Bubble/Threaded/Data/Content/HTMLString.swift

294 lines
13 KiB
Swift

//Made by Lumaa
import Foundation
import SwiftSoup
import SwiftUI
private enum CodingKeys: CodingKey {
case htmlValue, asMarkdown, asRawText, statusesURLs, links
}
public struct HTMLString: Codable, Equatable, Hashable, @unchecked Sendable {
public var htmlValue: String = ""
public var asMarkdown: String = ""
public var asRawText: String = ""
public var statusesURLs = [URL]()
public private(set) var links = [Link]()
public var asSafeMarkdownAttributedString: AttributedString = .init()
private var main_regex: NSRegularExpression?
private var underscore_regex: NSRegularExpression?
public init(from decoder: Decoder) {
var alreadyDecoded = false
do {
let container = try decoder.singleValueContainer()
htmlValue = try container.decode(String.self)
} catch {
do {
alreadyDecoded = true
let container = try decoder.container(keyedBy: CodingKeys.self)
htmlValue = try container.decode(String.self, forKey: .htmlValue)
asMarkdown = try container.decode(String.self, forKey: .asMarkdown)
asRawText = try container.decode(String.self, forKey: .asRawText)
statusesURLs = try container.decode([URL].self, forKey: .statusesURLs)
links = try container.decode([Link].self, forKey: .links)
} catch {
htmlValue = ""
}
}
if !alreadyDecoded {
// https://daringfireball.net/projects/markdown/syntax
// Pre-escape \ ` _ * ~ and [ as these are the only
// characters the markdown parser uses when it renders
// to attributed text. Note that ~ for strikethrough is
// not documented in the syntax docs but is used by
// AttributedString.
main_regex = try? NSRegularExpression(pattern: "([\\*\\`\\~\\[\\\\])", options: .caseInsensitive)
// don't escape underscores that are between colons, they are most likely custom emoji
underscore_regex = try? NSRegularExpression(pattern: "(?!\\B:[^:]*)(_)(?![^:]*:\\B)", options: .caseInsensitive)
asMarkdown = ""
do {
let document: Document = try SwiftSoup.parse(htmlValue)
handleNode(node: document)
document.outputSettings(OutputSettings().prettyPrint(pretty: false))
try document.select("br").after("\n")
try document.select("p").after("\n\n")
let html = try document.html()
var text = try SwiftSoup.clean(html, "", Whitelist.none(), OutputSettings().prettyPrint(pretty: false)) ?? ""
// Remove the two last line break added after the last paragraph.
if text.hasSuffix("\n\n") {
_ = text.removeLast()
_ = text.removeLast()
}
asRawText = text
if asMarkdown.hasPrefix("\n") {
_ = asMarkdown.removeFirst()
}
} catch {
asRawText = htmlValue
}
}
do {
let options = AttributedString.MarkdownParsingOptions(allowsExtendedAttributes: true,
interpretedSyntax: .inlineOnlyPreservingWhitespace)
asSafeMarkdownAttributedString = try AttributedString(markdown: asMarkdown, options: options)
} catch {
asSafeMarkdownAttributedString = AttributedString(stringLiteral: htmlValue)
}
}
public init(stringValue: String, parseMarkdown: Bool = false) {
htmlValue = stringValue
asMarkdown = stringValue
asRawText = stringValue
statusesURLs = []
if parseMarkdown {
do {
let options = AttributedString.MarkdownParsingOptions(allowsExtendedAttributes: true,
interpretedSyntax: .inlineOnlyPreservingWhitespace)
asSafeMarkdownAttributedString = try AttributedString(markdown: asMarkdown, options: options)
} catch {
asSafeMarkdownAttributedString = AttributedString(stringLiteral: htmlValue)
}
} else {
asSafeMarkdownAttributedString = AttributedString(stringLiteral: htmlValue)
}
}
public func encode(to encoder: Encoder) throws {
var container = encoder.container(keyedBy: CodingKeys.self)
try container.encode(htmlValue, forKey: .htmlValue)
try container.encode(asMarkdown, forKey: .asMarkdown)
try container.encode(asRawText, forKey: .asRawText)
try container.encode(statusesURLs, forKey: .statusesURLs)
try container.encode(links, forKey: .links)
}
private mutating func handleNode(node: SwiftSoup.Node) {
do {
if let className = try? node.attr("class") {
if className == "invisible" {
// don't display
return
}
if className == "ellipsis" {
// descend into this one now and
// append the ellipsis
for nn in node.getChildNodes() {
handleNode(node: nn)
}
asMarkdown += ""
return
}
}
if node.nodeName() == "p" {
if asMarkdown.count > 0 { // ignore first opening <p>
asMarkdown += "\n\n"
}
} else if node.nodeName() == "br" {
if asMarkdown.count > 0 { // ignore first opening <br>
asMarkdown += "\n"
}
} else if node.nodeName() == "a" {
let href = try node.attr("href")
if href != "" {
if let url = URL(string: href),
let _ = Int(url.lastPathComponent)
{
statusesURLs.append(url)
}
}
asMarkdown += "["
let start = asMarkdown.endIndex
// descend into this node now so we can wrap the
// inner part of the link in the right markup
for nn in node.getChildNodes() {
handleNode(node: nn)
}
let finish = asMarkdown.endIndex
var linkRef = href
// Try creating a URL from the string. If it fails, try URL encoding
// the string first.
