Reader: Add support for <br> tags in excerpt generator

Modules/Sources/WordPressKitModels/NSString+Summary.swift

@@ -5,76 +5,19 @@
 /// and convert HTML into plain text.
 ///
 extension NSString {
-
-    static let PostDerivedSummaryLength = 150
-
     /// Create a summary for the post based on the post's content.
     ///
     /// - Returns: A summary for the post.
     ///
     @objc
     public func wpkit_summarized() -> String {
-        let characterSet = CharacterSet(charactersIn: "\n")
-
-        return (self as String).strippingGutenbergContentForExcerpt()
-            .strippingShortcodes()
-            .makePlainText()
-            .trimmingCharacters(in: characterSet)
-            .wp_stringByEllipsizing(withMaxLength: NSString.PostDerivedSummaryLength, preserveWords: true)
+        GutenbergExcerptGenerator.firstParagraph(from: (self as String))
     }
-}
 
-private extension String {
-    func makePlainText() -> String {
-        let characterSet = NSCharacterSet.whitespacesAndNewlines
-
-        return self.strippingHTML()
+    @objc
+    public func wpkit_makePlainText() -> String {
+        self.strippingHTML()
             .stringByDecodingXMLCharacters()
-            .trimmingCharacters(in: characterSet)
-    }
-
-    /// Creates a new string by stripping all shortcodes from this string.
-    ///
-    func strippingShortcodes() -> String {
-        let pattern = "\\[[^\\]]+\\]"
-
-        return removingMatches(pattern: pattern, options: .caseInsensitive)
-    }
-
-    /// This method is the main entry point to generate excerpts for Gutenberg content.
-    ///
-    func strippingGutenbergContentForExcerpt() -> String {
-        return strippingGutenbergGalleries().strippingGutenbergVideoPress()
-    }
-
-    /// Strips Gutenberg galleries from strings.
-    ///
-    func strippingGutenbergGalleries() -> String {
-        let pattern = "(?s)<!--\\swp:gallery?(.*?)wp:gallery\\s-->"
-
-        return removingMatches(pattern: pattern, options: .caseInsensitive)
-    }
-
-    /// Strips VideoPress references from Gutenberg VideoPress and Video blocks.
-    ///
-    func strippingGutenbergVideoPress() -> String {
-        let pattern = "(?s)\n?<!--\\swp:video.*?(.*?)wp:video.*?\\s-->"
-
-        return removingMatches(pattern: pattern, options: .caseInsensitive)
-    }
-
-    /// Creates a new string by removing all matches of the specified regex.
-    ///
-    func removingMatches(pattern: String, options: NSRegularExpression.Options = []) -> String {
-        let range = NSRange(location: 0, length: self.utf16.count)
-        let regex: NSRegularExpression
-
-        do {
-            regex = try NSRegularExpression(pattern: pattern, options: options)
-        } catch {
-            return self
-        }
-
-        return regex.stringByReplacingMatches(in: self, options: .reportCompletion, range: range, withTemplate: "")
+            .trimmingCharacters(in: .whitespacesAndNewlines)
     }
 }

Modules/Sources/WordPressKitObjC/RemoteReaderPost.m

@@ -707,7 +707,7 @@ - (NSString *)createSummaryFromContent:(NSString *)string
  */
 - (NSString *)makePlainText:(NSString *)string
 {
-    return [string wpkit_summarized];
+    return [string wpkit_makePlainText];
 }
 
 /**

Modules/Sources/WordPressShared/Utility/GutenbergExcerptGenerator.swift

@@ -14,10 +14,10 @@ public struct GutenbergExcerptGenerator {
             return ""
         }
 
-        // Extract content
+        // Extract content while convering  <br>, <br/>, <br /> to newlines first
         let rawText = String(content[tagEnd.upperBound..<pEnd.lowerBound])
+            .replacingOccurrences(of: "<br\\s*/?>", with: " ", options: .regularExpression)
 
-        // Remove HTML tags AND shortcodes in one pass
         let range = NSRange(rawText.startIndex..., in: rawText)
         let text = (regex?.stringByReplacingMatches(in: rawText, options: [], range: range, withTemplate: "") ?? rawText)
             .stringByDecodingXMLCharacters()

