Merge pull request #865 from kiwix/853-snippets-using-libzim-only

Remove Fuzi, libxml2, Search snippets are using libzim only

Model/Brand.swift

@@ -63,7 +63,7 @@ enum Brand {
             // for custom apps, where we do not show this in settings, it should be disabled by default
             return .disabled
         }
-        return .firstSentence
+        return .matches
     }
 }
 

Support/br.lproj/Localizable.strings

@@ -177,5 +177,3 @@
 "enum.navigation_item.downloads" = "Pellgargadurioù";
 "enum.navigation_item.settings" = "Arventennoù";
 "enum.search_result_snippet_mode.disabled" = "Diweredekaet";
-"enum.search_result_snippet_mode.paragraph" = "Rannbennad kentañ";
-"enum.search_result_snippet_mode.sentence" = "Frazenn gentañ";

Support/dag.lproj/Localizable.strings

@@ -207,6 +207,4 @@
 "enum.navigation_item.downloads" = "Deebu";
 "enum.navigation_item.settings" = "Bintaɣindisi";
 "enum.search_result_snippet_mode.disabled" = "Zaɣisibu";
-"enum.search_result_snippet_mode.paragraph" = "Tuuli kuliga";
-"enum.search_result_snippet_mode.sentence" = "Tuuli yɛtɔɣili";
 "enum.search_result_snippet_mode.matches" = "Manchisi";

Support/de.lproj/Localizable.strings

@@ -236,6 +236,4 @@
 "enum.navigation_item.downloads" = "Downloads";
 "enum.navigation_item.settings" = "Einstellungen";
 "enum.search_result_snippet_mode.disabled" = "Deaktiviert";
-"enum.search_result_snippet_mode.paragraph" = "Erster Absatz";
-"enum.search_result_snippet_mode.sentence" = "Erster Satz";
 "enum.search_result_snippet_mode.matches" = "Treffer";

Support/en.lproj/Localizable.strings

@@ -276,6 +276,4 @@
 "enum.navigation_item.downloads" = "Downloads";
 "enum.navigation_item.settings" = "Settings";
 "enum.search_result_snippet_mode.disabled" = "Disabled";
-"enum.search_result_snippet_mode.paragraph" = "First Paragraph";
-"enum.search_result_snippet_mode.sentence" = "First Sentence";
 "enum.search_result_snippet_mode.matches" = "Matches";

Support/fr.lproj/Localizable.strings

@@ -238,6 +238,4 @@
 "enum.navigation_item.downloads" = "Téléchargements";
 "enum.navigation_item.settings" = "Paramètres";
 "enum.search_result_snippet_mode.disabled" = "Désactivé";
-"enum.search_result_snippet_mode.paragraph" = "Premier paragraphe";
-"enum.search_result_snippet_mode.sentence" = "Première phrase";
 "enum.search_result_snippet_mode.matches" = "Correspondances";

Support/ha.lproj/Localizable.strings

@@ -227,6 +227,4 @@
 "enum.navigation_item.downloads" = "Saukewa";
 "enum.navigation_item.settings" = "Saituttuka";
 "enum.search_result_snippet_mode.disabled" = "An kashe";
-"enum.search_result_snippet_mode.paragraph" = "Sakin Farko";
-"enum.search_result_snippet_mode.sentence" = "Hukuncin Farko";
 "enum.search_result_snippet_mode.matches" = "Matches";

Support/he.lproj/Localizable.strings

@@ -228,6 +228,4 @@
 "enum.navigation_item.downloads" = "הורדות";
 "enum.navigation_item.settings" = "הגדרות";
 "enum.search_result_snippet_mode.disabled" = "כבוי";
-"enum.search_result_snippet_mode.paragraph" = "פסקה ראשונה";
-"enum.search_result_snippet_mode.sentence" = "משפט ראשון";
 "enum.search_result_snippet_mode.matches" = "התאמות";

Support/ia.lproj/Localizable.strings

@@ -227,6 +227,4 @@
 "enum.navigation_item.downloads" = "Discargamentos";
 "enum.navigation_item.settings" = "Parametros";
 "enum.search_result_snippet_mode.disabled" = "Disactivate";
-"enum.search_result_snippet_mode.paragraph" = "Prime paragrapho";
-"enum.search_result_snippet_mode.sentence" = "Prime phrase";
 "enum.search_result_snippet_mode.matches" = "Correspondentias";

Support/ig.lproj/Localizable.strings

@@ -237,6 +237,4 @@
 "enum.navigation_item.downloads" = "Nbudata";
 "enum.navigation_item.settings" = "Ntọala";
 "enum.search_result_snippet_mode.disabled" = "Agbanyụrụ";
-"enum.search_result_snippet_mode.paragraph" = "Nkeji mbụ";
-"enum.search_result_snippet_mode.sentence" = "Ahịrịokwu mbụ";
 "enum.search_result_snippet_mode.matches" = "Ndị ndaba";

