ChWebsiteApp/CHDataManagement/Generator/PageContentGenerator.swift

import Foundation
import Ink
import Splash
import SwiftSoup

typealias VideoSource = (url: String, type: VideoFileType)

final class PageContentParser {

    private let pageLinkMarker = "page:"

    private let tagLinkMarker = "tag:"

    private static let codeHighlightFooter = "<script>hljs.highlightAll();</script>"

    private let swift = SyntaxHighlighter(format: HTMLOutputFormat())

    let results = PageGenerationResults()

    private let content: Content

    private let buttonHandler: ButtonCommandProcessor

    private let labelHandler: LabelsCommandProcessor

    private let audioPlayer: AudioPlayerCommandProcessor

    let language: ContentLanguage

    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
        self.buttonHandler = .init(content: content, results: results)
        self.labelHandler = .init(content: content, results: results)
        self.audioPlayer = .init(content: content, results: results)
    }

    func requestImages(_ generator: ImageGenerator) {
        for request in results.imagesToGenerate {
            generator.generateImageSet(
                for: request.image.id,
                maxWidth: CGFloat(request.size),
                maxHeight: CGFloat(request.size))
        }
    }

    func reset() {
        results.reset()
    }

    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 {
            results.requiredHeaders.insert(.codeHightlighting)
            results.requiredFooters.insert(PageContentParser.codeHighlightFooter)
            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) {
            return handlePageLink(file: file, html: html, markdown: markdown)
        }
        if file.hasPrefix(tagLinkMarker) {
            return handleTagLink(file: file, html: html, markdown: markdown)
        }
        results.externalLinks.insert(file)
        return html
    }

    private func handlePageLink(file: String, html: String, markdown: Substring) -> String {
        // 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.missing(page: pageId, markdown: markdown)
            // Remove link since the page can't be found
            return markdown.between("[", and: "]")
        }
        results.linkedPages.insert(page)
        let pagePath = page.absoluteUrl(for: language)
        return html.replacingOccurrences(of: textToChange, with: pagePath)
    }

    private func handleTagLink(file: String, html: String, markdown: Substring) -> String {
        // Retain links pointing to elements within a page
        let textToChange = file.dropAfterFirst("#")
        let tagId = textToChange.replacingOccurrences(of: tagLinkMarker, with: "")
        guard let tag = content.tag(tagId) else {
            results.missing(tag: tagId, markdown: markdown)
            // Remove link since the tag can't be found
            return markdown.between("[", and: "]")
        }
        results.linkedTags.insert(tag)
        let tagPath = content.absoluteUrlToTag(tag, language: language)
        return html.replacingOccurrences(of: textToChange, with: tagPath)
    }

    private func handleHTML(_: String, markdown: Substring) -> String {
        let result = String(markdown)
        #warning("Check HTML code in markdown for required resources")
        findImages(in: result)
        findLinks(in: result)
        findSourceSets(in: result)
        // Things to check: <img src=  <a href= <source>
        return result
    }

    private func findImages(in markdown: String) {
        do {
            // Parse the HTML string
            let document = try SwiftSoup.parse(markdown)

            // Select all 'img' elements
            let imgElements = try document.select("img")

            // Extract the 'src' attributes from each 'img' element
            let srcAttributes = try imgElements.array().compactMap { try $0.attr("src") }

            for src in srcAttributes {
                print("Found image in html: \(src)")
            }
        } catch {
            print("Error parsing HTML: \(error)")
        }
    }

    private func findLinks(in markdown: String) {
        do {
            // Parse the HTML string
            let document = try SwiftSoup.parse(markdown)

            // Select all 'img' elements
            let linkElements = try document.select("a")

            // Extract the 'src' attributes from each 'img' element
            let srcAttributes = try linkElements.array().compactMap { try $0.attr("href") }

            for src in srcAttributes {
                print("Found link in html: \(src)")
            }
        } catch {
            print("Error parsing HTML: \(error)")
        }
    }

    private func findSourceSets(in markdown: String) {
        do {
            // Parse the HTML string
            let document = try SwiftSoup.parse(markdown)

            // Select all 'img' elements
            let linkElements = try document.select("source")

