iOS NSAttributedString HTML Render 自行实现｜解决闪退与效能瓶颈

let data = "<div>Text</div>".data(using: .unicode)!
let attributed = try NSAttributedString(data: data, options: [.documentType: NSAttributedString.DocumentType.html, .characterEncoding: String.Encoding.utf8.rawValue], documentAttributes: nil)
let string = attributed.string

htmlString = "<div>Test</div>"
htmlString.replacingOccurrences(of: "<[^>]+>", with: "", options: .regularExpression, range: nil)

import UIKit
// Ref: https://github.com/malcommac/SwiftRichString
final class HTMLStripper: NSObject, XMLParserDelegate {

    private static let topTag = "source"
    private var xmlParser: XMLParser
    
    private(set) var storedString: String
    
    // The XML parser sometimes splits strings, which can break localization-sensitive
    // string transforms. Work around this by using the currentString variable to
    // accumulate partial strings, and then reading them back out as a single string
    // when the current element ends, or when a new one is started.
    private var currentString: String?
    
    // MARK: - Initialization

    init(string: String) throws {
        let xmlString = HTMLStripper.escapeWithUnicodeEntities(string)
        let xml = "<\(HTMLStripper.topTag)>\(xmlString)</\(HTMLStripper.topTag)>"
        guard let data = xml.data(using: String.Encoding.utf8) else {
            throw XMLParserInitError("Unable to convert to UTF8")
        }
        
        self.xmlParser = XMLParser(data: data)
        self.storedString = ""
        
        super.init()
        
        xmlParser.shouldProcessNamespaces = false
        xmlParser.shouldReportNamespacePrefixes = false
        xmlParser.shouldResolveExternalEntities = false
        xmlParser.delegate = self
    }
    
    /// Parse and generate attributed string.
    func parse() throws -> String {
        guard xmlParser.parse() else {
            let line = xmlParser.lineNumber
            let shiftColumn = (line == 1)
            let shiftSize = HTMLStripper.topTag.lengthOfBytes(using: String.Encoding.utf8) + 2
            let column = xmlParser.columnNumber - (shiftColumn ? shiftSize : 0)
            
            throw XMLParserError(parserError: xmlParser.parserError, line: line, column: column)
        }
        
        return storedString
    }
    
    // MARK: XMLParserDelegate
    
    @objc func parser(_ parser: XMLParser, didStartElement elementName: String, namespaceURI: String?, qualifiedName qName: String?, attributes attributeDict: [String: String]) {
        foundNewString()
    }
    
    @objc func parser(_ parser: XMLParser, didEndElement elementName: String, namespaceURI: String?, qualifiedName qName: String?) {
        foundNewString()
    }
    
    @objc func parser(_ parser: XMLParser, foundCharacters string: String) {
        currentString = (currentString ?? "").appending(string)
    }
    
    // MARK: Support Private Methods
    
    func foundNewString() {
        if let currentString = currentString {
            storedString.append(currentString)
            self.currentString = nil
        }
    }
    
    // handle html entity / html hex
    // Perform string escaping to replace all characters which is not supported by NSXMLParser
    // into the specified encoding with decimal entity.
    // For example if your string contains '&' character parser will break the style.
    // This option is active by default.
    // ref: https://github.com/malcommac/SwiftRichString/blob/e0b72d5c96968d7802856d2be096202c9798e8d1/Sources/SwiftRichString/Support/XMLStringBuilder.swift
    static func escapeWithUnicodeEntities(_ string: String) -> String {
        guard let escapeAmpRegExp = try? NSRegularExpression(pattern: "&(?!(#[0-9]{2,4}\\|[A-z]{2,6});)", options: NSRegularExpression.Options(rawValue: 0)) else {
            return string
        }
        
        let range = NSRange(location: 0, length: string.count)
        return escapeAmpRegExp.stringByReplacingMatches(in: string,
                                                        options: NSRegularExpression.MatchingOptions(rawValue: 0),
                                                        range: range,
                                                        withTemplate: "&amp;")
    }
}


let test = "我<br/><a href=\"http://google.com\">同意</a>提供<b><i>个</i>人</b>身分证字号／护照／居留<span style=\"color:#FF0000;font-size:20px;word-spacing:10px;line-height:10px\">证号码</span>，以供<i>跨境物流</i>方通关<span style=\"background-color:#00FF00;\">使用</span>，并已<img src=\"g.png\"/>了解跨境<br/>商品之物<p>流需</p>求"

let stripper = try HTMLStripper(string: test)
print(try! stripper.parse())

