Need to Scrap some data

Find the RESTAURANTS (R'S) with home delivery service, 1.- Spanish cities of 20.k to 50k inhabitants. I provide a list of the cities, which are 280. In these cities there are no aggregators or marketplaces, type jusiteat. You have to log out of google, for example the first 20 entries and that the bot generates the positives and negatives depending on the entries are R's at home or not. To do this, we must first generate some "decision rules" so that the bot does not generate false positives and false negatives. To learn, I need the bot to carve a city where I know the R's at home, and teach the bot to choose well. There will be a series of keywords to collect in each url, Once this is done, scan the 280 cities. So that Google does not make us, it must be done little by little, that it is scribbling at its own pace, for example XR's a day and that it is doing. I estimate around 1,000 R's that we can find. 1.- google sheet with list of cities. Those with 20-50 k inhabitants, selected. 2- If city = selected, the first 20 google entries are sculpted 3- According to some "Selection Rules", Positive is assigned = the url is from R's to Sun. Negatives = the url is not from R's to Sun. 4- With the first cities, I will refine the Selection Rules 5- When I discriminate well enough, scribble all the cities.