C# XPath来确定前辈或祖先是否具有内容，而不考虑级别

我正在研究一个以原始HTML阅读新闻文章的过程。新闻文章来源多种多样，因此不一致

目标是定位文章中的第一个图像，如果之前没有内容，则将图像去掉

我们的数据不一致（因为它来自不同的来源），例如

第一张图片前没有内容示例：

A: <p><a href="..."><img src="..." /></a></p><p>content</p>

B: <img src="..." /><p>content</p>

C: <div><p><img src="..." /></p></div><div><p>content</p></div>

D: <div><img src="..." /></div><div><p>content</p></div>

E: <div><p><a href="..."><img src="..." /></a>content</p></div>

及

若要查找图像的所有前辈和祖先，请分别查找无内容图像的所有前辈和祖先

            HtmlNode firstImageInArticle = rawContentModifier.DocumentNode.SelectSingleNode("(//img)[1]");
            if (firstImageInArticle == null) { // there are no images in this article - return HTML as is.
                return html;
            }

            // count all nodes and all empty nodes preceding the first image in the article
            HtmlNodeCollection allNodesPrecedingFirstImage = rawContentModifier.DocumentNode.SelectNodes("(//img)[1]/preceding::*|(//img)[1]/ancestor::*/@id");
            HtmlNodeCollection allEmptyNodesPrecedingFirstImage = rawContentModifier.DocumentNode.SelectNodes("(//img)[1]/preceding::*[not(text())]|(//img)[1]/ancestor::*[not(text())]/@id");

            // if there are no nodes preceding the first image, we have no content prior to the image and therefore should delete it
            if (allEmptyNodesPrecedingFirstImage == null || allNodesPrecedingFirstImage == null) {
                HtmlNode nodeToDelete = rawContentModifier.DocumentNode.SelectSingleNode(firstImageInArticle.XPath);

                // keep moving up the chain until we find the topmost parent that doesn't have any other children
                while (nodeToDelete.ParentNode.ChildNodes.Count == 1)
                {
                    nodeToDelete = nodeToDelete.ParentNode;
                }
                nodeToDelete.Remove();

                html = rawContentModifier.DocumentNode.InnerHtml;
                return html;                    
            }

            // if the number of empty nodes preceding the first image is equal to all nodes preceding the first image...
            // then we know that we basically have no content prior to the image and therefore can delete the first image
            if (allNodesPrecedingFirstImage.Count == allEmptyNodesPrecedingFirstImage.Count) {
                HtmlNode nodeToDelete = rawContentModifier.DocumentNode.SelectSingleNode(firstImageInArticle.XPath);

                // keep moving up the chain until we find the topmost parent that doesn't have any other children
                while (nodeToDelete.ParentNode.ChildNodes.Count == 1)
                {
                    nodeToDelete = nodeToDelete.ParentNode;
                }
                nodeToDelete.Remove();

                html = rawContentModifier.DocumentNode.InnerHtml;
                return html;
            }

编辑以添加示例E，该示例在我的代码中失败，因为p标记包含图像后面的内容，但不是自包含的

(//img)[1]/preceding::*[not(text())]|(//img)[1]/ancestor::*[not(text())]/@id"

            HtmlNode firstImageInArticle = rawContentModifier.DocumentNode.SelectSingleNode("(//img)[1]");
            if (firstImageInArticle == null) { // there are no images in this article - return HTML as is.
                return html;
            }

            // count all nodes and all empty nodes preceding the first image in the article
            HtmlNodeCollection allNodesPrecedingFirstImage = rawContentModifier.DocumentNode.SelectNodes("(//img)[1]/preceding::*|(//img)[1]/ancestor::*/@id");
            HtmlNodeCollection allEmptyNodesPrecedingFirstImage = rawContentModifier.DocumentNode.SelectNodes("(//img)[1]/preceding::*[not(text())]|(//img)[1]/ancestor::*[not(text())]/@id");

            // if there are no nodes preceding the first image, we have no content prior to the image and therefore should delete it
            if (allEmptyNodesPrecedingFirstImage == null || allNodesPrecedingFirstImage == null) {
                HtmlNode nodeToDelete = rawContentModifier.DocumentNode.SelectSingleNode(firstImageInArticle.XPath);

                // keep moving up the chain until we find the topmost parent that doesn't have any other children
                while (nodeToDelete.ParentNode.ChildNodes.Count == 1)
                {
                    nodeToDelete = nodeToDelete.ParentNode;
                }
                nodeToDelete.Remove();

                html = rawContentModifier.DocumentNode.InnerHtml;
                return html;                    
            }

            // if the number of empty nodes preceding the first image is equal to all nodes preceding the first image...
            // then we know that we basically have no content prior to the image and therefore can delete the first image
            if (allNodesPrecedingFirstImage.Count == allEmptyNodesPrecedingFirstImage.Count) {
                HtmlNode nodeToDelete = rawContentModifier.DocumentNode.SelectSingleNode(firstImageInArticle.XPath);

                // keep moving up the chain until we find the topmost parent that doesn't have any other children
                while (nodeToDelete.ParentNode.ChildNodes.Count == 1)
                {
                    nodeToDelete = nodeToDelete.ParentNode;
                }
                nodeToDelete.Remove();

                html = rawContentModifier.DocumentNode.InnerHtml;
                return html;
            }