Actually fix block element rendering

Sources/HTMLStreamer/AttributedStringConverter.swift

@@ -17,7 +17,7 @@ private typealias PlatformFont = UIFont
 private typealias PlatformFont = NSFont
 #endif
 
-public struct AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
+public struct AttributedStringConverter<Callbacks: HTMLConversionCallbacks>: BlockRenderer {
     private let configuration: AttributedStringConverterConfiguration
     private var fontCache: [FontTrait: PlatformFont] = [:]
     
@@ -26,7 +26,7 @@ public struct AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
     
     private var actionStack: [ElementAction] = []
     private var styleStack: [Style] = []
-    private var blockState = BlockState.unstarted
+    var blockState = BlockState.start
     private var currentElementIsEmpty = true
     private var previouslyFinishedListItem = false
     // The current run of text w/o styles changing
@@ -46,7 +46,7 @@ public struct AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
         
         actionStack = []
         styleStack = []
-        blockState = .unstarted
+        blockState = .start
         currentElementIsEmpty = true
         previouslyFinishedListItem = false
         currentRun = ""
@@ -126,26 +126,28 @@ public struct AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
             finishRun()
             styleStack.append(.monospace)
         case "pre":
-            startBlockIfNecessary()
+            startOrFinishBlock()
             finishRun()
             styleStack.append(.monospace)
         case "blockquote":
-            startBlockIfNecessary()
+            startOrFinishBlock()
             finishRun()
             styleStack.append(.blockquote)
         case "p":
-            startBlockIfNecessary()
+            startOrFinishBlock()
         case "ol":
-            startBlockIfNecessary()
+            startOrFinishBlock()
             finishRun()
             styleStack.append(.orderedList(nextElementOrdinal: 1))
         case "ul":
-            startBlockIfNecessary()
+            startOrFinishBlock()
             finishRun()
             styleStack.append(.unorderedList)
         case "li":
             if previouslyFinishedListItem {
                 currentRun.append("\n")
+            } else {
+                continueBlock()
             }
             let marker: String
             if case .orderedList(let nextElementOrdinal) = styleStack.last {
@@ -184,22 +186,22 @@ public struct AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
         case "pre":
             finishRun()
             removeLastStyle(.monospace)
-            finishBlockElement()
+            startOrFinishBlock()
         case "blockquote":
             finishRun()
             removeLastStyle(.blockquote)
-            finishBlockElement()
+            startOrFinishBlock()
         case "p":
-            finishBlockElement()
+            startOrFinishBlock()
         case "ol":
             finishRun()
             removeLastStyle(.orderedList)
-            finishBlockElement()
+            startOrFinishBlock()
             previouslyFinishedListItem = false
         case "ul":
             finishRun()
             removeLastStyle(.unorderedList)
-            finishBlockElement()
+            startOrFinishBlock()
             previouslyFinishedListItem = false
         case "li":
             finishRun()
@@ -209,42 +211,8 @@ public struct AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
         }
     }
     
-    private mutating func startBlockIfNecessary() {
+    mutating func insertBlockBreak() {
-        switch blockState {
-        case .unstarted:
-            blockState = .started(false)
-        case .started:
-            break
-        case .ongoing:
         currentRun.append("\n\n")
-            blockState = .started(true)
-        case .finished(let nonEmpty):
-            if nonEmpty {
-                currentRun.append("\n\n")
-            }
-            blockState = .started(nonEmpty)
-        }
-    }
-    
-    private mutating func continueBlock() {
-        switch blockState {
-        case .unstarted, .started(_):
-            blockState = .ongoing
-        case .ongoing:
-            break
-        case .finished(let nonEmpty):
-            if nonEmpty {
-                currentRun.append("\n\n")
-            }
-            blockState = .ongoing
-        }
-    }
-    
-    private mutating func finishBlockElement() {
-        if blockState == .started(true) && currentElementIsEmpty {
-            currentRun.removeLast(2)
-        }
-        blockState = .finished(blockState == .ongoing)
     }
     
     // Finds the last currently-open style of the given type.
@@ -452,13 +420,6 @@ private enum Style {
     }
 }
 
-enum BlockState: Equatable {
-    case unstarted
-    case started(Bool)
-    case ongoing
-    case finished(Bool)
-}
-
 extension Collection where Element == Attribute {
     public func attributeValue(for name: String) -> String? {
         first(where: { $0.name == name })?.value

Sources/HTMLStreamer/BlockState.swift

@@ -0,0 +1,45 @@
+//
+//  BlockState.swift
+//  HTMLStreamer
+//
+//  Created by Shadowfacts on 2/14/24.
+//
+
+import Foundation
+
+protocol BlockRenderer {
+    var blockState: BlockState { get set }
+    mutating func insertBlockBreak()
+}
+
+extension BlockRenderer {
+    mutating func startOrFinishBlock() {
+        switch blockState {
+        case .start:
+            break
+        case .nonEmptyBlock:
+            blockState = .emptyBlock
+        case .emptyBlock:
+            break
+        }
+    }
+    
+    mutating func continueBlock() {
+        switch blockState {
+        case .start:
+            blockState = .nonEmptyBlock
+        case .nonEmptyBlock:
+            break
+        case .emptyBlock:
+            insertBlockBreak()
+            blockState = .nonEmptyBlock
+        }
+    }
+    
+}
+
+enum BlockState: Equatable {
+    case start
+    case nonEmptyBlock
+    case emptyBlock
+}

