I’m having trouble with a newsletter from CaringBridge - the content in the original newsletter is a table with bgcolor="#7f1467", and then white text in a cell of the table. But the Newsblur sanitizer removes the bgcolor from the table and the result is white text on a white background.
Am I on the right track to think that this is a result of bgcolor not being included in the set of allowed_attributes in the Scrubber util?
Not trying to propose a particular solution yet - just making sure I understand correctly where this is happening in the code.
A follow-up: A while back, I had a separate thread on a related topic, but at the time I clearly misunderstood what was happening. The thread was here. I thought Newsblur was blocking/failing to load remote images.
Now, instead it appears to me that images that are contained in a <figure> element are being sanitized (deleted). This looks like it is a result of figure not being included in the allowed_tags set in the same util I linked before.
I am considering reaching out to the upstream library, but wanted to take the opportunity to get your feedback first, Samuel, to see if you agree I’ve correctly identified the issue here, or whether you’ve got any challenge. Thanks.
Background color should be stripped, IMO. But color should be preserved. If a newsletter is provided white on white, you can change the theme color temporarily to read the story. But NewsBlur should be normalizing content. In other words, the text color should also be stripped.
Thanks, @samuelclay .
Here’s an example Newsletter from CaringBridge. The font color is on a span:
<span style="font-family: Arial, Helvetica, Verdana, sans-serif; font-size: 14px; color: #ffffff; text-align: left; line-height: 24px;">
I’m not sure of the right terminology here - is the color a “sub attribute” of the style attribute? Anyway, looks like since style is a permitted attribute in scrubber, the style is permitted to set the text color to white.
The workaround you suggested works, I’ll try using that.
I guess that’d mean that the behavior regarding the <figure> tag is a more pressing thing for me personally, harder to work around at least since I have to visit the webpage where the content is hosted to work around that one.