Wp2ghost/CSharpGhostImporter/PostFetcher.cs

513 lines
23 KiB
C#
Raw Normal View History

2026-09-20 08:31:23 +02:00
using System.Text.Json;
namespace wp2ghost
{
// =============================================================================
// PostFetcher — Posts von WordPress abrufen, verarbeiten
// =============================================================================
static class PostFetcher
{
private static readonly HttpClient httpClient = new();
private static HashSet<int> ProcessedPosts = new();
private static List<GhostPost> GhostPosts = new();
private static string ProcessedPostsPath = "";
//private static string MarkdownDir = "";
private static string GhostPostsDir = "";
private static string BaseDir = "";
public static int ProcessedCount => ProcessedPosts.Count;
public static int PostCount => GhostPosts.Count;
public static IReadOnlyList<GhostPost> Posts => GhostPosts;
public static void Init(string baseDir)
{
BaseDir = baseDir;
ProcessedPostsPath = Path.Combine(baseDir, "processed_posts.txt");
//MarkdownDir = Path.Combine(baseDir, "markdown");
GhostPostsDir = Path.Combine(baseDir, "ghostPosts");
////if (!Directory.Exists(MarkdownDir))
//// Directory.CreateDirectory(MarkdownDir);
if (!Directory.Exists(GhostPostsDir))
Directory.CreateDirectory(GhostPostsDir);
LoadProcessedPosts();
}
public static void ClearProcessedPosts()
{
ProcessedPosts.Clear();
File.WriteAllText(ProcessedPostsPath, "wp_id,wp_slug,ghost_id" + Environment.NewLine);
Console.WriteLine("Abgefragte Posts wurden zurückgesetzt.");
}
public static void ShowInfo()
{
if (ProcessedPosts.Count > 0)
{
Console.WriteLine($"Letzte abgefragte Objekte: {string.Join(", ", ProcessedPosts.OrderByDescending(x => x).Take(10))}");
}
}
public static async Task FetchFromWordPressAsync(string wpBaseUrl, int maxPosts)
{
Console.WriteLine();
Console.WriteLine("Starte Abruf von WordPress...");
string wpHost = "";
try { wpHost = new Uri(wpBaseUrl).Host.ToLowerInvariant(); } catch { }
try
{
var allPostIds = new List<int>();
var postTypes = new[] { "post", "page" };
foreach (var postType in postTypes)
{
Console.WriteLine($"Hole {postType}-IDs von der WordPress-API...");
string idsUrl = $"{wpBaseUrl.TrimEnd('/')}/wp-json/custom/v1/posts-ids?post_type={postType}";
var idsResponse = await httpClient.GetStringAsync(idsUrl);
var idsObj = JsonSerializer.Deserialize<IdsResponse>(idsResponse);
if (idsObj?.Ids != null)
allPostIds.AddRange(idsObj.Ids);
}
if (allPostIds.Count == 0)
{
Console.WriteLine("Keine Beiträge oder Seiten gefunden.");
return;
}
2026-09-22 08:08:51 +02:00
ProcessedPosts.Clear();
LoadProcessedPosts();
2026-09-20 08:31:23 +02:00
var pendingIds = allPostIds.Where(id => !ProcessedPosts.Contains(id)).ToList();
Console.WriteLine($"Insgesamt gefunden: {allPostIds.Count}, Bereits abgefragt: {allPostIds.Count - pendingIds.Count}, Noch zu verarbeiten: {pendingIds.Count}");
if (pendingIds.Count == 0)
{
Console.WriteLine("Alle Beiträge und Seiten wurden bereits abgefragt. Nichts zu tun. (Nutze Option 5 zum Zurücksetzen)");
return;
}
