmlinfo. So far it is working well. I am adding in sorting and length limiting. If it seems viable I will pass on the code to you to determine if its something you want to include or not.

[stuckless]

Sounds good... can't wait to see it. BTW.... i the current serializer does actually send back primitive and arrays of primitives.... arrayRef is and objectRef are only returned for Sage objects that can't be serialized.

SageXMLWriter sounds like a good choice. Was this what jreichen was referrerring to in an earlier post? If so, then I didn't catch the full reference.

[jreichen]

Originally Posted by mikwilli

stuckless,
I didn't realize that the SAX parser was faster. I will check into using it.

It can be much faster and require less memory. But it can also be tricky to code because it is event-based and you have to track where you're at in the document. Plus it always parses the whole document. You can choose to ignore events after you find what you need, but the parser still runs through the whole document.

In Java, there is a newer parser type known as StAX (Streaming API for XML). .NET has something similar, but I'm not sure whether Python has it. It's efficient like a SAX parser because it doesn't build a DOM tree, but easier to code because you have more control over the parsing by asking for the next item when you're ready for it (it's a "pull" parser rather than a "push" parser). You can also stop the parsing when you find what you need (by not asking for any more items). Just an alternative for you to consider.

Originally Posted by stuckless

SageXMLWriter sounds like a good choice. Was this what jreichen was referrerring to in an earlier post? If so, then I didn't catch the full reference.

Yes, the web server uses nielm_sagexmlinfo.jar to create the XML docs. It also contains the DTD I mentioned.