ChWebsiteApp/CHDataManagement/Generator/PageContentGenerator.swift
2024-12-10 15:21:28 +01:00

508 lines
19 KiB
Swift

import Foundation
import Ink
import Splash
typealias VideoSource = (url: String, type: VideoFileType)
final class PageContentParser {
private let pageLinkMarker = "page:"
private let swift = SyntaxHighlighter(format: HTMLOutputFormat())
let results = PageGenerationResults()
private let content: Content
private let language: ContentLanguage
private var largeImageCount: Int = 0
var largeImageWidth: Int {
content.settings.pages.largeImageWidth
}
var thumbnailWidth: Int {
content.settings.pages.contentWidth
}
init(content: Content, language: ContentLanguage) {
self.content = content
self.language = language
}
func requestImages(_ generator: ImageGenerator) {
let thumbnailWidth = CGFloat(thumbnailWidth)
let largeImageWidth = CGFloat(largeImageWidth)
for image in results.files {
guard case .image = image.type else {
continue
}
generator.generateImageSet(
for: image.id,
maxWidth: thumbnailWidth, maxHeight: thumbnailWidth)
generator.generateImageSet(
for: image.id,
maxWidth: largeImageWidth, maxHeight: largeImageWidth)
}
}
func reset() {
results.reset()
largeImageCount = 0
}
func generatePage(from content: String) -> String {
reset()
let parser = MarkdownParser(modifiers: [
Modifier(target: .images, closure: processMarkdownImage),
Modifier(target: .codeBlocks, closure: handleCode),
Modifier(target: .links, closure: handleLink),
Modifier(target: .html, closure: handleHTML),
Modifier(target: .headings, closure: handleHeadlines)
])
return parser.html(from: content)
}
private func handleCode(html: String, markdown: Substring) -> String {
guard markdown.starts(with: "```swift") else {
return html // Just use normal code highlighting
}
// Highlight swift code using Splash
let code = markdown.between("```swift", and: "```").trimmed
return "<pre><code>" + swift.highlight(code) + "</pre></code>"
}
private func handleLink(html: String, markdown: Substring) -> String {
let file = markdown.between("(", and: ")")
if file.hasPrefix(pageLinkMarker) {
// Retain links pointing to elements within a page
let textToChange = file.dropAfterFirst("#")
let pageId = textToChange.replacingOccurrences(of: pageLinkMarker, with: "")
guard let page = content.page(pageId) else {
results.missingPages.insert(pageId)
// Remove link since the page can't be found
return markdown.between("[", and: "]")
}
results.linkedPages.insert(page)
let pagePath = content.pageLink(page, language: language)
return html.replacingOccurrences(of: textToChange, with: pagePath)
}
// TODO: Check that linked file exists
// if let filePath = page.nonAbsolutePathRelativeToRootForContainedInputFile(file) {
// // The target of the page link must be present after generation is complete
// results.expect(file: filePath, source: page.path)
// }
return html
}
private func handleHTML(html: String, markdown: Substring) -> String {
// TODO: Check HTML code in markdown for required resources
//print("[HTML] Found in page \(page.path):")
//print(markdown)
// Things to check:
// <img src=
// <a href=
//
return html
}
/**
Modify headlines by extracting an id from the headline and adding it into the html element
Format: ##<title>#<id>
The id is created by lowercasing the string, removing all special characters, and replacing spaces with scores
*/
private func handleHeadlines(html: String, markdown: Substring) -> String {
let id = markdown
.last(after: "#")
.trimmed
.filter { $0.isNumber || $0.isLetter || $0 == " " }
.lowercased()
.components(separatedBy: " ")
.filter { $0 != "" }
.joined(separator: "-")
let parts = html.components(separatedBy: ">")
return parts[0] + " id=\"\(id)\">" + parts.dropFirst().joined(separator: ">")
}
private func processMarkdownImage(html: String, markdown: Substring) -> String {
// First, check the content type, then parse the remaining arguments
// Notation:
// <abc?> -> Optional argument
// <abc...> -> Repeated argument (0 or more)
// ![url](<url>;<text>)
// ![image](<imageId>;<caption?>]
// ![video](<fileId>;<option1...>]
// ![svg](<fileId>;<<x>;<y>;<width>;<height>?>)
// ![download](<<fileId>,<text>,<download-filename?>;...)
