Add a parameter to be parsed as item's datetime stamp

[martinrotter]

[Owyn]

Would that be useful? 🤔
Since RSSGuard already sorts new entries right even without dates martinrotter/rssguard#621 (comment)

[RetroAbstract]

Would that be useful? 🤔

Yes. It would be useful in being able to limit the initial amount of articles CSS2RSS-based feeds fetch upon the first fetch after feed creation (or the first fetch after the database is cleaned & optimized) with article filters and properly limit the amount of articles in general with this upcoming feature.

Right now the article filter I use successfully limits the initial amount of non-CSS2RSS feed articles fetched but cannot limit the initial amount of CSS2RSS feed articles fetched as they all have the same timestamp, and the feature has unexpected behaviour as when for example limiting a CSS2RSS feed to:

• keeping 1 article, deleting important and non-read articles: RSS Guard will oscillate at random between the first and second item found at every fetch.
• keeping 2 articles, deleting important and non-read articles: RSS Guard will oscillate at random between the first and second item found at every fetch for the first article, and oscillate at random between the first, second and third item found at every fetch for the second article.

[Owyn]

Well, that can be done of course as a yet another parameter, but my worry is - will people actually bother to use it and match one more element with a new css selector every feed for this? (even if they'd want to limit the number of fetched articles with this mentioned feature)

[martinrotter]

well I myself would use that feature

it would be nice if the provided element was somehow magically parsed into date/time object, there is excellent python package "dateparser" which provides such automagic string -> date/time conversion function

[RetroAbstract]

Well, that can be done of course as a yet another parameter, but my worry is - will people actually bother to use it and match one more element with a new css selector every feed for this? (even if they'd want to limit the number of fetched articles with this mentioned feature)

I would, yes.

[Owyn]

@martinrotter articles keep getting remarked as non-read when their date changes even a little :-(

so with this feature they will be forever unread (cuz many dates are relative)

2024-01-21.16-47-14.mp4

Tried in 4.6.3 - same result

[Owyn]

Try the new update with the Date parsing selector (6th argument) - aim it at text nodes with the date or or elements with title or alt attributes containing Date

You'd need "No duplicates by URL" filter to not let items get remarked as unread every time currently (dunno why it's not a standard checkbox option)

function filterMessage() {
  if (msg.isAlreadyInDatabase(MessageObject.SameUrl)) {
    return MessageObject.Ignore;
  }
  else {
    return MessageObject.Accept;
  }
}

[RetroAbstract]

New date selector works relatively well, thanks alot :)

Couldn't get it to work for these pages even if they do have dates to their items:

https://jwfan.com/?cat=21
https://support.microsoft.com/en-us/topic/windows-10-update-history-8127c2c6-6edf-4fdf-8b9f-0f7be1ef3562

[martinrotter]

Calling script like this fails:

python3 scripts/css2rss.py ".clanek" "" "" "" "" ".date"

I only need to specify "item" selector and date selector (in this case with "date" class)

114, in <module>
    if sys.argv[2][0] == '!':
       ~~~~~~~~~~~^^^
IndexError: string index out of range

[RetroAbstract]

Calling script like this fails:

python3 scripts/css2rss.py ".clanek" "" "" "" "" ".date"

I only need to specify "item" selector and date selector (in this case with "date" class)

114, in <module>
    if sys.argv[2][0] == '!':
       ~~~~~~~~~~~^^^
IndexError: string index out of range

I think you need to use ~ instead of "" to skip selectors you don't need.