Yellow Pages

YellowPages through HasData

Two tools. hasdata_yellowpages_search_getSearchResults needs keyword and location and returns about thirty businesses a page. hasdata_yellowpages_place_getPlaceDetails opens one of them by its url.

sort takes default, distance, averageRating or name. domain takes the literal hostnames www.yellowpages.com or www.yellowpages.ca, so a country code is a rejected call.

The page is not the market

searchInformation.totalResults reports what the search actually found, and it runs far ahead of the thirty rows in front of you. Quote it alongside the page count, because reporting the page as the market understates it by an order of magnitude.

Ratings are the sparse field

A row reliably carries title, url, phone, categories, country and position. The address parts, workingHours, openState and website are there on most rows.

rating and reviews sit on a minority of listings, so a shortlist ranked by rating silently drops most of the page. Rank on what is present, say how many rows carried a rating, and treat an unrated business as unrated rather than as poor. reviewText carries a review snippet on some rows, which answers a quality question without opening anything, and badges marks the rest.

The detail call is grouped, and it loses fields

The place tool answers with overview, ratings, details, reviews and images rather than one flat object.

overview holds the title, url, phone, website, city, region, zipcode, country and breadcrumbs. It has no street line and no workingHours, both of which the search row already gave you, so carry those forward rather than expecting the detail call to repeat them.

Reviews carry author, authorUrl, text, helpfulCount, a date and a ratings object holding overall plus sometimes expertise, professionalism and promptness.

Review dates are written year, day, month, so 2019-24-07 is July 2019. Reading them as ISO produces impossible months and a nonsense ordering, so use the feed's own order for recency unless you parse them deliberately.

Paging

The request takes page as a number. The response reports where you are under pagination with currentPage, perPage, totalPages, nextPageUrl and otherPageUrls, so those URLs are for reading rather than values to send back. Take the next page by incrementing page.

Each successful call spends credits, and opening a business is a call of its own, so open the shortlist rather than the page.