Re: [backstage] Today's TV-Anytime tar is broken
At 11:26 07/03/2006 +, you wrote: Hi, Is there any news of this as there doesn't appear to be a file for today :-( Appart from this problem, the data is really great and I'm currently hacking Perl TV::Anytime to make use of the additional data that is hidden in these files. I'm make a patch available once it is fully working. There appears to be a problem with the FTP link used to get the data from the database machine to the web site itself. I've flagged the problem to our internet services people - hopefully the problem will be resolved soon. It seems to have corrupted yesterday's upload and completely failed on today's. Hywel I tried a manual push - that also failed. - Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html. Unofficial list archive: http://www.mail-archive.com/backstage@lists.bbc.co.uk/
Re: [backstage] BBC TV Listings Feed
At 13:50 24/01/2006 +, you wrote: Hi, The TV Listing feed isn't available yet for today (24-01-2006). Just wondered if your having any problems today generating the feed. Sorry - I inadvertently damaged the crontab that calls the daily data generator yesterday which meant it didn't run this morning. It's now fixed and the script was manually run to ensure that there's a set of valid data for today. Hywel - Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html. Unofficial list archive: http://www.mail-archive.com/backstage@lists.bbc.co.uk/
Re: [backstage] typo/error in TV-Anytime data
At 13:36 17/12/2005 +, you wrote: Hi, I think I found a typo/error in the latest TV-Anytime data: In 20051223BBCOne_pi.xml, there's this line... Synopsis length='short'![CDATA[The Long Goodbye: Owen returns to work, eager to start a new life with Chrissie. Elliot reveals his dark side when he loses a patient. Mickie realises that she needs to move on with her life. [ADS,SL]]]/Synopsis I believe ADS,SL should be AD,S,SL (or AD,SL)? The TV::Anytime parser chokes on the ADS... It looks like a typo. The synopses are typed in by hand and occasionally inevitably errors occur. Very often these errors are spotted and corrected in later database updates by the schedulers. Hywel - Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html. Unofficial list archive: http://www.mail-archive.com/backstage@lists.bbc.co.uk/
RE: [backstage] Timezone bug?
At 11:45 31/10/2005 +, you wrote: From an internal perspective Yes! seperate out Series, Episode numbers and Ep Titles into clear fields. It's really useful if you were to use TV anytime data to populate programme support sites like, say, bbc.co.uk/buffy or /theoffice - and becomes really important for following ongoing series. It's something that the Radio Times do in a very patchy way. TV-Anytime can represent episode numbers using the EpisodeOf tag: e.g. EpisodeOf index='2'/ would denote the second episode in a series. The problem however is that although this information is available within the BBC, it's not provided in the data stream I use to generate the TV-Anytime files, with the exception of BBC7, who put this information at the end of the synopsis. This data source was intended for the generation of EPG guides - TV-Anytime on the other hand has a much bigger remit. We enhance the basic guide data with as much information as possible to provide the TV-Anytime data - but if we're not provided with the information in the first place, there's only so much that I can do! You could, for instance, make a query engine that lets you build a personal schedule around watching a show in story order across all the various channels and repeats That's exactly what the tag is meant for - to ensure the order of a series is maintained by an automated recording/displaying service. Hywel - Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html. Unofficial list archive: http://www.mail-archive.com/backstage@lists.bbc.co.uk/
RE: [backstage] Timezone bug?
At 14:32 31/10/2005 +, you wrote: [Taking this back on list as need the attention of the TV anytime guys] LOL - If anything, I'm Ms Web Editorial - I'm a content producer with small grasp of technical stuff, who just happens to be reasonably good at translating technical stuff into useful content ideas. I've done a lot of making TV programme based content on bbc.co.uk, and a bit of thinking about data models for that. The set-to-record from trailer is a great idea - and you're right, should be possible / doable. The problem is, i think, that CRID isn't widely used across the whole industry - i don't know if TV anytime is used by sky? There's also a problem that somehow, the trails (BBC speak for trailer) would have to be synched with some kind of CRID broadcasting widget. I may be wrong, but I think that trails are added on a very ad-hoc basis in the transmission suite - they can be added, swapped or removed at the last second (literally!) to keep the schedules running on time, and are played in from digibeta. (This might be from harddisk these days... it's a year or two since I've chatted to anyone in TX (BBC Speak for Transmission) in the bar.) Oy! Kingswood boys! What do you think to this, and can you explain whether the idea would work, and what the issues are, in terms I can follow? We already have a demonstration of programme selection by trailer, using a modified Pioneer Freeview box and a transport stream containing TV-Anytime data and signalling. This was on show to the public in 2004's IBC as well as being demonstrated many times to industry (e.g. BBC RD's Open Days this year). There are many obstacles to getting this as a live service of course - the least of this being the way trailers are changed at last minute, but we've shown it can be done with the existing TVA standard and technology. We've also experimented with linking to the live playout system and getting triggers for trailers that way. Hywel - Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html. Unofficial list archive: http://www.mail-archive.com/backstage@lists.bbc.co.uk/
Re: [backstage] TV Data bug..
