Wp2ghost/CSharpGhostImporter/HtmlConverter.cs

using System.Globalization;
using System.Text.RegularExpressions;
using HtmlAgilityPack;

namespace wp2ghost
{
    // =============================================================================
    // HtmlGhostConverter — WordPress HTML zu Ghost-kompatibles HTML konvertieren
    // =============================================================================
    public static class HtmlGhostConverter
    {
        private static readonly Dictionary<string, string?> WpClassMap = new(StringComparer.OrdinalIgnoreCase)
        {
            // Block-Elemente → Ghost-Äquivalent oder null (Klasse entfernen)
            { "wp-block-paragraph",       null },
            { "wp-block-heading",         null },
            { "wp-block-list",            null },
            { "wp-block-quote",           null },
            { "wp-block-code",            null },
            { "wp-block-separator",       null },
            { "wp-block-group",           null },
            { "wp-block-columns",         null },
            { "wp-block-column",          null },
            { "wp-block-freeform",        null },
            { "wp-block-verse",           null },
            { "wp-block-preformatted",    null },

            // Ghost-Kompatible Klassen
            { "wp-block-image",           "kg-card kg-image-card" },
            { "wp-block-gallery",         "kg-card kg-gallery-card" },
            { "wp-block-pullquote",       "kg-card kg-pullquote" },
            { "wp-block-embed",           "kg-card kg-embed-card" },
            { "wp-block-embed-wrapper",   "kg-card kg-embed-card" },
            { "wp-block-video",           "kg-card kg-embed-card" },
            { "wp-block-audio",           "kg-card kg-embed-card" },
            { "wp-block-cover",           "kg-card kg-header-card" },
            { "wp-block-media-text",      "kg-card kg-width-wide" },
            { "wp-block-table",           "kg-card kg-table-card" },
            { "wp-block-button",          "kg-card kg-button-card" },
            { "wp-block-buttons",         "kg-card kg-button-card" },
            { "wp-block-file",            "kg-card kg-file-card" },

            // Sub-Elemente
            { "wp-block-button__link",    "gh-btn" },
            { "wp-block-button__link__inline", "gh-btn" },
            { "wp-element-button",        "gh-btn" },
            { "has-text-align-center",    "kg-width-wide" },
            { "has-text-align-left",      "" },
            { "has-text-align-right",     "" },
            { "is-style-wide",            "kg-width-wide" },
            { "is-style-full",            "kg-width-wide" },
            { "is-style-large",           "kg-width-wide" },

            // Theme-spezifisch
            { "wp-block-post-title",      null },
            { "wp-block-post-content",    null },
            { "wp-block-site-title",      null },
            { "wp-block-site-tagline",    null },
            { "wp-block-query",           null },
            { "wp-block-comments",        null },
            { "wp-block-comment-template", null },
            { "wp-block-template-part",   null },

            // CoBlocks
            { "wp-block-coblocks-gallery-stacked", "kg-card kg-gallery-card" },
            { "wp-block-coblocks-gallery-collage", "kg-card kg-gallery-card" },
            { "wp-block-coblocks-gallery-masonry", "kg-card kg-gallery-card" },
            { "wp-block-coblocks-gallery-thumbnails", "kg-card kg-gallery-card" },
            { "wp-block-coblocks-alert",  "kg-card kg-alert-card" },
            { "wp-block-coblocks-accordion", "kg-card kg-accordion-card" },
            { "wp-block-coblocks-accordion-item", "kg-accordion-item" },
            { "wp-block-coblocks-row",    "kg-width-wide" },
            { "wp-block-coblocks-column", "" },
            { "wp-block-coblocks-click-to-tweet", "kg-card kg-click-to-tweet" },
            { "wp-block-coblocks-social-profiles", "kg-card kg-social-link" },