// 我同意提供个人身分证 字号／护照／居留证号码，以供跨境物流方通关使用，并已了解跨境商品之物流需求

protocol HTMLTagParser {
    static var tag: String { get } // 宣告想解析的 Tag Name, e.g. a
    var storedHTMLAttributes: [String: String]? { get set } // Attributed 解析结果将存放于此, e.g. href,style
    var style: AttributedStringStyle? { get } // 此 Tag 想套用的样式
    
    func render(attributedString: inout NSMutableAttributedString) // 实现渲染 HTML to attributedString 的逻辑
}

protocol AttributedStringStyle {
    var font: UIFont? { get set }
    var color: UIColor? { get set }
    var backgroundColor: UIColor? { get set }
    var wordSpacing: CGFloat? { get set }
    var paragraphStyle: NSParagraphStyle? { get set }
    var customs: [NSAttributedString.Key: Any]? { get set } // 万能设定口，建议确定可支援规格后将其抽象出来，并关闭此开口
    func render(attributedString: inout NSMutableAttributedString)
}


// abstract implement
extension AttributedStringStyle {
    func render(attributedString: inout NSMutableAttributedString) {
        let range = NSMakeRange(0, attributedString.length)
        if let font = font {
            attributedString.addAttribute(NSAttributedString.Key.font, value: font, range: range)
        }
        if let color = color {
            attributedString.addAttribute(NSAttributedString.Key.foregroundColor, value: color, range: range)
        }
        if let backgroundColor = backgroundColor {
            attributedString.addAttribute(NSAttributedString.Key.backgroundColor, value: backgroundColor, range: range)
        }
        if let wordSpacing = wordSpacing {
            attributedString.addAttribute(NSAttributedString.Key.kern, value: wordSpacing as Any, range: range)
        }
        if let paragraphStyle = paragraphStyle {
            attributedString.addAttribute(NSAttributedString.Key.paragraphStyle, value: paragraphStyle, range: range)
        }
        if let customAttributes = customs {
            attributedString.addAttributes(customAttributes, range: range)
        }
    }
}

// only support tag attributed down below
// can set color,font seize,line height,word spacing,background color

enum HTMLStyleAttributedParser: String {
    case color = "color"
    case fontSize = "font-size"
    case lineHeight = "line-height"
    case wordSpacing = "word-spacing"
    case backgroundColor = "background-color"
    
    func render(attributedString: inout NSMutableAttributedString, value: String) -> Bool {
        let range = NSMakeRange(0, attributedString.length)
        switch self {
        case .color:
            if let color = convertToiOSColor(value) {
                attributedString.addAttribute(NSAttributedString.Key.foregroundColor, value: color, range: range)
                return true
            }
        case .backgroundColor:
            if let color = convertToiOSColor(value) {
                attributedString.addAttribute(NSAttributedString.Key.backgroundColor, value: color, range: range)
                return true
            }
        case .fontSize:
            if let size = convertToiOSSize(value) {
                attributedString.addAttribute(NSAttributedString.Key.font, value: UIFont.systemFont(ofSize: CGFloat(size)), range: range)
                return true
            }
        case .lineHeight:
            if let size = convertToiOSSize(value) {
                let paragraphStyle = NSMutableParagraphStyle()
                paragraphStyle.lineSpacing = size
                attributedString.addAttribute(NSAttributedString.Key.paragraphStyle, value: paragraphStyle, range: range)
                return true
            }
        case .wordSpacing:
            if let size = convertToiOSSize(value) {
                attributedString.addAttribute(NSAttributedString.Key.kern, value: size, range: range)
                return true
            }
        }
        
        return false
    }
    
    // convert 36px -> 36
    private func convertToiOSSize(_ string: String) -> CGFloat? {
        guard let regex = try? NSRegularExpression(pattern: "^([0-9]+)"),
              let firstMatch = regex.firstMatch(in: string, options: [], range: NSRange(location: 0, length: string.utf16.count)),
              let range = Range(firstMatch.range, in: string),
              let size = Float(String(string[range])) else {
            return nil
        }
        return CGFloat(size)
    }
    