Sources/HTMLStreamer/TextConverter.swift

@@ -7,7 +7,7 @@
 
 import Foundation
 
-public struct TextConverter<Callbacks: HTMLConversionCallbacks> {
+public struct TextConverter<Callbacks: HTMLConversionCallbacks>: BlockRenderer {
     
     private let configuration: TextConverterConfiguration
     
@@ -15,7 +15,7 @@ public struct TextConverter<Callbacks: HTMLConversionCallbacks> {
     private var str: String!
     
     private var actionStack: [ElementAction] = []
-    private var blockState = BlockState.unstarted
+    var blockState = BlockState.start
     private var currentElementIsEmpty = true
     private var currentRun = ""
     
@@ -31,7 +31,7 @@ public struct TextConverter<Callbacks: HTMLConversionCallbacks> {
         tokenizer = Tokenizer(chars: html.unicodeScalars.makeIterator())
         str = ""
         
-        blockState = .unstarted
+        blockState = .start
         currentElementIsEmpty = true
         currentRun = ""
         
@@ -80,7 +80,7 @@ public struct TextConverter<Callbacks: HTMLConversionCallbacks> {
                 currentRun.append(" ")
             }
         case "pre", "blockquote", "p", "ol", "ul":
-            startBlockIfNecessary()
+            startOrFinishBlock()
         default:
             break
         }
@@ -89,65 +89,19 @@ public struct TextConverter<Callbacks: HTMLConversionCallbacks> {
     private mutating func handleEndTag(_ name: String) {
         switch name {
         case "pre", "blockquote", "p", "ol", "ul":
-            finishBlockElement()
+            startOrFinishBlock()
             finishRun()
         default:
             break
         }
     }
     
-    private mutating func startBlockIfNecessary() {
+    mutating func insertBlockBreak() {
-        switch blockState {
-        case .unstarted:
-            blockState = .started(false)
-        case .started:
-            break
-        case .ongoing:
         if configuration.insertNewlines {
             currentRun.append("\n\n")
         } else {
             currentRun.append(" ")
         }
-            blockState = .started(true)
-        case .finished(let nonEmpty):
-            if nonEmpty {
-                if configuration.insertNewlines {
-                    currentRun.append("\n\n")
-                } else {
-                    currentRun.append(" ")
-                }
-            }
-            blockState = .started(nonEmpty)
-        }
-    }
-    
-    private mutating func continueBlock() {
-        switch blockState {
-        case .unstarted, .started(_):
-            blockState = .ongoing
-        case .ongoing:
-            break
-        case .finished(let nonEmpty):
-            if nonEmpty {
-                if configuration.insertNewlines {
-                    currentRun.append("\n\n")
-                } else {
-                    currentRun.append(" ")
-                }
-            }
-            blockState = .ongoing
-        }
-    }
-    
-    private mutating func finishBlockElement() {
-        if blockState == .started(true) && currentElementIsEmpty {
-            if configuration.insertNewlines {
-                currentRun.removeLast(2)
-            } else {
-                currentRun.removeLast(1)
-            }
-        }
-        blockState = .finished(blockState == .ongoing)
     }
 
     private mutating func finishRun() {

Tests/HTMLStreamerTests/AttributedStringConverterTests.swift

@@ -206,20 +206,11 @@ final class AttributedStringConverterTests: XCTestCase {
     }
     
     func testMultipleBlockElements() {
-        let result = NSMutableAttributedString()
+        let result = NSAttributedString(string: "a\n\nb", attributes: [
-        result.append(NSAttributedString(string: "a", attributes: [
             .font: italicFont,
             .paragraphStyle: blockquoteParagraphStyle,
-        ]))
+            .foregroundColor: color,
-        result.append(NSAttributedString(string: "\n\n", attributes: [
+        ])
-            .font: font,
-            .paragraphStyle: NSParagraphStyle.default,
-        ]))
-        result.append(NSAttributedString(string: "b", attributes: [
-            .font: italicFont,
-            .paragraphStyle: blockquoteParagraphStyle,
-        ]))
-        result.addAttribute(.foregroundColor, value: color, range: NSRange(location: 0, length: result.length))
         XCTAssertEqual(convert("<blockquote>a</blockquote><blockquote>b</blockquote>"), result)
     }
     
@@ -321,12 +312,12 @@ final class AttributedStringConverterTests: XCTestCase {
     
     func testFollowedByList() {
         let result = NSMutableAttributedString()
-        result.append(NSAttributedString(string: "a\n\n", attributes: [
+        result.append(NSAttributedString(string: "a", attributes: [
             .font: font,
             .paragraphStyle: NSParagraphStyle.default,
             .foregroundColor: color,
         ]))
-        result.append(NSAttributedString(string: "\t1.\tb\n\t2.\tc", attributes: [
+        result.append(NSAttributedString(string: "\n\n\t1.\tb\n\t2.\tc", attributes: [
             .font: font,
             .paragraphStyle: listParagraphStyle,
             .foregroundColor: color,
@@ -356,4 +347,19 @@ final class AttributedStringConverterTests: XCTestCase {
         XCTAssertEqual(convert("</span>"), .init())
     }
     
+    func testWTF() {
+        let result = NSMutableAttributedString()
+        result.append(NSAttributedString(string: "a", attributes: [
+            .font: italicFont,
+            .paragraphStyle: blockquoteParagraphStyle,
+            .foregroundColor: color,
+        ]))
+        result.append(NSAttributedString(string: "\n\nb", attributes: [
+            .font: font,
+            .paragraphStyle: NSParagraphStyle.default,
+            .foregroundColor: color,
+        ]))
+        XCTAssertEqual(convert(#"<blockquote><p>a</p></blockquote><p>b</p>"#), result)
+    }
+    
 }