// ![box](<title>;<body>)
// ![model](<file>;<description>)
// ![page](<pageId>)
// ![external](<<url>;<text>...>
// ![html](<fileId>)
let argumentList = markdown.between(first: "](", andLast: ")").removingPercentEncoding ?? markdown.between(first: "](", andLast: ")")
let arguments = argumentList.components(separatedBy: ";")
let rawCommand = markdown.between("![", and: "]").trimmed
guard rawCommand != "" else {
return handleImage(arguments)
}
guard let convertedCommand = rawCommand.removingPercentEncoding,
let command = ShorthandMarkdownKey(rawValue: convertedCommand) else {
// Treat unknown commands as normal links
results.warnings.append("Unknown markdown command '\(rawCommand)'")
return html
}
switch command {
case .image:
return handleImage(arguments)
case .hikingStatistics:
return handleHikingStatistics(arguments)
case .downloadButtons:
return handleDownloadButtons(arguments)
case .video:
return handleVideo(arguments)
case .externalLink:
return handleExternalButtons(arguments)
/*
case .includedHtml:
return handleExternalHTML(file: content)
case .box:
return handleSimpleBox(content: content)
case .pageLink:
return handlePageLink(pageId: content)
case .model:
return handle3dModel(content: content)
*/
default:
results.warnings.append("Unhandled command '\(command.rawValue)'")
return ""
}
}
private func handleImage(_ arguments: [String]) -> String {
// [image](<imageId>;<caption?>]
guard (1...2).contains(arguments.count) else {
results.invalidCommandArguments.append((.image , arguments))
return ""
}
let imageId = arguments[0]
guard let image = content.image(imageId) else {
results.missingFiles.insert(imageId)
return ""
}
results.files.insert(image)
let caption = arguments.count == 2 ? arguments[1] : nil
let altText = image.getDescription(for: language)
let path = content.pathToImage(image)
let thumbnail = FeedEntryData.Image(
rawImagePath: path,
width: thumbnailWidth,
height: thumbnailWidth,
altText: altText)
let largeImage = FeedEntryData.Image(
rawImagePath: path,
width: largeImageWidth,
height: largeImageWidth,
altText: altText)
return PageImage(
imageId: imageId.replacingOccurrences(of: ".", with: "-"),
thumbnail: thumbnail,
largeImage: largeImage,
caption: caption).content
}
private func handleHikingStatistics(_ arguments: [String]) -> String {
guard (1...5).contains(arguments.count) else {
results.invalidCommandArguments.append((.hikingStatistics, arguments))
return ""
}
let time = arguments[0].trimmed
let elevationUp = arguments.count > 1 ? arguments[1].trimmed : nil
let elevationDown = arguments.count > 2 ? arguments[2].trimmed : nil
let distance = arguments.count > 3 ? arguments[3].trimmed : nil
let calories = arguments.count > 4 ? arguments[4].trimmed : nil
return HikingStatistics(
time: time,
elevationUp: elevationUp,
elevationDown: elevationDown,
distance: distance,
calories: calories)
.content
}
private func handleDownloadButtons(_ arguments: [String]) -> String {
// ![download](<<fileId>,<text>,<download-filename?>;...)
