I'm using feedparser for working with RSS.
I'm getting regularly (e.g. every 15 minutes) RSS channel with items and store it. In the channels there aren't often any new items. So, it's unefficient.
Is there a way to detect quickly if there are some new items in the channel and if not, do nothing with this channel?
thank you
For RSS 2.0, the channel element has an optional lastBuildDate eleement. For atom, there's a similar "atom:updated" element, but the standard does state that this is when "an entry or feed was modified in a way the publisher considers significant. Therefore, not all modifications necessarily result in a changed atom:updated value".
There's also a PubDate element in RSS 2.0, also optional, but lastBuildDate should be the one to use, assuming it's there and the publisher is using it correctly.
You can store the previous one and compare the newly retrieved value with the old one.
Added material on feedparser:
For feedparser, see feed-updated_parsed and feed-updated.
Related
I have a large set of podcast feed URLs which I'm periodically polling to check for updates. I'm really struggling to find a robust way to detect if a feed has changed that doesn't have any false positives. I'd like to be able to detect not just if there is a new episode, but also if an existing episode was updated.
RSS and Atom feeds provide pubDate, lastBuildDate or updated elements. However, I'm finding these frequently misused so that the feed is actually inserting the current date time into these fields each request. This makes them difficult to rely on to detect changes.
My next thought was to strip all date information from the podcasts, then MD5 hash the feed contents. I can then compare the feed hashes to detect changes to the feeds.
This seems to work for about 90% of the cases. However, there are still hundreds of podcasts that insert dynamic data into their feeds.
One podcast has the following as their podcast cover art:
http://erikglassman.hipcast.com/albumart/1000.1439649026.jpg
Where 1439649026 is what I assume is a timestamp. This second number changes with each request of their feed.
This is starting to seem like a losing battle. If I can't reliably trust the date fields of a podcast feed, and if some percentage of podcasts insert dynamic data into their feed text, how can I reliably detect changes to a feed in a robust way?
Everything you say is true, so it's not a good idea to try to detect changes at the feed level, instead look for them at the item level.
That generally works, if it doesn't the feed can't be used by anyone, so the source of the feed is likely to have fixed any problem. That's why I think it works so well.
I've been writing feed readers as long as they have existed, my current product is called River4, it's available as open source, MIT License, so you can use it as example code, for this and other issues.
This is where it checks if an item is new:
https://github.com/scripting/river4/blob/master/river4.js#L1411
That might move around as the code changes, so look for a routine called getItemGuid. It shows you how to get a value that uniquely identifies the item. I use this code for my podcatcher, http://podcatch.com/, and it seems to catch the new items, and doesn't get false positives.
Hope this helps! :-)
I am building RSS feed for the first time and I have some simple, direct questions that I was unable to find on the web, well at list in a sense that would be clear to me. Can you help me understand following
Which items should I include in RSS generation? should I always put in all the articles or what is the criteria when I query my articles for the feed?
What value should I set for pubDate? The specification says "The publication date for the content in the channel. For example, the New York Times publishes on a daily basis, the publication date flips once every 24 hours. That's when the pubDate of the channel changes.". I do not quite understand how to apply this to my feed. I have new articles daily, should I set the pubDate to let say 06:00 AM today and update it every day?
lastBuildDate: if I understand this right is the date of the latest updated item?
Which items should I include in RSS generation?
You should have one generic feed with all the new articles you post (for example: news). Additionally if you got your webpage split into categories, or you have some specific feeds (eg. calendar of the events) then it's good to create additional separate RSS for each one of them
What value should I set for pubDate? I do not quite understand how to apply this to my feed. I have new articles daily, should I set the pubDate to let say 06:00 AM today and update it every day?
Always set pubDate to the time when your news/articles went online. So if you have new articles daily pubDate should be a date when they were released to the public. Not random hour in the morning. Not the moment when you started writing them.
lastBuildDate: if I understand this right is the date of the latest updated item?
lastBuildDate is the most recent date when any of the results was posted or modified. Usually you should skip it - especially if your lastBuildDate will be simply a most recent pubDate. It's an optional parameter.
I use lastBuildDate only for calendar RSS feeds to show when the calendar was updated (as in calendars you not only add new entries but also often edit existing).
