public Entry Parse(string textSample)
{
var entry = new Entry();
textSample = textSample.Trim();
var lines = textSample.Split(new[] {Environment.NewLine}, StringSplitOptions.RemoveEmptyEntries).Select(l => l.Trim()).ToList();
if (lines.Count < 4)
{
return Empty;
}
MapFrom(lines, entry);
MapSent(lines, entry);
MapTitle(lines, entry);
MapLink(lines, entry);
lines.RemoveRange(0, 4);
lines.RemoveRange(lines.Count - 1, 1);
entry.Body = string.Join("\n\n", lines);
entry.Body = entry.Body.Replace("\n\n[embedded content]\n\n", "\n\n");
RegexOptions options = RegexOptions.None;
Regex regex = new Regex(@"[ ]{2,}", options);
entry.Body = regex.Replace(entry.Body, @" ");
return entry;
}
private void MapTitle(List<string> lines, Entry entry)
{
entry.Title = TrimTag("Subject", lines[3]);
if (entry.Title.Trim() == "Subject:")
{
entry.Title = "";
}
}
public void ProcessSingleItem(Entry entry, Publication publication)
{
_currentArticleId = _currentArticleId + 1;
entry.Writer.Id = _writers.CreateOrRetrieveId(entry.Writer);
entry.Id = _currentArticleId;
entry.PublicationId = publication.Id;
var entrySql = _sqlGen.GenerateSqlFor(entry);
_writer.WriteLine(entrySql);
}
public string GenerateSqlFor(Entry item)
{
return string.Format("\n" + ArticleInsert,
EncodeMySqlString(item.Id.ToString()),
EncodeMySqlString(item.Title),
EncodeMySqlString(item.Body.TrimStart()),
EncodeMySqlString(item.Uri),
EncodeMySqlString(item.Timestamp.ToString()),
EncodeMySqlString(item.PublicationId.ToString()),
EncodeMySqlString(item.Writer.Id.ToString()));
}
public string GenerateSqlFor(Entry item)
{
return string.Format(ArticleInsert,
item.Id.ToEscapedString(),
item.Title.ToEscapedString(),
item.Body.TrimStart().ToEscapedString(),
item.Uri.ToEscapedString(),
item.Timestamp.ToEscapedString(),
item.PublicationId.ToEscapedString(),
item.Writer.Id.ToEscapedString());
}
private void MapFrom(List<string> lines, Entry entry)
{
var trimmed = TrimTag("From", lines[0]);
var trimmedParts = trimmed.Split(new[] { '(' });
if (trimmedParts.Length <= 1)
{
entry.Writer = new Writer {Name = trimmedParts.First().Trim(), Email = string.Empty};
return;
}
entry.Writer = new Writer
{
Email = trimmedParts.First().Trim(),
Name = trimmedParts.Skip(1).First().Trim().Replace(")", "")
};
}
public Entry Parse(string textSample)
{
var entry = new Entry();
textSample = textSample.Trim();
var lines = textSample.Split(new[] {Environment.NewLine}, StringSplitOptions.RemoveEmptyEntries).Select(l => l.Trim()).ToList();
if (lines.Count < 4)
{
return Empty;
}
MapFrom(lines, entry);
MapSent(lines, entry);
MapTitle(lines, entry);
MapLink(lines, entry);
lines.RemoveRange(0, 4);
lines.RemoveRange(lines.Count - 1, 1);
entry.Body = string.Join("\n\n", lines);
SanitiseBody(entry);
return entry;
}
private void MapSent(List<string> lines, Entry entry)
{
entry.Timestamp = DateTime.Parse(TrimTag("Sent", lines[1]), new DateTimeFormatInfo(), DateTimeStyles.None);
}
private void MapLink(List<string> lines, Entry entry)
{
var articleSlugLine = lines.Last();
var cutDown = articleSlugLine.Replace("View article... <", "").Replace(">", "");
entry.Uri = cutDown.Trim();
}
private static void SanitiseBody(Entry entry)
{
entry.Body = entry.Body.Replace("\n\n[embedded content]\n\n", "\n\n");
entry.Body = new Regex(@"[ ]{2,}", RegexOptions.None).Replace(entry.Body, @" ");
}
public void SetUp()
{
var parser = new RssFeedToSql.DataSources.DirectoryAndTextFile.TextFileDataParser();
_entry = parser.Parse(_textSample);
}
