From 38d57b3f795871e787d93f2d459fabb5dd00468d Mon Sep 17 00:00:00 2001
From: Shadowfacts <me@shadowfacts.net>
Date: Fri, 24 Nov 2023 21:59:30 -0500
Subject: [PATCH] Attributed string conversion

---
 Package.swift                                 |   4 +
 .../AttributedStringConverter.swift           | 320 ++++++++++++++++++
 .../AttributedStringConverterTests.swift      | 195 +++++++++++
 3 files changed, 519 insertions(+)
 create mode 100644 Sources/HTMLStreamer/AttributedStringConverter.swift
 create mode 100644 Tests/HTMLStreamerTests/AttributedStringConverterTests.swift

diff --git a/Package.swift b/Package.swift
index f2a208b..d261095 100644
--- a/Package.swift
+++ b/Package.swift
@@ -5,6 +5,10 @@ import PackageDescription
 
 let package = Package(
     name: "HTMLStreamer",
+    platforms: [
+        .iOS(.v13),
+        .macOS(.v10_15),
+    ],
     products: [
         // Products define the executables and libraries a package produces, making them visible to other packages.
         .library(
diff --git a/Sources/HTMLStreamer/AttributedStringConverter.swift b/Sources/HTMLStreamer/AttributedStringConverter.swift
new file mode 100644
index 0000000..8459f3d
--- /dev/null
+++ b/Sources/HTMLStreamer/AttributedStringConverter.swift
@@ -0,0 +1,320 @@
+//
+//  AttributedStringConverter.swift
+//  HTMLStreamer
+//
+//  Created by Shadowfacts on 11/24/23.
+//
+
+#if os(iOS)
+import UIKit
+#elseif os(macOS)
+import AppKit
+#endif
+
+struct AttributedStringConverter<Callbacks: AttributedStringCallbacks> {
+    private let configuration: AttributedStringConverterConfiguration
+    private var tokenizer: Tokenizer<String.Iterator>
+    private let str = NSMutableAttributedString()
+    
+    private var actionStack: InlineArray3<ElementAction> = []
+    private var styleStack: InlineArray3<Style> = []
+    // The current run of text w/o styles changing
+    private var currentRun: String = ""
+    
+    init(html: String, configuration: AttributedStringConverterConfiguration) where Callbacks == DefaultCallbacks {
+        self.init(html: html, configuration: configuration, callbacks: DefaultCallbacks.self)
+    }
+    
+    init(html: String, configuration: AttributedStringConverterConfiguration, callbacks _: Callbacks.Type = Callbacks.self) {
+        self.configuration = configuration
+        self.tokenizer = Tokenizer(chars: html.makeIterator())
+    }
+    
+    mutating func convert() -> NSAttributedString {
+        while let token = tokenizer.next() {
+            switch token {
+            case .character(let c):
+                currentRun.append(c)
+            case .comment:
+                // ignored
+                continue
+            case .startTag(let name, let selfClosing, let attributes):
+                let action = Callbacks.elementAction(name: name, attributes: attributes)
+                actionStack.append(action)
+                // self closing tags are ignored since they have no content
+                if !selfClosing {
+                    handleStartTag(name, attributes: attributes)
+                }
+            case .endTag(let name):
+                handleEndTag(name)
+                // if we have a non-default action for the current element, the run finishes here
+                if actionStack.last != .default {
+                    finishRun()
+                }
+                actionStack.removeLast()
+            case .doctype:
+                // ignored
+                continue
+            }
+        }
+        
+        finishRun()
+        
+        return str
+    }
+    
+    private mutating func handleStartTag(_ name: String, attributes: InlineArray3<HTMLStreamer.Attribute>) {
+        switch name {
+        case "br":
+            currentRun.append("\n")
+        case "a":
+            // we need to always insert in attribute, because we need to always have one
+            // to remove from the stack in handleEndTag
+            // but we only need to finish the run if we have a URL, since otherwise
+            // the final attribute run won't be affected
+            let url = attributes.attributeValue(for: "href").flatMap(Callbacks.makeURL(string:))
+            if url != nil {
+                finishRun()
+            }
+            styleStack.append(.link(url))
+        case "em", "i":
+            finishRun()
+            styleStack.append(.italic)
+        case "strong", "b":
+            finishRun()
+            styleStack.append(.bold)
+        case "del":
+            finishRun()
+            styleStack.append(.strikethrough)
+        case "code":