    // convert html hex color #ffffff to UIKit Color
    private func convertToiOSColor(_ hexString: String) -> UIColor? {
        var cString: String = hexString.trimmingCharacters(in: .whitespacesAndNewlines).uppercased()

        if cString.hasPrefix("#") {
            cString.remove(at: cString.startIndex)
        }

        if (cString.count) != 6 {
            return nil
        }

        var rgbValue: UInt64 = 0
        Scanner(string: cString).scanHexInt64(&rgbValue)

        return UIColor(
            red: CGFloat((rgbValue & 0xFF0000) >> 16) / 255.0,
            green: CGFloat((rgbValue & 0x00FF00) >> 8) / 255.0,
            blue: CGFloat(rgbValue & 0x0000FF) / 255.0,
            alpha: CGFloat(1.0)
        )
    }
}

extension HTMLTagParser {

    func render(attributedString: inout NSMutableAttributedString) {
        defaultStyleRender(attributedString: &attributedString)
    }
    
    func defaultStyleRender(attributedString: inout NSMutableAttributedString) {
        // setup default style to NSMutableAttributedString
        style?.render(attributedString: &attributedString)
        
        // setup & override HTML style (style="color:red;background-color:black") to NSMutableAttributedString if is exists
        // any html tag can have style attribute
        if let style = storedHTMLAttributes?["style"] {
            let styles = style.split(separator: ";").map { $0.split(separator: ":") }.filter { $0.count == 2 }
            for style in styles {
                let key = String(style[0])
                let value = String(style[1])
                
                if let styleAttributed = HTMLStyleAttributedParser(rawValue: key), styleAttributed.render(attributedString: &attributedString, value: value) {
                    print("Unsupport style attributed or value[\(key):\(value)]")
                }
            }
        }
    }
}

// Ref: https://github.com/malcommac/SwiftRichString
final class HTMLToAttributedStringParser: NSObject {
    
    private static let topTag = "source"
    private var xmlParser: XMLParser?
    
    private(set) var attributedString: NSMutableAttributedString = NSMutableAttributedString()
    private(set) var supportedTagRenders: [HTMLTagParser] = []
    private let defaultStyle: AttributedStringStyle
    
    /// Styles applied at each fragment.
    private var renderingTagRenders: [HTMLTagParser] = []

    // The XML parser sometimes splits strings, which can break localization-sensitive
    // string transforms. Work around this by using the currentString variable to
    // accumulate partial strings, and then reading them back out as a single string
    // when the current element ends, or when a new one is started.
    private var currentString: String?
    
    // MARK: - Initialization

    init(defaultStyle: AttributedStringStyle) {
        self.defaultStyle = defaultStyle
        super.init()
    }
    
    func register(_ tagRender: HTMLTagParser) {
        if let index = supportedTagRenders.firstIndex(where: { type(of: $0).tag == type(of: tagRender).tag }) {
            supportedTagRenders.remove(at: index)
        }
        supportedTagRenders.append(tagRender)
    }
    
    /// Parse and generate attributed string.
    func parse(string: String) throws -> NSAttributedString {
        var xmlString = HTMLToAttributedStringParser.escapeWithUnicodeEntities(string)
        
        // make sure <br/> format is correct XML
        // because Web may use <br> to present <br/>, but <br> is not a vaild XML
        xmlString = xmlString.replacingOccurrences(of: "<br>", with: "<br/>")
        
        let xml = "<\(HTMLToAttributedStringParser.topTag)>\(xmlString)</\(HTMLToAttributedStringParser.topTag)>"
        guard let data = xml.data(using: String.Encoding.utf8) else {
            throw XMLParserInitError("Unable to convert to UTF8")
        }
        
        let xmlParser = XMLParser(data: data)
        xmlParser.shouldProcessNamespaces = false
        xmlParser.shouldReportNamespacePrefixes = false
        xmlParser.shouldResolveExternalEntities = false
        xmlParser.delegate = self
        self.xmlParser = xmlParser
        
        attributedString = NSMutableAttributedString()
        
