From 95edb408fad0f48f3ac19e202429261c16f15372 Mon Sep 17 00:00:00 2001
From: Shadowfacts <me@shadowfacts.net>
Date: Sun, 17 Mar 2024 10:56:49 -0400
Subject: [PATCH] Fix replace callback with nested element, fix skip/replace at
 beginning of block skipping block break

Fixes shadowfacts/Tusker#461
---
 .../AttributedStringConverter.swift           | 29 +++++++----
 .../HTMLConversionCallbacks.swift             |  8 ---
 Sources/HTMLStreamer/TextConverter.swift      | 29 +++++++----
 .../AttributedStringConverterTests.swift      | 50 +++++++++++++++++++
 4 files changed, 90 insertions(+), 26 deletions(-)

diff --git a/Sources/HTMLStreamer/AttributedStringConverter.swift b/Sources/HTMLStreamer/AttributedStringConverter.swift
index f9dd781..393e6d0 100644
--- a/Sources/HTMLStreamer/AttributedStringConverter.swift
+++ b/Sources/HTMLStreamer/AttributedStringConverter.swift
@@ -24,7 +24,19 @@ public class AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
     private var tokenizer: Tokenizer<String.UnicodeScalarView.Iterator>!
     private var str: NSMutableAttributedString!
     
-    private var actionStack: [ElementAction] = []
+    private var actionStack: [ElementAction] = [] {
+        didSet {
+            hasSkipOrReplaceElementAction = actionStack.contains(where: {
+                switch $0 {
+                case .skip, .replace(_):
+                    true
+                default:
+                    false
+                }
+            })
+        }
+    }
+    private var hasSkipOrReplaceElementAction = false
     private var styleStack: [Style] = []
     private var blockStateMachine = BlockStateMachine(blockBreak: "", lineBreak: "", listIndentForContentOutsideItem: "", append: { _ in }, removeChar: {})
     private var currentElementIsEmpty = true
@@ -59,13 +71,15 @@ public class AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
             switch token {
             case .character(let c):
                 currentElementIsEmpty = false
-                if blockStateMachine.continueBlock(char: c) {
+                if blockStateMachine.continueBlock(char: c),
+                   !hasSkipOrReplaceElementAction {
                     currentRun.unicodeScalars.append(c)
                 }
             case .characterSequence(let s):
                 currentElementIsEmpty = false
                 for c in s.unicodeScalars {
-                    if blockStateMachine.continueBlock(char: c) {
+                    if blockStateMachine.continueBlock(char: c),
+                       !hasSkipOrReplaceElementAction {
                         currentRun.unicodeScalars.append(c)
                     }
                 }
@@ -281,13 +295,10 @@ public class AttributedStringConverter<Callbacks: HTMLConversionCallbacks> {
     }()
     
     private func finishRun() {
-        if actionStack.contains(.skip) {
-            currentRun = ""
-            return
-        } else if case .append(let s) = actionStack.last {
+        if case .append(let s) = actionStack.last {
             currentRun.append(s)
-        } else if case .replace(let replacement) = actionStack.first(where: \.isReplace) {
-            currentRun = replacement
+        } else if case .replace(let replacement) = actionStack.last {
+            currentRun.append(replacement)
         }
         
         guard !currentRun.isEmpty else {
diff --git a/Sources/HTMLStreamer/HTMLConversionCallbacks.swift b/Sources/HTMLStreamer/HTMLConversionCallbacks.swift
index a2db8ab..6e55898 100644
--- a/Sources/HTMLStreamer/HTMLConversionCallbacks.swift
+++ b/Sources/HTMLStreamer/HTMLConversionCallbacks.swift
@@ -17,14 +17,6 @@ public enum ElementAction: Equatable {
     case skip
     case replace(String)
     case append(String)
-    
-    var isReplace: Bool {
-        if case .replace(_) = self {
-            true
-        } else {
-            false
-        }
-    }
 }
 
 public extension HTMLConversionCallbacks {
diff --git a/Sources/HTMLStreamer/TextConverter.swift b/Sources/HTMLStreamer/TextConverter.swift
index 267bb1b..f446c62 100644
--- a/Sources/HTMLStreamer/TextConverter.swift
+++ b/Sources/HTMLStreamer/TextConverter.swift
@@ -13,7 +13,19 @@ public class TextConverter<Callbacks: HTMLConversionCallbacks> {
     private var tokenizer: Tokenizer<String.UnicodeScalarView.Iterator>!
     private var str: String!
     
