Negative range positions in malformed HTML fragment

[KennyWongPFPT]

Hello, please see below a test program that tries to extract the text node range positions from the malformed fragment foo<p/>far. Notice the malformed tag <p/>.

import org.jsoup.nodes.*;
import org.jsoup.parser.*;
import org.jsoup.select.*;

public class Test {
    public static void main(String[] args) {
        HtmlTreeBuilder treeBuilder = new HtmlTreeBuilder();
        Parser parser = new Parser(treeBuilder);
        parser.setTrackPosition(true);
        Document document = parser.parseInput("foo<p/>bar", "");
        NodeTraversor.traverse((Node node, int depth) -> {
            if (node instanceof TextNode textNode) {
                Range sourceRange = textNode.sourceRange();
                System.out.printf("text=%s start=%d end=%d%n",
                    textNode.text(),
                    sourceRange.start().pos(),
                    sourceRange.end().pos());
            }
        }, document);
    }
}

With release 1.16.1, all positions are negative:

% java -cp ~/.m2/repository/org/jsoup/jsoup/1.16.1/jsoup-1.16.1.jar Test.java
text=foo start=-1 end=-1
text=bar start=-1 end=-1

With release 1.18.1, it's a little better, except for the -1 start position for the bar text immediately following the malformed tag.

% java -cp ~/.m2/repository/org/jsoup/jsoup/1.18.1/jsoup-1.18.1.jar Test.java
text=foo start=0 end=3
text=bar start=-1 end=10

[jhy]

Thanks for the clear report, fixed!