private static void ReportGraph(EntryGraph graph, int indent = 0)
{
Console.WriteLine((new String('\t', indent)) + graph.Entry.Id);
indent++;
foreach (var child in graph.Children)
{
ReportGraph(child, indent);
}
}
private static void ReportPost(Entry post, EntryGraph graph)
{
Console.WriteLine("Id: " + post.Id);
Console.WriteLine("Published: " + post.Published);
Console.WriteLine("Updated: " + post.Updated);
Console.WriteLine("Title: " + post.Title);
Console.WriteLine("Url: " + post.Url);
Console.WriteLine("Tags: " + String.Join(", ", post.Tags));
Console.WriteLine("Author: " + post.Author.Name);
Console.WriteLine("Content Length: " + post.Content.Length);
ReportGraph(graph);
}
private static void BuildRss(Rss comments, Blog meta, EntryGraph graph)
{
if (graph.Children != null && graph.Children.Count > 0)
{
comments.Channel.Items.Add(new Item
{
Title = meta.Title,
AbsoluteUrl = DESTINATION_BLOG_PATH + meta.Url,
Content = "",
ThreadIdentified = meta.Id,
Published = meta.Published,
Status = "open",
Comments = BuildComments(graph.Children, "")
});
}
}
public static void Export(EntryGraph graph, Rss comments)
{
var debug = false;
Console.WriteLine(graph.Entry.Url);
Blog meta = new Blog();
meta.Id = Guid.NewGuid().ToString();
meta.Url = graph.Entry.Url;
meta.Author = graph.Entry.Author.Name;
meta.Published = graph.Entry.Published;
meta.Modified = graph.Entry.Updated;
meta.Title = graph.Entry.Title;
//meta.Image;
//meta.Description;
meta.Enabled = true;
meta.Redirects = new List <Redirect>
{
new Redirect
{
Url = graph.Entry.BloggerUrl,
RedirectType = HttpStatusCode.MovedPermanently,
RedirectByParameter = true,
RedirectByRoute = false
}
};
string contentFolder = CreateContentFolder(graph.Entry.Url);
string mediaFolder = CreateMediaFolder(graph.Entry.Url);
string mediaPath = MEDIABASEPATH + graph.Entry.Url;
var md = graph.Entry.Content;
// Add Carriage Returns
if (debug)
{
Console.WriteLine("Add Carriage Returns");
}
md = md.Replace("<br />", Environment.NewLine);
// Map <hx>
if (debug)
{
Console.WriteLine("Map <hx>");
}
md = Regex.Replace(md, "<h1>(.*?)</h1>", x =>
{
return("# " + x.Groups[1] + Environment.NewLine); // + new String('-', x.Groups[1].Length) + Environment.NewLine;
});
md = Regex.Replace(md, "<h2>(.*?)</h2>", x =>
{
return("## " + x.Groups[1] + Environment.NewLine); // + new String('-', x.Groups[1].Length) + Environment.NewLine;
});
md = Regex.Replace(md, "<h3>(.*?)</h3>", x =>
{
return("### " + x.Groups[1] + Environment.NewLine); // + new String('-', x.Groups[1].Length) + Environment.NewLine;
});
// Replace non-breaking spaces
if (debug)
{
Console.WriteLine("Replace non-breaking spaces");
}
md = md.Replace(" ", " ");
// Remove any meta Tags
if (debug)
{
Console.WriteLine("Remove any meta Tags");
}
md = Regex.Replace(md, "<meta (.*?)/>" + Environment.NewLine, "");
// Map blogger image
if (debug)
{
Console.WriteLine("Map blogger image");
}
md = Regex.Replace(md, "<div class=\"separator\"(.*?)<a href=\"(.*?)\"(.*?)<img (.*?)src=\"(.*?)\"(.*?)</div>", x =>
{
return("![Image](" + ProcessImage(mediaFolder, mediaPath, x.Groups[2].Value) + ")" + Environment.NewLine);
});
// Remove summary breaks
if (debug)
{
Console.WriteLine("Remove summary breaks");
}