            // Jetpack
            { "wp-block-jetpack-tiled-gallery", "kg-card kg-gallery-card" },
            { "wp-block-jetpack-slideshow", "kg-card kg-gallery-card" },
            { "wp-block-jetpack-story",   "kg-card kg-gallery-card" },
        };

        public static string Convert(string html)
        {
            if (string.IsNullOrWhiteSpace(html)) return string.Empty;

            var htmlDoc = new HtmlDocument();
            htmlDoc.LoadHtml(html);

            int totalGalleries = 0;
            totalGalleries += ConvertClassicGalleries(htmlDoc);
            totalGalleries += ConvertGutenbergGalleries(htmlDoc);
            totalGalleries += ConvertJetpackGalleries(htmlDoc);
            totalGalleries += ConvertCoBlocksGalleries(htmlDoc);

            if (totalGalleries > 0)
                Console.WriteLine($"   [Gallery] {totalGalleries} Galerie(n) konvertiert");

            ConvertSingleImages(htmlDoc);
            ConvertFigures(htmlDoc);
            ConvertWpClasses(htmlDoc);
            AddTargetBlankToExternalLinks(htmlDoc);

            return htmlDoc.DocumentNode.OuterHtml;
        }

        // =========================================================================
        // Gallery-Typ 1: WordPress Classic Gallery
        // HTML: <div class='gallery gallery-columns-3'>
        //         <figure class='gallery-item'>
        //           <div class='gallery-icon portrait'>
        //             <a href='...'><img src='thumb' data-orig-file='full.jpg' /></a>
        //           </div>
        //         </figure>
        //       </div>
        // =========================================================================
        private static int ConvertClassicGalleries(HtmlDocument htmlDoc)
        {
            var galleryNodes = htmlDoc.DocumentNode.SelectNodes(
                "//div[contains(@class, 'gallery') and not(contains(@class, 'wp-block'))]"
            );
            if (galleryNodes == null) return 0;

            int count = 0;
            foreach (var gallery in galleryNodes.ToList())
            {
                var items = gallery.SelectNodes(".//figure[contains(@class, 'gallery-item')]");
                if (items == null || items.Count == 0) continue;

                var ghostGallery = htmlDoc.CreateElement("figure");
                SetClasses(ghostGallery, "kg-card kg-gallery-card");
                AddWideClass(ghostGallery, gallery);

                var container = htmlDoc.CreateElement("div");
                container.AddClass("kg-gallery-container");

                var wrappers = new List<HtmlNode>();
                var imageCaptions = new List<string>();

                foreach (var item in items)
                {
                    var img = item.SelectSingleNode(".//img");
                    if (img == null) continue;

                    // Bild-Titel aus data-image-caption Attribut extrahieren
                    string dataCaption = img.GetAttributeValue("data-image-caption", "").Trim();
                    if (!string.IsNullOrEmpty(dataCaption))
                        imageCaptions.Add(dataCaption);

                    // Auch figcaption innerhalb des gallery-items prüfen
                    var figCap = item.SelectSingleNode(".//figcaption");
                    if (figCap != null && !string.IsNullOrWhiteSpace(figCap.InnerText))
                    {
                        string capText = figCap.InnerText.Trim();
                        if (!string.IsNullOrEmpty(capText) && !imageCaptions.Contains(capText))
                            imageCaptions.Add(capText);
                    }

                    string src = GetBestImageUrl(img);
                    if (string.IsNullOrEmpty(src)) continue;

                    wrappers.Add(BuildGalleryImage(htmlDoc, img, src));
                }

                if (wrappers.Count == 0) continue;

                AppendGalleryRows(htmlDoc, container, wrappers, GetGalleryColumns(gallery));
                ghostGallery.AppendChild(container);

                // Gallery-Bild-Titel übertragen, wenn vorhanden
                if (imageCaptions.Count > 0)
                {
                    // Ersten nicht-leeren Titel nehmen als Gallery-Kapitel
                    string firstCaption = imageCaptions[0];
                    
                    var newCaption = htmlDoc.CreateElement("figcaption");
                    newCaption.InnerHtml = firstCaption;
                    ghostGallery.AppendChild(newCaption);
                    ghostGallery.AddClass("kg-card-hascaption");
                }

                gallery.ParentNode?.ReplaceChild(ghostGallery, gallery);
                count++;
            }
            return count;
        }

