Java Code Examples for org.jsoup.nodes.Document#html()

private String addResourcesInMessage(final MimeMessageHelper mailMessage, final String htmlText) throws Exception {
    final Document document = Jsoup.parse(htmlText);

    final List<String> resources = new ArrayList<>();

    final Elements imageElements = document.getElementsByTag("img");
    resources.addAll(imageElements.stream()
            .filter(imageElement -> imageElement.hasAttr("src"))
            .filter(imageElement -> !imageElement.attr("src").startsWith("http"))
            .map(imageElement -> {
                final String src = imageElement.attr("src");
                imageElement.attr("src", "cid:" + src);
                return src;
            })
            .collect(Collectors.toList()));

    final String html = document.html();
    mailMessage.setText(html, true);

    for (final String res : resources) {
        final FileSystemResource templateResource = new FileSystemResource(new File(templatesPath, res));
        mailMessage.addInline(res, templateResource, getContentTypeByFileName(res));
    }

    return html;
}
 

static String toHtml(Spanned spanned, Context context) {
    HtmlEx converter = new HtmlEx(context);
    String html = converter.toHtml(spanned, TO_HTML_PARAGRAPH_LINES_CONSECUTIVE);

    // @Google: why convert size to and from in a different way?
    Document doc = JsoupEx.parse(html);
    for (Element element : doc.select("span")) {
        String style = element.attr("style");
        if (style.startsWith("font-size:")) {
            int colon = style.indexOf(':');
            int semi = style.indexOf("em;", colon);
            if (semi > colon)
                try {
                    String hsize = style.substring(colon + 1, semi).replace(',', '.');
                    float size = Float.parseFloat(hsize);
                    element.tagName(size < 1.0f ? "small" : "big");
                    element.attributes().remove("style");
                } catch (NumberFormatException ex) {
                    Log.e(ex);
                }
        }
    }

    return doc.html();
}
 

/**
 * 将html中外联的css变成内联,并去掉外联样式
 * @author Frodez
 * @date 2019-03-21
 */
@Override
public String revert(String html) {
	Assert.notNull(html, "html must not be null");
	try {
		Document document = Jsoup.parse(html);
		Elements links = document.select("link[href]");
		Elements htmlElement = document.select("html");
		for (Element iter : links) {
			String path = iter.attr("href");
			if (!path.endsWith(".css")) {
				continue;
			}
			htmlElement.prepend(StrUtil.concat("<style type=\"text/css\">", FileUtil.readString(ResourceUtils
				.getFile(StrUtil.concat(FreemarkerRender.getLoaderPath(), path))), "</style>"));
		}
		links.remove();
		return document.html();
	} catch (Exception e) {
		log.error("[frodez.util.renderer.reverter.CSSReverter.revert]", e);
		return html;
	}
}
 

private String addResourcesInMessage(final MimeMessageHelper mailMessage, final String htmlText) throws Exception {
    final Document document = Jsoup.parse(htmlText);

    final List<String> resources = new ArrayList<>();

    final Elements imageElements = document.getElementsByTag("img");
    resources.addAll(imageElements.stream()
            .filter(imageElement -> imageElement.hasAttr("src"))
            .filter(imageElement -> !imageElement.attr("src").startsWith("http"))
            .map(imageElement -> {
                final String src = imageElement.attr("src");
                imageElement.attr("src", "cid:" + src);
                return src;
            })
            .collect(Collectors.toList()));

    final String html = document.html();
    mailMessage.setText(html, true);

    for (final String res : resources) {
        final FileSystemResource templateResource = new FileSystemResource(new File(templatesPath, res));
        mailMessage.addInline(res, templateResource, getContentTypeByFileName(res));
    }

    return html;
}
 

public boolean save(@NotNull Document document) throws Exception {
	convertURLsToBase(document);
	String documentAsString = document.html();

	if (documentAsString.length() <= skipLength) {
		System.out.println("Skipping:" + url);
		return false;
	}

	File outputFile = new File((basePath != null ? basePath + "/" : "") + saveName);
	outputFile.createNewFile();
	System.out.println("Saving " + url + " to " + outputFile);


	FileOutputStream fos = new FileOutputStream(outputFile);
	fos.write(documentAsString.getBytes());
	fos.flush();
	fos.close();

	if (latch != null) {
		latch.countDown();
	}

	return true;
}
 

private String addResourcesInMessage(final MimeMessageHelper mailMessage, final String htmlText) throws Exception {
    final Document document = Jsoup.parse(htmlText);

    final List<String> resources = new ArrayList<>();

    final Elements imageElements = document.getElementsByTag("img");
    resources.addAll(imageElements.stream()
            .filter(imageElement -> imageElement.hasAttr("src"))
            .filter(imageElement -> !imageElement.attr("src").startsWith("http"))
            .map(imageElement -> {
                final String src = imageElement.attr("src");
                imageElement.attr("src", "cid:" + src);
                return src;
            })
            .collect(Collectors.toList()));

    final String html = document.html();
    mailMessage.setText(html, true);

    for (final String res : resources) {
        final FileSystemResource templateResource = new FileSystemResource(new File(templatesPath, res));
        mailMessage.addInline(res, templateResource, getContentTypeByFileName(res));
    }

    return html;
}
 

public boolean save(@NotNull Document document) throws Exception {
	convertURLsToBase(document);
	String documentAsString = document.html();

	if (documentAsString.length() <= skipLength) {
		System.out.println("Skipping:" + url);
		return false;
	}

