Preview picture for each article

[nicosomb]

As @moafred (from Bookeen team) told me yesterday, we can use og:image meta to get a preview picture for each article.
This tag is used by twitter and facebook to generate preview in posts.

For example, with this article http://www.lemonde.fr/pixels/article/2014/12/18/affaire-sony-pictures-la-strategie-anti-piratage-d-hollywood-revelee_4542100_4408996.html, we have this meta :

<meta property="og:image" content="http://s1.lemde.fr/image/2014/12/17/600x315/4542146_3_8afb_le-legendaire-signe-surplombant-hollywood-en_443f1e6556ca1af14830f9d266203bcf.jpg" />

It can be useful for generating nice ePUB files (instead of wallabag logo).

@tcitworld Are you interested to have a look to this feature?

[tcitworld]

Yup, seems quite good.

[tcitworld]

To do :

• look at how we can get the property information from fulltextrss class : extend Poche class ?
• if Download pictures is activated, download...
• integrate picture in themes ! (Pocket-like)
• pass the link to epub generation

[nicosomb]

We firstly need the last action in fact.
For the integration in the theme, we can wait. As you said in an other issue, we also have to think to the mobile view, etc.

[tcitworld]

I was looking if FTRSS already has something like that, but it doesn't seem so (I only looked the 3.2 version). The picture they produce comes only from pure RSS feeds (<itunes:image>, <atom:logo>, <atom:icon>, <image><title> or `dc:title) via Simple Pie, not regular html pages.
I could build the detection right in the makefulltextfeed.php file, or in wallabag itself. Are you interested, @fivefilters ?

[tcitworld]

Wondering if it's better to extend the Readability class or the ContentExtrator one.

[fivefilters]

We don't do anything special with images at the moment. Might try to implement it at some point. Will most likely go into ContentExtractor rather than Readability. But also makes sense to have it in the Readability class.

By the way, might want to look at how the Newspaper project handles it - https://github.com/codelucas/newspaper/blob/74f349340391122cfd726eb2b049e109a9922c2c/newspaper/extractors.py#L298 - seems they look for the main image in the following order:

1. <meta property="og:image" content="[URL]">
2. <link rel="icon" href="[URL]">
3. <link rel="img_src" href="[URL]">
4. <meta name="og:image" content="[URL]">

Not sure why that particular order though...

[tcitworld]

Thanks a lot for the link !
I'll start the work with the ContentExtractor class.

[nicosomb]

for your information, @fivefilters implemented this feature in last version http://blog.fivefilters.org/post/121443274427/full-text-rss-35

[j0k3r]

This mean I need to add this feature to graby :)

[j0k3r]

Picture is now attached to each entry. We need to display it now.

[nicosomb]

what do you think of this?

[j0k3r]

Great !