if (maxPosts > 0)
pendingIds = pendingIds.Take(maxPosts).ToList();
Console.WriteLine($"Es werden {pendingIds.Count} Beiträge/Seiten verarbeitet.");
int internalId = 1;
if (GhostPosts.Count > 0)
{
var lastId = GhostPosts.Select(gp => int.TryParse(gp.Id, out int id) ? id : 0).Max();
internalId = lastId + 1;
}
int processed = 0;
foreach (int wpId in pendingIds)
{
Console.WriteLine($"[{++processed}/{pendingIds.Count}] -> Beitrag/Seite ID: {wpId}...");
string detailUrl = $"{wpBaseUrl.TrimEnd('/')}/wp-json/custom/v1/post-details/{wpId}";
2026-09-22 08:08:51 +02:00
//string postUrl = $"{wpBaseUrl.TrimEnd('/')}/wp-json/wp/v2/posts/{wpId}?_embed";
2026-09-20 08:31:23 +02:00
try
{
var detailJson = await httpClient.GetStringAsync(detailUrl);
var wpPost = JsonSerializer.Deserialize<WpPostResponse>(detailJson);
2026-09-22 08:08:51 +02:00
//var postJson = await httpClient.GetStringAsync(postUrl);
2026-09-20 08:31:23 +02:00
if (wpPost == null)
{
Console.WriteLine($" Übersprungen (null).");
MarkProcessed(wpId, null);
continue;
}
string htmlContent = wpPost.Content ?? "";
var parsedDate = DateTime.MinValue;
var postDate = DateTime.TryParse(wpPost.Date, out parsedDate) ? parsedDate : new DateTime(2026, 9, 1);
// Featured Image: falls von der WP API nicht gesetzt, ersten Bild-URL aus HTML nehmen
if (string.IsNullOrEmpty(wpPost.FeaturedImage))
{
var firstImgMatch = System.Text.RegularExpressions.Regex.Match(htmlContent, @"<img\s+[^>]*src=[""']([^""']+)[""'][^>]*>", System.Text.RegularExpressions.RegexOptions.IgnoreCase);
if (firstImgMatch.Success)
{
string firstImgUrl = firstImgMatch.Groups[1].Value;
if (IsInternalUrl(firstImgUrl, wpHost))
{
wpPost.FeaturedImage = firstImgUrl;
Console.WriteLine($" [Featured Image] Aus HTML extrahiert (da API keine lieferte): {Path.GetFileName(new Uri(firstImgUrl).AbsolutePath)}");
}
}
}
2026-09-20 08:31:23 +02:00
// Alle internen Bild-URLs aus dem HTML extrahieren (keine externen Links)
var htmlImageUrls = ExtractImageUrls(htmlContent, wpHost);
// Bevorzugte Media-Dateien ermitteln
var preferredMedia = GetPreferredMediaFiles(wpPost.MediaFiles, htmlImageUrls, wpHost);
// Prüfe: Referenzierte Bilder im HTML, die NICHT in MediaFiles sind
if (htmlImageUrls.Count > 0 && wpPost.MediaFiles != null)
{
var mediaFileNames = wpPost.MediaFiles.Select(m => GetFileNameFromUrl(m).ToLowerInvariant()).ToHashSet();
foreach (var imgUrl in htmlImageUrls)
{
string imgFileName = GetFileNameFromUrl(imgUrl).ToLowerInvariant();
if (!mediaFileNames.Contains(imgFileName))
{
Console.WriteLine($" [Warnung] Bild im HTML nicht in MediaFiles: {Path.GetFileName(new Uri(imgUrl).AbsolutePath)}");
}
}
}
if (preferredMedia.Count > 0)
{
foreach (string mediaUrl in preferredMedia)
{
var downloaded = await ImageManager.DownloadAsync(mediaUrl, postDate);
if (downloaded.HasValue)
{
string ghostMediaPath = $"__GHOST_URL__/content/images/{downloaded.Value.RelativePath}";
htmlContent = htmlContent.Replace(mediaUrl, ghostMediaPath);
}
}
}
// WordPress HTML zu Ghost-kompatibles HTML konvertieren (Galerien, Figures, etc.)