	File outputFile = new File((basePath != null ? basePath + "/" : "") + saveName);
	outputFile.createNewFile();
	System.out.println("Saving " + url + " to " + outputFile);


	FileOutputStream fos = new FileOutputStream(outputFile);
	fos.write(documentAsString.getBytes());
	fos.flush();
	fos.close();

	if (latch != null) {
		latch.countDown();
	}

	return true;
}
 

private String parse(String str) {
    Document document = Jsoup.parse(str, "", Parser.xmlParser());
    String result = "";
    switch (outType) {
        case TYPE_TEXT:
            result = document.text();
            break;
        case TYPE_HTML:
            result = document.html();
            break;
        default:
            result = document.text();
            break;
    }
    return result;
}
 

private String parseBody(String body) {
    Matcher m = PatternExtensions.compile("br \\/>\\s*(<fieldset>[\\S\\s]*<.form>)").matcher(body);
    if (m.find()) {
        body = "<form>" + m.group(1);
        //body =  + "</form><input type=\"button\" value=\"asdghjk\" onclick=\"jsonElem();\">";
        body = body.replaceAll("<td class=\"row1\" width=\"30%\"><b>О себе:</b>[\\s\\S]*?</td>",
                "<td class=\"row1\" width=\"30%\"><b>О себе</b></td>");
        body = body.replaceAll("<td width=\"30%\" class=\"row1\" style='padding:6px;'><b>Город</b>[\\s\\S]*?</td>",
                "<td class=\"row1\" width=\"30%\" style='padding:6px;'><b>Город</b></td>");
        body = body.replaceAll("legend", "h2").replaceAll("<fieldset>", "<div class=\"field\">").replaceAll("</fieldset>", "</div>");
        Document doc = Jsoup.parse(body);
        doc.select(".formbuttonrow .button").remove();
        doc.select(".formbuttonrow").append("<input type=\"button\" value=\"Сохранить\" onclick=\"jsonElem();\">");
        doc.select("textarea").first().attr("maxlength", "500");
        body = doc.html();
    }
    return body;
}
 

/**
* 处理 对话消息中的图片
* @param message
* @return
*/
  public static String filterChatMessage(String message){
  	Document document = Jsoup.parse(message) ;
  	Elements pngs = document.select("img[src]");
  	for (Element element : pngs) {
  		String imgUrl = element.attr("src");
  		if(imgUrl.indexOf("/res/image") >= 0){
  			element.attr("class", "ukefu-media-image") ;
  		}
  	}
  	return document.html() ;
  }
 

private String getUpdatedFileContent(List<Vacancy> vacancies) {

        Document document = null;
        try {
            document = getDocument();

            Element templateOriginal = document.getElementsByClass("template").first();
            Element copyTemplate = templateOriginal.clone();
            copyTemplate.removeAttr("style");
            copyTemplate.removeClass("template");
            document.select("tr[class=vacancy]").remove().not("tr[class=vacancy template");

            for (Vacancy vacancy : vacancies) {
                Element localClone = copyTemplate.clone();
                localClone.getElementsByClass("city").first().text(vacancy.getCity());
                localClone.getElementsByClass("companyName").first().text(vacancy.getCompanyName());
                localClone.getElementsByClass("salary").first().text(vacancy.getSalary());
                Element link =localClone.getElementsByTag("a").first();
                link.text(vacancy.getTitle());
                link.attr("href", vacancy.getUrl());

                templateOriginal.before(localClone.outerHtml());
            }
        } catch (IOException e) {
            e.printStackTrace();
            return "Some exception occurred";
        }
        return document.html();
    }
 

private String removeUnknownMacroElements(String result) {
    Document document = Jsoup.parse(result);
    document.getElementsByClass("wysiwyg-unknown-macro").stream()
        .filter(element -> element.attr("src").contains("livingdoc"))
        .forEach(element -> element.remove());

    return document.html();
}
 

private static String addStyleAttributes(String htmlString) {
    Document graphDoc = Jsoup.parse(htmlString);
    Element htmlNode = graphDoc.getElementsByTag("html").get(0);
    htmlNode.attr("style", "height: 100%");
    Element bodyNode = graphDoc.getElementsByTag("body").get(0);
    bodyNode.attr("style", "height: 90%; margin:0;");
    return graphDoc.html();
}
 

@Test
public void givenMultiPagePdf_renderOnlyFirstPage_outputHtmlOnlyHasFirstPage() throws Exception
{
    Document htmlDoc = convertWithPageRange(testPath + "3-page-document.pdf", 0, 1);
    String htmlText = htmlDoc.html();

    Assert.assertThat(htmlText, containsString("#1"));

    Assert.assertThat(htmlText, not(containsString("#2")));
    Assert.assertThat(htmlText, not(containsString("#3")));
}