        guard xmlParser.parse() else {
            let line = xmlParser.lineNumber
            let shiftColumn = (line == 1)
            let shiftSize = HTMLToAttributedStringParser.topTag.lengthOfBytes(using: String.Encoding.utf8) + 2
            let column = xmlParser.columnNumber - (shiftColumn ? shiftSize : 0)
            
            throw XMLParserError(parserError: xmlParser.parserError, line: line, column: column)
        }
        
        return attributedString
    }
}

// MARK: Private Method

private extension HTMLToAttributedStringParser {
    func enter(element elementName: String, attributes: [String: String]) {
        // elementName = tagName, EX: a,span,div...
        guard elementName != HTMLToAttributedStringParser.topTag else {
            return
        }
        
        if let index = supportedTagRenders.firstIndex(where: { type(of: $0).tag == elementName }) {
            var tagRender = supportedTagRenders[index]
            tagRender.storedHTMLAttributes = attributes
            renderingTagRenders.append(tagRender)
        }
    }
    
    func exit(element elementName: String) {
        if !renderingTagRenders.isEmpty {
            renderingTagRenders.removeLast()
        }
    }
    
    func foundNewString() {
        if let currentString = currentString {
            // currentString != nil ,ex: <i>currentString</i>
            var newAttributedString = NSMutableAttributedString(string: currentString)
            if !renderingTagRenders.isEmpty {
                for (key, tagRender) in renderingTagRenders.enumerated() {
                    // Render Style
                    tagRender.render(attributedString: &newAttributedString)
                    renderingTagRenders[key].storedHTMLAttributes = nil
                }
            } else {
                defaultStyle.render(attributedString: &newAttributedString)
            }
            attributedString.append(newAttributedString)
            self.currentString = nil
        } else {
            // currentString == nil ,ex: <br/>
            var newAttributedString = NSMutableAttributedString()
            for (key, tagRender) in renderingTagRenders.enumerated() {
                // Render Style
                tagRender.render(attributedString: &newAttributedString)
                renderingTagRenders[key].storedHTMLAttributes = nil
            }
            attributedString.append(newAttributedString)
        }
    }
}

// MARK: Helper

extension HTMLToAttributedStringParser {
    // handle html entity / html hex
    // Perform string escaping to replace all characters which is not supported by NSXMLParser
    // into the specified encoding with decimal entity.
    // For example if your string contains '&' character parser will break the style.
    // This option is active by default.
    // ref: https://github.com/malcommac/SwiftRichString/blob/e0b72d5c96968d7802856d2be096202c9798e8d1/Sources/SwiftRichString/Support/XMLStringBuilder.swift
    static func escapeWithUnicodeEntities(_ string: String) -> String {
        guard let escapeAmpRegExp = try? NSRegularExpression(pattern: "&(?!(#[0-9]{2,4}\\|[A-z]{2,6});)", options: NSRegularExpression.Options(rawValue: 0)) else {
            return string
        }
        
        let range = NSRange(location: 0, length: string.count)
        return escapeAmpRegExp.stringByReplacingMatches(in: string,
                                                        options: NSRegularExpression.MatchingOptions(rawValue: 0),
                                                        range: range,
                                                        withTemplate: "&amp;")
    }
}

// MARK: XMLParserDelegate

extension HTMLToAttributedStringParser: XMLParserDelegate {
    func parser(_ parser: XMLParser, didStartElement elementName: String, namespaceURI: String?, qualifiedName qName: String?, attributes attributeDict: [String: String]) {
        foundNewString()
        enter(element: elementName, attributes: attributeDict)
    }
    
    func parser(_ parser: XMLParser, didEndElement elementName: String, namespaceURI: String?, qualifiedName qName: String?) {
        foundNewString()
        guard elementName != HTMLToAttributedStringParser.topTag else {
            return
        }
        
        exit(element: elementName)
    }
    
    func parser(_ parser: XMLParser, foundCharacters string: String) {
        currentString = (currentString ?? "").appending(string)
    }
}

let test = "我<br/><a href=\"http://google.com\">同意</a>提供<b><i>个</i>人</b>身分证字号／护照／居留<span style=\"color:#FF0000;font-size:20px;word-spacing:10px;line-height:10px\">证号码</span>，以供<i>跨境物流</i>方通关<span style=\"background-color:#00FF00;\">使用</span>，并已<img src=\"g.png\"/>了解跨境<br/>商品之物<p>流需</p>求"
let render = HTMLToAttributedStringParser(defaultStyle: DefaultTextStyle())
render.register(ATagParser())
render.register(BoldTagParser())
render.register(SpanTagParser())
//...
print(try! render.parse(string: test))

