169 lines
6.2 KiB
Swift
169 lines
6.2 KiB
Swift
/*
|
|
Copyright 2024 New Vector Ltd.
|
|
Copyright 2020 The Matrix.org Foundation C.I.C
|
|
|
|
SPDX-License-Identifier: AGPL-3.0-only
|
|
Please see LICENSE in the repository root for full details.
|
|
*/
|
|
|
|
import Foundation
|
|
import Down
|
|
import libcmark
|
|
|
|
@objc public protocol MarkdownToHTMLRendererProtocol: NSObjectProtocol {
|
|
func renderToHTML(markdown: String) -> String?
|
|
}
|
|
|
|
@objcMembers
|
|
public class MarkdownToHTMLRenderer: NSObject {
|
|
|
|
fileprivate var options: DownOptions = []
|
|
|
|
// Do not expose an initializer with options, because `DownOptions` is not ObjC compatible.
|
|
public override init() {
|
|
super.init()
|
|
}
|
|
}
|
|
|
|
extension MarkdownToHTMLRenderer: MarkdownToHTMLRendererProtocol {
|
|
|
|
public func renderToHTML(markdown: String) -> String? {
|
|
do {
|
|
let ast = try DownASTRenderer.stringToAST(markdown, options: options)
|
|
defer {
|
|
cmark_node_free(ast)
|
|
}
|
|
ast.repairLinks()
|
|
return try DownHTMLRenderer.astToHTML(ast, options: options)
|
|
} catch {
|
|
MXLog.error("[MarkdownToHTMLRenderer] renderToHTML failed")
|
|
return nil
|
|
}
|
|
}
|
|
|
|
}
|
|
|
|
@objcMembers
|
|
public class MarkdownToHTMLRendererHardBreaks: MarkdownToHTMLRenderer {
|
|
|
|
public override init() {
|
|
super.init()
|
|
options = .hardBreaks
|
|
}
|
|
|
|
}
|
|
|
|
// MARK: - AST-handling private extensions
|
|
private extension CMarkNode {
|
|
/// Formatting symbol associated with given note type
|
|
/// Note: this is only defined for node types that are handled in repairLinks
|
|
var formattingSymbol: String {
|
|
switch self.type {
|
|
case CMARK_NODE_EMPH:
|
|
return "_"
|
|
case CMARK_NODE_STRONG:
|
|
return "__"
|
|
default:
|
|
return ""
|
|
}
|
|
}
|
|
|
|
/// Repairs links that were broken down by markdown formatting.
|
|
/// Should be used on the first node of libcmark's AST
|
|
/// (e.g. the object returned by DownASTRenderer.stringToAST).
|
|
func repairLinks() {
|
|
let iterator = cmark_iter_new(self)
|
|
var text = ""
|
|
var isInParagraph = false
|
|
var previousNode: CMarkNode?
|
|
var orphanNodes: [CMarkNode] = []
|
|
var shouldUnlinkFormattingMode = false
|
|
var event: cmark_event_type?
|
|
while event != CMARK_EVENT_DONE {
|
|
event = cmark_iter_next(iterator)
|
|
|
|
guard let node = cmark_iter_get_node(iterator) else { return }
|
|
|
|
if node.type == CMARK_NODE_PARAGRAPH {
|
|
if event == CMARK_EVENT_ENTER {
|
|
isInParagraph = true
|
|
} else {
|
|
isInParagraph = false
|
|
text = ""
|
|
}
|
|
}
|
|
|
|
if isInParagraph {
|
|
switch node.type {
|
|
case CMARK_NODE_SOFTBREAK,
|
|
CMARK_NODE_LINEBREAK:
|
|
text = ""
|
|
case CMARK_NODE_TEXT:
|
|
if let literal = node.literal {
|
|
text += literal
|
|
// Reset text if it ends up with a whitespace.
|
|
if text.last?.isWhitespace == true {
|
|
text = ""
|
|
}
|
|
// Only the last part could be a link conflicting with next node.
|
|
text = String(text.split(separator: " ").last ?? "")
|
|
}
|
|
case CMARK_NODE_EMPH where previousNode?.type == CMARK_NODE_TEXT,
|
|
CMARK_NODE_STRONG where previousNode?.type == CMARK_NODE_TEXT:
|
|
if event == CMARK_EVENT_ENTER {
|
|
if !text.containedUrls.isEmpty,
|
|
let childLiteral = node.pointee.first_child.literal {
|
|
// If current text is a link, the formatted text is reverted back to a
|
|
// plain text as a part of the link.
|
|
let symbol = node.formattingSymbol
|
|
let nonFormattedText = "\(symbol)\(childLiteral)\(symbol)"
|
|
let replacementTextNode = cmark_node_new(CMARK_NODE_TEXT)
|
|
cmark_node_set_literal(replacementTextNode, nonFormattedText)
|
|
cmark_node_insert_after(previousNode, replacementTextNode)
|
|
// Set child literal to empty string so we dont read it.
|
|
// This avoids having to re-create the main
|
|
// iterator in the middle of the process.
|
|
cmark_node_set_literal(node.pointee.first_child, "")
|
|
let newIterator = cmark_iter_new(node)
|
|
_ = cmark_iter_next(newIterator)
|
|
cmark_node_unlink(node)
|
|
orphanNodes.append(node)
|
|
let nextNode = cmark_iter_get_node(newIterator)
|
|
cmark_node_insert_after(previousNode, nextNode)
|
|
shouldUnlinkFormattingMode = true
|
|
}
|
|
} else {
|
|
if shouldUnlinkFormattingMode {
|
|
cmark_node_unlink(node)
|
|
orphanNodes.append(node)
|
|
shouldUnlinkFormattingMode = false
|
|
}
|
|
}
|
|
default:
|
|
break
|
|
}
|
|
}
|
|
previousNode = node
|
|
}
|
|
|
|
// Free all nodes removed from the AST.
|
|
// This is done as a last step to avoid messing
|
|
// up with the main itertor.
|
|
for orphanNode in orphanNodes {
|
|
cmark_node_free(orphanNode)
|
|
}
|
|
}
|
|
}
|
|
|
|
private extension String {
|
|
/// Returns array of URLs detected inside the String.
|
|
var containedUrls: [NSTextCheckingResult] {
|
|
guard let detector = try? NSDataDetector(types: NSTextCheckingResult.CheckingType.link.rawValue),
|
|
let percentEncoded = self.addingPercentEncoding(withAllowedCharacters: .urlHostAllowed) else {
|
|
return []
|
|
}
|
|
|
|
return detector.matches(in: percentEncoded, options: [], range: NSRange(location: 0, length: percentEncoded.utf16.count))
|
|
}
|
|
}
|