Sur la base des autres réponses et des commentaires sur cette question, il semble que la plupart des gens qui viennent ici recherchent vraiment une solution générale qui fournira une représentation en texte brut bien formatée d'un document HTML. Je sais que je l'étais.
Heureusement, JSoup fournit déjà un exemple assez complet de la façon d'y parvenir: HtmlToPlainText.java
L'exemple FormattingVisitorpeut facilement être modifié selon vos préférences et traite de la plupart des éléments de bloc et de l'habillage de ligne.
Pour éviter la pourriture des liens, voici la solution de Jonathan Hedley dans son intégralité:
package org.jsoup.examples;
import org.jsoup.Jsoup;
import org.jsoup.helper.StringUtil;
import org.jsoup.helper.Validate;
import org.jsoup.nodes.Document;
import org.jsoup.nodes.Element;
import org.jsoup.nodes.Node;
import org.jsoup.nodes.TextNode;
import org.jsoup.select.Elements;
import org.jsoup.select.NodeTraversor;
import org.jsoup.select.NodeVisitor;
import java.io.IOException;
/**
* HTML to plain-text. This example program demonstrates the use of jsoup to convert HTML input to lightly-formatted
* plain-text. That is divergent from the general goal of jsoup's .text() methods, which is to get clean data from a
* scrape.
* <p>
* Note that this is a fairly simplistic formatter -- for real world use you'll want to embrace and extend.
* </p>
* <p>
* To invoke from the command line, assuming you've downloaded the jsoup jar to your current directory:</p>
* <p><code>java -cp jsoup.jar org.jsoup.examples.HtmlToPlainText url [selector]</code></p>
* where <i>url</i> is the URL to fetch, and <i>selector</i> is an optional CSS selector.
*
* @author Jonathan Hedley, jonathan@hedley.net
*/
public class HtmlToPlainText {
private static final String userAgent = "Mozilla/5.0 (jsoup)";
private static final int timeout = 5 * 1000;
public static void main(String... args) throws IOException {
Validate.isTrue(args.length == 1 || args.length == 2, "usage: java -cp jsoup.jar org.jsoup.examples.HtmlToPlainText url [selector]");
final String url = args[0];
final String selector = args.length == 2 ? args[1] : null;
// fetch the specified URL and parse to a HTML DOM
Document doc = Jsoup.connect(url).userAgent(userAgent).timeout(timeout).get();
HtmlToPlainText formatter = new HtmlToPlainText();
if (selector != null) {
Elements elements = doc.select(selector); // get each element that matches the CSS selector
for (Element element : elements) {
String plainText = formatter.getPlainText(element); // format that element to plain text
System.out.println(plainText);
}
} else { // format the whole doc
String plainText = formatter.getPlainText(doc);
System.out.println(plainText);
}
}
/**
* Format an Element to plain-text
* @param element the root element to format
* @return formatted text
*/
public String getPlainText(Element element) {
FormattingVisitor formatter = new FormattingVisitor();
NodeTraversor traversor = new NodeTraversor(formatter);
traversor.traverse(element); // walk the DOM, and call .head() and .tail() for each node
return formatter.toString();
}
// the formatting rules, implemented in a breadth-first DOM traverse
private class FormattingVisitor implements NodeVisitor {
private static final int maxWidth = 80;
private int width = 0;
private StringBuilder accum = new StringBuilder(); // holds the accumulated text
// hit when the node is first seen
public void head(Node node, int depth) {
String name = node.nodeName();
if (node instanceof TextNode)
append(((TextNode) node).text()); // TextNodes carry all user-readable text in the DOM.
else if (name.equals("li"))
append("\n * ");
else if (name.equals("dt"))
append(" ");
else if (StringUtil.in(name, "p", "h1", "h2", "h3", "h4", "h5", "tr"))
append("\n");
}
// hit when all of the node's children (if any) have been visited
public void tail(Node node, int depth) {
String name = node.nodeName();
if (StringUtil.in(name, "br", "dd", "dt", "p", "h1", "h2", "h3", "h4", "h5"))
append("\n");
else if (name.equals("a"))
append(String.format(" <%s>", node.absUrl("href")));
}
// appends text to the string builder with a simple word wrap method
private void append(String text) {
if (text.startsWith("\n"))
width = 0; // reset counter if starts with a newline. only from formats above, not in natural text
if (text.equals(" ") &&
(accum.length() == 0 || StringUtil.in(accum.substring(accum.length() - 1), " ", "\n")))
return; // don't accumulate long runs of empty spaces
if (text.length() + width > maxWidth) { // won't fit, needs to wrap
String words[] = text.split("\\s+");
for (int i = 0; i < words.length; i++) {
String word = words[i];
boolean last = i == words.length - 1;
if (!last) // insert a space if not the last word
word = word + " ";
if (word.length() + width > maxWidth) { // wrap and reset counter
accum.append("\n").append(word);
width = word.length();
} else {
accum.append(word);
width += word.length();
}
}
} else { // fits as is, without need to wrap text
accum.append(text);
width += text.length();
}
}
@Override
public String toString() {
return accum.toString();
}
}
}