Support/igl.lproj/Localizable.strings

@@ -64,6 +64,4 @@
 "enum.navigation_item.downloads" = "Downloadu";
 "enum.navigation_item.settings" = "Eñoche";
 "enum.search_result_snippet_mode.disabled" = "Nyo ló";
-"enum.search_result_snippet_mode.paragraph" = "Óla ubia nei";
-"enum.search_result_snippet_mode.sentence" = "Óla ejio dudu";
 "enum.search_result_snippet_mode.matches" = "A chana";

Support/lb.lproj/Localizable.strings

@@ -168,6 +168,4 @@
 "enum.navigation_item.new" = "Nei";
 "enum.navigation_item.settings" = "Astellungen";
 "enum.search_result_snippet_mode.disabled" = "Desaktivéiert";
-"enum.search_result_snippet_mode.paragraph" = "Éischten Abschnitt";
-"enum.search_result_snippet_mode.sentence" = "Éischte Saz";
 "enum.search_result_snippet_mode.matches" = "Treffer";

Support/mk.lproj/Localizable.strings

@@ -228,6 +228,4 @@
 "enum.navigation_item.downloads" = "Преземања";
 "enum.navigation_item.settings" = "Нагодувања";
 "enum.search_result_snippet_mode.disabled" = "Оневозможено";
-"enum.search_result_snippet_mode.paragraph" = "Прв пасус";
-"enum.search_result_snippet_mode.sentence" = "Прва реченица";
 "enum.search_result_snippet_mode.matches" = "Совпаѓања";

Support/pl.lproj/Localizable.strings

@@ -167,5 +167,3 @@
 "enum.navigation_item.downloads" = "Pobierz";
 "enum.navigation_item.settings" = "Ustawienia";
 "enum.search_result_snippet_mode.disabled" = "Wyłączony";
-"enum.search_result_snippet_mode.paragraph" = "Pierwszy paragraf";
-"enum.search_result_snippet_mode.sentence" = "Pierwsze zdanie";

Support/qqq.lproj/Localizable.strings

@@ -224,7 +224,5 @@
 "enum.navigation_item.new" = "Side navigation menu item title";
 "enum.navigation_item.downloads" = "Side navigation menu item title";
 "enum.navigation_item.settings" = "Side navigation menu item title";
-"enum.search_result_snippet_mode.disabled" = "Picker option title in settings. Search snippet: Disabled | First Paragraph | First Sentence | Matches";
-"enum.search_result_snippet_mode.paragraph" = "Picker option title in settings. Search snippet: Disabled | First Paragraph | First Sentence | Matches";
-"enum.search_result_snippet_mode.sentence" = "Picker option title in settings. Search snippet: Disabled | First Paragraph | First Sentence | Matches";
-"enum.search_result_snippet_mode.matches" = "Picker option title in settings. Search snippet: Disabled | First Paragraph | First Sentence | Matches";
+"enum.search_result_snippet_mode.disabled" = "Picker option title in settings. Search snippet: Disabled | Matches";
+"enum.search_result_snippet_mode.matches" = "Picker option title in settings. Search snippet: Disabled | Matches";

Support/ru.lproj/Localizable.strings

@@ -233,6 +233,4 @@
 "enum.navigation_item.downloads" = "Скачать";
 "enum.navigation_item.settings" = "Настройки";
 "enum.search_result_snippet_mode.disabled" = "Выключен";
-"enum.search_result_snippet_mode.paragraph" = "Формат абзаца";
-"enum.search_result_snippet_mode.sentence" = "Первое предложение";
 "enum.search_result_snippet_mode.matches" = "Совпадения";

Support/skr-arab.lproj/Localizable.strings

@@ -131,5 +131,3 @@
 "enum.navigation_item.downloads" = "ڈاؤن لوڈ";
 "enum.navigation_item.settings" = "ترتیباں";
 "enum.search_result_snippet_mode.disabled" = "غیر فعال تھی ڳیا";
-"enum.search_result_snippet_mode.paragraph" = "پہلا پیرا";
-"enum.search_result_snippet_mode.sentence" = "پہلا جملہ";

Support/sl.lproj/Localizable.strings

@@ -225,6 +225,4 @@
 "enum.navigation_item.downloads" = "Prenosi";
 "enum.navigation_item.settings" = "Nastavitve";
 "enum.search_result_snippet_mode.disabled" = "Onemogočeno";
-"enum.search_result_snippet_mode.paragraph" = "Prvi odstavek";
-"enum.search_result_snippet_mode.sentence" = "Prvi stavek";
 "enum.search_result_snippet_mode.matches" = "Ujemanja";