htmlContent = HtmlGhostConverter.Convert(htmlContent);
// Lokale Bildpfade für JSON-Datei sammeln
var imageFiles = new List<string>();
if (preferredMedia.Count > 0)
{
foreach (string mediaUrl in preferredMedia)
{
if (ImageManager.TryGetRelativePath(mediaUrl, out string relPath))
imageFiles.Add(relPath);
}
}
// Featured Image verarbeiten
string featureImageGhostPath = "";
if (!string.IsNullOrEmpty(wpPost.FeaturedImage))
{
if (ImageManager.TryGetRelativePath(wpPost.FeaturedImage, out string featRel))
{
featureImageGhostPath = $"__GHOST_URL__/content/images/{featRel}";
}
else
{
var featDownload = await ImageManager.DownloadAsync(wpPost.FeaturedImage, postDate);
if (featDownload.HasValue)
{
featureImageGhostPath = $"__GHOST_URL__/content/images/{featDownload.Value.RelativePath}";
if (!imageFiles.Contains(featDownload.Value.RelativePath))
imageFiles.Add(featDownload.Value.RelativePath);
}
}
}
var ghostPost = new GhostPost
{
Id = internalId.ToString(),
Title = wpPost.Title,
Slug = wpPost.Slug,
Html = htmlContent,
Status = "published",
CreatedAt = postDate.ToString("o"),
PublishedAt = postDate.ToString("o")
};
GhostPosts.Add(ghostPost);
// JSON-Datei in ghostPosts/ speichern
2026-09-22 08:08:51 +02:00
var ghostPostFile = new GhostPostFile
2026-09-20 08:31:23 +02:00
{
WpId = wpId,
WpSlug = wpPost.Slug,
GhostId = ghostPost.Id,
Title = wpPost.Title,
Slug = wpPost.Slug,
Excerpt = wpPost.Excerpt,
Status = "published",
CreatedAt = postDate.ToString("o"),
PublishedAt = postDate.ToString("o"),
FeatureImage = featureImageGhostPath,
Categories = wpPost.Categories ?? new(),
2026-09-22 08:08:51 +02:00
Tags = wpPost.Tags ?? new(),
2026-09-20 08:31:23 +02:00
Keywords = wpPost.Keywords ?? "",
Comments = wpPost.Comments ?? new(),
ImageFiles = imageFiles,
Html = htmlContent
};
2026-09-22 08:08:51 +02:00
string jsonPath = Path.Combine(GhostPostsDir, $"{ghostPostFile.WpId}.json");
File.WriteAllText(jsonPath, JsonSerializer.Serialize(ghostPostFile, new JsonSerializerOptions { WriteIndented = true }));
// Cache sofort speichern (nicht erst beim Beenden)
ImageManager.SaveCache();
2026-09-20 08:31:23 +02:00
MarkProcessed(wpId, wpPost.Slug);
internalId++;
}
catch (Exception ex)
{
Console.WriteLine($" Fehler beim Abrufen von ID {wpId}: {ex.Message}");
MarkProcessed(wpId, null);
}
}
Console.WriteLine();
Console.WriteLine("=========================================");
Console.WriteLine($" Abruf abgeschlossen! {GhostPosts.Count} Beiträge geladen.");
Console.WriteLine($" Abgefragte Posts insgesamt: {ProcessedPosts.Count}");
Console.WriteLine(" Nutze Option 2 um die Ghost ZIP zu erstellen.");
Console.WriteLine("=========================================");
}
catch (Exception ex)
{
Console.WriteLine($"\nFehler beim Abruf: {ex.Message}");
Console.WriteLine($"Stack Trace: {ex.StackTrace}");
}
}
//public static void SaveAllMarkdownFiles()
//{
// foreach (var post in GhostPosts)
// {
// if (string.IsNullOrEmpty(post.Slug)) continue;
// string mdPath = Path.Combine(MarkdownDir, $"{post.Slug}.md");
// if (!string.IsNullOrEmpty(post.Markdown))
// File.WriteAllText(mdPath, post.Markdown);
// }
// Console.WriteLine($"Markdown-Dateien gespeichert: {GhostPosts.Count} Dateien unter {MarkdownDir}");
//}
private static List<string> ExtractImageUrls(string html, string wpHost)
{
var urls = new List<string>();