Modules/Tests/WordPressSharedTests/GutenbergExcerptGeneratorTests.swift

@@ -14,21 +14,29 @@ struct GutenbergPostExcerptGeneratorTests {
         let content = "<!-- wp:gallery {\"ids\":[2315,2309,2308]} --><figure class=\"wp-block-gallery columns-3 is-cropped\"><ul class=\"blocks-gallery-grid\"><li class=\"blocks-gallery-item\"><figure><img src=\"https://diegotest4.files.wordpress.com/2020/01/img_0005-1-1.jpg\" data-id=\"2315\" class=\"wp-image-2315\"/><figcaption class=\"blocks-gallery-item__caption\">Asdasdasd</figcaption></figure></li><li class=\"blocks-gallery-item\"><figure><img src=\"https://diegotest4.files.wordpress.com/2020/01/img_0111-1-1.jpg\" data-id=\"2309\" class=\"wp-image-2309\"/><figcaption class=\"blocks-gallery-item__caption\">Asdasdasd</figcaption></figure></li><li class=\"blocks-gallery-item\"><figure><img src=\"https://diegotest4.files.wordpress.com/2020/01/img_0004-1.jpg\" data-id=\"2308\" class=\"wp-image-2308\"/><figcaption class=\"blocks-gallery-item__caption\">Adsasdasdasd</figcaption></figure></li></ul></figure><!-- /wp:gallery --><p>Some Content</p>"
 
         let summary = GutenbergExcerptGenerator.firstParagraph(from: content, maxLength: 150)
-        #expect(summary == "Some Content")
+        #expect(summary == "Some Content…")
     }
 
     @Test func summaryForContentWithGallery2() {
         let content = "<p>Before</p>\n<!-- wp:gallery {\"ids\":[2315,2309,2308]} --><figure class=\"wp-block-gallery columns-3 is-cropped\"><ul class=\"blocks-gallery-grid\"><li class=\"blocks-gallery-item\"><figure><img src=\"https://diegotest4.files.wordpress.com/2020/01/img_0005-1-1.jpg\" data-id=\"2315\" class=\"wp-image-2315\"/><figcaption class=\"blocks-gallery-item__caption\">Asdasdasd</figcaption></figure></li><li class=\"blocks-gallery-item\"><figure><img src=\"https://diegotest4.files.wordpress.com/2020/01/img_0111-1-1.jpg\" data-id=\"2309\" class=\"wp-image-2309\"/><figcaption class=\"blocks-gallery-item__caption\">Asdasdasd</figcaption></figure></li><li class=\"blocks-gallery-item\"><figure><img src=\"https://diegotest4.files.wordpress.com/2020/01/img_0004-1.jpg\" data-id=\"2308\" class=\"wp-image-2308\"/><figcaption class=\"blocks-gallery-item__caption\">Adsasdasdasd</figcaption></figure></li></ul></figure><!-- /wp:gallery --><p>After</p>"
 
         let summary = GutenbergExcerptGenerator.firstParagraph(from: content, maxLength: 150)
-        #expect(summary == "Before")
+        #expect(summary == "Before…")
     }
 
     @Test
     func testVideoPressBlock() {
         let content = "<p>Before</p>\n<!-- wp:videopress/video {\"title\":\"demo\",\"description\":\"\",\"id\":5297,\"guid\":\"AbCDe\",\"videoRatio\":56.333333333333336,\"privacySetting\":2,\"allowDownload\":false,\"rating\":\"G\",\"isPrivate\":true,\"duration\":1673} -->\n<figure class=\"wp-block-videopress-video wp-block-jetpack-videopress jetpack-videopress-player\"><div class=\"jetpack-videopress-player__wrapper\">\nhttps://videopress.com/v/AbCDe?resizeToParent=true&amp;cover=true&amp;preloadContent=metadata&amp;useAverageColor=true\n</div></figure>\n<!-- /wp:videopress/video -->\n<p>After</p>"
 
         let summary = GutenbergExcerptGenerator.firstParagraph(from: content, maxLength: 150)
-        #expect(summary == "Before")
+        #expect(summary == "Before…")
+    }
+
+    @Test func testPostWithBRTags() {
+        let content = #"<p class="wp-block-paragraph">Yes,<br>look behind<br>in remembrance and with gratitude.</p><p class="wp-block-paragraph">Then,<br>stay present,<br>or miss memories in the making.</p>"#
+
+        let summary = GutenbergExcerptGenerator.firstParagraph(from: content, maxLength: 150)
+        print(summary)
+        #expect(summary == "Yes, look behind in remembrance and with gratitude.")
     }
 }