+            finishRun()
+            styleStack.append(.monospace)
+        case "pre":
+            startBlockElement()
+            finishRun()
+            styleStack.append(.monospace)
+        case "blockquote":
+            startBlockElement()
+            finishRun()
+            styleStack.append(.blockquote)
+        case "p":
+            startBlockElement()
+        default:
+            break
+        }
+    }
+    
+    private mutating func startBlockElement() {
+        if str.length != 0 || !currentRun.isEmpty {
+            currentRun.append("\n\n")
+        }
+    }
+    
+    private mutating func handleEndTag(_ name: String) {
+        switch name {
+        case "a":
+            if case .link(.some(_)) = styleStack.last {
+                finishRun()
+            }
+            removeLastStyle(.link)
+        case "em", "i":
+            finishRun()
+            removeLastStyle(.italic)
+        case "strong", "b":
+            finishRun()
+            removeLastStyle(.bold)
+        case "del":
+            finishRun()
+            removeLastStyle(.strikethrough)
+        case "code":
+            finishRun()
+            removeLastStyle(.monospace)
+        case "pre":
+            finishRun()
+            removeLastStyle(.monospace)
+        case "blockquote":
+            finishRun()
+            removeLastStyle(.blockquote)
+        default:
+            break
+        }
+    }
+    
+    // needed to correctly handle mis-nested tags
+    private mutating func removeLastStyle(_ type: Style.StyleType) {
+        var i = styleStack.index(before: styleStack.endIndex)
+        while i >= styleStack.startIndex {
+            if styleStack[i].type == type {
+                styleStack.remove(at: i)
+                return
+            }
+            styleStack.formIndex(before: &i)
+        }
+    }
+    
+    private lazy var blockquoteParagraphStyle: NSParagraphStyle = {
+        let style = configuration.paragraphStyle.mutableCopy() as! NSMutableParagraphStyle
+        style.headIndent = 32
+        style.firstLineHeadIndent = 32
+        return style
+    }()
+    
+    private mutating func finishRun() {
+        guard !currentRun.isEmpty else {
+            return
+        }
+        
+        if actionStack.contains(.skip) {
+            currentRun = ""
+            return
+        } else if case .replace(let replacement) = actionStack.first(where: \.isReplace) {
+            currentRun = replacement
+        }
+        
+        var attributes = [NSAttributedString.Key: Any]()
+        var currentFontTraits = Set<FontTrait>()
+        for style in styleStack {
+            switch style {
+            case .bold:
+                currentFontTraits.insert(.bold)
+            case .italic:
+                currentFontTraits.insert(.italic)
+            case .monospace:
+                currentFontTraits.insert(.monospace)
+            case .link(let url):
+                if let url {
+                    attributes[.link] = url
+                }
+            case .strikethrough:
+                attributes[.strikethroughStyle] = NSUnderlineStyle.single.rawValue
+            case .blockquote:
+                attributes[.paragraphStyle] = blockquoteParagraphStyle
+                currentFontTraits.insert(.italic)
+            }
+        }
+        
+        let baseFont = currentFontTraits.contains(.monospace) ? configuration.monospaceFont : configuration.font
+        var descriptor = baseFont.fontDescriptor
+        if currentFontTraits.contains(.bold) && currentFontTraits.contains(.italic),
+           let boldItalic = descriptor.withSymbolicTraits([.traitBold, .traitItalic]) {
+            descriptor = boldItalic
+        } else if currentFontTraits.contains(.bold),
+           let bold = descriptor.withSymbolicTraits(.traitBold) {
+            descriptor = bold
+        } else if currentFontTraits.contains(.italic),
+           let italic = descriptor.withSymbolicTraits(.traitItalic) {
+            descriptor = italic
+        }
+        #if os(iOS)
+        attributes[.font] = UIFont(descriptor: descriptor, size: 0)
+        #elseif os(macOS)
+        attributes[.font] = NSFont(descriptor: descriptor, size: 0)
+        #endif
+        
+        if !attributes.keys.contains(.paragraphStyle) {
+            attributes[.paragraphStyle] = configuration.paragraphStyle
+        }
+        
+        str.append(NSAttributedString(string: currentRun, attributes: attributes))
+        currentRun = ""
+    }
+}
+
+protocol AttributedStringCallbacks {
+    static func makeURL(string: String) -> URL?
