ReadViewSDK/Sources/RDReaderView/EPUBTextRendering/RDEPUBTextBookBuilder.swift

import UIKit

public struct RDEPUBTextPage: Equatable {
    public var absolutePageIndex: Int
    public var chapterIndex: Int
    public var spineIndex: Int
    public var href: String
    public var chapterTitle: String
    public var pageIndexInChapter: Int
    public var totalPagesInChapter: Int
    public var content: NSAttributedString
    public var contentRange: NSRange
    public var pageStartOffset: Int
    public var pageEndOffset: Int
}

public struct RDEPUBTextChapter: Equatable {
    public var chapterIndex: Int
    public var spineIndex: Int
    public var href: String
    public var title: String
    public var attributedContent: NSAttributedString
    public var fragmentOffsets: [String: Int]
    public var pages: [RDEPUBTextPage]
}

public struct RDEPUBTextBook: Equatable {
    public var chapters: [RDEPUBTextChapter]
    public var pages: [RDEPUBTextPage]

    public init(chapters: [RDEPUBTextChapter], pages: [RDEPUBTextPage]) {
        self.chapters = chapters
        self.pages = pages
    }

    public func page(at pageNumber: Int) -> RDEPUBTextPage? {
        guard pageNumber > 0, pages.indices.contains(pageNumber - 1) else {
            return nil
        }
        return pages[pageNumber - 1]
    }

    public func pageNumber(for location: RDEPUBLocation, resolver: RDEPUBResourceResolver, bookIdentifier: String?) -> Int? {
        guard let normalizedLocation = resolver.normalizedLocation(location, bookIdentifier: bookIdentifier),
              let chapter = chapters.first(where: { $0.href == normalizedLocation.href }) else {
            return nil
        }

        let targetOffset: Int
        if let fragment = normalizedLocation.fragment, let fragmentOffset = chapter.fragmentOffsets[fragment] {
            targetOffset = fragmentOffset
        } else {
            let lastOffset = max(chapter.attributedContent.length - 1, 0)
            targetOffset = min(lastOffset, max(0, Int(round(Double(lastOffset) * normalizedLocation.navigationProgression))))
        }

        if let page = chapter.pages.first(where: { targetOffset >= $0.pageStartOffset && targetOffset <= $0.pageEndOffset }) {
            return page.absolutePageIndex + 1
        }

        return chapter.pages.last.map { $0.absolutePageIndex + 1 }
    }

    public func location(forPageNumber pageNumber: Int, bookIdentifier: String?) -> RDEPUBLocation? {
        guard let page = page(at: pageNumber),
              let chapter = chapters.first(where: { $0.chapterIndex == page.chapterIndex }) else {
            return nil
        }

        let totalLength = max(chapter.attributedContent.length - 1, 1)
        return RDEPUBLocation(
            bookIdentifier: bookIdentifier,
            href: page.href,
            progression: Double(page.pageStartOffset) / Double(totalLength),
            lastProgression: Double(page.pageEndOffset) / Double(totalLength),
            fragment: nil
        )
    }
}

public final class RDEPUBTextBookBuilder {
    private let renderer: RDEPUBTextRenderer

    public init(renderer: RDEPUBTextRenderer) {
        self.renderer = renderer
    }

    public convenience init() {
        self.init(renderer: RDEPUBDTCoreTextRenderer())
    }

    public func build(
        parser: RDEPUBParser,
        publication: RDEPUBPublication,
        pageSize: CGSize,
        style: RDEPUBTextRenderStyle
    ) throws -> RDEPUBTextBook {
        var chapters: [RDEPUBTextChapter] = []
        var flatPages: [RDEPUBTextPage] = []

        for (spineIndex, item) in publication.spine.enumerated() where item.linear {
            guard item.mediaType.contains("html") || item.mediaType.contains("xhtml"),
                  let rawHTML = parser.htmlString(forRelativePath: item.href) else {
                continue
            }

            let chapterTitle = resolvedChapterTitle(for: item, toc: publication.tableOfContents)
            let normalizedHTML = normalizeHTML(rawHTML)
            let rendered = try renderer.renderChapter(
                html: normalizedHTML,
                baseURL: parser.fileURL(forRelativePath: item.href)?.deletingLastPathComponent(),
                style: style
            )

            let plainText = rendered.attributedString.string.trimmingCharacters(in: .whitespacesAndNewlines)
            if shouldSkipChapter(item: item, text: plainText) {
                continue
            }

            let chapterIndex = chapters.count
            let content = NSMutableAttributedString(attributedString: rendered.attributedString)
            let pageRanges = content.length > 0 ? content.ss_pageRanges(size: pageSize) : []
            let effectivePageRanges = pageRanges.isEmpty && content.length > 0
                ? [NSRange(location: 0, length: content.length)]
                : pageRanges

            let pages = effectivePageRanges.enumerated().map { localPageIndex, range in
                RDEPUBTextPage(
                    absolutePageIndex: flatPages.count + localPageIndex,
                    chapterIndex: chapterIndex,
                    spineIndex: spineIndex,
                    href: item.href,
                    chapterTitle: chapterTitle,
                    pageIndexInChapter: localPageIndex,
                    totalPagesInChapter: effectivePageRanges.count,
                    content: content.attributedSubstring(from: range),
                    contentRange: range,
                    pageStartOffset: range.location,
                    pageEndOffset: range.location + max(range.length - 1, 0)
                )
            }

            chapters.append(
                RDEPUBTextChapter(
                    chapterIndex: chapterIndex,
                    spineIndex: spineIndex,
                    href: item.href,
                    title: chapterTitle,
                    attributedContent: content.copy() as! NSAttributedString,
                    fragmentOffsets: rendered.fragmentOffsets,
                    pages: pages
                )
            )
            flatPages.append(contentsOf: pages)
        }

        return RDEPUBTextBook(chapters: chapters, pages: flatPages)
    }

    private func resolvedChapterTitle(for item: RDEPUBSpineItem, toc: [EPUBTableOfContentsItem]) -> String {
        if let title = flattenedTOCItems(from: toc).first(where: { tocItem in
            tocItem.href.components(separatedBy: "#").first == item.href
        })?.title.trimmingCharacters(in: .whitespacesAndNewlines), !title.isEmpty {
            return title
        }
        let trimmedTitle = item.title.trimmingCharacters(in: .whitespacesAndNewlines)
        return trimmedTitle.isEmpty ? item.href : trimmedTitle
    }

    private func flattenedTOCItems(from items: [EPUBTableOfContentsItem]) -> [EPUBTableOfContentsItem] {
        items.flatMap { item in
            [item] + flattenedTOCItems(from: item.children)
        }
    }

    private func shouldSkipChapter(item: RDEPUBSpineItem, text: String) -> Bool {
        let lowercasedHref = item.href.lowercased()
        if text.isEmpty && (lowercasedHref.contains("cover") || lowercasedHref.contains("title")) {
            return true
        }
        return false
    }

    private func normalizeHTML(_ html: String) -> String {
        var cleanedHTML = html
        let replacements: [(pattern: String, template: String)] = [
            (#"<hr\s+lang="zh-CN">分页符</hr>"#, ""),
            (#"\r"#, "\n"),
            (#"\n+"#, "\n")
        ]

        for replacement in replacements {
            if let regex = try? NSRegularExpression(pattern: replacement.pattern, options: [.caseInsensitive]) {
                cleanedHTML = regex.stringByReplacingMatches(
                    in: cleanedHTML,
                    options: [],
                    range: NSRange(location: 0, length: cleanedHTML.utf16.count),
                    withTemplate: replacement.template
                )
            }
        }

        return cleanedHTML
    }
}