        // =========================================================================
        // Gallery-Typ 2: WordPress Gutenberg Gallery
        // =========================================================================
        private static int ConvertGutenbergGalleries(HtmlDocument htmlDoc)
        {
            var galleryNodes = htmlDoc.DocumentNode.SelectNodes(
                "//figure[contains(@class, 'wp-block-gallery')] | " +
                "//div[contains(@class, 'wp-block-gallery')] | " +
                "//ul[contains(@class, 'wp-block-gallery')]"
            );
            if (galleryNodes == null) return 0;

            int count = 0;
            foreach (var gallery in galleryNodes.ToList())
            {
                var imgs = gallery.SelectNodes(".//img");
                if (imgs == null || imgs.Count == 0) continue;

                var ghostGallery = htmlDoc.CreateElement("figure");
                SetClasses(ghostGallery, "kg-card kg-gallery-card");
                AddWideClass(ghostGallery, gallery);

                // Neue Format-Prüfung: data-carousel-extra Attribut speichern
                string carouselExtra = gallery.GetAttributeValue("data-carousel-extra", "");

                var container = htmlDoc.CreateElement("div");
                container.AddClass("kg-gallery-container");

                var wrappers = new List<HtmlNode>();
                foreach (var img in imgs)
                {
                    string src = GetBestImageUrl(img);
                    if (string.IsNullOrEmpty(src)) continue;

                    wrappers.Add(BuildGalleryImage(htmlDoc, img, src));
                }

                if (wrappers.Count == 0) continue;

                AppendGalleryRows(htmlDoc, container, wrappers, GetGalleryColumns(gallery));
                ghostGallery.AppendChild(container);

                // Alle individuellen Bildunterschriften aus der Gutenberg-Gallery sammeln
                var captionNodes = gallery.SelectNodes(".//figcaption");
                if (captionNodes != null && captionNodes.Count > 0)
                {
                    var distinctCaptions = captionNodes
                        .Select(n => n.InnerText.Trim())
                        .Where(t => !string.IsNullOrWhiteSpace(t))
                        .Distinct()
                        .ToList();

                    if (distinctCaptions.Count > 0)
                    {
                        var newCaption = htmlDoc.CreateElement("figcaption");
                        newCaption.InnerHtml = string.Join(" · ", distinctCaptions);
                        ghostGallery.AppendChild(newCaption);
                        ghostGallery.AddClass("kg-card-hascaption");
                    }
                }

                // data-carousel-extra als Ghost-Attribut übertragen (falls vorhanden)
                if (!string.IsNullOrEmpty(carouselExtra))
                {
                    var extraAttr = htmlDoc.CreateAttribute("data-carousel-extra");
                    extraAttr.Value = carouselExtra;
                    ghostGallery.Attributes.Append(extraAttr);
                }

                gallery.ParentNode?.ReplaceChild(ghostGallery, gallery);
                count++;
            }
            return count;
        }

        // =========================================================================
        // Gallery-Typ 3: Jetpack Tiled Gallery
        // =========================================================================
        private static int ConvertJetpackGalleries(HtmlDocument htmlDoc)
        {
            var galleryNodes = htmlDoc.DocumentNode.SelectNodes(
                "//div[contains(@class, 'wp-block-jetpack-tiled-gallery')] | " +
                "//figure[contains(@class, 'wp-block-jetpack-tiled-gallery')]"
            );
            if (galleryNodes == null) return 0;

            int count = 0;
            foreach (var gallery in galleryNodes.ToList())
            {
                var imgs = gallery.SelectNodes(".//img");
                if (imgs == null || imgs.Count == 0) continue;

                var ghostGallery = htmlDoc.CreateElement("figure");
                SetClasses(ghostGallery, "kg-card kg-gallery-card");
                AddWideClass(ghostGallery, gallery);

                var container = htmlDoc.CreateElement("div");
                container.AddClass("kg-gallery-container");

                var wrappers = new List<HtmlNode>();
                foreach (var img in imgs)
                {
                    string src = GetBestImageUrl(img);
                    if (string.IsNullOrEmpty(src)) continue;

                    wrappers.Add(BuildGalleryImage(htmlDoc, img, src));
                }

                if (wrappers.Count == 0) continue;

