orangecoding / fredy

:heart: Fredy - [F]ind [R]eal [E]states [D]amn Eas[y] - Fredy will constantly search for new listings on sites like Immoscout or Immowelt and send new results to you, so that you can focus on more important things in life ;)
http://www.orange-coding.net
MIT License
231 stars 58 forks source link

crawler for immowelt is grabbing to much #105

Closed qudiqudi closed 2 days ago

qudiqudi commented 1 month ago

Describe the bug immowelt shows listings outside the filter criteria below the matched listings. Seperated by an info line:

Screenshot 2024-08-10 at 13 41 59

Fredy, grabs these listings below that line as potential findings for the assigned job, which is incorrect. It's mostly outside the defined search radius.

Looks like it should stop at the alertBoxSeparator div class:

Screenshot 2024-08-10 at 13 52 58
orangecoding commented 1 month ago

Immowelt is fucked up in general. Their source code is horribly broken at significant places. Whoever is maintains their websites, I often times thought their only purpose in life is giving me headaches ;)

I'll see what I can do..

orangecoding commented 1 month ago

Hey @qudiqudi my latest change should mitigate the issue. Plz check it out, but be aware of the breaking change (See Release Notes)

If it works, please close the issue

qudiqudi commented 1 month ago

Unfortunately, it still grabs listings below the divider.

Screenshot 2024-09-06 at 09 06 57

And it looks like they A/B testing the website layout, yesterday it looked different.

orangecoding commented 1 month ago

This is the reason why I said they are fucked up. They change their source with every internal build. Bad for the user (caching) but even worst for us

orangecoding commented 1 month ago

Im on vacation now. Will check afterwards. Maybe it has settled by then... ;)

qudiqudi commented 6 days ago

Fredy is still ignoring the "Weitere Ergebnisse" divider with immowelt.

orangecoding commented 3 days ago

Yeah. I'm currently busy with family related stuff. As soon as I can I will jump on it. But how about you give it a shot yourself and create a pr? ;)

orangecoding commented 2 days ago

@qudiqudi That does fix the issue. https://github.com/orangecoding/fredy/commit/6d2ab5f9581cd689446501304c0cdbcd803b72a3

Next time a pr would be nice. Thanks ;)