At 11:04 30/10/2005, you wrote: Hi, Program: crid://bbc.co.uk/1103146355 and crid://bbc.co.uk/1103146353 Are the same program, both on BBC 4 Unfortunately to get 100% accuracy, the programmes should all be matched by hand, possibly by the schedulers. Unfortunately this isn't currently done and repeat programmes are found using a simple algorithm, which occasionally fails as it has in this case. The opposite can occasionally happen when two programmes are matched as repeats, when they're not - usually caused by generic saynopses. There is also a know issue with films that traverse the news, with the two halves not being paired correctly. This is a test environment - it s believed that far more accurate matching will be done should a full broadcast TV-Anytime service be set up. Hywel [EMAIL PROTECTED] - Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html. Unofficial list archive: http://www.mail-archive.com/backstage@lists.bbc.co.uk/
Re: [backstage] Timezone bug?
At 14:23 30/10/2005, you wrote: Hello, backstagers: It looks like the data generator on the BBC end is having timezone issues -- all data I have for today (from the 20051029 tarball) is an hour off. For a while I thought that it was on my end, but no, the actual XML files get it wrong. Also, I noticed that at least for some shows, the time in the ProgramURL does not agree with the time in the PublishedStartTime by a few minutes. What does this difference represent? Are the PublishedStartTimes simply inaccurate but look nice, or is there something more interesting going on? Also, rather unrelatedly, would it be possible to get the name of the episode in a seperate element from the free-text descriptions? Thank you, -=- James Mastros I can't see anything wrong with the times in the data sets both for today and the 29th's tarball. Remember that they're all in UTC (hence the Z at the end of the time), so anything up until 31st October appears to be out of synch by an hour. I'll look into it in more detail tomorrow to ensure there isn't a problem, but at the end of the day, it's up to the end application to sort out daylight savings. The time difference you're seeing is the difference between the published time and the actual time the playout server was going to show the programme. Sometimes programmes may start early and at other times late. More often or not, these days this is not accidental - prerecorded BBC television services mostly come off servers these days so the timing can be down to the second. Having said that, the figures you're seeing are only a snapshot of the estimated start times at 8am in the morning when the files are generated. These times shift and change throughout the day. The only anchors you'll notice are the 1, 6 and 10 o'clock news on BBC One which start at the exact time published. In the future, a TV-Anytime service may be broadcast along with a tv/radio digital service or provided as a live Internet feed and this exact time could be used to, say start a PVR or stream recorder. As to separating the name - I'll look into the TV-Anytime standard to see if there's a field for this (I don't recollect one) when I have that hefty tome on my desk at work tomorrow. The synopsis provided is in fact a slightly cleaned up version of the one that goes out to Sky and Freeview, so to extract the title from the synopsis, I'd have to somehow work out which part of the text is the episode title. If there's a pattern, it's easy to write a filter. Where different schedulers do it in different ways, that's where it gets complicated. Hywel [EMAIL PROTECTED] - Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html. Unofficial list archive: http://www.mail-archive.com/backstage@lists.bbc.co.uk/
[backstage] TV-Anytime regional opt-out files
In view of the stagnant thread, I thought I'd add something new to the TV-Anytime data so that people can have something new to experiment with. It hasn't been rigorously rested and may include some errors - but this is a stab at representing regional opt-outs using TV-Anytime. At the moment I'm only providing English regional opt-outs for BBC One, but hopefully it won't be too much extra work to extend to the Nations and BBC2 Nations. Here's the way I've implemented it: 1. In the ServiceInformationTable, the main BBCOne feed is still London - this actually reflects how the BBC have provided sustaining feeds to the regions for a long time. 2. The English regions are also in the ServiceInformationTable - these have an extra element describing the parent service as being the BBCOne feed. In other words, when there's no data provided specifically for the regional feed, assume the parent feed's data is to be used. 3. ProgramInformationTable: The metadata for all the regions is found in the BBCOne file. As always, these are indexed by their CRIDs. 