                AppendGalleryRows(htmlDoc, container, wrappers, GetGalleryColumns(gallery));
                ghostGallery.AppendChild(container);

                gallery.ParentNode?.ReplaceChild(ghostGallery, gallery);
                count++;
            }
            return count;
        }

        // =========================================================================
        // Gallery-Typ 4: CoBlocks Gallery
        // =========================================================================
        private static int ConvertCoBlocksGalleries(HtmlDocument htmlDoc)
        {
            var galleryNodes = htmlDoc.DocumentNode.SelectNodes(
                "//figure[contains(@class, 'wp-block-coblocks-gallery-masonry')] | " +
                "//figure[contains(@class, 'wp-block-coblocks-gallery-stacked')] | " +
                "//figure[contains(@class, 'wp-block-coblocks-gallery-collage')] | " +
                "//figure[contains(@class, 'wp-block-coblocks-gallery-thumbnails')]"
            );
            if (galleryNodes == null) return 0;

            int count = 0;
            foreach (var gallery in galleryNodes.ToList())
            {
                var imgs = gallery.SelectNodes(".//img");
                if (imgs == null || imgs.Count == 0) continue;

                var ghostGallery = htmlDoc.CreateElement("figure");
                SetClasses(ghostGallery, "kg-card kg-gallery-card");
                AddWideClass(ghostGallery, gallery);

                var container = htmlDoc.CreateElement("div");
                container.AddClass("kg-gallery-container");

                var wrappers = new List<HtmlNode>();
                foreach (var img in imgs)
                {
                    string src = GetBestImageUrl(img);
                    if (string.IsNullOrEmpty(src)) continue;

                    wrappers.Add(BuildGalleryImage(htmlDoc, img, src));
                }

                if (wrappers.Count == 0) continue;

                AppendGalleryRows(htmlDoc, container, wrappers, GetGalleryColumns(gallery));
                ghostGallery.AppendChild(container);

                gallery.ParentNode?.ReplaceChild(ghostGallery, gallery);
                count++;
            }
            return count;
        }

        // =========================================================================
        // Galerie-Helfer: Spaltenanzahl, Wide-Klasse, Bild-Wrapper, Zeilen
        // =========================================================================
        private static int GetGalleryColumns(HtmlNode gallery)
        {
            string classes = gallery.GetAttributeValue("class", "");
            var match = Regex.Match(classes, @"columns-(\d+)");
            if (match.Success && int.TryParse(match.Groups[1].Value, out int cols) && cols >= 1 && cols <= 9)
                return cols;
            return 3;
        }

        private static void AddWideClass(HtmlNode ghostGallery, HtmlNode originalGallery)
        {
            string origClass = originalGallery.GetAttributeValue("class", "");
            if (origClass.Contains("is-style-wide") || origClass.Contains("alignwide"))
                ghostGallery.AddClass("kg-width-wide");
        }

        private static HtmlNode BuildGalleryImage(HtmlDocument htmlDoc, HtmlNode originalImg, string src)
        {
            var imgWrapper = htmlDoc.CreateElement("div");
            imgWrapper.AddClass("kg-gallery-image");

            var newImg = htmlDoc.CreateElement("img");
            newImg.SetAttributeValue("src", src);

            double w = 0, h = 0;

            string origSize = originalImg.GetAttributeValue("data-orig-size", "");
            if (!string.IsNullOrEmpty(origSize))
            {
                var parts = origSize.Split(',');
                if (parts.Length == 2 &&
                    double.TryParse(parts[0].Trim(), NumberStyles.Float, CultureInfo.InvariantCulture, out double ow) &&
                    double.TryParse(parts[1].Trim(), NumberStyles.Float, CultureInfo.InvariantCulture, out double oh))
                {
                    w = ow;
                    h = oh;
                }
            }