var url = URL(string: href)
if url == nil {
url = URL(string: href, encodePath: true)
}
if let linkUrl = url {
linkRef = linkUrl.absoluteString
let displayString = asMarkdown[start ..< finish]
links.append(Link(linkUrl, displayString: String(displayString)))
}
asMarkdown += "]("
asMarkdown += linkRef
asMarkdown += ")"
return
} else if node.nodeName() == "#text" {
var txt = node.description
if let underscore_regex, let main_regex {
// This is the markdown escaper
txt = main_regex.stringByReplacingMatches(in: txt, options: [], range: NSRange(location: 0, length: txt.count), withTemplate: "\\\\$1")
txt = underscore_regex.stringByReplacingMatches(in: txt, options: [], range: NSRange(location: 0, length: txt.count), withTemplate: "\\\\$1")
}
// Strip newlines and line separators - they should be being sent as <br>s
asMarkdown += txt.replacingOccurrences(of: "\n", with: "").replacingOccurrences(of: "\u{2028}", with: "")
} else if node.nodeName() == "ul" {
// Unordered (bulleted) list
// SwiftUI's Text won't display these in an AttributedString, but we can at least improve the appearance
asMarkdown += "\n\n"
for nn in node.getChildNodes() {
asMarkdown += "- "
handleNode(node: nn)
asMarkdown += "\n"
}
return
} else if node.nodeName() == "ol" {
// Ordered (numbered) list
// Same thing, won't display in a Text, but this is just an attempt to improve the appearance
asMarkdown += "\n\n"
var curNumber = 1
for nn in node.getChildNodes() {
asMarkdown += "\(curNumber). "
handleNode(node: nn)
asMarkdown += "\n"
curNumber += 1
}
return
}
for n in node.getChildNodes() {
handleNode(node: n)
}
} catch {}
}
public struct Link: Codable, Hashable, Identifiable {
public var id: Int { hashValue }
public let url: URL
public let displayString: String
public let type: LinkType
public let title: String
init(_ url: URL, displayString: String) {
self.url = url
self.displayString = displayString
switch displayString.first {
case "@":
type = .mention
title = displayString
case "#":
type = .hashtag
title = String(displayString.dropFirst())
default:
type = .url
var hostNameUrl = url.host ?? url.absoluteString
if hostNameUrl.hasPrefix("www.") {
hostNameUrl = String(hostNameUrl.dropFirst(4))
}
title = hostNameUrl
}
}
public enum LinkType: String, Codable {
case url
case mention
case hashtag
}
}
}
public extension URL {
// It's common to use non-ASCII characters in URLs even though they're technically
// invalid characters. Every modern browser handles this by silently encoding
// the invalid characters on the user's behalf. However, trying to create a URL
// object with un-encoded characters will result in nil so we need to encode the
// invalid characters before creating the URL object. The unencoded version
// should still be shown in the displayed status.
init?(string: String, encodePath: Bool) {
var encodedUrlString = ""
if encodePath,
string.starts(with: "http://") || string.starts(with: "https://"),
var startIndex = string.firstIndex(of: "/")
{
startIndex = string.index(startIndex, offsetBy: 1)
// We don't want to encode the host portion of the URL
if var startIndex = string[startIndex...].firstIndex(of: "/") {
encodedUrlString = String(string[...startIndex])
while let endIndex = string[string.index(after: startIndex)...].firstIndex(of: "/") {
let componentStartIndex = string.index(after: startIndex)
encodedUrlString = encodedUrlString + (string[componentStartIndex ... endIndex].addingPercentEncoding(withAllowedCharacters: .urlPathAllowed) ?? "")
startIndex = endIndex
}
// The last part of the path may have a query string appended to it
let componentStartIndex = string.index(after: startIndex)
if let queryStartIndex = string[componentStartIndex...].firstIndex(of: "?") {
encodedUrlString = encodedUrlString + (string[componentStartIndex ..< queryStartIndex].addingPercentEncoding(withAllowedCharacters: .urlPathAllowed) ?? "")
encodedUrlString = encodedUrlString + (string[queryStartIndex...].addingPercentEncoding(withAllowedCharacters: .urlQueryAllowed) ?? "")
} else {
encodedUrlString = encodedUrlString + (string[componentStartIndex...].addingPercentEncoding(withAllowedCharacters: .urlPathAllowed) ?? "")
}
}
}
if encodedUrlString.isEmpty {
encodedUrlString = string
}
self.init(string: encodedUrlString)
}
}