|
| 1 | +import Foundation |
| 2 | +import SwiftSoup |
| 3 | +import WebKit |
| 4 | +import WebScrapper |
| 5 | + |
| 6 | +struct AppleDocumentationSearchService: SearchService { |
| 7 | + func search(query: String) async throws -> WebSearchResult { |
| 8 | + let queryEncoded = query |
| 9 | + .addingPercentEncoding(withAllowedCharacters: .urlQueryAllowed) ?? "" |
| 10 | + let url = URL(string: "https://developer.apple.com/search/?q=\(queryEncoded)")! |
| 11 | + |
| 12 | + let scrapper = await WebScrapper() |
| 13 | + let html = try await scrapper.fetch(url: url) { document in |
| 14 | + DeveloperDotAppleResultParser.validate(document: document) |
| 15 | + } |
| 16 | + |
| 17 | + return try DeveloperDotAppleResultParser.parse(html: html) |
| 18 | + } |
| 19 | +} |
| 20 | + |
| 21 | +enum DeveloperDotAppleResultParser { |
| 22 | + static func validate(document: SwiftSoup.Document) -> Bool { |
| 23 | + guard let _ = try? document.select("ul.search-results").first |
| 24 | + else { return false } |
| 25 | + return true |
| 26 | + } |
| 27 | + |
| 28 | + static func parse(html: String) throws -> WebSearchResult { |
| 29 | + let document = try SwiftSoup.parse(html) |
| 30 | + let searchResult = try? document.select("ul.search-results").first |
| 31 | + |
| 32 | + guard let searchResult else { return .init(webPages: []) } |
| 33 | + |
| 34 | + var results: [WebSearchResult.WebPage] = [] |
| 35 | + for element in searchResult.children() { |
| 36 | + if let titleElement = try? element.select("p.result-title"), |
| 37 | + let link = try? titleElement.select("a").attr("href"), |
| 38 | + !link.isEmpty |
| 39 | + { |
| 40 | + let title = (try? titleElement.text()) ?? "" |
| 41 | + let snippet = (try? element.select("p.result-description").text()) |
| 42 | + ?? (try? element.select("ul.breadcrumb-list").text()) |
| 43 | + ?? "" |
| 44 | + results.append(WebSearchResult.WebPage( |
| 45 | + urlString: { |
| 46 | + if link.hasPrefix("/") { |
| 47 | + return "https://developer.apple.com\(link)" |
| 48 | + } |
| 49 | + return link |
| 50 | + }(), |
| 51 | + title: title, |
| 52 | + snippet: snippet |
| 53 | + )) |
| 54 | + } |
| 55 | + } |
| 56 | + |
| 57 | + return WebSearchResult(webPages: results) |
| 58 | + } |
| 59 | +} |
| 60 | + |
0 commit comments