            if (w <= 0 || h <= 0)
            {
                if (double.TryParse(originalImg.GetAttributeValue("width", ""), NumberStyles.Float, CultureInfo.InvariantCulture, out double aw) &&
                    double.TryParse(originalImg.GetAttributeValue("height", ""), NumberStyles.Float, CultureInfo.InvariantCulture, out double ah))
                {
                    w = aw;
                    h = ah;
                }
            }

            if (w > 0 && h > 0)
            {
                newImg.SetAttributeValue("width", ((int)w).ToString(CultureInfo.InvariantCulture));
                newImg.SetAttributeValue("height", ((int)h).ToString(CultureInfo.InvariantCulture));
                string ratio = (w / h).ToString("G6", CultureInfo.InvariantCulture);
                imgWrapper.SetAttributeValue("style", $"flex: {ratio} 1 0%;");
            }
            else
            {
                imgWrapper.SetAttributeValue("style", "flex: 1 1 0%;");
            }

            newImg.SetAttributeValue("loading", "lazy");
            newImg.SetAttributeValue("alt", originalImg.GetAttributeValue("alt", ""));

            imgWrapper.AppendChild(newImg);
            return imgWrapper;
        }

        private static void AppendGalleryRows(HtmlDocument htmlDoc, HtmlNode container, List<HtmlNode> wrappers, int columns)
        {
            for (int i = 0; i < wrappers.Count; i += columns)
            {
                var row = htmlDoc.CreateElement("div");
                row.AddClass("kg-gallery-row");

                int end = Math.Min(i + columns, wrappers.Count);
                for (int j = i; j < end; j++)
                    row.AppendChild(wrappers[j]);

                container.AppendChild(row);
            }
        }


        // =========================================================================
        // Helper: Beste Bild-URL aus img-Element holen
        // Priorität: data-orig-file → data-large-file → srcset (letzter/Widest) → src
        // =========================================================================
        private static string GetBestImageUrl(HtmlNode img)
        {
            // 1. data-orig-file (Vollbild)
            string src = img.GetAttributeValue("data-orig-file", "");
            if (!string.IsNullOrEmpty(src)) return src;

            // 2. data-large-file
            src = img.GetAttributeValue("data-large-file", "");
            if (!string.IsNullOrEmpty(src)) return src;

            // 3. srcset (größte Variante nehmen)
            string srcset = img.GetAttributeValue("srcset", "");
            if (!string.IsNullOrEmpty(srcset))
            {
                src = GetLargestFromSrcset(srcset);
                if (!string.IsNullOrEmpty(src)) return src;
            }

            // 4. src (Fallback)
            return img.GetAttributeValue("src", "");
        }

        private static string GetLargestFromSrcset(string srcset)
        {
            string bestUrl = "";
            int bestWidth = 0;

            foreach (string part in srcset.Split(',', StringSplitOptions.RemoveEmptyEntries))
            {
                string trimmed = part.Trim();
                string[] tokens = trimmed.Split(' ', StringSplitOptions.RemoveEmptyEntries);
                if (tokens.Length < 2) continue;

                string url = tokens[0];
                string descriptor = tokens[1];

                if (int.TryParse(descriptor.Replace("w", ""), out int width) && width > bestWidth)
                {
                    bestWidth = width;
                    bestUrl = url;
                }
            }

            return bestUrl;
        }

        // =========================================================================
        // Einzelbilder (nicht in Galerien)
        // =========================================================================
        private static void ConvertSingleImages(HtmlDocument htmlDoc)
        {
            var singleImgNodes = htmlDoc.DocumentNode.SelectNodes("//img");
            if (singleImgNodes == null) return;

            foreach (var img in singleImgNodes.ToList())
            {
                if (img.Ancestors("figure").Any(a => HasClass(a, "kg-gallery-card")))
                    continue;

                string src = GetBestImageUrl(img);
                if (string.IsNullOrEmpty(src)) continue;

