Matt Berseth

   1: static void Main(string[] args)

   2: {

   3:     //  I exported the entry table from the MySQL admin page

   4:     var entries =

   5:         from e in XDocument.Load(@"..\..\blog.xml").Descendants("mt_entry")

   6:         select new

   7:         {

   8:             id = e.Descendants("entry_id").First().Value,

   9:             name = e.Descendants("entry_basename").First().Value,

  10:             postDate = Convert.ToDateTime(e.Descendants("entry_created_on").First().Value)

  11:         };

  12:  

  13:     if(!Directory.Exists("entries"))

  14:     {

  15:         Directory.CreateDirectory("entries");

  16:     }

  17:  

  18:     WebClient wc = new WebClient();

  19:     foreach (var entry in entries)

  20:     {

  21:         string url = string.Format("http://mattberseth.com/blog/{0}/{1}/{2}.html", entry.postDate.Year, string.Format("{0:00}", entry.postDate.Month), entry.name);

  22:  

  23:         //  download the url and save it.

  24:         wc.DownloadFile(url, Path.Combine("entries", string.Format("{0}.html", entry.name)));

  25:     }

  26:  

  27:     //  move each file into its own folder

  28:     foreach (string file in Directory.GetFiles("entries"))

  29:     {

  30:         //  create the directory

  31:         Directory.CreateDirectory(Path.Combine("entries", Path.GetFileNameWithoutExtension(file)));

  32:  

  33:         //  move the file into the dir

  34:         File.Move(file, Path.Combine(Path.Combine("entries", Path.GetFileNameWithoutExtension(file)), Path.GetFileName(file)));

  35:     }

  36: }

   1: static void Main(string[] args)

   2: {

   3:     //  extract the comments xml and save it into the folder

   4:     foreach (string dir in Directory.GetDirectories("entries"))

   5:     {

   6:         string entryContents = File.ReadAllText(Directory.GetFiles(dir, "*.html")[0]);

   7:  

   8:         int commentsStartIndex = entryContents.IndexOf("<div id=\"comments\" class=\"comments\">");

   9:         int commentsEndIndex = entryContents.IndexOf("<form method=\"post\" action=\"http://mattberseth.com/blog-mt/mt-comments.fcgi\"");

  10:  

  11:         string commentsXml = entryContents.Substring(commentsStartIndex, (commentsEndIndex - commentsStartIndex)) + "</div>";

  12:  

  13:         try

  14:         {

  15:             XDocument.Parse(commentsXml).Save(Path.Combine(dir, "comments.xml"), SaveOptions.None);

  16:         }

  17:         catch (Exception ex)

  18:         {

  19:             //  error

  20:             Console.WriteLine(dir);

  21:  

  22:             //  just save the file as text

  23:             File.WriteAllText(Path.Combine(dir, "comments.xml"), commentsXml);

  24:         }

  25:     }

  26: }

   1: static void Main(string[] args)

   2: {

   3:     int commentID = 1;

   4:     //  extract the comments xml and save it into the folder

   5:     foreach (string dir in Directory.GetDirectories("entries"))

   6:     {

   7:         string commentsXml = File.ReadAllText(Directory.GetFiles(dir, "*.xml")[0]);

   8:  

   9:         if(File.Exists(Path.Combine(dir, "insert.sql")))

  10:         {

  11:             File.Delete(Path.Combine(dir, "insert.sql"));

  12:         }

  13:  

  14:         //  find all of the comments in the document

  15:         foreach(XElement e in XDocument.Parse(commentsXml).Descendants().Where(x => x.HasAttributes && x.Attribute("class") != null && x.Attribute("class").Value == "comment"))

  16:         {

  17:             //  the name of the person leaving the comment

  18:             string comment_author;

  19:             //  the date the comment was entered

  20:             DateTime comment_date;

  21:             //  the payload

  22:             string comment_content = null;

  23:             //  the authors url

  24:             string comment_url = null;

  25:             //  the author emails

  26:             string comment_email = null;

  27:  

  28:             //  the comment header contains both the comment_author and the comment_date.

  29:             XElement commentHeader = e.Elements().First();

  30:  

  31:             //  the last anchor is always the comment date

  32:             comment_date = DateTime.Parse(commentHeader.Elements().Last().Value.Trim());

  33:  

  34:             //  if there are 2 anchors, the first one contains the comment_url as the href

  35:             //  and the payload is the name, otherwise the comment_author is the text between

  36:             //  the Posted by: and on tokens.