+    static func elementAction(name: String, attributes: InlineArray3<Attribute>) -> ElementAction
+}
+
+enum ElementAction: Equatable {
+    case `default`
+    case skip
+    case replace(String)
+    
+    var isReplace: Bool {
+        if case .replace(_) = self {
+            true
+        } else {
+            false
+        }
+    }
+}
+
+extension AttributedStringCallbacks {
+    static func makeURL(string: String) -> URL? {
+        URL(string: string)
+    }
+    static func elementAction(name: String, attributes: InlineArray3<Attribute>) -> ElementAction {
+        .default
+    }
+}
+
+struct DefaultCallbacks: AttributedStringCallbacks {
+}
+
+struct AttributedStringConverterConfiguration {
+    #if os(iOS)
+    var font: UIFont
+    var monospaceFont: UIFont
+    var color: UIColor
+    #elseif os(macOS)
+    var font: NSFont
+    var monospaceFont: NSFont
+    var color: NSColor
+    #endif
+    var paragraphStyle: NSParagraphStyle
+}
+
+#if os(macOS)
+private extension NSFontDescriptor {
+    func withSymbolicTraits(_ traits: SymbolicTraits) -> NSFontDescriptor? {
+        let descriptor: NSFontDescriptor = self.withSymbolicTraits(traits)
+        return descriptor
+    }
+}
+private extension NSFontDescriptor.SymbolicTraits {
+    static var traitBold: Self { .bold }
+    static var traitItalic: Self { .italic }
+}
+#endif
+
+private enum FontTrait {
+    case bold
+    case italic
+    case monospace
+}
+
+private enum Style {
+    case bold
+    case italic
+    case monospace
+    case link(URL?)
+    case strikethrough
+    case blockquote
+    
+    var type: StyleType {
+        switch self {
+        case .bold:
+            return .bold
+        case .italic:
+            return .italic
+        case .monospace:
+            return .monospace
+        case .link(_):
+            return .link
+        case .strikethrough:
+            return .strikethrough
+        case .blockquote:
+            return .blockquote
+        }
+    }
+    
+    enum StyleType {
+        case bold, italic, monospace, link, strikethrough, blockquote
+    }
+}
+
+extension Collection where Element == Attribute {
+    func attributeValue(for name: String) -> String? {
+        first(where: { $0.name == name })?.value
+    }
+}
diff --git a/Tests/HTMLStreamerTests/AttributedStringConverterTests.swift b/Tests/HTMLStreamerTests/AttributedStringConverterTests.swift
new file mode 100644
index 0000000..73dbd89
--- /dev/null
+++ b/Tests/HTMLStreamerTests/AttributedStringConverterTests.swift
@@ -0,0 +1,195 @@
+//
+//  AttributedStringConverterTests.swift
+//  
+//
+//  Created by Shadowfacts on 11/24/23.
+//
+
+import XCTest
+@testable import HTMLStreamer
+
+final class AttributedStringConverterTests: XCTestCase {
+    
+    #if os(iOS)
+    private let font = UIFont.systemFont(ofSize: 13)
+    private let monospaceFont = UIFont.monospacedSystemFont(ofSize: 13, weight: .regular)
+    #elseif os(macOS)
+    private let font = NSFont.systemFont(ofSize: 13)
+    private lazy var italicFont = NSFont(descriptor: font.fontDescriptor.withSymbolicTraits(.italic), size: 13)!