                // Beste Variante als src setzen; srcset/sizes verweisen auf WordPress-Größenvarianten
                img.SetAttributeValue("src", src);
                img.Attributes.Remove("srcset");
                img.Attributes.Remove("sizes");

                var parentFigure = img.Ancestors("figure").FirstOrDefault();
                if (parentFigure != null)
                {
                    SetClasses(parentFigure, "kg-card kg-image-card");
                    img.AddClass("kg-image");
                }
                else
                {
                    var wrapper = htmlDoc.CreateElement("figure");
                    SetClasses(wrapper, "kg-card kg-image-card");
                    img.AddClass("kg-image");

                    var imgParent = img.ParentNode;
                    if (imgParent != null)
                    {
                        imgParent.InsertBefore(wrapper, img);
                        imgParent.RemoveChild(img);
                        wrapper.AppendChild(img);
                    }
                }
            }
        }

        // =========================================================================
        // Figure ohne kg-card → kg-width-wide
        // =========================================================================
        private static void ConvertFigures(HtmlDocument htmlDoc)
        {
            var figures = htmlDoc.DocumentNode.SelectNodes("//figure");
            if (figures == null) return;

            foreach (var figure in figures.ToList())
            {
                if (HasClass(figure, "kg-card")) continue;
                SetClasses(figure, "kg-width-wide kg-card-width-wide");
            }
        }

        // =========================================================================
        // WP-Klassen mappen
        // =========================================================================
        private static void ConvertWpClasses(HtmlDocument htmlDoc)
        {
            var allNodes = htmlDoc.DocumentNode.SelectNodes("//*[contains(@class, 'wp-block')]");
            if (allNodes == null) return;

            foreach (var node in allNodes.ToList())
            {
                string classes = node.GetAttributeValue("class", "");
                if (string.IsNullOrEmpty(classes)) continue;

                var classList = classes.Split(' ', StringSplitOptions.RemoveEmptyEntries).ToList();
                var newClasses = new List<string>();
                bool changed = false;

                foreach (string cls in classList)
                {
                    if (WpClassMap.TryGetValue(cls, out string? ghostClass))
                    {
                        changed = true;
                        if (ghostClass != null && ghostClass.Length > 0)
                            newClasses.AddRange(ghostClass.Split(' '));
                    }
                    else
                    {
                        newClasses.Add(cls);
                    }
                }

                if (changed)
                {
                    string finalClasses = string.Join(" ", newClasses.Where(c => c.Length > 0).Distinct());
                    if (finalClasses.Length > 0)
                        node.SetAttributeValue("class", finalClasses);
                    else
                        node.Attributes.Remove("class");
                }
            }

            // Separator
            var separators = htmlDoc.DocumentNode.SelectNodes("//div[contains(@class, 'wp-block-separator')]");
            if (separators != null)
            {
                foreach (var sep in separators.ToList())
                {
                    var hr = htmlDoc.CreateElement("hr");
                    foreach (var attr in sep.Attributes)
                        hr.SetAttributeValue(attr.Name, attr.Value);
                    sep.ParentNode?.ReplaceChild(hr, sep);
                }
            }
        }

        // =========================================================================
        // Externe Links mit target="_blank" versehen
        // =========================================================================
        private static void AddTargetBlankToExternalLinks(HtmlDocument htmlDoc)
        {
            var anchors = htmlDoc.DocumentNode.SelectNodes("//a[@href]");
            if (anchors == null) return;

            foreach (var anchor in anchors.ToList())
            {
                string href = anchor.GetAttributeValue("href", "").Trim();
                if (string.IsNullOrEmpty(href)) continue;

                // Nur absolute externe URLs (http/https)
                if (href.StartsWith("http://") || href.StartsWith("https://"))
                {
                    anchor.SetAttributeValue("target", "_blank");
                    anchor.SetAttributeValue("rel", "noopener noreferrer");
                }
            }
        }

        private static bool HasClass(HtmlNode node, string className)
        {
            string classes = node.GetAttributeValue("class", "");
            return classes.Split(' ', StringSplitOptions.RemoveEmptyEntries)
                          .Any(c => c.Equals(className, StringComparison.OrdinalIgnoreCase));
        }

        private static void SetClasses(HtmlNode node, string classes)
        {
            node.SetAttributeValue("class", classes);
        }
    }
}