  37:             if (commentHeader.Elements().Count(x => x.Name == "a") == 2)

  38:             {

  39:                 XElement element = commentHeader.Elements().First(x => x.Name == "a");

  40:                 comment_author = element.Value.Trim();

  41:                 comment_url = element.Attribute("href").Value.Trim();

  42:  

  43:                 if (comment_url == "http://mattberseth.com")

  44:                 {

  45:                     //  its me!

  46:                     comment_email = "matt@mattberseth.com";

  47:                 }

  48:             }

  49:             else

  50:             {

  51:                 const string beginToken = "Posted by:\n";

  52:                 const string endToken = "on\n";

  53:  

  54:                 int beginIndex = commentHeader.Value.IndexOf(beginToken) + beginToken.Length;

  55:                 int length = commentHeader.Value.LastIndexOf(endToken) - beginIndex;

  56:  

  57:                 comment_author = commentHeader.Value.Substring(beginIndex, length).Trim();

  58:             }

  59:  

  60:             //  the content in the comment-content portion is the payload.  not sure what to

  61:             //  do about the HTML that is embedded ...

  62:             foreach (XElement paragraph in e.Elements().Last().Elements().Where(x => x.Name == "p"))

  63:             {

  64:                 comment_content += string.Format("{0}\n\n", paragraph.Value).Replace("\n", "\\n");

  65:             }

  66:  

  67:             Console.WriteLine("comment_author: {0}", comment_author);

  68:             Console.WriteLine("comment_date: {0}", comment_date);

  69:             Console.WriteLine("comment_url: {0}", comment_url);

  70:             Console.WriteLine("comment_content: {0}", comment_content);

  71:  

  72:             //  get the mapping from entries to id's

  73:             Dictionary<string, int> map = GetEntryToIDMap();

  74:             

  75:             //  Format:

  76:             //  {0} = comment_id

  77:             //  {1} = comment_author

  78:             //  {2} = comment_email

  79:             //  {3} = comment_entry_id

  80:             //  {4} = comment_text

  81:             //  {5} = comment_url

  82:             //  {6} = comment_created_on

  83:             //  INSERT INTO `mt_comment` VALUES ({0}, '{1}', 1, NULL, {2}, 1, NULL, NULL, NULL, 1, '2000-01-01 00:00:00', '{3}', '{4}', 1, '{5} 00:00:00', NULL, '{5} 00:00:00', NULL);

  84:             string insert = string.Format(

  85:                 "INSERT INTO `mt_comment` VALUES ({0}, '{1}', 1, NULL, {2}, {3}, NULL, NULL, NULL, 1, '2000-01-01 00:00:00', '{4}', {5}, 1, '{6} 00:00:00', NULL, '{6} 00:00:00', NULL);",

  86:                 commentID++,

  87:                 comment_author.Replace("'", string.Empty),

  88:                 comment_email == null ? "NULL" : string.Format("'{0}'", comment_email),

  89:                 map[Path.GetFileName(dir)],

  90:                 comment_content.Replace("'", string.Empty),

  91:                 comment_url == null ? "NULL" : string.Format("'{0}'", comment_url.Replace("'", string.Empty)),

  92:                 comment_date.ToString("yyyy-MM-dd"));

  93:  

  94:             File.AppendAllText(Path.Combine(dir, "insert.sql"), insert);

  95:             File.AppendAllText(Path.Combine(dir, "insert.sql"), Environment.NewLine);

  96:  

  97:             //  todo: escape single quotes in the comment text

  98:             //  todo: escape single quotes in the comment author

  99:             //  todo: escape single quotes in the url

 100:  

 101:             Console.WriteLine();

 102:             Console.WriteLine();

 103:         }

 104:     }

 105: }

   1: static void Main(string[] args)

   2: {

   3:     int offset = 0;

   4:     foreach (string insert in Directory.GetFiles("entries", "*.sql", SearchOption.AllDirectories))

   5:     {

   6:         long fileSize = new FileInfo(insert).Length;

   7:         long currentCombinedFileSize = File.Exists(string.Format("combined-{0}.sql", offset)) ? new FileInfo(string.Format("combined-{0}.sql", offset)).Length : 0;

   8:  

   9:         if (currentCombinedFileSize + fileSize > 40000)

  10:         {

  11:             offset++;

  12:         }

  13:  

  14:         File.AppendAllText(string.Format("combined-{0}.sql", offset), File.ReadAllText(insert));

  15:         File.AppendAllText(string.Format("combined-{0}.sql", offset), string.Empty);

  16:     }

  17: }