The AllMusic tagsource was broken #677
Conversation
The format of the AllMusic page has clearly changed a little since it was written: - album and artist are now in h1 and h2 tags not in div tags - the class of the sidebar album container is no longer "album-contain" but "album-contain gallery-launch". For which the HTML parser needed a patch to support CSS class matching a la: https://tanzu.vmware.com/content/blog/xpath-css-class-matching - A clear bug remained in which album_page was overwrritten by an encoding - Another bug existed in which a string was sought (.find) in a byte-string.
|
There are some small whitespace changes that crept in, I suspect because of my IDE tidying up formatting. If a problem, let me know. This one took a little while to crack, as the AllMusic tagsource was still crashing when expanding one of the returned search results. The crux of it is that the AllMusic pages have changed a little since it was written, and a few old Python2 carry over bugs I suspect remained. The biggest puzzle to me is why we're using lxml xpath searches (quite kludgy for doing CSS matches requiring a bit of non lucid code - I would comment that in the code base actually and will if you agree) rather than BeautifulSoup say. But I worked on a minimalist patch to make it work. Now I can use AllMusic ;-). |
|
@bernd-wechner could you please look into diacritic marks in the tag source - they don't pull through correctly: #721 Whilst your fix gets the tag source working it doesn't handle diacritic marks correctly. Thanks |
The format of the AllMusic page has clearly changed a little since it was written:
https://tanzu.vmware.com/content/blog/xpath-css-class-matching