// Result:
// 我{
//     NSColor = "UIExtendedGrayColorSpace 0 1";
//     NSFont = "\".SFNS-Regular 14.00 pt. P [] (0x13a012970) fobj=0x13a012970, spc=3.79\"";
//     NSParagraphStyle = "Alignment 4, LineSpacing 3, ParagraphSpacing 0, ParagraphSpacingBefore 0, HeadIndent 0, TailIndent 0, FirstLineHeadIndent 0, LineHeight 0/0, LineHeightMultiple 0, LineBreakMode 0, Tabs (\n    28L,\n    56L,\n    84L,\n    112L,\n    140L,\n    168L,\n    196L,\n    224L,\n    252L,\n    280L,\n    308L,\n    336L\n), DefaultTabInterval 0, Blocks (\n), Lists (\n), BaseWritingDirection -1, HyphenationFactor 0, TighteningForTruncation NO, HeaderLevel 0 LineBreakStrategy 0 PresentationIntents (\n) ListIntentOrdinal 0 CodeBlockIntentLanguageHint ''";
// }同意{
//     NSColor = "UIExtendedSRGBColorSpace 0 0 1 1";
//     NSFont = "\".SFNS-Regular 14.00 pt. P [] (0x13a012970) fobj=0x13a012970, spc=3.79\"";
//     NSLink = "http://google.com";
//     NSUnderline = 1;
// }提供{
//     NSColor = "UIExtendedGrayColorSpace 0 1";
//     NSFont = "\".SFNS-Regular 14.00 pt. P [] (0x13a012970) fobj=0x13a012970, spc=3.79\"";
//     NSParagraphStyle = "Alignment 4, LineSpacing 3, ParagraphSpacing 0, ParagraphSpacingBefore 0, HeadIndent 0, TailIndent 0, FirstLineHeadIndent 0, LineHeight 0/0, LineHeightMultiple 0, LineBreakMode 0, Tabs (\n    28L,\n    56L,\n    84L,\n    112L,\n    140L,\n    168L,\n    196L,\n    224L,\n    252L,\n    280L,\n    308L,\n    336L\n), DefaultTabInterval 0, Blocks (\n), Lists (\n), BaseWritingDirection -1, HyphenationFactor 0, TighteningForTruncation NO, HeaderLevel 0 LineBreakStrategy 0 PresentationIntents (\n) ListIntentOrdinal 0 CodeBlockIntentLanguageHint ''";
// }个{
//     NSColor = "UIExtendedGrayColorSpace 0 1";
//     NSFont = "\".SFNS-Bold 14.00 pt. P [] (0x13a013870) fobj=0x13a013870, spc=3.46\"";
//     NSUnderline = 1;
// }人身分证字号／护照／居留{
//     NSColor = "UIExtendedGrayColorSpace 0 1";
//     NSFont = "\".SFNS-Regular 14.00 pt. P [] (0x13a012970) fobj=0x13a012970, spc=3.79\"";
//     NSParagraphStyle = "Alignment 4, LineSpacing 3, ParagraphSpacing 0, ParagraphSpacingBefore 0, HeadIndent 0, TailIndent 0, FirstLineHeadIndent 0, LineHeight 0/0, LineHeightMultiple 0, LineBreakMode 0, Tabs (\n    28L,\n    56L,\n    84L,\n    112L,\n    140L,\n    168L,\n    196L,\n    224L,\n    252L,\n    280L,\n    308L,\n    336L\n), DefaultTabInterval 0, Blocks (\n), Lists (\n), BaseWritingDirection -1, HyphenationFactor 0, TighteningForTruncation NO, HeaderLevel 0 LineBreakStrategy 0 PresentationIntents (\n) ListIntentOrdinal 0 CodeBlockIntentLanguageHint ''";
// }证号码{
//     NSColor = "UIExtendedSRGBColorSpace 1 0 0 1";
//     NSFont = "\".SFNS-Regular 20.00 pt. P [] (0x13a015fa0) fobj=0x13a015fa0, spc=4.82\"";
//     NSKern = 10;