4. ProgramLocationTable: Again, a single BBCOne file is provided. The Schedule table for the BBCOne sustaining service is initially listed, followed by the regional opt-outs. Only the programmes that differ are listed here, with the parent service being assumed to otherwise. 5. ContentReferencingTable: Here, for each CRID, the locations for all regions are listed. When services are opted out, they're not listed under the main BBCOne service. This isn't the only way to represent regional opt-outs in files, but I thought I'd experiment with this format, to see if it's a viable method of distributing a complete service list for a broadcaster which has regional opt-outs. The first file of this format is found in the sub directory test from where the data is usually put and is called 20051025.tar.gz The data automatically provided every day hasn't changed. For now, I'll only produce an occasional set of files in that directory for you to get the gist. If someone requires a more regular service, that can be arranged... Hywel - Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html. Unofficial list archive: http://www.mail-archive.com/backstage@lists.bbc.co.uk/
[backstage] New, more experimental data to play with
Folks, A request was made a while ago that a separate feed of data that's closer to what we're using on the development end be made available, which may be more liable to instant change as new features appear. This has now been done. I haven't yet automated the generation of this feed so for now it won't be updated daily, but this is also in the pipeline. The files can be found in a sub-directory of the tvradio directory: http://backstage.bbc.co.uk/feeds/tvradio/test/ Currently, there are only two major differences between this feed and the main Backstage one - this one now uses the more recent 2004 classification schema and also supports almost all local BBC Radio stations. One issue with regard to local radio stations is that occasionally, sister stations get together to broadcast the same programmes. Ideally, and by definition, these programmes should have the same CRID, regardless of where they're broadcast, however this is not currently supported. Each station in the current model is treated completely separately, but I'm looking into the possibility of providing a more accurate model in future (and with this in mind, regional telly schedules would be desirable too, again something I'm looking into). Hywel - Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html. Unofficial list archive: http://www.mail-archive.com/backstage@lists.bbc.co.uk/
Re: [backstage] BUG: dodgy data..
At 01:42 18/09/2005, you wrote: Hi, Program crids: 285169583, 285170277 and 285171923 are in fact all the same program Two Pints of Lager and a Packet of Crisps, All with the same description Purgatory: Gaz tries to get over his guilt by doing one special thing for Donna. Janet holds a party for Louise's graduation. 285171923 also seems to have 21 audio channels - which although impressive might not be true :) Cheers Leo Thanks for keeping us informed about these anomalies - most appreciated. Many of these happen unfortunately further up the chain from where the data is generated but these two certainly sound like they shouldn't have slipped the net. The first one is strange - a programme with the same synopsis and title on BBC Three should have been allocated the same CRID - I suspect that isn't a chain problem and should hopefully be something I can track down and fix myself. I see no reason why any channels should have 21 channels of audio (well, I think I've worked out what's going on - read on)! Any errors that come through from our scheduling department that are obviously bogus (e.g. 5.1 surround or any other signification other than stereo or mono) should be trapped and are given a default value within the permitted range. I'll check the code on Monday to see how on earth that happened! Almost certainly a case statement gong wrong and giving it both 2 and 1 channels I suspect... -- Hywel Williams [EMAIL PROTECTED] Backstage Email Account @ Home - Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html. Unofficial list archive: http://www.mail-archive.com/backstage@lists.bbc.co.uk/
Re: [backstage] Some changes to the TV-Anytime data
