Fix HTML issues

This commit is contained in:
Justin Mazzocchi 2020-09-26 14:09:37 -07:00
parent 582ff3623c
commit 083e960166
No known key found for this signature in database
GPG key ID: E223E6937AAFB01C

View file

@ -17,7 +17,13 @@ extension HTML: Codable {
let container = try decoder.singleValueContainer() let container = try decoder.singleValueContainer()
raw = try container.decode(String.self) raw = try container.decode(String.self)
attributed = HTMLParser(string: raw).parse()
if let cachedAttributedString = Self.attributedStringCache.object(forKey: raw as NSString) {
attributed = cachedAttributedString
} else {
attributed = HTMLParser(string: raw).parse()
Self.attributedStringCache.setObject(attributed, forKey: raw as NSString)
}
} }
public func encode(to encoder: Encoder) throws { public func encode(to encoder: Encoder) throws {
@ -27,6 +33,10 @@ extension HTML: Codable {
} }
} }
private extension HTML {
static var attributedStringCache = NSCache<NSString, NSAttributedString>()
}
// https://docs.joinmastodon.org/spec/activitypub/#sanitization // https://docs.joinmastodon.org/spec/activitypub/#sanitization
private class HTMLParser: NSObject { private class HTMLParser: NSObject {
@ -48,7 +58,9 @@ private class HTMLParser: NSObject {
private static let closingContainerTag = "</\(containerTag)>" private static let closingContainerTag = "</\(containerTag)>"
init(string: String) { init(string: String) {
rawString = Self.openingContainerTag + string + Self.closingContainerTag rawString = Self.openingContainerTag
.appending(string.replacingOccurrences(of: "<br>", with: "<br/>"))
.appending(Self.closingContainerTag)
parser = XMLParser(data: Data(rawString.utf8)) parser = XMLParser(data: Data(rawString.utf8))
parseStopColumn = rawString.count - Self.closingContainerTag.count parseStopColumn = rawString.count - Self.closingContainerTag.count
@ -81,7 +93,7 @@ extension HTMLParser: XMLParserDelegate {
attributesStack.append(attributeDict) attributesStack.append(attributeDict)
if elementName == "a", let hrefString = attributeDict["href"], let href = URL(string: hrefString) { if elementName == "a", let hrefString = attributeDict["href"], let href = URL(string: hrefString) {
currentLink = Link(href: href, location: constructedString.count) currentLink = Link(href: href, location: constructedString.utf16.count)
} else if elementName == "br" { } else if elementName == "br" {
constructedString.append("\n") constructedString.append("\n")
} }
@ -98,7 +110,7 @@ extension HTMLParser: XMLParserDelegate {
} }
if elementName == "a", var link = currentLink { if elementName == "a", var link = currentLink {
link.length = constructedString.count - link.location link.length = constructedString.utf16.count - link.location
links.insert(link) links.insert(link)
} else if elementName == "p", parser.columnNumber < parseStopColumn { } else if elementName == "p", parser.columnNumber < parseStopColumn {
constructedString.append("\n\n") constructedString.append("\n\n")