+    private lazy var boldFont = NSFont(descriptor: font.fontDescriptor.withSymbolicTraits(.bold), size: 13)!
+    private lazy var boldItalicFont = NSFont(descriptor: font.fontDescriptor.withSymbolicTraits([.bold, .italic]), size: 13)!
+    private let monospaceFont = NSFont.monospacedSystemFont(ofSize: 13, weight: .regular)
+    #endif
+    private let blockquoteParagraphStyle: NSParagraphStyle = {
+        let style = NSParagraphStyle.default.mutableCopy() as! NSMutableParagraphStyle
+        style.headIndent = 32
+        style.firstLineHeadIndent = 32
+        return style
+    }()
+    
+    private func convert(_ html: String) -> NSAttributedString {
+        convert(html, callbacks: DefaultCallbacks.self)
+    }
+    
+    private func convert<Callbacks: AttributedStringCallbacks>(_ html: String, callbacks _: Callbacks.Type = Callbacks.self) -> NSAttributedString {
+        let config = AttributedStringConverterConfiguration(
+            font: font,
+            monospaceFont: monospaceFont,
+            color: .black,
+            paragraphStyle: .default
+        )
+        var converter = AttributedStringConverter<Callbacks>(html: html, configuration: config)
+        return converter.convert()
+    }
+
+    func testConvertBR() {
+        XCTAssertEqual(convert("a<br>b"), NSAttributedString(string: "a\nb", attributes: [
+            .font: font,
+            .paragraphStyle: NSParagraphStyle.default,
+        ]))
+    }
+    
+    func testConvertA() {
+        XCTAssertEqual(convert("<a href='https://example.com'>link</a>"), NSAttributedString(string: "link", attributes: [
+            .font: font,
+            .paragraphStyle: NSParagraphStyle.default,
+            .link: URL(string: "https://example.com")!,
+        ]))
+        XCTAssertEqual(convert("<a>link</a>"), NSAttributedString(string: "link", attributes: [
+            .font: font,
+            .paragraphStyle: NSParagraphStyle.default,
+        ]))
+    }
+    
+    func testConvertP() {
+        XCTAssertEqual(convert("<p>a</p><p>b</p>"), NSAttributedString(string: "a\n\nb", attributes: [
+            .font: font,
+            .paragraphStyle: NSParagraphStyle.default,
+        ]))
+    }
+    
+    func testConvertEm() {
+        XCTAssertEqual(convert("<em>hello</em>"), NSAttributedString(string: "hello", attributes: [
+            .font: italicFont,
+            .paragraphStyle: NSParagraphStyle.default,
+        ]))
+        XCTAssertEqual(convert("<i>hello</i>"), NSAttributedString(string: "hello", attributes: [
+            .font: italicFont,
+            .paragraphStyle: NSParagraphStyle.default,
+        ]))
+    }
+    
+    func testConvertStrong() {
+        XCTAssertEqual(convert("<strong>hello</strong>"), NSAttributedString(string: "hello", attributes: [
+            .font: boldFont,
+            .paragraphStyle: NSParagraphStyle.default,
+        ]))
+        XCTAssertEqual(convert("<b>hello</b>"), NSAttributedString(string: "hello", attributes: [
+            .font: boldFont,
+            .paragraphStyle: NSParagraphStyle.default,
+        ]))
+    }
+    
+    func testConvertBoldItalic() {
+        XCTAssertEqual(convert("<strong><em>hello</em></strong>"), NSAttributedString(string: "hello", attributes: [
+            .font: boldItalicFont,
+            .paragraphStyle: NSParagraphStyle.default,
+        ]))
+    }
+    
+    func testIncorrectNesting() {
+        let result = NSMutableAttributedString()
+        result.append(NSAttributedString(string: "bold ", attributes: [
+            .font: boldFont,
+        ]))
+        result.append(NSAttributedString(string: "both", attributes: [
+            .font: boldItalicFont,
+        ]))