At 11:15 15/09/2005 +0100, you wrote: At 18:32 14/09/2005 +0100, you wrote: Some bug fixes and modifications have been made to the data set - these are outlined below. 2. When audio description is being provided for a programme, this is indicated with an extra AudioAtrributes entry with an AudioLanguage purpose of type 1 (Audio description for the visually impaired): AudioAttributes NumOfChannels1/NumOfChannels AudioLanguage purpose='urn:tva:metadata:cs:AudioPurposeCS:2004:1'E N-UK/AudioLanguage /AudioAttributes delurk Ahh - this reminds me of something I meant to ask ! /delurk From other interest areas, I know the Beeb are involved in the W3C Timed Text initiative. I think one of the contacts is David Kirby in the RD section of BBC. For those of you who don't know, this is an XML format for storing subtitles or captions (if you're in the USA). So here's a thought : Supposing each broadcast program is subtitled - most are, to meet disability guidelines for deaf viewers. How about publishing the subtitle files as part of backstage, in W3C TT format (maybe in the final draft format until the complete spec is ratified). Then in the TV data feeds, publish a linkage to the subtitle file. So, not actually embedded in the TV data feeds, but as a separate dataset. Now we'd have something similar to the google video and Blinkx text searches. I do this on my Captionkit website, but of course it's a manual process to enter the captions in the editor. As the BBC already pay people to create and store this data, it's definitely available and I think it would be a massive benefit to publish it. If you did that I'd for-sure drum up a demo project in under a week (probably a PHP API) to make use of those, and link a subtitle search to program information for repeats etc. Food for thought ? It's an excellent idea, and technically quite possible.The big hurdle unfortunately is rights. Obviously, a broadcaster wouldn't publish subtitles for programmes that haven't been shown yet - the tabloid press would have a field day publishing the plot of EastEnders days before it's broadcast! After/during broadcast? Definitely technically possible and has already been demonstrated on both pre-recorded and live output, but at the moment the rights to do so are prohibitive for a public service. Hywel - Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html. Unofficial list archive: http://www.mail-archive.com/backstage@lists.bbc.co.uk/
Re: [backstage] Some changes to the TV-Anytime data
At 10:23 15/09/2005 +0200, you wrote: Are you just changing the genres or are you also introducing some 2004 elements with namespace and all? TVA's atrocious practice of changing namespaces when they change versions could introduce all manners of problems for implementations if the upgrade isn't planned. Since the genre schemas are backwards compatible, the initial change would probably be cosmetic in just changing the 2002 to 2004 in the data. New genres introduced in the 2004 schema will eventually be used. In fact, 2004 descriptors are already in use for AudioPurpse in the audio description information for the simple reason that they didn't exist in the 2002 classification schema! Hywel - Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html. Unofficial list archive: http://www.mail-archive.com/backstage@lists.bbc.co.uk/
[backstage] Some changes to the TV-Anytime data
Some bug fixes and modifications have been made to the data set - these are outlined below. 1. The character set being used is now correctly identified as ISO-8859-9. This was always the case but the ISO-8859-1 ident was incorrectly displayed. 2. When audio description is being provided for a programme, this is indicated with an extra AudioAtrributes entry with an AudioLanguage purpose of type 1 (Audio description for the visually impaired): AudioAttributes NumOfChannels1/NumOfChannels AudioLanguage purpose='urn:tva:metadata:cs:AudioPurposeCS:2004:1'E N-UK/AudioLanguage /AudioAttributes 3. The missing ServiceGenre items in the ServiceInformationTable have now been added. 4. CBeebies' identifier was misspelt as CeeBeebies in the ServiceInformationTable (and everywhere else!) This has now been corrected. Note that this has also affected the filename used to represent the channel's data. - Some forthcoming changes/issues. 1. It has been pointed out that the format of the DVB locator used in the ContentReferencing Tables and the ProgramLocation tables don't conform to the DVB standard! Essentially the divider between the date and time should use -- rather than / to be DVB compliant. e.g. dvb://233a.4000.4740;[EMAIL PROTECTED]:00:00Z--PT01H00M While this is dead easy to fix, it may cause some software already written to use the current method to fail so this is forewarning that this bug will be fixed soon. 2. More channels have been requested. I'm currently testing the inclusion of almost all BBC regional radio stations in the schedules. The main issue is that at present they don't represent occasions when more than one regional station is broadcasting the same programme. 3. Currently, the 2002 genre classification is being used instead of the more recent 2004. While again something very easy to change since they're very similar, this again could have issues for software expecting the 2002. - Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html. Unofficial list archive: http://www.mail-archive.com/backstage@lists.bbc.co.uk/