-    private var actionStack: [ElementAction] = []
+    private var actionStack: [ElementAction] = [] {
+        didSet {
+            hasSkipOrReplaceElementAction = actionStack.contains(where: {
+                switch $0 {
+                case .skip, .replace(_):
+                    true
+                default:
+                    false
+                }
+            })
+        }
+    }
+    private var hasSkipOrReplaceElementAction = false
     var blockStateMachine = BlockStateMachine(blockBreak: "", lineBreak: "", listIndentForContentOutsideItem: "", append: { _ in }, removeChar: {})
     private var currentElementIsEmpty = true
     private var currentRun = ""
@@ -46,13 +58,15 @@ public class TextConverter<Callbacks: HTMLConversionCallbacks> {
             switch token {
             case .character(let scalar):
                 currentElementIsEmpty = false
-                if blockStateMachine.continueBlock(char: scalar) {
+                if blockStateMachine.continueBlock(char: scalar),
+                   !hasSkipOrReplaceElementAction {
                     currentRun.unicodeScalars.append(scalar)
                 }
             case .characterSequence(let string):
                 currentElementIsEmpty = false
                 for c in string.unicodeScalars {
-                    if blockStateMachine.continueBlock(char: c) {
+                    if blockStateMachine.continueBlock(char: c),
+                       !hasSkipOrReplaceElementAction {
                         currentRun.unicodeScalars.append(c)
                     }
                 }
@@ -137,13 +151,10 @@ public class TextConverter<Callbacks: HTMLConversionCallbacks> {
     }
     
     private func finishRun() {
-        if actionStack.contains(.skip) {
-            currentRun = ""
-            return
-        } else if case .append(let s) = actionStack.last {
+        if case .append(let s) = actionStack.last {
             currentRun.append(s)
-        } else if case .replace(let replacement) = actionStack.first(where: \.isReplace) {
-            currentRun = replacement
+        } else if case .replace(let replacement) = actionStack.last {
+            currentRun.append(replacement)
         }
         
         guard !currentRun.isEmpty else {
diff --git a/Tests/HTMLStreamerTests/AttributedStringConverterTests.swift b/Tests/HTMLStreamerTests/AttributedStringConverterTests.swift
index 9a5eb3d..f1e2829 100644
--- a/Tests/HTMLStreamerTests/AttributedStringConverterTests.swift
+++ b/Tests/HTMLStreamerTests/AttributedStringConverterTests.swift
@@ -264,6 +264,12 @@ final class AttributedStringConverterTests: XCTestCase {
             .paragraphStyle: NSParagraphStyle.default,
             .foregroundColor: color,
         ]))
+        let replaceNested = convert("<span class='replace'><b>a</b></span>", callbacks: Callbacks.self)
+        XCTAssertEqual(replaceNested, NSAttributedString(string: "…", attributes: [
+            .font: font,
+            .paragraphStyle: NSParagraphStyle.default,
+            .foregroundColor: color,
+        ]))
         let appended = convert("<span class='append'>test</span>", callbacks: Callbacks.self)
         XCTAssertEqual(appended, NSAttributedString(string: "test…", attributes: [
             .font: font,
@@ -612,4 +618,48 @@ final class AttributedStringConverterTests: XCTestCase {
         XCTAssertEqual(convert("<ol><li><pre>a</pre></li></ol>"), result)
     }
     
+    func testInvisibleAtBeginningOfParagraphDoesNotPreventParagraphBreak() {
+        struct Invisible: HTMLConversionCallbacks {
+            static func elementAction(name: String, attributes: [Attribute]) -> ElementAction {
+                if attributes.attributeValue(for: "class") == "invisible" {
+                    .skip
+                } else {
+                    .default
+                }
+            }
+        }
+        
+        let result = NSAttributedString(string: "a\n\nc", attributes: [
+            .font: font,
+            .paragraphStyle: NSParagraphStyle.default,
+            .foregroundColor: color,
+        ])
+        let html = """
+<p>a</p><p><span class="invisible">b</span><span class="ellipsis">c</span></p>
+"""
+        XCTAssertEqual(convert(html, callbacks: Invisible.self), result)
+    }
+    
+    func testReplaceAtBeginningOfParagraphDoesNotPreventParagraphBreak() {
+        struct Replace: HTMLConversionCallbacks {
+            static func elementAction(name: String, attributes: [Attribute]) -> ElementAction {
+                if attributes.attributeValue(for: "class") == "replace" {
+                    .replace("c")
+                } else {
+                    .default
+                }
+            }
+        }
+        
+        let result = NSAttributedString(string: "a\n\nc", attributes: [
+            .font: font,
+            .paragraphStyle: NSParagraphStyle.default,
+            .foregroundColor: color,
+        ])
+        let html = """
+<p>a</p><p><span class="replace">b</span></p>
+"""
+        XCTAssertEqual(convert(html, callbacks: Replace.self), result)
+    }
+    
 }