From b33be0f178d31dfb2a797ce66dc8ab7cc1db62d3 Mon Sep 17 00:00:00 2001
From: Shadowfacts <me@shadowfacts.net>
Date: Tue, 26 Dec 2023 14:29:40 -0500
Subject: [PATCH] Actually fix how whitespace is inserted between block
 elements

---
 .../AttributedStringConverter.swift           | 33 ++++++--------
 Sources/HTMLStreamer/TextConverter.swift      | 44 +++++++++----------
 .../AttributedStringConverterTests.swift      | 15 +++++++
 .../TextConverterTests.swift                  |  4 ++
 4 files changed, 53 insertions(+), 43 deletions(-)

diff --git a/Sources/HTMLStreamer/AttributedStringConverter.swift b/Sources/HTMLStreamer/AttributedStringConverter.swift
index a287c4d..9846804 100644
--- a/Sources/HTMLStreamer/AttributedStringConverter.swift
+++ b/Sources/HTMLStreamer/AttributedStringConverter.swift
@@ -27,6 +27,7 @@ public struct AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
     private var actionStack: [ElementAction] = []
     private var styleStack: [Style] = []
     private var previouslyFinishedBlockElement = false
+    private var currentElementIsEmpty = true
     // The current run of text w/o styles changing
     private var currentRun: String = ""
     
@@ -49,13 +50,19 @@ public struct AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
         while let token = tokenizer.next() {
             switch token {
             case .character(let c):
+                currentElementIsEmpty = false
+                previouslyFinishedBlockElement = false
                 currentRun.unicodeScalars.append(c)
             case .characterSequence(let s):
+                currentElementIsEmpty = false
+                previouslyFinishedBlockElement = false
                 currentRun.append(s)
             case .comment:
                 // ignored
                 continue
             case .startTag(let name, let selfClosing, let attributes):
+                currentElementIsEmpty = true
+                previouslyFinishedBlockElement = false
                 let action = Callbacks.elementAction(name: name, attributes: attributes)
                 actionStack.append(action)
                 handleStartTag(name, selfClosing: selfClosing, attributes: attributes)
@@ -72,6 +79,9 @@ public struct AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
             }
         }
         
+        if previouslyFinishedBlockElement {
+            currentRun.removeLast(2)
+        }
         finishRun()
         
         return str
@@ -111,21 +121,15 @@ public struct AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
             finishRun()
             styleStack.append(.monospace)
         case "pre":
-            startBlockElement()
             finishRun()
             styleStack.append(.monospace)
         case "blockquote":
-            startBlockElement()
             finishRun()
             styleStack.append(.blockquote)
-        case "p":
-            startBlockElement()
         case "ol":
-            startBlockElement()
             finishRun()
             styleStack.append(.orderedList(nextElementOrdinal: 1))
         case "ul":
-            startBlockElement()
             finishRun()
             styleStack.append(.unorderedList)
         case "li":
@@ -147,13 +151,6 @@ public struct AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
         }
     }
     
-    private mutating func startBlockElement() {
-        if str.length != 0 || !currentRun.isEmpty {
-            previouslyFinishedBlockElement = false
-            currentRun.append("\n\n")
-        }
-    }
-    
     private mutating func handleEndTag(_ name: String) {
         switch name {
         case "a":
@@ -181,6 +178,8 @@ public struct AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
             finishRun()
             removeLastStyle(.blockquote)
             finishBlockElement()
+        case "p":
+            finishBlockElement()
         case "ol":
             finishRun()
             removeLastStyle(.orderedList)
@@ -197,8 +196,9 @@ public struct AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
     }
     
     private mutating func finishBlockElement() {
-        if str.length != 0 {
+        if !currentElementIsEmpty {
             previouslyFinishedBlockElement = true
+            currentRun.append("\n\n")
         }
     }
     
@@ -251,11 +251,6 @@ public struct AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
             currentRun = replacement
         }
 
-        if previouslyFinishedBlockElement {
-            previouslyFinishedBlockElement = false
-            currentRun.insert(contentsOf: "\n\n", at: currentRun.startIndex)
-        }
-        
         var attributes = [NSAttributedString.Key: Any]()
         var paragraphStyle = configuration.paragraphStyle
         var currentFontTraits: FontTrait = []
diff --git a/Sources/HTMLStreamer/TextConverter.swift b/Sources/HTMLStreamer/TextConverter.swift
index 3f89151..1237625 100644
--- a/Sources/HTMLStreamer/TextConverter.swift
+++ b/Sources/HTMLStreamer/TextConverter.swift
@@ -16,6 +16,7 @@ public struct TextConverter<Callbacks: HTMLConversionCallbacks> {
     
     private var actionStack: [ElementAction] = []
     private var previouslyFinishedBlockElement = false
+    private var currentElementIsEmpty = true
     private var currentRun = ""
     