            // Extract the 'src' attributes from each 'img' element
            let srcAttributes = try linkElements.array().compactMap { try $0.attr("srcset") }

            for src in srcAttributes {
                print("Found source set in html: \(src)")
            }
        } catch {
            print("Error parsing HTML: \(error)")
        }
    }

    /**
     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 percentDecoded(_ string: String) -> String {
        guard let decoded = string.removingPercentEncoding else {
            print("Invalid string: \(string)")
            return string
        }
        return decoded
    }

    private func processMarkdownImage(html: String, markdown: Substring) -> String {
        //
        let argumentList = percentDecoded(markdown.between(first: "](", andLast: ")"))
        let arguments = argumentList.components(separatedBy: ";")


        let rawCommand = percentDecoded(markdown.between("![", and: "]").trimmed)
        guard rawCommand != "" else {
            return handleImage(arguments, markdown: markdown)
        }

        guard let command = ShorthandMarkdownKey(rawValue: rawCommand) else {
            // Treat unknown commands as normal links
            results.invalid(command: nil, markdown)
            return html
        }

        switch command {
        case .image:
            return handleImage(arguments, markdown: markdown)
        case .labels:
            return labelHandler.process(arguments, markdown: markdown)
        case .buttons:
            return buttonHandler.process(arguments, markdown: markdown)
        case .video:
            return handleVideo(arguments, markdown: markdown)
        case .pageLink:
            return handlePageLink(arguments, markdown: markdown)
        case .includedHtml:
            return handleExternalHtml(arguments, markdown: markdown)
        case .box:
            return handleSimpleBox(arguments, markdown: markdown)
        case .model:
            return handleModel(arguments, markdown: markdown)
        case .svg:
            return handleSvg(arguments, markdown: markdown)
        case .audioPlayer:
            return audioPlayer.process(arguments, markdown: markdown)
        default:
            results.invalid(command: nil, markdown)
            return ""
        }
    }

    /**
     Format: `[image](<imageId>;<caption?>]`
     */
    private func handleImage(_ arguments: [String], markdown: Substring) -> String {
        guard (1...2).contains(arguments.count) else {
            results.invalid(command: .image, markdown)
            return ""
        }
        let imageId = arguments[0]

        guard let image = content.image(imageId) else {
            results.missing(file: imageId, markdown: markdown)
            return ""
        }
        results.files.insert(image)

        let caption = arguments.count == 2 ? arguments[1] : nil
        let altText = image.getDescription(for: language)

        let path = image.absoluteUrl

        guard !image.type.isSvg else {
            return SvgImage(imagePath: path, altText: altText).content
        }

        let thumbnail = FeedEntryData.Image(
            rawImagePath: path,
            width: thumbnailWidth,
            height: thumbnailWidth,
            altText: altText)
        results.imagesToGenerate.insert(.init(size: thumbnailWidth, image: image))

        let largeImage = FeedEntryData.Image(
            rawImagePath: path,
            width: largeImageWidth,
            height: largeImageWidth,
            altText: altText)
        results.imagesToGenerate.insert(.init(size: largeImageWidth, image: image))

        return PageImage(
            imageId: imageId.replacingOccurrences(of: ".", with: "-"),
            thumbnail: thumbnail,
            largeImage: largeImage,
            caption: caption).content
    }

    /**
     Format: `![video](<fileId>;<option1...>]`
     */
    private func handleVideo(_ arguments: [String], markdown: Substring) -> String {
        guard arguments.count >= 1 else {
            results.invalid(command: .video, markdown)
            return ""
        }
        let fileId = arguments[0].trimmed

        let options = arguments.dropFirst().compactMap { convertVideoOption($0, markdown: markdown) }

        guard let file = content.file(fileId) else {
            results.missing(file: fileId, markdown: markdown)
            return ""
        }
        results.files.insert(file)

        guard let videoType = file.type.videoType?.htmlType else {
            results.invalid(command: .video, markdown)
            return ""
        }

        return ContentPageVideo(
            filePath: file.absoluteUrl,
            videoType: videoType,
            options: options)
        .content
    }

