/// <summary>
/// Validates the file.
/// </summary>
/// <param name="file">The file.</param>
/// <param name="parser">The parser.</param>
/// <returns></returns>
private bool ValidateFile(FileInfo file, Parser parser)
{
var fileSystemHelper = new FileSystemHelper();
m_fileContent = new StringBuilder();
var sr = new StreamReader(fileSystemHelper.GetFileNameFromFileInfo(file), Encoding.Default);
m_fileContent.Append(sr.ReadToEnd());
return parser == Parser.Comment
? m_fileContent.ToString().StartsWith("VERSION|") &&
m_fileContent.ToString().Contains("|NAME|") &&
m_fileContent.ToString().Contains("|CONTENT|")
: m_fileContent.ToString().StartsWith("VERSION|") &&
m_fileContent.ToString().Contains("|SUBJECT|") &&
m_fileContent.ToString().Contains("|CONTENT|");
}
/// <summary>
/// Converts the blog to SQL.
/// </summary>
/// <param name="sourcePath">The source path.</param>
/// <param name="targetPath">The target path.</param>
public void ConvertBlogToSql(string sourcePath, string targetPath)
{
// Create the categories
m_categories = ParseCategories(CategoryList);
// Go through the folder structure, looking for txt and GZ files
// There are three levels under content:
// Year (usually YY)
// Month (usually MM)
var fileSystemHelper = new FileSystemHelper();
WpPostTable = "wp_posts";
WpCommentTable = "wp_comments";
WpMetaTable = "wp_postmeta";
WpTermTable = "wp_terms";
WpTermRelationshipTable = "wp_term_relationships";
WpTermTaxonomyTable = "wp_term_taxonomy";
var di = new DirectoryInfo(sourcePath); // the content folder
foreach (var yearFolder in di.GetDirectories())
{
if (yearFolder.ToString().Trim().ToLower() == "counter" || yearFolder.ToString().Trim().ToLower() == "static")
{
// Static Pages
if (yearFolder.ToString().Trim().ToLower() == "static")
{
foreach (var file in yearFolder.GetFiles())
{
PrepareFiles(fileSystemHelper, file, Parser.Static);
}
}
}
else
{
// Blog entries
foreach (var monthFolder in yearFolder.GetDirectories())
{
foreach (var file in monthFolder.GetFiles())
{
PrepareFiles(fileSystemHelper, file, Parser.BlogEntry);
var commentFolderName = string.Format("{0}\\{1}\\comments", file.DirectoryName,
file.Name.Replace(".gz", string.Empty).Replace(".txt",
string.
Empty));
if (Directory.Exists(commentFolderName) && Directory.GetFiles(commentFolderName).Length > 0)
{
var directoryInfo = new DirectoryInfo(commentFolderName);
foreach (var fileInfo in directoryInfo.GetFiles())
{
m_comments = new List<string>();
m_commentEntries = new List<PhpCommentEntity>();
PrepareFiles(fileSystemHelper, fileInfo, Parser.Comment);
// Add comment to appropriate blog entry
foreach (var b in m_blogEntries)
{
if (b.FileName ==
file.Name.Replace(".gz", string.Empty).Replace(".txt", string.Empty))
{
foreach (var c in m_commentEntries)
{
if (b.Comments == null)
{
b.Comments = new List<PhpCommentEntity>();
}
var phpCommentEntity =
b.Comments.Find(
item =>
item.Date == c.Date && item.Content == c.Content &&
item.Name == c.Name);
if (phpCommentEntity == null)
{
b.Comments.Add(c);
}
}
break;
}
}
}
}
}
}
}
}
// Empty lists for GC
if (m_statics != null)
{
m_statics.Clear();
}
if (m_blogs != null)
{
m_blogs.Clear();
}
if (m_comments != null)
{
m_comments.Clear();
}
// Now go through the list of BlogEntities and convert them
m_postEntities = new List<WpPostEntity>();
if (m_blogEntries == null)
{
m_blogEntries = new List<PhpBlogEntity>();
}
foreach (var b in m_blogEntries)
{
m_postEntities.Add(ConvertBlogEntity(b));
}
if (m_staticEntries == null)
{
m_staticEntries = new List<PhpStaticEntity>();
}
foreach (var s in m_staticEntries)
{
m_postEntities.Add(ConvertStaticEntity(s));
}
// Write to the SQL StringBuilder
var sb = GenerateSql(m_postEntities);
// Write to the SQL file
fileSystemHelper.FileWrite(sb, targetPath);
}
/// <summary>
/// Prepares the files.
/// </summary>
/// <param name="fileSystemHelper">The file system helper.</param>
/// <param name="file">The file.</param>
/// <param name="parser">The parser.</param>
private void PrepareFiles(FileSystemHelper fileSystemHelper, FileInfo file, Parser parser)
{
List<string> files = new List<string>();
switch (parser)
{
case Parser.BlogEntry:
files = m_blogs;
break;
case Parser.Comment:
files = m_comments;
break;
case Parser.Static:
files = m_statics;
break;
}
if (files == null)
{
files = new List<string>();
}
if (!files.Contains(fileSystemHelper.GetFileNameFromFileInfo(file)))
{
FileInfo unzippedFile;
if (!file.Extension.Contains(".gz"))
unzippedFile = file;
else
{
var fileName = fileSystemHelper.GetFileNameFromFileInfo(file).Replace(".gz", string.Empty);
unzippedFile = !File.Exists(fileName) ? fileSystemHelper.UnzipFile(file) : file;
}
bool result;
switch (parser)
{
case Parser.Comment:
result = ParseComment(unzippedFile);
break;
case Parser.Static:
result = ParseStaticPage(unzippedFile);
break;
//case Parser.File:
default:
result = ParseBlogEntry(unzippedFile);
break;
}
if (result)
{
if (!files.Contains(fileSystemHelper.GetFileNameFromFileInfo(unzippedFile)))
{
files.Add(fileSystemHelper.GetFileNameFromFileInfo(unzippedFile));
}
}
}
switch (parser)
{
case Parser.Static:
m_statics = files;
break;
case Parser.Comment:
m_comments = files;
break;
default:
m_blogs = files;
break;
}
}