var imageExts = new[] { ".jpg", ".jpeg", ".png", ".gif", ".webp", ".svg", ".bmp", ".tiff", ".tif" };
// 1) <img src="..."> extrahieren
var imgMatches = System.Text.RegularExpressions.Regex.Matches(html, @"<img\s+[^>]*src=[""']([^""']+)[""']", System.Text.RegularExpressions.RegexOptions.IgnoreCase);
foreach (System.Text.RegularExpressions.Match m in imgMatches)
{
string url = m.Groups[1].Value;
if (!string.IsNullOrEmpty(url) && IsInternalUrl(url, wpHost) && !urls.Contains(url))
urls.Add(url);
}
// 2) data-orig-file (Vollbild) extrahieren
var origFileMatches = System.Text.RegularExpressions.Regex.Matches(html, @"data-orig-file=[""']([^""']+)[""']", System.Text.RegularExpressions.RegexOptions.IgnoreCase);
foreach (System.Text.RegularExpressions.Match m in origFileMatches)
{
string url = m.Groups[1].Value;
if (!string.IsNullOrEmpty(url) && IsInternalUrl(url, wpHost) && !urls.Contains(url))
urls.Add(url);
}
// 3) data-large-file extrahieren
var largeFileMatches = System.Text.RegularExpressions.Regex.Matches(html, @"data-large-file=[""']([^""']+)[""']", System.Text.RegularExpressions.RegexOptions.IgnoreCase);
foreach (System.Text.RegularExpressions.Match m in largeFileMatches)
{
string url = m.Groups[1].Value;
if (!string.IsNullOrEmpty(url) && IsInternalUrl(url, wpHost) && !urls.Contains(url))
urls.Add(url);
}
// 4) srcset (größte Variante pro img) extrahieren
var srcsetMatches = System.Text.RegularExpressions.Regex.Matches(html, @"srcset=[""']([^""']+)[""']", System.Text.RegularExpressions.RegexOptions.IgnoreCase);
foreach (System.Text.RegularExpressions.Match m in srcsetMatches)
{
string srcset = m.Groups[1].Value;
string widestUrl = GetWidestFromSrcset(srcset);
if (!string.IsNullOrEmpty(widestUrl) && IsInternalUrl(widestUrl, wpHost) && !urls.Contains(widestUrl))
urls.Add(widestUrl);
}
// 5) <a href="..."> mit Bildendung (Figure-Links auf Originalbild)
var aMatches = System.Text.RegularExpressions.Regex.Matches(html, @"<a\s+[^>]*href=[""']([^""']+)[""'][^>]*>", System.Text.RegularExpressions.RegexOptions.IgnoreCase);
foreach (System.Text.RegularExpressions.Match m in aMatches)
{
string url = m.Groups[1].Value;
if (string.IsNullOrEmpty(url)) continue;
try
{
string ext = Path.GetExtension(new Uri(url).AbsolutePath).ToLowerInvariant();
if (imageExts.Contains(ext) && IsInternalUrl(url, wpHost) && !urls.Contains(url))
urls.Add(url);
}
catch { }
}
return urls;
}
private static string GetWidestFromSrcset(string srcset)
{
string bestUrl = "";
int bestWidth = 0;
foreach (string part in srcset.Split(',', StringSplitOptions.RemoveEmptyEntries))
{
string[] tokens = part.Trim().Split(' ', StringSplitOptions.RemoveEmptyEntries);
if (tokens.Length < 2) continue;
if (int.TryParse(tokens[1].Replace("w", ""), out int w) && w > bestWidth)
{
bestWidth = w;
bestUrl = tokens[0];
}
}
return bestUrl;
}
private static bool IsInternalUrl(string url, string wpHost)
{
if (string.IsNullOrEmpty(wpHost)) return true;
try
{
var uri = new Uri(url);
return uri.Host.ToLowerInvariant().EndsWith(wpHost) || wpHost.EndsWith(uri.Host.ToLowerInvariant());
}
catch
{
return url.StartsWith("/") || url.StartsWith("wp-content");
}
}
private static string NormalizeImageUrl(string url)
{
try
{
var uri = new Uri(url);
string path = uri.AbsolutePath;
// Entferne -scaled UND Dimensionspostfixe für den Vergleich
path = System.Text.RegularExpressions.Regex.Replace(path, @"-scaled(?=\.\w+$)", "");
