apify
Return an error if HTML parsing of results is broken
When Google changed the layout, we had an issue that we were returning invalid results to our users rather than failing (see https://github.com/apify/actor-google-search-scraper/issues/71). We should have some sanity checks in the code to ensure we're parsing the HTML correctly and then fail and return an error to the user, rather than returning an invalid result.
I've discussed this issue with potential customers and this was quite important for them.
This is hard to do correctly. What means "broken"? If paid results are missing but organic are correct, is it broken? What about vice versa? Etc. Isn't it easier for the caller to do this rudimentary check on their side?
Just for context: Google usually changes only a few selectors so it breaks only parts of the parser.
What should happen is that we should have a test running often and a maintainer that fixes this within hours. (both is in progress)
But it would be good to hear it from the customers.
Well, the test could check that either organic results are present, or there is an info message that no results were found. That would make it fairly bulletproof, what do you think?
Of course, having regular tests would be best...