let buttons: [ContentButtons.Item] = arguments.compactMap { button in
let parts = button.components(separatedBy: ",")
guard (2...3).contains(parts.count) else {
results.invalidCommandArguments.append((.downloadButtons, parts))
return nil
}
let file = parts[0].trimmed
let title = parts[1].trimmed
let downloadName = parts.count > 2 ? parts[2].trimmed : nil
// Ensure that file is available
guard let filePath = content.pathToFile(file) else {
results.missingFiles.insert(file)
return nil
}
return ContentButtons.Item(icon: .download, filePath: filePath, text: title, downloadFileName: downloadName)
}
return ContentButtons(items: buttons).content
}
private func handleVideo(_ arguments: [String]) -> String {
// ![video](<fileId>;<option1...>]
guard arguments.count >= 1 else {
results.invalidCommandArguments.append((.video, arguments))
return ""
}
let fileId = arguments[0].trimmed
let options = arguments.dropFirst().compactMap(convertVideoOption)
guard let file = content.file(id: fileId) else {
results.missingFiles.insert(fileId)
return ""
}
results.files.insert(file)
guard let videoType = file.type.videoType?.htmlType else {
results.warnings.append("Unknown video file type for \(fileId)")
return ""
}
let filePath = content.pathToFile(file)
return ContentPageVideo(
filePath: filePath,
videoType: videoType,
options: options)
.content
}
private func convertVideoOption(_ videoOption: String) -> VideoOption? {
guard let optionText = videoOption.trimmed.nonEmpty else {
return nil
}
guard let option = VideoOption(rawValue: optionText) else {
results.invalidCommandArguments.append((.video, [optionText]))
return nil
}
if case let .poster(imageId) = option {
if let image = content.image(imageId) {
results.files.insert(image)
let link = content.pathToImage(image)
let width = 2*thumbnailWidth
let fullLink = WebsiteImage.imagePath(source: link, width: width, height: width)
return .poster(image: fullLink)
} else {
results.missingFiles.insert(imageId)
return nil // Image file not present, so skip the option
}
}
if case let .src(videoId) = option {
if let video = content.video(videoId) {
results.files.insert(video)
let link = content.pathToVideo(video)
// TODO: Set correct video path?
return .src(link)
} else {
results.missingFiles.insert(videoId)
return nil // Video file not present, so skip the option
}
}
return option
}
/*
private func handleGif(file: String, altText: String) -> String {
let imagePath = page.pathRelativeToRootForContainedInputFile(file)
results.require(file: imagePath, source: page.path)
guard let size = results.getImageSize(atPath: imagePath, source: page.path) else {
return ""
}
let width = Int(size.width)
let height = Int(size.height)
return factory.html.image(file: file, width: width, height: height, altText: altText)
}
private func handleSvg(file: String, area: String?) -> String {
let imagePath = page.pathRelativeToRootForContainedInputFile(file)
results.require(file: imagePath, source: page.path)
guard let size = results.getImageSize(atPath: imagePath, source: page.path) else {
return "" // Missing image warning already produced
}
let width = Int(size.width)
let height = Int(size.height)
var altText = "image " + file.lastComponentAfter("/")
guard let area = area else {
return factory.html.image(file: file, width: width, height: height, altText: altText)
}
let parts = area.components(separatedBy: ",").map { $0.trimmed }
switch parts.count {
case 1:
return factory.html.image(file: file, width: width, height: height, altText: parts[0])
case 4:
break
case 5:
altText = parts[4]
default:
results.warning("Invalid area string for svg image", source: page.path)
return factory.html.image(file: file, width: width, height: height, altText: altText)
}
guard let x = Int(parts[0]),
let y = Int(parts[1]),
let partWidth = Int(parts[2]),
let partHeight = Int(parts[3]) else {
results.warning("Invalid area string for svg image", source: page.path)
return factory.html.image(file: file, width: width, height: height, altText: altText)
}
let part = SVGSelection(x, y, partWidth, partHeight)