path = System.Text.RegularExpressions.Regex.Replace(path, @"-\d+x\d+(?=\.\w+$)", "");
return path;
}
catch
{
return url;
}
}
private static string GetFileNameFromUrl(string url)
{
try
{
return Path.GetFileName(new Uri(url).AbsolutePath);
}
catch
{
return url;
}
}
private static List<string> GetPreferredMediaFiles(List<string>? mediaFiles, List<string> htmlImageUrls, string wpHost)
{
var result = new List<string>();
var imageExts = new[] { ".jpg", ".jpeg", ".png", ".gif", ".webp", ".svg", ".bmp", ".tiff", ".tif" };
// Alle internen Bild-URLs zusammenführen: MediaFiles + HTML-referenzierte
var allUrls = new List<string>();
if (mediaFiles != null)
allUrls.AddRange(mediaFiles.Where(u => IsInternalUrl(u, wpHost)));
foreach (var u in htmlImageUrls)
{
if (!allUrls.Contains(u))
allUrls.Add(u);
}
if (allUrls.Count == 0)
return result;
// Dateinamen aus den HTML-Links extrahieren
var htmlFileNames = htmlImageUrls.Select(u => GetFileNameFromUrl(u).ToLowerInvariant()).ToHashSet();
// Gruppiere nach normalisiertem Dateinamen (ohne -scaled und Dimensionspostfixe)
var grouped = new Dictionary<string, List<string>>();
foreach (var url in allUrls)
{
string normalized = NormalizeImageUrl(url);
if (!grouped.ContainsKey(normalized))
grouped[normalized] = new List<string>();
grouped[normalized].Add(url);
}
foreach (var group in grouped)
{
// Nur Hauptbilder berücksichtigen (keine WordPress-Sizes wie -300x200, keine -scaled)
var mainImages = group.Value.Where(u => ImageManager.IsMainImage(u)).ToList();
if (mainImages.Count == 0)
{
Console.WriteLine($" [Bild] Übersprungen (kein Hauptbild): {Path.GetFileName(new Uri(@group.Value[0]).AbsolutePath)}");
continue;
}
if (mainImages.Count == 1)
{
result.Add(mainImages[0]);
}
else
{
// Prüfe ob eine der Varianten als <a href> im HTML referenziert wird
string? linked = mainImages.FirstOrDefault(u => htmlFileNames.Contains(GetFileNameFromUrl(u).ToLowerInvariant()));
if (linked != null)
{
result.Add(linked);
var skipped = mainImages.Where(u => u != linked).ToList();
foreach (var s in skipped)
Console.WriteLine($" [Bild] Übersprungen (nicht verlinkt): {Path.GetFileName(new Uri(s).AbsolutePath)}");
}
else
{
// Keine Verlinkung: Original (ohne -scaled) bevorzugen
var preferred = mainImages.FirstOrDefault(u => !u.Contains("-scaled")) ?? mainImages[0];
result.Add(preferred);
var skipped = mainImages.Where(u => u != preferred).ToList();
foreach (var s in skipped)
Console.WriteLine($" [Bild] Übersprungen (scaled): {Path.GetFileName(new Uri(s).AbsolutePath)}");
}
}
}
return result;
}
private static void MarkProcessed(int wpId, string? wpSlug = null)
{
ProcessedPosts.Add(wpId);
SaveProcessedPosts(wpId, wpSlug);
}
private static void LoadProcessedPosts()
{
if (!File.Exists(ProcessedPostsPath)) return;
try
{
var lines = File.ReadAllLines(ProcessedPostsPath);
foreach (var line in lines)
{
if (string.IsNullOrWhiteSpace(line) || line.StartsWith("wp_id")) continue;
var parts = line.Split(',');
if (parts.Length >= 1 && int.TryParse(parts[0].Trim(), out int postId))
ProcessedPosts.Add(postId);
}
}
catch (Exception ex)
{
Console.WriteLine($"Fehler beim Laden der abgefragten Posts: {ex.Message}");
}
}
private static void SaveProcessedPosts(int wpId, string? wpSlug)
{
string entry = $"{wpId},{wpSlug ?? ""}";
using var writer = File.AppendText(ProcessedPostsPath);
writer.WriteLine(entry);
}
}
}