    private func convertVideoOption(_ videoOption: String, markdown: Substring) -> VideoOption? {
        guard let optionText = videoOption.trimmed.nonEmpty else {
            return nil
        }
        guard let option = VideoOption(rawValue: optionText) else {
            results.invalid(command: .video, markdown)
            return nil
        }
        if case let .poster(imageId) = option {
            if let image = content.image(imageId) {
                results.files.insert(image)
                let width = 2*thumbnailWidth
                let fullLink = WebsiteImage.imagePath(source: image.absoluteUrl, width: width, height: width)
                return .poster(image: fullLink)
            } else {
                results.missing(file: imageId, markdown: markdown)
                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 = video.absoluteUrl
                // TODO: Set correct video path?
                return .src(link)
            } else {
                results.missing(file: videoId, markdown: markdown)
                return nil // Video file not present, so skip the option
            }
        }
        return option
    }

    /**
     Format: `![html](<fileId>)`
     */
    private func handleExternalHtml(_ arguments: [String], markdown: Substring) -> String {
        guard arguments.count == 1 else {
            results.invalid(command: .includedHtml, markdown)
            return ""
        }
        let fileId = arguments[0]
        guard let file = content.file(fileId) else {
            results.missing(file: fileId, markdown: markdown)
            return ""
        }
        return file.textContent()
    }

    /**
     Format: `![box](<title>;<body>)`
     */
    private func handleSimpleBox(_ arguments: [String], markdown: Substring) -> String {
        guard arguments.count > 1 else {
            results.invalid(command: .box, markdown)
            return ""
        }
        let title = arguments[0]
        let text = arguments.dropFirst().joined(separator: ";")
        return ContentBox(title: title, text: text).content
    }

    /**
     Format: `![page](<pageId>)`
     */
    private func handlePageLink(_ arguments: [String], markdown: Substring) -> String {
        guard arguments.count == 1 else {
            results.invalid(command: .pageLink, markdown)
            return ""
        }
        let pageId = arguments[0]

        guard let page = content.page(pageId) else {
            results.missing(page: pageId, markdown: markdown)
            return ""
        }
        guard !page.isDraft else {
            // Prevent linking to unpublished content
            return ""
        }

        let localized = page.localized(in: language)
        let url = page.absoluteUrl(for: language)
        let title = localized.linkPreviewTitle ?? localized.title
        let description = localized.linkPreviewDescription ?? ""

        let image = localized.linkPreviewImage.map { image in
            let size = content.settings.pages.pageLinkImageSize
            results.files.insert(image)
            results.imagesToGenerate.insert(.init(size: size, image: image))

            return RelatedPageLink.Image(
                url: image.absoluteUrl,
                description: image.getDescription(for: language),
                size: size)
        }

        return RelatedPageLink(
            title: title,
            description: description,
            url: url,
            image: image)
            .content
    }

    /**
     Format: `![model](<file>)`
     */
    private func handleModel(_ arguments: [String], markdown: Substring) -> String {
        guard arguments.count == 1 else {
            results.invalid(command: .model, markdown)
            return ""
        }
        let fileId = arguments[0]
        guard fileId.hasSuffix(".glb") else {
            results.invalid(command: .model, markdown)
            return ""
        }

        guard let file = content.file(fileId) else {
            results.missing(file: fileId, markdown: markdown)
            return ""
        }
        results.files.insert(file)
        results.requiredHeaders.insert(.modelViewer)

        let description = file.getDescription(for: language)
        return ModelViewer(file: file.absoluteUrl, description: description).content
    }

    private func handleSvg(_ arguments: [String], markdown: Substring) -> String {
        guard arguments.count == 5 else {
            results.invalid(command: .svg, markdown)
            return ""
        }

        guard let x = Int(arguments[1]),
              let y = Int(arguments[2]),
              let partWidth = Int(arguments[3]),
              let partHeight = Int(arguments[4]) else {
            results.invalid(command: .svg, markdown)
            return ""
        }

        let imageId = arguments[0]

        guard let image = content.image(imageId) else {
            results.missing(file: imageId, markdown: markdown)
            return ""
        }
        guard case .image(let imageType) = image.type,
              imageType == .svg else {
            results.invalid(command: .svg, markdown)
            return ""
        }

        return PartialSvgImage(
            imagePath: image.absoluteUrl,
            altText: image.getDescription(for: language),
            x: x,
            y: y,
            width: partWidth,
            height: partHeight)
        .content
    }
}

/*
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)
}
*/