     public init(configuration: TextConverterConfiguration = .init()) where Callbacks == DefaultCallbacks {
@@ -33,10 +34,16 @@ public struct TextConverter<Callbacks: HTMLConversionCallbacks> {
         while let token = tokenizer.next() {
             switch token {
             case .character(let scalar):
+                currentElementIsEmpty = false
+                previouslyFinishedBlockElement = false
                 currentRun.unicodeScalars.append(scalar)
             case .characterSequence(let string):
+                currentElementIsEmpty = false
+                previouslyFinishedBlockElement = false
                 currentRun.append(string)
             case .startTag(let name, let selfClosing, let attributes):
+                currentElementIsEmpty = true
+                previouslyFinishedBlockElement = false
                 let action = Callbacks.elementAction(name: name, attributes: attributes)
                 actionStack.append(action)
                 handleStartTag(name, selfClosing: selfClosing, attributes: attributes)
@@ -51,6 +58,13 @@ public struct TextConverter<Callbacks: HTMLConversionCallbacks> {
             }
         }
         
+        if previouslyFinishedBlockElement {
+            if configuration.insertNewlines {
+                currentRun.removeLast(2)
+            } else {
+                currentRun.removeLast(1)
+            }
+        }
         finishRun()
         
         return str
@@ -64,25 +78,11 @@ public struct TextConverter<Callbacks: HTMLConversionCallbacks> {
             } else {
                 currentRun.append(" ")
             }
-        case "pre", "blockquote", "p", "ol", "ul":
-            startBlockElement()
-            finishRun()
         default:
             break
         }
     }
     
-    private mutating func startBlockElement() {
-        if !str.isEmpty {
-            previouslyFinishedBlockElement = false
-            if configuration.insertNewlines {
-                currentRun.append("\n\n")
-            } else {
-                currentRun.append(" ")
-            }
-        }
-    }
-    
     private mutating func handleEndTag(_ name: String) {
         switch name {
         case "pre", "blockquote", "p", "ol", "ul":
@@ -94,8 +94,13 @@ public struct TextConverter<Callbacks: HTMLConversionCallbacks> {
     }
     
     private mutating func finishBlockElement() {
-        if !str.isEmpty {
+        if !currentElementIsEmpty {
             previouslyFinishedBlockElement = true
+            if configuration.insertNewlines {
+                currentRun.append("\n\n")
+            } else {
+                currentRun.append(" ")
+            }
         }
     }
     
@@ -111,15 +116,6 @@ public struct TextConverter<Callbacks: HTMLConversionCallbacks> {
             currentRun = replacement
         }
         
-        if previouslyFinishedBlockElement {
-            previouslyFinishedBlockElement = false
-            if configuration.insertNewlines {
-                currentRun.insert(contentsOf: "\n\n", at: currentRun.startIndex)
-            } else {
-                currentRun.insert(" ", at: currentRun.startIndex)
-            }
-        }
-        
         str.append(currentRun)
         currentRun = ""
     }
diff --git a/Tests/HTMLStreamerTests/AttributedStringConverterTests.swift b/Tests/HTMLStreamerTests/AttributedStringConverterTests.swift
index 3cbc26b..041ae59 100644
--- a/Tests/HTMLStreamerTests/AttributedStringConverterTests.swift
+++ b/Tests/HTMLStreamerTests/AttributedStringConverterTests.swift
@@ -290,4 +290,19 @@ final class AttributedStringConverterTests: XCTestCase {
         ]))
     }
     
+    func testEmptyBlockElements() {
+        let result = NSMutableAttributedString()
+        result.append(NSAttributedString(string: "inside\nquote", attributes: [
+            .font: italicFont,
+            .paragraphStyle: blockquoteParagraphStyle,
+            .foregroundColor: color,
+        ]))
+        result.append(NSAttributedString(string: "\n\nafter", attributes: [
+            .font: font,
+            .paragraphStyle: NSParagraphStyle.default,
+            .foregroundColor: color,
+        ]))
+        XCTAssertEqual(convert("<p></p><blockquote><span>inside<br>quote</span></blockquote><span>after</span><p></p>"), result)
+    }
+    
 }
diff --git a/Tests/HTMLStreamerTests/TextConverterTests.swift b/Tests/HTMLStreamerTests/TextConverterTests.swift
index 485aa60..78faee4 100644
--- a/Tests/HTMLStreamerTests/TextConverterTests.swift
+++ b/Tests/HTMLStreamerTests/TextConverterTests.swift
@@ -63,4 +63,8 @@ final class TextConverterTests: XCTestCase {
         XCTAssertEqual(replaced, "…")
     }
     
+    func testEmptyBlockElements() {
+        XCTAssertEqual(convert("<p></p><blockquote><span>inside<br>quote</span></blockquote><span>after</span><p></p>"), "inside\nquote\n\nafter")
+    }
+    
 }