You should put every article, but the best is to provide different feeds for different categories, even search keywords. You can build it like any dynamic page, with a querystring.
that's not super important, you can put whatever. I don't think may feed readers use it.
theoretically it's the date the content changed. So the date of the latest updated item should work.
Something super important, since people are going to do polling on this page (meaning a lot of requests on the page)
- Cache it on your server
- Serve and Etag header and/or a LastModifiedDate. That way your server can respond with just a "not modified" if the client has it in cache already.
What is the correct response an RSS client should have when it encounters a feed that has multiple items with the same guid/identifier?
Currently in my application, any items that use an existing guid won't be cached or displayed because it believes it already has that item.
In this example feed a lot of items share this id:
tag:blizzard.com,2010-10-22:diablo3:feed:en-us:1
According to w3 when there are duplicate entries in an RSS feed:
Atom Processors MAY choose to display all of them or some subset of them. One typical behavior would be to display only the entry with the latest atom: updated timestamp.
I would go with the spec and display only the entry with the latest updated timestamp. Don't forget to send an email to Blizzard support and have them get their RSS validated - just don't threaten to keep them out of the next raid.
Take care.
I think your app is doing it right. Don't get fancy. If you've already seen an item with that guid you don't present it a second time. You should contact whe webmaster for the feed if possible and alert them to the problem.
Does each item have a unique URL? If so, fall back to using the URL.
I'm writing an RSS feed. Let's say it's for a list of entries as in a blog.
How do I handle updating the feed? I mean, let's assume that The feed always displays the last 10 entries.
If someone subscribes now, he'll get the last 10 entries (1..10)... what if there are for example 2 new articles, and then what will his feed reader do? Because I will return the articles (2..12).
Do I have to do any special handling to start from a certain article in the feed, or do I just always put the last 10 and this will be fine
Returning the last n articles will be fine. Because you assign a unique identifier to each article (you do, right?) the feed reader can easily keep track of what it has already seen or not.
The feed reader will probably watch to see how often new articles appear, to help determine how often it checks for new articles.
I have the feeling, in every RSS.xml file, both the pubDate and the lastBuildDate match.
I am sure that this one, is not always true...
So firstly, what is the difference between those two above?
Secondly, the RSS readers, sort the content by Date, based on the pubDate or the lastBuildDate?
pubDate:
The original publication date for the channel or item. (optional)
lastBuildDate:
The most recent time the content of the channel was modified. (optional)
Here are some docs for the optional items in the RSS 2.0 spec.
Answers here are all over the place. Some people are getting confused by the fact that item has a pubDate as well. I believe the OP is specifically asking about the difference between lastBuildDate and pubDate at the channel level.
From the best of my understanding of the RSS spec, which is notorious for ambiguous explanations, lastBuildDate would be the last time the feed was created. For example, if you cache a copy of it on your server for some period of time, lastBuildDate would the time that cached copy was created.
pubDate, on the other hand, seems to be basically the last time any actual content within the feed has changed. For the most part it's pretty much going to be the latest pubDate value from the items in the feed, since generally, the feed content is only changing when some new item gets published. However, it could also be a date when you made some change to the channel, itself, such as changing the channel title, description, etc.
lastBuildDate specifies the last date/time the entry was modified. pubDate specifies the actual publication date/time.
The reason you see these as generally the same is because by the time you get the RSS feed, there hasn't been any edit to the article.
I can't find the RSS spec on this unfortunately, but I am pretty positive that's what they are.
By RSS 2.0 specification, it seems they are roughly equivalent:
lastBuildDate:
The last time the content of the channel changed.
pubDate:
The publication date for the content in the channel. ...
The difference is subtle: They tell us about the method that was used. In case of <pubDate>, the channel is published manually or in fixed period. In case of <lastBuildDate>, the channel is built automatically upon new article being added on the website, adding it as new item.
While the other answers here do provide some good information, I feel the need to elaborate just a little bit for any future visitors.
pubDate
The publication date for the content in the channel. For example, the New York Times publishes on a daily basis, the publication date flips once every 24 hours. That's when the pubDate of the channel changes.
lastBuildDate
The last time the content of the channel changed.
So, taking the New York Times as an example again, the <pubDate> is the date the feed was published while the <lastBuildDate> would be the date the content inside the feed changed. In the end, I would view the <pubDate> as the date the feed is published and the <lastBuildDate> as the date any content in the feed was last modified.