Stackoverflow while rendering markdown from HTML

[chibenwa]

Describe the bug

I was considering using Flexmark as a HTML => text/plain engine for Apache James

(We currently rely on an homegrown Jsoup based parser)

I did throw our test suite at flexmark-html2md-converter and we encountered a Strackoverflow error with the following test:

    @Test
    public void deeplyNestedHtmlShouldNotThrowStackOverflow() {
        final int count = 2048;
        String html = Strings.repeat("<div>", count) +  "<p>para1</p><p>para2</p>" + Strings.repeat("</div>", count);
        String expectedPlainText = "para1\n\npara2\n\n";
        
        assertThat(FlexmarkHtmlConverter.builder().build().convert(html))
            .isEqualTo(expectedPlainText);
    }

java.lang.StackOverflowError
	at java.base/java.util.Vector.addElement(Vector.java:616)
	at java.base/java.util.Stack.push(Stack.java:68)
	at com.vladsch.flexmark.html2md.converter.FlexmarkHtmlConverter$MainHtmlConverter.pushState(FlexmarkHtmlConverter.java:1151)
	at com.vladsch.flexmark.html2md.converter.FlexmarkHtmlConverter.processHtmlTree(FlexmarkHtmlConverter.java:1696)
	at com.vladsch.flexmark.html2md.converter.FlexmarkHtmlConverter$MainHtmlConverter.renderChildren(FlexmarkHtmlConverter.java:1146)
	at com.vladsch.flexmark.html2md.converter.internal.HtmlConverterCoreNodeRenderer.processDiv(HtmlConverterCoreNodeRenderer.java:537)
	at com.vladsch.flexmark.html2md.converter.HtmlNodeRendererHandler.render(HtmlNodeRendererHandler.java:19)
	at com.vladsch.flexmark.html2md.converter.FlexmarkHtmlConverter$MainHtmlConverter.renderNode(FlexmarkHtmlConverter.java:1135)
	at com.vladsch.flexmark.html2md.converter.FlexmarkHtmlConverter$MainHtmlConverter.render(FlexmarkHtmlConverter.java:1050)
	at com.vladsch.flexmark.html2md.converter.FlexmarkHtmlConverter.processHtmlTree(FlexmarkHtmlConverter.java:1707)
	at com.vladsch.flexmark.html2md.converter.FlexmarkHtmlConverter$MainHtmlConverter.renderChildren(FlexmarkHtmlConverter.java:1146)
	at com.vladsch.flexmark.html2md.converter.internal.HtmlConverterCoreNodeRenderer.processDiv(HtmlConverterCoreNodeRenderer.java:537)
	at com.vladsch.flexmark.html2md.converter.HtmlNodeRendererHandler.render(HtmlNodeRendererHandler.java:19)

The rendrering is tree based, and rendering it uses the Java stack with recursion.

This feels familiar as we have had a similar issue with our homegrown jsoup-based parser that relied on similar mechanisms. We did overcome this limitation by replacing recursion with stacks (in and out) and loops.