return factory.html.svgImage(file: file, part: part, altText: altText)
}
private func handleFile(file: String, fileExtension: String) -> String {
results.warning("Unhandled file \(file) with extension \(fileExtension)", source: page.path)
return ""
}
*/
private func handleExternalButtons(_ arguments: [String]) -> String {
// ![external](<<url>;<text>...>
guard arguments.count >= 1 else {
results.invalidCommandArguments.append((.externalLink, arguments))
return ""
}
let buttons: [ContentButtons.Item] = arguments.compactMap { button in
let parts = button.components(separatedBy: ",")
guard parts.count == 2 else {
results.invalidCommandArguments.append((.externalLink, parts))
return nil
}
let rawUrl = parts[0].trimmed
guard let url = rawUrl.addingPercentEncoding(withAllowedCharacters: .urlQueryAllowed) else {
results.invalidCommandArguments.append((.externalLink, parts))
return nil
}
let title = parts[1].trimmed
return .init(
icon: .externalLink,
filePath: url,
text: title)
}
return ContentButtons(items: buttons).content
}
/*
private func handleExternalHTML(file: String) -> String {
let path = page.pathRelativeToRootForContainedInputFile(file)
return results.getContentOfRequiredFile(at: path, source: page.path) ?? ""
}
private func handleSimpleBox(content: String) -> String {
let parts = content.components(separatedBy: ";")
guard parts.count > 1 else {
results.warning("Invalid box specification", page: page)
return ""
}
let title = parts[0]
let text = parts.dropFirst().joined(separator: ";")
return factory.makePlaceholder(title: title, text: text)
}
private func handlePageLink(pageId: String) -> String {
guard let linkedPage = siteRoot.find(pageId) else {
// Checking the page path will add it to the missing pages
_ = results.getPagePath(for: pageId, source: page.path, language: language)
// Remove link since the page can't be found
return ""
}
guard linkedPage.state == .standard else {
// Prevent linking to unpublished content
return ""
}
var content = [PageLinkTemplate.Key: String]()
content[.title] = linkedPage.title(for: language)
content[.altText] = ""
let fullThumbnailPath = linkedPage.thumbnailFilePath(for: language).destination
// Note: Here we assume that the thumbnail was already used elsewhere, so already generated
let relativeImageUrl = page.relativePathToOtherSiteElement(file: fullThumbnailPath)
let metadata = linkedPage.localized(for: language)
if linkedPage.state.hasThumbnailLink {
let fullPageUrl = linkedPage.fullPageUrl(for: language)
let relativePageUrl = page.relativePathToOtherSiteElement(file: fullPageUrl)
content[.url] = "href=\"\(relativePageUrl)\""
}
content[.image] = relativeImageUrl.dropAfterLast(".")
if let suffix = metadata.thumbnailSuffix {
content[.title] = factory.html.make(title: metadata.title, suffix: suffix)
} else {
content[.title] = metadata.title
}
let path = linkedPage.makePath(language: language, from: siteRoot)
content[.path] = factory.pageLink.makePath(components: path)
content[.description] = metadata.relatedContentText
if let parent = linkedPage.findParent(from: siteRoot), parent.thumbnailStyle == .large {
content[.className] = " related-page-link-large"
}
// We assume that the thumbnail images are already required by overview pages.
return factory.pageLink.generate(content)
}
private func handle3dModel(content: String) -> String {
let parts = content.components(separatedBy: ";")
guard parts.count > 1 else {
results.warning("Invalid 3d model specification", page: page)
return ""
}
let file = parts[0]
guard file.hasSuffix(".glb") else {
results.warning("Invalid 3d model file \(file) (must be .glb)", page: page)
return ""
}
// Ensure that file is available
let filePath = page.pathRelativeToRootForContainedInputFile(file)
results.require(file: filePath, source: page.path)
// Add required file to head
headers.insert(.modelViewer)
let description = parts.dropFirst().joined(separator: ";")
return """
<model-viewer alt="\(description)" src="\(file)" ar shadow-intensity="1" camera-controls touch-action="pan-y"></model-viewer>
"""
}
*/
}