+        result.append(NSAttributedString(string: " italic", attributes: [
+            .font: italicFont,
+        ]))
+        result.addAttribute(.paragraphStyle, value: NSParagraphStyle.default, range: NSRange(location: 0, length: result.length))
+        XCTAssertEqual(convert("<strong>bold <em>both</strong> italic</em>"), result)
+    }
+    
+    func testDel() {
+        XCTAssertEqual(convert("<del>blah</del>"), NSAttributedString(string: "blah", attributes: [
+            .font: font,
+            .paragraphStyle: NSParagraphStyle.default,
+            .strikethroughStyle: NSUnderlineStyle.single.rawValue,
+        ]))
+    }
+    
+    func testCode() {
+        XCTAssertEqual(convert("<code>wee</code>"), NSAttributedString(string: "wee", attributes: [
+            .font: monospaceFont,
+            .paragraphStyle: NSParagraphStyle.default,
+        ]))
+    }
+    
+    func testPre() {
+        XCTAssertEqual(convert("<pre>wee</pre>"), NSAttributedString(string: "wee", attributes: [
+            .font: monospaceFont,
+            .paragraphStyle: NSParagraphStyle.default,
+        ]))
+    }
+    
+    func testBlockquote() {
+        XCTAssertEqual(convert("<blockquote>hello</blockquote>"), NSAttributedString(string: "hello", attributes: [
+            .font: italicFont,
+            .paragraphStyle: blockquoteParagraphStyle,
+        ]))
+        XCTAssertEqual(convert("<blockquote><b>hello</b></blockquote>"), NSAttributedString(string: "hello", attributes: [
+            .font: boldItalicFont,
+            .paragraphStyle: blockquoteParagraphStyle,
+        ]))
+    }
+    
+    func testSelfClosing() {
+        XCTAssertEqual(convert("<b />asdf"), NSAttributedString(string: "asdf", attributes: [
+            .font: font,
+            .paragraphStyle: NSParagraphStyle.default,
+        ]))
+    }
+    
+    func testMakeURLCallback() {
+        struct Callbacks: AttributedStringCallbacks {
+            static func makeURL(string: String) -> URL? {
+                URL(string: "https://apple.com")
+            }
+        }
+        let result = convert("<a href='https://example.com'>test</a>", callbacks: Callbacks.self)
+        XCTAssertEqual(result, NSAttributedString(string: "test", attributes: [
+            .font: font,
+            .paragraphStyle: NSParagraphStyle.default,
+            .link: URL(string: "https://apple.com")!,
+        ]))
+    }
+    
+    func testElementActionCallback() {
+        struct Callbacks: AttributedStringCallbacks {
+            static func elementAction(name: String, attributes: InlineArray3<Attribute>) -> ElementAction {
+                let clazz = attributes.attributeValue(for: "class")
+                if clazz == "invisible" {
+                    return .skip
+                } else if clazz == "ellipsis" {
+                    return .replace("…")
+                } else {
+                    return .default
+                }
+            }
+        }
+        let skipped = convert("<span class='invisible'>test</span>", callbacks: Callbacks.self)
+        XCTAssertEqual(skipped, NSAttributedString())
+        let skipNestped = convert("<span class='invisible'><b>test</b></span>", callbacks: Callbacks.self)
+        XCTAssertEqual(skipNestped, NSAttributedString())
+        let skipNestped2 = convert("<b><span class='invisible'>test</span></b>", callbacks: Callbacks.self)
+        XCTAssertEqual(skipNestped2, NSAttributedString())
+        let replaced = convert("<span class='ellipsis'>test</span>", callbacks: Callbacks.self)
+        XCTAssertEqual(replaced, NSAttributedString(string: "…", attributes: [
+            .font: font,
+            .paragraphStyle: NSParagraphStyle.default,
+        ]))
+    }
+    
+}