//     NSParagraphStyle = "Alignment 4, LineSpacing 10, ParagraphSpacing 0, ParagraphSpacingBefore 0, HeadIndent 0, TailIndent 0, FirstLineHeadIndent 0, LineHeight 0/0, LineHeightMultiple 0, LineBreakMode 0, Tabs (\n    28L,\n    56L,\n    84L,\n    112L,\n    140L,\n    168L,\n    196L,\n    224L,\n    252L,\n    280L,\n    308L,\n    336L\n), DefaultTabInterval 0, Blocks (\n), Lists (\n), BaseWritingDirection -1, HyphenationFactor 0, TighteningForTruncation NO, HeaderLevel 0 LineBreakStrategy 0 PresentationIntents (\n) ListIntentOrdinal 0 CodeBlockIntentLanguageHint ''";
// }，以供跨境物流方通关{
//     NSColor = "UIExtendedGrayColorSpace 0 1";
//     NSFont = "\".SFNS-Regular 14.00 pt. P [] (0x13a012970) fobj=0x13a012970, spc=3.79\"";
//     NSParagraphStyle = "Alignment 4, LineSpacing 3, ParagraphSpacing 0, ParagraphSpacingBefore 0, HeadIndent 0, TailIndent 0, FirstLineHeadIndent 0, LineHeight 0/0, LineHeightMultiple 0, LineBreakMode 0, Tabs (\n    28L,\n    56L,\n    84L,\n    112L,\n    140L,\n    168L,\n    196L,\n    224L,\n    252L,\n    280L,\n    308L,\n    336L\n), DefaultTabInterval 0, Blocks (\n), Lists (\n), BaseWritingDirection -1, HyphenationFactor 0, TighteningForTruncation NO, HeaderLevel 0 LineBreakStrategy 0 PresentationIntents (\n) ListIntentOrdinal 0 CodeBlockIntentLanguageHint ''";
// }使用{
//     NSBackgroundColor = "UIExtendedSRGBColorSpace 0 1 0 1";
//     NSColor = "UIExtendedGrayColorSpace 0 1";
//     NSFont = "\".SFNS-Regular 14.00 pt. P [] (0x13a012970) fobj=0x13a012970, spc=3.79\"";
//     NSParagraphStyle = "Alignment 4, LineSpacing 3, ParagraphSpacing 0, ParagraphSpacingBefore 0, HeadIndent 0, TailIndent 0, FirstLineHeadIndent 0, LineHeight 0/0, LineHeightMultiple 0, LineBreakMode 0, Tabs (\n    28L,\n    56L,\n    84L,\n    112L,\n    140L,\n    168L,\n    196L,\n    224L,\n    252L,\n    280L,\n    308L,\n    336L\n), DefaultTabInterval 0, Blocks (\n), Lists (\n), BaseWritingDirection -1, HyphenationFactor 0, TighteningForTruncation NO, HeaderLevel 0 LineBreakStrategy 0 PresentationIntents (\n) ListIntentOrdinal 0 CodeBlockIntentLanguageHint ''";
// }，并已了解跨境商品之物流需求{
//     NSColor = "UIExtendedGrayColorSpace 0 1";
//     NSFont = "\".SFNS-Regular 14.00 pt. P [] (0x13a012970) fobj=0x13a012970, spc=3.79\"";
//     NSParagraphStyle = "Alignment 4, LineSpacing 3, ParagraphSpacing 0, ParagraphSpacingBefore 0, HeadIndent 0, TailIndent 0, FirstLineHeadIndent 0, LineHeight 0/0, LineHeightMultiple 0, LineBreakMode 0, Tabs (\n    28L,\n    56L,\n    84L,\n    112L,\n    140L,\n    168L,\n    196L,\n    224L,\n    252L,\n    280L,\n    308L,\n    336L\n), DefaultTabInterval 0, Blocks (\n), Lists (\n), BaseWritingDirection -1, HyphenationFactor 0, TighteningForTruncation NO, HeaderLevel 0 LineBreakStrategy 0 PresentationIntents (\n) ListIntentOrdinal 0 CodeBlockIntentLanguageHint ''";
// }