Support/sv.lproj/Localizable.strings

@@ -231,6 +231,4 @@
 "enum.navigation_item.downloads" = "Nedladdningar";
 "enum.navigation_item.settings" = "Inställningar";
 "enum.search_result_snippet_mode.disabled" = "Inaktiverad";
-"enum.search_result_snippet_mode.paragraph" = "Första stycket";
-"enum.search_result_snippet_mode.sentence" = "Första meningen";
 "enum.search_result_snippet_mode.matches" = "Matchningar";

Support/tr.lproj/Localizable.strings

@@ -233,6 +233,4 @@
 "enum.navigation_item.downloads" = "İndirilenler";
 "enum.navigation_item.settings" = "Ayarlar";
 "enum.search_result_snippet_mode.disabled" = "Devre dışı";
-"enum.search_result_snippet_mode.paragraph" = "İlk Paragraf";
-"enum.search_result_snippet_mode.sentence" = "İlk Cümle";
 "enum.search_result_snippet_mode.matches" = "Eşleşmeler";

Support/zh-hans.lproj/Localizable.strings

@@ -230,6 +230,4 @@
 "enum.navigation_item.downloads" = "下载";
 "enum.navigation_item.settings" = "设置";
 "enum.search_result_snippet_mode.disabled" = "已停用";
-"enum.search_result_snippet_mode.paragraph" = "第一段";
-"enum.search_result_snippet_mode.sentence" = "第一句";
 "enum.search_result_snippet_mode.matches" = "相符";

Support/zh-hant.lproj/Localizable.strings

@@ -229,6 +229,4 @@
 "enum.navigation_item.downloads" = "下載";
 "enum.navigation_item.settings" = "設定";
 "enum.search_result_snippet_mode.disabled" = "已停用";
-"enum.search_result_snippet_mode.paragraph" = "第一段";
-"enum.search_result_snippet_mode.sentence" = "第一句";
 "enum.search_result_snippet_mode.matches" = "相符";

SwiftUI/Model/Enum.swift

@@ -271,8 +271,13 @@ enum NavigationItem: Hashable, Identifiable {
     }
 }
 
+/// Note: The cases were reduced from:
+/// `case disabled, firstParagraph, firstSentence, matches`
+/// which (due to enum values) accurately migrating our users, as of our intents
+/// DO NOT change the order of the cases, as that might cause migration problems from version 3.4.0
+/// see: https://github.com/kiwix/kiwix-apple/issues/853
 enum SearchResultSnippetMode: String, CaseIterable, Identifiable, Defaults.Serializable {
-    case disabled, firstParagraph, firstSentence, matches
+    case disabled, matches
 
     var id: String { rawValue }
 
@@ -280,10 +285,6 @@ enum SearchResultSnippetMode: String, CaseIterable, Identifiable, Defaults.Seria
         switch self {
         case .disabled:
             return "enum.search_result_snippet_mode.disabled".localized
-        case .firstParagraph:
-            return "enum.search_result_snippet_mode.paragraph".localized
-        case .firstSentence:
-            return "enum.search_result_snippet_mode.sentence".localized
         case .matches:
             return "enum.search_result_snippet_mode.matches".localized
         }