md = md.Replace("<a name='more'></a>" + Environment.NewLine + Environment.NewLine, DESCRIPTION_BREAK_TAG);
md = md.Replace("<a name='more'></a>" + Environment.NewLine, DESCRIPTION_BREAK_TAG);
md = md.Replace("<a name='more'></a>", DESCRIPTION_BREAK_TAG);
// Remove RFC Weekly header
if (debug)
{
Console.WriteLine("Remove RFC Weekly header");
}
Regex rfcWeeklyHeaderRegex = new Regex("<a href=\"(.*?)Hopefully someone else will also find useful.", RegexOptions.Singleline);
md = rfcWeeklyHeaderRegex.Replace(md, "");
// Map Anchors
if (debug)
{
Console.WriteLine("Map Anchors");
}
md = Regex.Replace(md, "<a (.*?)>(.*?)</a>", x =>
{
var sb = new StringBuilder();
sb.Append("[");
sb.Append(x.Groups[2]);
sb.Append("]");
sb.Append("(");
sb.Append(Regex.Matches(x.Groups[1].ToString(), "href=\"(.*?)\"")[0].Groups[1]);
sb.Append(")");
return(sb.ToString());
});
// Map Lists
if (debug)
{
Console.WriteLine("Map Lists");
}
md = md.Replace("<ul>", Environment.NewLine);
md = md.Replace("<li>", "* ");
md = md.Replace("</li>", Environment.NewLine);
md = md.Replace("</ul>", Environment.NewLine);
// Gists
if (debug)
{
Console.WriteLine("Gists");
}
md = Regex.Replace(md, "<script src=\"(.*?)\">(.*?)</script>", x => "%[" + x.Groups[1].Value + "]");
// Remove Spread the love
if (debug)
{
Console.WriteLine("Remove Spread the love");
}
md = Regex.Replace(md, "<div id=\"SpreadTheLove\">(.*?)</div>", "");
// Remove Divs
if (debug)
{
Console.WriteLine("Remove Divs");
}
md = Regex.Replace(md, "<div(.*?)>", "");
md = md.Replace("</div>", "");
// Replace html encoding
if (debug)
{
Console.WriteLine("Replace html encoding");
}
md = md.Replace(">", ">");
// Remove empty lines at the start & end
if (debug)
{
Console.WriteLine("Remove empty lines at the start & end");
}
while (md.StartsWith(Environment.NewLine))
{
md = md.Substring(Environment.NewLine.Length);
}
while (md.EndsWith(Environment.NewLine + Environment.NewLine))
{
md = md.Substring(0, md.Length - Environment.NewLine.Length);
}
// Load the Description (then remove the tag)
if (debug)
{
Console.WriteLine("Load the Description (then remove the tag)");
}
if (md.Contains(DESCRIPTION_BREAK_TAG))
{
meta.Description = md.Substring(0, md.IndexOf(DESCRIPTION_BREAK_TAG));
md = md.Replace(DESCRIPTION_BREAK_TAG, "");
}
// Set up the Keywords
if (debug)
{
Console.WriteLine("Setup the keywords");
}
meta.KeyWords = graph.Entry.Tags;
// RFC Weekly specific actions
if (debug)
{
Console.WriteLine("RFC Weekly specific actions");
}
if (graph.Entry.Title.ToLower().Contains("rfc") && graph.Entry.Title.ToLower().Contains("weekly"))
{
// Add the image & description
meta.Image = "/media/blog/rfc-weekly/RFCWeeklyTwitterCard.png";
meta.Description = "RFC Weekly - a summary of things that I find interesting. It is an indulgence; its the weekly update that I would like to receive. Unfortunately no-one else is producing it so I figured I best get on with it. Hopefully someone else will also find useful.";
// Add as a key word
if (!meta.KeyWords.Contains("RFCWeekly"))
{
meta.KeyWords.Add("RFCWeekly");
}
}
if (debug)
{
Console.WriteLine("Save to file");
}
SaveMarkdown(contentFolder, graph.Entry.Url, md);
SaveMeta(contentFolder, graph.Entry.Url, meta);
BuildRss(comments, meta, graph);
}
