Handle block elements better (again)

Sources/HTMLStreamer/AttributedStringConverter.swift

@@ -26,7 +26,7 @@ public struct AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
     
     private var actionStack: [ElementAction] = []
     private var styleStack: [Style] = []
-    private var previouslyFinishedBlockElement = false
+    private var blockState = BlockState.unstarted
     private var currentElementIsEmpty = true
     private var previouslyFinishedListItem = false
     // The current run of text w/o styles changing
@@ -46,26 +46,26 @@ public struct AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
         
         actionStack = []
         styleStack = []
-        previouslyFinishedBlockElement = false
+        blockState = .unstarted
         currentElementIsEmpty = true
+        previouslyFinishedListItem = false
         currentRun = ""
         
         while let token = tokenizer.next() {
             switch token {
             case .character(let c):
                 currentElementIsEmpty = false
-                previouslyFinishedBlockElement = false
+                continueBlock()
                 currentRun.unicodeScalars.append(c)
             case .characterSequence(let s):
                 currentElementIsEmpty = false
-                previouslyFinishedBlockElement = false
+                continueBlock()
                 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)
@@ -82,9 +82,6 @@ public struct AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
             }
         }
         
-        if previouslyFinishedBlockElement {
-            currentRun.removeLast(2)
-        }
         finishRun()
         
         return str
@@ -124,15 +121,21 @@ public struct AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
             finishRun()
             styleStack.append(.monospace)
         case "pre":
+            startBlockIfNecessary()
             finishRun()
             styleStack.append(.monospace)
         case "blockquote":
+            startBlockIfNecessary()
             finishRun()
             styleStack.append(.blockquote)
+        case "p":
+            startBlockIfNecessary()
         case "ol":
+            startBlockIfNecessary()
             finishRun()
             styleStack.append(.orderedList(nextElementOrdinal: 1))
         case "ul":
+            startBlockIfNecessary()
             finishRun()
             styleStack.append(.unorderedList)
         case "li":
@@ -201,13 +204,44 @@ public struct AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
         }
     }
     
-    private mutating func finishBlockElement() {
+    private mutating func startBlockIfNecessary() {
-        if !currentElementIsEmpty {
+        switch blockState {
-            previouslyFinishedBlockElement = true
+        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.
     // We can't just use the last one because we need to handle mis-nested tags.
     private mutating func removeLastStyle(_ type: Style.StyleType) {
@@ -413,6 +447,13 @@ 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/TextConverter.swift

@@ -15,7 +15,7 @@ public struct TextConverter<Callbacks: HTMLConversionCallbacks> {
     private var str: String!
     
     private var actionStack: [ElementAction] = []
-    private var previouslyFinishedBlockElement = false
+    private var blockState = BlockState.unstarted
     private var currentElementIsEmpty = true
     private var currentRun = ""
     
@@ -31,7 +31,7 @@ public struct TextConverter<Callbacks: HTMLConversionCallbacks> {
         tokenizer = Tokenizer(chars: html.unicodeScalars.makeIterator())
         str = ""
         
-        previouslyFinishedBlockElement = false
+        blockState = .unstarted
         currentElementIsEmpty = true
         currentRun = ""
         
@@ -39,15 +39,14 @@ public struct TextConverter<Callbacks: HTMLConversionCallbacks> {
             switch token {
             case .character(let scalar):
                 currentElementIsEmpty = false
-                previouslyFinishedBlockElement = false
+                continueBlock()
                 currentRun.unicodeScalars.append(scalar)
             case .characterSequence(let string):
                 currentElementIsEmpty = false
-                previouslyFinishedBlockElement = false
+                continueBlock()
                 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)
@@ -62,13 +61,6 @@ public struct TextConverter<Callbacks: HTMLConversionCallbacks> {
             }
         }
         
-        if previouslyFinishedBlockElement {
-            if configuration.insertNewlines {
-                currentRun.removeLast(2)
-            } else {
-                currentRun.removeLast(1)
-            }
-        }
         finishRun()
         
         return str
@@ -82,6 +74,8 @@ public struct TextConverter<Callbacks: HTMLConversionCallbacks> {
             } else {
                 currentRun.append(" ")
             }
+        case "pre", "blockquote", "p", "ol", "ul":
+            startBlockIfNecessary()
         default:
             break
         }
@@ -90,24 +84,67 @@ public struct TextConverter<Callbacks: HTMLConversionCallbacks> {
     private mutating func handleEndTag(_ name: String) {
         switch name {
         case "pre", "blockquote", "p", "ol", "ul":
-            finishRun()
             finishBlockElement()
+            finishRun()
         default:
             break
         }
     }
     
-    private mutating func finishBlockElement() {
+    private mutating func startBlockIfNecessary() {
-        if !currentElementIsEmpty {
+        switch blockState {
-            previouslyFinishedBlockElement = true
+        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() {
         if actionStack.contains(.skip) {
             currentRun = ""

Tests/HTMLStreamerTests/AttributedStringConverterTests.swift

@@ -319,7 +319,7 @@ final class AttributedStringConverterTests: XCTestCase {
         XCTAssertEqual(convert("<p></p><blockquote><span>inside<br>quote</span></blockquote><span>after</span><p></p>"), result)
     }
     
-    func testParagraphFollowedByList() {
+    func testFollowedByList() {
         let result = NSMutableAttributedString()
         result.append(NSAttributedString(string: "a\n\n", attributes: [
             .font: font,
@@ -332,6 +332,8 @@ final class AttributedStringConverterTests: XCTestCase {
             .foregroundColor: color,
         ]))
         XCTAssertEqual(convert("<p>a</p><ol><li>b</li><li>c</li></ol>"), result)
+        XCTAssertEqual(convert("<span>a</span><ol><li>b</li><li>c</li></ol>"), result)
+        XCTAssertEqual(convert("a<ol><li>b</li><li>c</li></ol>"), result)
     }
     
 }