SwiftUI/Model/HTMLParser.swift

@@ -1,108 +0,0 @@
-// This file is part of Kiwix for iOS & macOS.
-//
-// Kiwix is free software; you can redistribute it and/or modify it
-// under the terms of the GNU General Public License as published by
-// the Free Software Foundation; either version 3 of the License, or
-// any later version.
-//
-// Kiwix is distributed in the hope that it will be useful, but
-// WITHOUT ANY WARRANTY; without even the implied warranty of
-// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
-// General Public License for more details.
-//
-// You should have received a copy of the GNU General Public License
-// along with Kiwix; If not, see https://www.gnu.org/licenses/.
-
-import CoreLocation
-import NaturalLanguage
-#if os(macOS)
-import AppKit
-#elseif os(iOS)
-import UIKit
-#endif
-
-import Fuzi
-
-class HTMLParser {
-    private let document: HTMLDocument
-
-    #if os(macOS)
-    static private let boldFont = NSFont.systemFont(ofSize: 12.0, weight: .medium)
-    #elseif os(iOS)
-    static private let boldFont = UIFont.systemFont(ofSize: 12.0, weight: .medium)
-    #endif
-
-    init(document: HTMLDocument) {
-        self.document = document
-    }
-
-    convenience init(html: String) throws {
-        self.init(document: try HTMLDocument(string: html))
-    }
-
-    convenience init(data: Data) throws {
-        self.init(document: try HTMLDocument(data: data))
-    }
-
-    convenience init(url: URL) throws {
-        guard let content = ZimFileService.shared.getURLContent(url: url) else { throw NSError() }
-        try self.init(data: content.data)
-    }
-
-    var title: String? { document.title }
-
-    func getFirstParagraph() -> NSAttributedString? {
-        guard let firstParagraph = document.firstChild(xpath: "//p") else { return nil }
-        let snippet = NSMutableAttributedString()
-        for child in firstParagraph.childNodes(ofTypes: [.Text, .Element]) {
-            if let element = child as? Fuzi.XMLElement, element.attributes["class"]?.contains("mw-ref") == true {
-                continue
-            } else if let element = child as? Fuzi.XMLElement {
-                let attributedSting = NSAttributedString(
-                    string: element.stringValue.replacingOccurrences(of: "\n", with: ""),
-                    attributes: element.tag == "b" ? [.font: HTMLParser.boldFont] : nil
-                )
-                snippet.append(attributedSting)
-            } else {
-                let text = child.stringValue.replacingOccurrences(of: "\n", with: "")
-                snippet.append(NSAttributedString(string: text))
-            }
-        }
-        return snippet.length > 0 ? snippet : nil
-    }
-
-    func getFirstSentence(languageCode: String?) -> NSAttributedString? {
-        guard let firstParagraph = self.getFirstParagraph() else { return nil }
-        let text = firstParagraph.string
-        var firstSentence: NSAttributedString?
-
-        let tokenizer = NLTokenizer(unit: .sentence)
-        if let languageCode = languageCode {tokenizer.setLanguage(NLLanguage(languageCode))}
-        tokenizer.string = text
-        tokenizer.enumerateTokens(in: text.startIndex..<text.endIndex) { range, _ in
-            firstSentence = firstParagraph.attributedSubstring(from: NSRange(range, in: firstParagraph.string))
-            return false
-        }
-        return firstSentence
-    }
-
-    func getFirstImagePath() -> String? {
-        guard let firstImage = document.firstChild(xpath: "//img") else { return nil }
-        return firstImage.attributes["src"]
-    }
-
-//    func getGeoCoordinate() -> CLLocationCoordinate2D? {
-//        do {
-//            let elements = try document.select("head > meta[name='geo.position']")
-//            let content = try elements.first()?.attr("content")
-//            guard let parts = content?.split(separator: ";"), parts.count == 2,
-//                let lat = Double(parts[0]), let lon = Double(parts[1]) else { return nil }
-//            return CLLocationCoordinate2D(latitude: lat, longitude: lon)
-//        } catch { return nil }
-//    }
-
-    class func parseBodyFragment(_ bodyFragment: String) -> NSAttributedString? {
-        let html = "<!DOCTYPE html><html><head></head><body><p>\(bodyFragment)</p></body></html>"
-        return (try? HTMLParser(html: html))?.getFirstParagraph()
-    }
-}

SwiftUI/Model/SearchOperation/SearchOperation.swift

@@ -34,15 +34,14 @@ extension SearchOperation {
                 guard !self.isCancelled else { return }
 
                 switch snippetMode {
-                case .firstParagraph:
-                    guard let parser = try? HTMLParser(url: result.url) else { return }
-                    result.snippet = parser.getFirstParagraph()
-                case .firstSentence:
-                    guard let parser = try? HTMLParser(url: result.url) else { return }
-                    result.snippet = parser.getFirstSentence(languageCode: nil)
                 case .matches:
-                    guard let html = result.htmlSnippet else { return }
-                    result.snippet = HTMLParser.parseBodyFragment(html)
+                    guard let html = result.htmlSnippet,
+                          let data = html.data(using: .utf8) else { return }
+                    result.snippet = try? NSAttributedString(
+                        data: data,
+                        options: [.documentType: NSAttributedString.DocumentType.html],
+                        documentAttributes: nil
+                    )
                 case .disabled:
                     break
                 }

project.yml

@@ -45,9 +45,6 @@ settings:
       VALIDATE_PRODUCT: YES
 
 packages:
-  Fuzi:
-    url: https://github.com/cezheng/Fuzi
-    majorVersion: 3.0.0
   Defaults:
     url: https://github.com/sindresorhus/Defaults
     majorVersion: 6.0.0
@@ -75,7 +72,6 @@ targetTemplates:
       - sdk: WebKit.framework
       - sdk: NotificationCenter.framework
       - sdk: QuickLook.framework
-      - package: Fuzi
       - package: Defaults
       - package: OrderedCollections
     sources: