/**
 * Downloads a web page and all its assets. The main function of this class is to
 * download a web page and convert its resources (images, CSS files, etc.) to base64 format.
 * The reason for this is to create a self-contained HTML file that can be opened without
 * requiring any external resources, making it easier to view offline or distribute.
 */
public class WebPageDownloader extends WebPageLoader {
    private final Encoder resourceEncoder;
    public WebPageDownloader(String targetUrl) {
        this(targetUrl, null);
    }
    public WebPageDownloader(String targetUrl, HttpHeadersSpec httpHeadersSpec) {
        this(targetUrl, null, httpHeadersSpec);
    }
    public WebPageDownloader(String targetUrl, String cssSelector, HttpHeadersSpec httpHeadersSpec) {
        super(targetUrl, cssSelector, httpHeadersSpec);
        this.resourceEncoder = new Encoder();
    }
    /**
     * Downloads a web page and its resources, and converts them to a single HTML string.
     * This function is the main entry point for this class.
     */
    public byte[] downloadPageWithAssets() throws IOException, InterruptedException {
        WebClient webClient = getAvailableWebClient();
        try {
            HtmlPage page = prepareAndLoadPage(webClient);
            waitForSelectorIfNeeded(page);
            URL baseUrl = page.getFullyQualifiedUrl(page.getBaseURI());
            encodePageResources(page, baseUrl);
            removeJavascriptTags(page);
            String finalHtml = page.asXml();
            return finalHtml.getBytes(StandardCharsets.UTF_8);
        } finally {
            releaseWebClient(webClient);
        }
    }
    private void waitForSelectorIfNeeded(HtmlPage page) throws InterruptedException {
        String cssSelector = getCssSelector();
        if (cssSelector != null) {
            long startTime = System.currentTimeMillis();
            waitForSelector(page);
            long executionTime = System.currentTimeMillis() - startTime;
            System.out.println("Waited for CSS selector: '" + cssSelector + "'  for " + executionTime + " ms");
        }
    }
    private void encodePageResources(HtmlPage page, URL baseUrl) {
        List<HtmlElement> elements = page.getByXPath("//*");
        for (HtmlElement element : elements) {
            encodeResourceAttributes(element, baseUrl);
        }
    }
    private void encodeResourceAttributes(HtmlElement element, URL baseUrl) {
        String src = element.getAttribute("src");
        String href = element.getAttribute("href");
        String style = element.getAttribute("style");
        if (!src.isEmpty()) {
            encodeResourceAttribute(element, src, baseUrl, "src");
        } else if (!href.isEmpty()) {
            encodeResourceAttribute(element, href, baseUrl, "href");
        } else if (!style.isEmpty()) {
            encodeResourceStyle(element, style, baseUrl);
        }
    }
    private void encodeResourceAttribute(HtmlElement element, String url, URL baseUrl, String attributeName) {
        String resourceUrl = UrlUtils.resolveUrl(baseUrl, url);
        if (resourceUrl.endsWith(".css")) {
            replaceCssLinkWithStyle(element, resourceUrl);
        } else {
            encodeAttribute(element, resourceUrl, attributeName);
        }
    }
    private void replaceCssLinkWithStyle(HtmlElement element, String resourceUrl) {
        try {
            String cssContent = loadResourceContent(resourceUrl);
            String encodedCssContent = encodeResourceInStyle(cssContent, new URL(resourceUrl));
            insertStyleBeforeElement(element, encodedCssContent);
            element.remove();
        } catch (IOException e) {
            System.err.println(e.getMessage());
        }
    }
    private void encodeAttribute(HtmlElement element, String resourceUrl, String attributeName) {
        String dataUrlBase64 = resourceEncoder.encode(resourceUrl);
        element.setAttribute(attributeName, dataUrlBase64);
    }
    private String loadResourceContent(String resourceUrl) throws IOException {
        InputStream cssInStream = resourceEncoder.getUrlAsStream(resourceUrl);
        byte[] cssInBytes = cssInStream.readAllBytes();
        return new String(cssInBytes, StandardCharsets.UTF_8);
    }
    private void insertStyleBeforeElement(HtmlElement element, String cssContent) {
        DomNode parentNode = element.getParentNode();
        HtmlPage page = element.getHtmlPageOrNull();
        HtmlStyle newElement = (HtmlStyle) page.createElement("style");
        newElement.setAttribute("type", "text/css");
        newElement.setTextContent(cssContent);
        parentNode.insertBefore(newElement, element);
    }
    private void encodeResourceStyle(HtmlElement element, String style, URL baseUrl) {
        String encodedStyle = encodeResourceInStyle(style, baseUrl);
        element.setAttribute("style", encodedStyle);
    }
    /**
     * Encodes URLs inside a CSS or style attribute. This is to make sure the CSS rules are
     * self-contained and do not require external resources.
     */
    private String encodeResourceInStyle(String style, URL baseUrl) {
        Pattern urlPattern = Pattern.compile("url\\(['\"]?(.*?)['\"]?\\)");
        Matcher matcher = urlPattern.matcher(style);
        StringBuilder encodedStyle = new StringBuilder();
        int lastIndex = 0;
        while (matcher.find()) {
            String resourceUrl = UrlUtils.resolveUrl(baseUrl, matcher.group(1));
            // Append everything before the current match
            encodedStyle.append(style.substring(lastIndex, matcher.start()));
            if (resourceUrl.endsWith(".css")) {
                encodedStyle.append(encodeCssResource(matcher, resourceUrl));
            } else {
                encodedStyle.append(encodeNonCssResource(matcher, resourceUrl));
            }
            lastIndex = matcher.end();
        }
        // Append everything after the last match
        encodedStyle.append(style.substring(lastIndex));
        return encodedStyle.toString();
    }
    private String encodeCssResource(Matcher matcher, String resourceUrl) {
        try {
            InputStream cssInStream = resourceEncoder.getUrlAsStream(resourceUrl);
            byte[] cssInBytes = cssInStream.readAllBytes();
            String cssText = new String(cssInBytes, StandardCharsets.UTF_8);
            return "url(" + encodeResourceInStyle(cssText, new URL(resourceUrl)) + ")";
        } catch (IOException e) {
            throw new RuntimeException(e);
        }
    }
    private String encodeNonCssResource(Matcher matcher, String resourceUrl) {
        String dataUrlBase64 = resourceEncoder.encode(resourceUrl);
        return "url(" + dataUrlBase64 + ")";
    }
    private void removeJavascriptTags(HtmlPage page) {
        List<HtmlElement> scripts = page.getByXPath("//script");
        for(HtmlElement script : scripts){
            script.remove();
        }
    }
}